RU2795549C2 - Method and device for intra-frame prediction - Google Patents

Abstract

FIELD: image encoding and decoding technology.

SUBSTANCE: invention is related in particular to a method and apparatus for encoding/decoding intra prediction. A device for intra-frame prediction is proposed, which is capable of obtaining the intra-frame prediction mode of the current block, determining among the plurality of pixel lines a pixel line for intra-frame prediction of the current block, and performing intra-frame prediction of the current block based on the intra-frame prediction mode and the pixel line. In addition, the first reference pixel of the determined pixel line can be selectively filtered, and the prediction pixel of the current block can be selectively corrected, and thus the intra-frame prediction accuracy can be improved.

EFFECT: increase of efficiency of image compression.

7 cl, 35 dwg, 8 tbl

Description

FIELD OF TECHNOLOGY TO WHICH THE INVENTION RELATES

[0001] The present invention relates to a technology for encoding and decoding an image, and in particular, to a method and apparatus for encoding/decoding intra prediction.

BACKGROUND OF THE INVENTION

[0002] Recently, the demand for high resolution and high quality images, such as high definition (HD) images and ultra high definition (UHD) images, has been increasing in various applications, and therefore, highly efficient image compression techniques have been developed.

[0003] As for the image compression technology, there may be various technologies such as inter-image prediction technology in which the value of a pixel included in the current frame is predicted based on the frame before or after the current frame, intra prediction technology in which a value of a pixel included in the current frame is predicted using the pixel information in the current frame, and an entropy encoding technique for assigning a short code to information having a high occurrence rate and assigning a long code to information having a low occurrence rate. These image compression technologies can be used to efficiently compress and transmit or store image data.

SUMMARY OF THE INVENTION

Technical task

[0004] The present invention relates to an image encoding and decoding technology, and more particularly to a method and apparatus for intra prediction encoding/decoding.

Technical solution

[0005] According to the method and apparatus for intra prediction according to the present invention, the intra prediction mode of the current block can be obtained, among the plurality of pixel lines, a pixel line for intra prediction of the current block can be determined, and based on the intra prediction mode and the determined pixel line intra-prediction of the current block can be performed.

[0006] According to the method and apparatus for intra prediction according to the present invention, filtering can be performed on the first reference pixel of a certain pixel line.

[0007] According to the method and apparatus for intra prediction according to the present invention, filtering can be selectively performed based on a first flag indicating whether the first reference pixel for intra prediction has been filtered.

[0008] According to the method and device for intra prediction according to the present invention, the first flag can be obtained from the device for decoding based on the coding parameter of the current block, and the coding parameter can contain at least one of the block size, component type, intra mode prediction, or whether intra-prediction was applied to sub-block elements.

[0009] According to the method and apparatus for intra prediction according to the present invention, a prediction pixel of a current block can be corrected according to intra prediction.

[0010] In accordance with the method and apparatus for intra prediction according to the present invention, the step of performing the correction may further include: determining at least one of the second reference pixel or a weighted value for correction based on the prediction pixel position of the current block.

[0011] According to the method and apparatus for intra prediction according to the present invention, the correction can be selectively performed considering at least one of the position of a pixel line of the current block, the intra prediction mode of the current block, or whether intra prediction has been performed in subblock elements of the current block.

[0012] According to the method and apparatus for intra prediction according to the present invention, intra prediction may be performed on sub-block elements of the current block, and sub-blocks may be determined based on at least one of a second flag indicating whether to perform division, information about partition direction or partition number information.

[0013] According to the method and apparatus for intra prediction according to the present invention, an intra prediction mode can be obtained based on a predetermined default mode or a plurality of most likely (MPM) candidate modes.

technical effect

[0014] According to the present invention, coding/decoding efficiency can be improved by predicting in sub-block elements.

[0015] According to the present invention, by intra prediction based on a plurality of pixel lines, intra prediction encoding/decoding efficiency can be improved.

[0016] According to the present invention, by performing filtering on a reference pixel, intra prediction encoding/decoding efficiency can be improved.

[0017] According to the present invention, by correcting an intra prediction pixel, it is possible to improve the efficiency of intra prediction encoding/decoding.

[0018] According to the present invention, by obtaining an intra prediction mode based on a default mode or an MPM candidate, the coding/decoding efficiency of the intra prediction mode can be improved.

BRIEF DESCRIPTION OF THE ERAPHY MATERIALS

[0019] FIG. 1 is a block diagram of an image encoding apparatus according to an embodiment of the present invention.

[0020] FIG. 2 is a block diagram of an image decoding apparatus according to an embodiment of the present invention.

[0021] In FIG. 3 is a schematic diagram showing the tree form of the block.

[0022] FIG. 4 is a schematic diagram depicting a typed block shape.

[0023] FIG. 5 is a schematic diagram showing various block shapes that can be obtained using the block splitting section according to the present invention.

[0024] In FIG. 6 is a schematic diagram for explaining a tree division according to an embodiment of the present invention.

[0025] In FIG. 7 is a schematic diagram for explaining tree division according to an embodiment of the present invention.

[0026] FIG. 8 shows a block splitting method according to an embodiment of the present invention.

[0027] FIG. 9 is a circuit diagram showing an intra prediction mode predefined in the image encoding/decoding apparatus.

[0028] FIG. 10 shows an example of comparing pixels from different color spaces to obtain correlation information.

[0029] FIG. 11 is a schematic diagram for explaining the configuration of reference pixels for intra prediction.

[0030] FIG. 12 is a circuit diagram for explaining the range of a reference pixel for intra prediction.

[0031] In FIG. 13 is a diagram showing a block adjacent to a current block for generating a prediction block.

[0032] FIG. 14 and 15 show some examples of confirming the division information of each block.

[0033] FIG. 16 is a circuit diagram showing various situations of block division.

[0034] FIG. 17 shows an example of block division according to an embodiment of the present invention.

[0035] FIG. 18 shows various examples of intra prediction mode candidate groups for setting blocks generating prediction information (the prediction blocks in this example are 2N×N blocks).

[0036] FIG. 19 shows various examples of intra prediction mode candidate groups for setting blocks generating prediction information (the prediction blocks in this example are N×2N blocks).

[0037] FIG. 20 shows an example of block division according to an embodiment of the present invention.

[0038] FIG. 21 and 22 show various examples of intra prediction mode candidate groups for setting blocks generating prediction information.

[0039] FIG. 23-25 show examples of generating a prediction block according to the adjacent block prediction mode.

[0040] FIG. 26 is a schematic diagram of the relationship between the current block and the adjacent block.

[0041] In FIG. 27 and 28 show intra prediction considering the directivity of the prediction mode.

[0042] FIG. 29 is a schematic diagram for explaining the configuration of reference pixels for intra prediction.

[0043] FIG. 30-35 are schematic diagrams with respect to the configuration of reference pixels.

DETAILED DESCRIPTION

[0044] The method and apparatus for intra prediction according to the present invention are configured to obtain the intra prediction mode of the current block, determine among the plurality of pixel lines a pixel line for intra prediction of the current block, and perform intra prediction of the current block based on the intra prediction mode and the determined pixel line.

[0045] In accordance with the method and device for intra prediction according to the present invention, the first reference pixel from a certain line of pixels can be filtered.

[0046] According to the method and apparatus for intra prediction according to the present invention, filtering can be selectively performed based on a first flag indicating whether the first reference pixel for intra prediction has been filtered.

[0047] According to the method and device for intra prediction according to the present invention, the first flag can be obtained from the device for decoding based on the coding parameter of the current block, and the coding parameter can contain at least one of the block size, component type, intra mode prediction, or whether intra-prediction was applied to sub-block elements.

[0048] According to the method and apparatus for intra prediction according to the present invention, a prediction pixel of a current block can be corrected according to intra prediction.

[0049] In accordance with the method and apparatus for intra prediction according to the present invention, the correction operation may further include: determining at least one of the second reference pixel or a weighted value for correction based on the prediction pixel position of the current block.

[0050] According to the method and apparatus for intra prediction according to the present invention, the correction operation can be selectively performed considering at least one of the position of a pixel line of the current block, the intra prediction mode of the current block, or whether intra prediction has been performed in sub-block elements current block.

[0051] According to the method and apparatus for intra prediction according to the present invention, intra prediction may be performed on sub-block elements of the current block, and sub-blocks may be determined based on at least one of a second flag indicating whether to perform division, information about partition direction or partition number information.

[0052] According to the method and apparatus for intra prediction according to the present invention, an intra prediction mode can be obtained based on a predetermined default mode or a plurality of MPM candidates.

[0053] Various modifications may be made to the present invention and various embodiments may be provided, and specific embodiments will be described in detail with reference to the accompanying drawings. However, it should be understood that the present invention is not limited to these specific embodiments, but is to include all alterations, equivalents, and substitutions that are within the spirit and scope of the present invention.

[0054] Terms such as first, second, A, and B may be used to describe various elements, but these elements should not be limited to such terms. These terms are only meant to distinguish one element from another. For example, the first element may be referred to as the second element, and similarly, the second element may be referred to as the first element without departing from the scope of the present invention. The term "and/or" may include a combination of a plurality of related listed objects or any of a plurality of related listed objects.

[0055] It should be understood that if an element is called "connected" or "associated" with another element, the element may be directly connected to or associated with another element. However, it should be understood that there may be another element between them. On the other hand, if an element is said to be "directly connected" or "directly connected" to another element, it should be understood that there is no other element between them.

[0056] The terms used in this application are used only to describe specific embodiments and are not intended to limit the present invention. Singular forms include plural forms unless the context clearly indicates otherwise. In this application, a term such as "include" or "have" is intended to mean the presence of features, numbers, steps, acts, components, parts, or combinations thereof, described in the description, and it should be understood that this term does not exclude the possibility of the presence or adding one or more other features or numbers, steps, actions, components, parts, or combinations thereof.

[0057] Unless otherwise indicated, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by a person skilled in the art to which the present invention pertains. Terms such as those defined in a commonly used dictionary should be interpreted as corresponding to the meanings of the relevant technology and should not be interpreted as ideal or overly formal meanings unless they are explicitly defined in this application.

[0058] Video encoding and decoding devices may be user terminals such as a personal computer (PC), laptop, personal digital assistant (PDA), portable media player (PMP), PlayStation Portable game console (PSP), wireless communication terminal , smartphone, TV, virtual reality (VR) device, augmented reality (AR) device, mixed reality (MR) device, head-mounted display (HMD), and smart glasses, or server terminals such as an application server and a service server, and may include includes various devices such as a communication device containing a communication modem for communicating with various wired/wireless communication devices or networks, a memory device for storing various programs and data for encoding or decoding an image, or for performing prediction between or within images. images, for encoding or decoding, a processor for executing a program for performing calculation and control operations, etc. In addition, an image encoded as a bitstream by an image encoding device can be transmitted to the image decoding device in real time or non-real time via a wired or wireless network such as the Internet, a local area network, a wireless LAN, a network WiBro or mobile communication network, or through various communication interfaces such as cable and universal serial bus (USB), is decoded by the image decoding device and restored and reproduced as an image.

[0059] In addition, an image encoded as a bitstream by an image encoding device can be transmitted from an encoding device to a decoding device via a computer-readable storage medium.

[0060] The above-described image encoding device and image decoding device, respectively, may be separate devices. However, the devices may be implemented as a single image encoding/decoding device according to the implementation. In this case, some components of the image encoding device are essentially the same technical elements as some components of the image decoding device, and can be implemented with at least the same structure or perform at least the same function as some components. image decoding devices.

[0061] Therefore, redundant descriptions of the respective technical elements will be omitted in the detailed description of the following technical elements and their working principles.

[0062] In addition, since the image decoding apparatus corresponds to a computing device that applies the image encoding method performed by the image encoding apparatus to decoding, the following description will focus on the image encoding apparatus.

[0063] The computing device may include a storage device that stores a program or software module that implements an image encoding method and/or an image decoding method, and a processor associated with the storage device for executing the program. In addition, an image encoding apparatus may be referred to as an encoder, and an image decoding apparatus may be referred to as a decoder.

[0064] In general, a picture may be made as a series of still pictures, and these still pictures can be classified by group of frames (GOP) elements, and each still picture can be called a frame. In this case, a frame may represent one of a frame or a field in a progressive or interlaced signal, and an image may be expressed as a "frame" if encoding/decoding is performed in frame units, and expressed as a "field" if encoding /decoding is done in field elements. The present invention discusses and describes a progressive signal. However, the present invention can also be applied to an interlaced signal. In a higher level concept, elements such as GOP and sequence may exist, and each frame may be divided into predefined regions such as slices, tiles, and blocks. In addition, one GOP may contain elements such as an I frame, a P frame, and a B frame. /decoded by performing a process such as motion estimation and motion compensation using a reference frame. In general, an I frame and a P frame may be used as reference frames in the case of a P frame, and an I frame and a P frame may be used as reference frames in the case of a B frame. However, the above definition can be changed by encoding/decoding setting.

[0065] In this case, the frame for encoding/decoding is called a reference frame, and the block or pixel related to it is called a reference block or reference pixel. In addition, the reference data may be not only a pixel value in the spatial domain, but also a coefficient value in the frequency domain and various types of encoding/decoding information generated and determined during the encoding/decoding process. Examples thereof may be information related to intra prediction or motion related information in the prediction section, information related to transformation in the transform section/inverse transform section, information related to quantization in the quantization section/inverse quantization section, information related to encoding/decoding (context information) in the encoding section/decoding section, information associated with the filter in the filter section in the loop, and so on.

[0066] The smallest element constituting an image may be a pixel, and the number of bits used to represent one pixel is called bit depth. In general, the bit depth may be 8 bits, and according to the encoding settings, a bit depth of more than 8 bits may be supported. As the bit depth, at least one bit depth according to the color space may be supported. In addition, at least one color space may be made in accordance with the color format of the image. One or more frames having a certain size or one or more frames having different sizes may be provided according to the color format. For example, in the case of YCbCr 4:2:0, one luminance component (Y in this example) and two color difference components (Cb/Cr in this example) may be provided. In this example, the component ratio of the color difference components and the luminance component may be a 1:2 ratio in width and height. As another example, in the case of 4:4:4, the width and height may be the same in aspect ratio. In the case of having one or more color spaces, as in the example above, the frame can be divided into the respective color spaces.

[0067] In the present invention, the description will be given based on some color spaces (Y in this example) of some color formats (YCbCr in this example), and the same or similar application (setting depending on the particular color space) can be applied to others. color spaces (Cb and Cr in this example) according to the color format. However, some differences can be made to each color space (regardless of the setting for a particular color space). In other words, a setting dependent on each color space may mean having a setting associated with or dependent on the component ratio of each component (e.g., 4:2:0, 4:2:2, 4:4:4, etc.). and an independent setting for each color space may mean that there is only a setting for the corresponding color space that does not take into account or does not depend on the component ratio of each component. In the present invention, depending on the encoder/decoder, some configurations may have independent or dependent settings.

[0068] The setting information or syntax element required in the image encoding process may be defined at the element level, such as video, sequence, frame, slice, tile, block, etc., which is included in the bitstream in elements, such as Video Parameter Set (VPS), Sequence Parameter Set (SPS), Frame Parameter Set (PPS), slice header, tile header, block header, etc., and is passed to the encoder, and the decoder can perform analysis in the elements of that the same level to recover the setting information transmitted from the encoder, and use the recovered setting information in the image decoding process. In addition, related information can be transmitted as a bitstream in the form of additional optimization information (SEI) or metadata, and it can be parsed and used. Each option set has a unique ID (identifier) value, and the bottom option set can have the value of the ID of the top option set to refer to. For example, the lower parameter set may refer to the information of the upper parameter set having the same ID value among one or more upper parameter sets. Among the examples of the various elements mentioned above, when one element includes one or more other elements, the corresponding element may be referred to as the top element, and the included element may be referred to as the bottom element.

[0069] The setting information generated in an element may include content regarding an independent setting for each element, or include content regarding a setting dependent on the previous, next, or top element. In this case, the dependent setting can be understood as indicating the setting information of the corresponding element as flag information indicating that the setting of the previous, next or top element is observed (for example, 1-bit flag, the setting is observed in the case of 1 and not observed in the case of 0) . The description of setting information in the present invention will focus on an example of independent setting. However, an example of adding or replacing content depending on the setting information of the previous or next element relative to the current element or the top element can be included.

[0070] FIG. 1 is a block diagram of an image encoding apparatus according to an embodiment of the present invention. In FIG. 2 is a block diagram of an image decoding apparatus according to an embodiment of the present invention.

[0071] With reference to FIG. 1, the image encoding apparatus may comprise a prediction section, a subtraction section, a transform section, a quantization section, an inverse quantization section, an inverse transform section, an adder section, an in-loop filter section, a memory, and/or an encoding section. Some of the above components may optionally be included, some or all of the components may be included selectively as needed, and some other configurations not shown in the figure may be included.

[0072] With reference to FIG. 2, the image decoding apparatus may comprise a decoding section, a prediction section, an inverse quantization section, an inverse transform section, an addition section, an in-loop filter section, and/or a storage device. Some of the above components may optionally be included, some or all of the components may be included selectively as needed, and some other configurations not shown in the figure may be included.

[0073] The image encoding device and the image decoding device may be separate devices, however, according to the implementation, may be composed as an image encoding/decoding device. In this case, some configurations of the image encoding device are essentially the same technical elements as some components of the image decoding device, and can be implemented with at least the same structure or perform at least the same function as some components. image decoding devices. Therefore, redundant descriptions of the respective technical elements will be omitted from the detailed description of the following technical elements and their working principles. Since the image decoding device corresponds to a computing device that applies the image encoding method performed by the image encoding device to decoding, the following description will focus on the image encoding device. An image encoding device may be referred to as an encoder and a video decoding device may be referred to as a decoder.

[0074] The prediction section may include an intra prediction section that performs intra prediction and an inter-picture prediction section that performs inter-picture prediction. In intra prediction, an intra prediction mode may be determined by configuring a block pixel adjacent to the current block as a reference pixel, and using the intra prediction mode, a prediction block may be generated. In inter-picture prediction, a prediction block may be generated by determining motion information of the current block using one or more reference pictures and performing motion compensation using the motion information. One of intra-picture prediction and inter-picture prediction (coding unit or prediction unit) may be specified for use with respect to the current block, and specific information (e.g., intra prediction mode, motion vector, reference video, etc.) may be specified. ) according to each prediction method. In this case, the processing element in which prediction is performed and the processing element in which the prediction method and specific content are determined can be determined according to the encoding/decoding setting. For example, a prediction method, a prediction mode, and so on. are defined at a prediction element (or coding element), and prediction is performed at a prediction block element (or coding element or transform element).

[0075] The subtraction section generates a residual block by subtracting a prediction block from the current block. That is, the subtraction section calculates the difference between the pixel value of each pixel in the current block to be encoded and the predicted pixel value of each pixel in the prediction block, which is generated by the prediction section, to generate a residual block, which is a residual signal in the image.

[0076] The conversion section can convert a signal belonging to the spatial domain into a signal belonging to the frequency domain, and the signal resulting from the conversion process is called a converted coefficient. For example, a transformation block containing a transformed coefficient can be obtained by transforming a residual block having a residual signal received from the subtractor section, and the input signal is determined according to the encoding setting and is not limited to the residual signal.

[0077] The transform section may transform the residual block using a transform scheme such as Hadamard transform, discrete sine transform (DST-based transform), or discrete cosine transform (DCT-based transform). However, the conversion scheme is not limited to this, and various conversion schemes obtained by improving and modifying this conversion scheme can be used.

[0078] For example, the transformation may support at least one transformation scheme, and each transformation scheme may support at least one detailed transformation scheme. In this case, the at least one detailed transformation scheme may be a transformation scheme in which the basis vector part is configured differently in each transformation method. For example, how transform technologies can support DST-based and DCT-based transforms. For DST, detailed mapping schemes such as DST-I, DST-II, DST-III, DST-V, DST-VI, DST-VII, and DST-VIII can be supported. For DCT, detailed mapping schemes such as DCT-I, DCT-II, DCT-III, DCT-V, DCT-VI, DCT-VII and DCT-VIII can be supported.

[0079] One of the transformations (for example, a transformation scheme and a detailed transformation scheme) may be set as a basic transformation scheme. Therefore, additional conversion schemes (eg, a plurality of conversion schemes and a plurality of detailed conversion schemes) may be supported. You can define support for an additional transformation scheme in sequence, frame, slice, tile, and so on elements. so that related information can be generated in a given element, and if an additional mapping scheme is supported, mapping schema selection information can be defined to generate related information in elements such as blocks.

[0080] The conversion may be performed in the horizontal/vertical direction. For example, using a basis vector in the transformation, a full 2D transformation can be performed by performing a 1D transformation in the horizontal direction and performing a 1D transformation in the vertical direction, thereby converting a pixel value in the spatial domain to a pixel value in the frequency domain.

[0081] In addition, the transformation in the horizontal/vertical direction can be performed adaptively. In more detail, the performance of an adaptive transform may be determined in accordance with at least one coding setting. For example, in the case of internal conversion, if the conversion mode is the horizontal mode, DCT-I may be applied in the horizontal direction, and DST-I may be applied in the vertical direction. If the conversion mode is the vertical mode, DST-VI may be applied in the horizontal direction, and DCT-VI may be applied in the vertical direction. In the case of the diagonal down left mode, DCT-II may be applied in the horizontal direction, and DCT-V may be applied in the vertical direction. In the case of diagonal down right mode, DST-I may be applied in the horizontal direction, and DST-VI may be applied in the vertical direction.

[0082] The size and shape of each transformation block may be determined according to the encoding cost of each transformation block size and shape candidate, and information such as the determined image data of each transformation block and the determined size and shape of each transformation block may be encoded.

[0083] A square transform in a transform shape may be specified as a basic transform shape, and additional transform shapes (eg, rectangular shapes) may be supported. Support for additional transformation forms can be defined in sequence, frame, slice, tile, etc. elements, related information can be generated in a given element, and transformation form selection information can be defined to generate related information in elements such as blocks. .

[0084] In addition, according to the encoding information, support for the transform block format can be determined. In this case, the encoding information may correspond to a slice type, an encoding mode, a block size and shape, a block division mode, and so on. That is, one transformation form can be supported according to at least one kind of encoding information, and a plurality of transformation forms can be supported according to at least one kind of encoding information. The first case may be an implicit case and the last case may be an explicit case. In the explicit case, adaptive selection information indicating the best candidate group among the plurality of candidate groups can be generated and included in the bitstream. It is understood that in the present invention, including the present example, when explicitly generating encoding information, corresponding information is included in the bitstream in various elements, and the decoder parses the related information in various elements to recover the decoded information. In addition, it is understood that when processing information about encoding/decoding implicitly, the encoder and the decoder perform processing using the same method and rule.

[0085] As an example, support for converting rectangular shapes can be defined according to the slice type. In the case of an I slice, the supported transform may be a square transform, and in the case of P/B slices, the transform may be a square or rectangular transform.

[0086] As an example, support for converting rectangular shapes may be determined according to the coding mode. In the case of intra prediction, the supported transform may be a square transform, and in the case of inter-picture prediction, the supported transform may be a square or rectangular transform.

[0087] As an example, support for converting rectangular shapes can be determined according to the block size and type. The transform shape supported in a predefined size block or larger may be a square shape transform, and the transform shape supported in a predefined size block or smaller may be a square or rectangular shape transform.

[0088] As an example, support for converting rectangular shapes can be defined according to the block splitting mode. If the block to be transformed is a block obtained by the quad-tree partitioning mode, the supported transformation form may be a square transformation, and if the block is a block obtained by the binary tree-based partitioning mode, the supported form the transformation may be a square or rectangular transformation.

[0089] The above example is an example of supporting a transform form according to one piece of encoding information, and multiple pieces of information can be combined to participate in setting supported additional transform forms. The above example is not limited to the above examples, but is directed to examples in which additional transform forms are supported according to different encoding settings, and various modification examples can be implemented.

[0090] According to the encoding settings or image characteristics, the conversion process may be omitted. For example, the conversion process (including the inverse process) may be omitted according to the encoding settings (lossless compression conditions are assumed in this example). In another example, the conversion process may be omitted if performing compression by conversion is not applied according to the characteristics of the image. In this case, the omitted transform may be the entire element, or one of the horizontal and vertical elements may be omitted, and it may be determined whether such omission is supported according to the block size and shape.

[0091] For example, in a setting in which horizontal and vertical transformation omission are associated, if the transformation omission flag is 1, the transformation is not performed in the horizontal and vertical directions, and if the flag is 0, the transformation is performed in the horizontal and vertical directions. In a setting where horizontal and vertical conversion operations are performed independently, if the first conversion omission flag is 1, the conversion is not performed in the horizontal direction, and if the flag is 0, the conversion is performed in the horizontal direction. If the second conversion omit flag is 1, no conversion is performed in the vertical direction, and if the flag is 0, the conversion is performed in the vertical direction.

[0092] In case the block size is in range A, omission of the transform may be supported, and in the case of range B, omission of the transform may not be supported. For example, if the horizontal block length is greater than M or the vertical block length is greater than N, the transform omitted flag may not be supported, and if the horizontal block length is less than m or the vertical block length is less than n, the transform omitted flag may be supported. M(m) and N(n) may be the same or different. Transform related settings can be defined in sequence, frame, slice, etc. elements.

[0093] If additional transformation schemes are supported, transformation scheme settings may be determined in accordance with at least one piece of encoding information. In this case, the encoding information may correspond to a slice type, an encoding mode, a block size and shape, a prediction mode, and so on.

[0094] As an example, support for a transform scheme may be determined according to an encoding mode. For intra prediction, the supported transform schemes may be DCT-I, DCT-III, DCT-VI, DST-II, and DST-III, and for inter-picture prediction, the supported transform schemes may be DCT-II, DCT-III, and DST -III.

[0095] As an example, support for a transformation scheme may be determined according to a slice type. In the case of slice I, the supported transform schemes may be DCT-I, DCT-II, and DCT-III, in the case of slice P, the supported transform schemes may be DCT-V, DST-V, and DST-VI, and in the case of slice B, the supported transform schemes may be DCT-I, DCT-II and DST-III.

[0096] As an example, support for a transform scheme may be determined according to a prediction mode. The transformation schemes supported in prediction mode A may be DCT-1 and DCT-II, the transformation schemes supported in prediction mode B may be DCT-1 and DST-1, and the transformation scheme supported in prediction mode C may be DCT-I. In this case, prediction modes A and B may be directional modes, and prediction mode C may be non-directional mode.

[0097] As an example, support for a mapping scheme may be determined according to the block size and type. The transformation scheme supported for blocks with sizes that are greater than a certain size may be DCT-II, the transformation schemes supported for blocks with sizes that are less than a certain size may be DCT-II and DST-V, and transformation schemes supported for blocks with sizes greater than or less than a certain size may be DCT-I, DCT-II, and DST-I. In addition, square-shaped conversion circuits may be DCT-1 and DCT-II, and square-shaped conversion circuits may be DCT-1 and DST-1.

[0098] The above examples are examples of supporting a transformation scheme according to one piece of encoding information, and multiple pieces of information can be combined to participate in customizing the supported additional transformation technologies. The examples are not limited to the examples above, but can also be converted to other examples. In addition, the transformation section can pass information necessary for generating a transformation block to the encoding section for encoding information, including the information in a bitstream, and transmitting to the decoder, and the decoding section of the decoder analyzes the information and uses the information for an inverse transformation process.

[0099] The quantization section is configured to quantize the input signal, and the signal resulting from the quantization process is called a quantized coefficient. For example, a quantized coefficient quantization block may be obtained by quantizing a residual block containing a residual transform coefficient received from the transform section. In this case, the input signal is determined according to the encoding setting, which is not limited by the residual transform coefficient.

[00100] The quantization section is configured to quantize the transformed residual block using quantization technology such as uniform threshold quantization with dead zone and weighted quantization matrix. However, the quantization technology is not limited to this, and various quantization technologies obtained by improving and modifying this quantization technology can be used.

[00101] In addition, the quantization section is configured to transmit information necessary for generating a quantization block to the coding section, whereby the information is encoded, the information is included in the bitstream, and transmitted to the decoder, and the decoding section of the decoder is configured to analyze the information and use information for the inverse quantization process.

[00102] In the above example, the description has been made on the assumption that the residual block is subjected to transformation and quantization by a transformation section and a quantization section. However, the residual signal of the residual block may be transformed to generate a residual block having a transformed coefficient, and the quantization process may not be performed. Alternatively, the residual signal of the residual block may not be converted to a transformed coefficient, and only the quantization process may be performed. Alternatively, neither the transform process nor the quantization process may be performed. This can be determined based on the encoder settings.

[00103] The encoding section scans the quantized coefficient, the transformed coefficient, or the residual signal of the generated residual block in accordance with at least one scan order (e.g., zigzag scan, vertical scan, horizontal scan, etc.), generates a sequence of quantized coefficients, a sequence transformed coefficients or signal sequence, and may perform coding using at least one entropy coding technique. In this case, the scan order information may be determined according to the coding setting (eg, coding mode, prediction mode, etc.), and related information may be generated implicitly or explicitly. For example, one of a plurality of scan orders may be selected according to the intra prediction mode. In this case, the scanning pattern can be set to one of various patterns such as zigzag pattern, diagonal lines, and raster.

[00104] In addition, encoding data including encoding information transmitted from each component may be generated and output as a bit stream, which may be implemented by a multiplexer (MUX). In this case, encoding may be performed using a method such as exponential Golomb, context adaptive variable length coding (CAVLC), or context adaptive binary arithmetic coding (CABAC) as the coding scheme. However, the coding scheme is not limited to this, and various coding schemes obtained by improving and modifying this coding scheme can be used.

[00105] When performing entropy encoding (in this example, CABAC is assumed) on residual block data and a syntax element such as information generated during encoding/decoding, the entropy encoding apparatus may comprise a binarizer, a context modeler, and a binary arithmetic encoder. In this case, the binary arithmetic encoder may comprise a normal coding engine and a bypass coding engine. In this case, the engine for normal coding may be a section executed on the context modeler, and the engine for bypass coding may be a means executed independently of the context simulator.

[00106] Because the syntax element input to the entropy encoding device may not be a binary value when the syntax elements are not binary values, the binarizer can binarize the syntax elements and output a binary number string containing 0 or 1 In this case, the binary number is a bit containing 0 or 1, and it can be encoded using a binary arithmetic encoder. In this case, one of the engine for normal coding and the engine for bypass coding may be selected based on the occurrence probability of 0 and 1, which may be determined according to the encoding/decoding setting. The engine for bypass encoding can be used when the syntax element is data in which the frequency of occurrence of 0 and 1 is the same, and the engine for normal encoding can be used in other cases that can be referenced when the subsequent engine for normal encoding is performed via context modeling ( or contextual information updates).

[00107] In this case, the context is information about the probability of occurrence of a binary number, and context modeling is a process of estimating the probability of a binary number necessary for binary arithmetic encoding using a binary number, which is obtained by binarization, as an input value . A binary number syntax element, an index that is the position of a binary number in a string of binary numbers, the probability that the binary number is included in an adjacent block, etc. can be used to evaluate the probability, and at least one context table can be used for this. For example, a plurality of context tables may be used as information for some flags according to a combination of whether the flags of neighboring blocks are used.

[00108] Various methods may be used when converting a syntax element to binary form. For example, the methods can be divided into fixed-length binarization and variable-length binarization. In the case of variable length binarization, one can use unary binarization (truncated unary binarization), truncated Rais binarization, exponential Golomb binarization of order k, truncated binarization, and etc. Also, signed binarization or unsigned binarization can be performed according to the value range of the syntax element. The binarization process for the syntax element carried out in the present invention can be performed, including not only the binarization mentioned in the above example, but also other additional binarization methods.

[00109] The inverse quantization section and the inverse transform section can be implemented by performing in reverse order the processes of the transform section and the quantization section. For example, the inverse quantization section is configured to inversely quantize the quantized transformed coefficient generated by the quantization section, and the inverse transformation section may inversely transform the inversely quantized transformed coefficient to generate a reconstructed residual block.

[00110] The addition section reconstructs the current block by adding the prediction block and the restored residual block. The reconstructed block may be stored in the memory and used as reference data (prediction section, filter element, etc.).

[00111] The in-loop filter section may provide at least one post-processing filtering component such as a deblocking filter, sample adaptive shift (SAO), and adaptive loop filter (ALF). The deblocking filter can remove block distortion that occurs at the boundary between blocks in the reconstructed image. ALF can perform filtering based on the value obtained by comparing the reconstructed image and the input image. In more detail, filtering can be performed based on the value obtained by comparing the image restored after filtering the block using the deblocker filter and the input image. Alternatively, filtering may be performed based on a value obtained by comparing the image reconstructed after filtering the block with SAO and the input video.

[00112] The recovered block or frame may be stored in a storage device. The reconstructed block or frame stored in the memory may be provided to a prediction section that performs intra or inter-picture prediction. More specifically, a storage space in the form of a bitstream queue compressed by an encoder may be treated as a coded frame buffer (CPB), and a space for storing a decoded image in frame elements may be treated as a decoded frame buffer (DPB). In the case of the CPB, the decoding items are stored according to the decoding order, the decoding operation is emulated in the encoder, the bitstream compressed during the emulation process can be stored, the bitstream output from the CPB is restored by the decoding process, the reconstructed image is stored in the DPB, and the frames stored in the DPB may be referred to in a subsequent picture encoding and decoding process.

[00113] The decoding section may be implemented by performing the process in the encoding section in reverse order. For example, a sequence of quantized coefficients, a sequence of transformed coefficients, or a sequence of signals may be received from a bitstream and decoded, and decoding data including decoding information may be parsed and transmitted to each component.

[00114] On the other hand, the block dividing section may also be included in the image encoding apparatus and the image decoding apparatus shown in FIG. 1 and 2, although she is not depicted on them. Information about the base bin may be obtained from the frame division section, and the base bin may represent the base (or start) bin used for prediction, transformation, quantization, and so on. during image encoding/decoding. In this case, coding units may be composed into one luma coding unit and two chroma component coding units according to a color format (YCbCr in this example), and the size of each block can be determined according to the color format. In the following examples, a block-based description (in this example, the luma component) will be implemented. In this case, the block is assumed to be an element that can be obtained after defining each element, and description will be given on the assumption that similar settings can be applied to other types of blocks.

[00115] A block division section may be defined in connection with each component of an image encoding apparatus and an image decoding apparatus, and a block size and shape may be determined by this process. In this case, the predetermined block may be defined differently depending on the configuration, and may correspond to a prediction block in the case of a prediction section, a transformation block in the case of a prediction section, and a quantization block in the case of a quantization section. The present invention is not limited to this, and block elements according to other components can be further defined. The size and shape of the block can be determined by the horizontal and vertical lengths of the block.

[00116] In a block division section, a block may be expressed as M×N, and the maximum and minimum values of each block may be obtained within a certain range. For example, if the block shape supports a square, and the maximum block value is 256×256, and the minimum block value is 8×8, it is possible to obtain a block having a size of 2 ^m ×2 ^m (in this example, m is an integer from 3 to 8, for example, 8x8, 16x16, 32x32, 64x64, ^{128x128, or 256x256), a block having a size of 2mx2m (} in this example, m is an integer from 4 to 128), or a block having a size m×m (in this example, m is an integer from 8 to 256). Alternatively, if the shape of the block supports square and rectangle, and it has the same range as in the above example, it is possible to obtain a block having a size of 2 ^m × 2 ⁿ (in this example, m and n are integers from 3 to 8, e.g. 8x8, 8x16, 16x8, 16x16, 16x32, 32x16, 32x32, 32x64, 64x32, 64x64, 64x128, 128x64, 128x128, 128x256, 256x128, or 256x256 assuming the maximum width-to-height ratio is 2:1, and depending on the encoding/decoding setting, there may be no width-to-height ratio limit, or there may there is a maximum value of this ratio). Alternatively, it is possible to obtain a block having a size of 2m×2n (in this example, m and n are integers from 4 to 128). Alternatively, it is possible to obtain a block having a size of m×n (in this example, the type is integers from 8 to 256).

[00117] Received blocks may be determined according to encoding/decoding settings (eg, block type, partitioning scheme, partitioning setting, etc.). For example, as a coding block, a block having a size of ^2m × ²ⁿ can be obtained, as a prediction block, a block having a size of 2m×2n or m×n can be obtained, and as a transformation block, a block having a size of ^2m ×2 can be obtained. ²ⁿ . Information about block size, range, etc. (eg information related to index and multiplier, etc.) can be generated based on the settings.

[00118] A range (defined in this example as a maximum value and a minimum value) may be determined according to the block type. In addition, for some blocks, the block range information may be explicitly generated, and for some blocks, the block range information may be implicitly defined. For example, the related information may be explicitly generated in the coding and transform block, and the related information may be implicitly processed in the prediction block.

[00119] In the explicit case, at least one piece of range information may be generated. For example, in the case of a coding block, the range information may generate maximum value and minimum value information. Alternatively, the information may be generated based on the difference between the maximum value and a predetermined minimum value (eg, 8) (eg, generated based on a setting, value information of the difference between the maximum value and minimum value indices, etc.). In addition, multiple range information can be generated for the horizontal and vertical lengths of the rectangular box.

[00120] In the implicit case, the range information may be determined based on encoding/decoding settings (eg, block type, partitioning scheme, partitioning setting, etc.). For example, in the case of a prediction block, information about the maximum value and minimum value can be obtained using a candidate group (in this example, M×N and m/2×n/2), which can be obtained by setting the division of the prediction block (for example, division based on a quaternary tree + split depth 0) in a coding block (for example, the maximum coding block size is M×N and the minimum coding block size is m×n), which is the top element.

[00121] The size and shape of the initial (or initial) block of the block division section may be determined by the top element. In the case of a coding block, the source block may be a basic coding block obtained from a frame split section, in the case of a prediction block, the source block may be a coding block, and in the case of a transform block, the source block may be a coding block or a prediction block, which may be determined according to with encoding/decoding settings. For example, if the coding mode is intra-prediction, the top element of the transform block may be the prediction block, and if the coding mode is inter-picture prediction, the prediction block may be an element independent of the transform block. The original block may be divided into smaller blocks as initial partition elements. If the optimal size and shape is determined according to the division of each block, the block can be determined as the source block of the bottom element. For example, in the first case, the block may be a coding block, and in the latter case (lower element), it may be a prediction block or a transformation block. If the source block of the bottom element is defined as in the example above, a splitting process can be performed to find a block having the optimal size and shape as the top element.

[00122] To sum up, the block splitting section is configured to split the base coding unit (or the largest coding unit) into at least one coding unit (or bottom coding unit). In addition, the coding element may be divided into at least one prediction element and may be divided into at least one transformation element. A coding unit may be divided into at least one coding block, a coding block may be divided into at least one prediction block, and may be divided into at least one transform block. The prediction element may be divided into at least one prediction block, and the transform element may be divided into at least one transformation block.

[00123] As in the above example, if a block having an optimal size and shape is found by the mode determination process, mode information (eg, split information, etc.) can be generated for that block. The mode information may be included in the bitstream along with the information generated by the component to which the given block belongs (eg, prediction related information, transform related information, etc.) and transmitted to the decoder and can be subjected to analysis in the same level element in the decoder and used in the image decoding process.

[00124] In the example described below, the separation scheme will be described, and the description will be given on the assumption that the original block is square. However, in the case of a rectangular shape, the same or similar example is allowed.

[00125] The block division section is configured to support various division modes. For example, tree partitioning or typed partitioning may be supported, and other methods may be used. In the case of a tree partition, a partition flag may be used to generate partition information, and in the case of a typed partition, index information for block shapes included in a predefined candidate group can be used to generate partition information.

[00126] FIG. 3 is a schematic diagram showing the tree form of the block.

[00127] With reference to FIG. 3, "a" depicts one 2N×2N block not split, "b" depicts 2N×N blocks obtained by some partition flag (in this example, horizontal split of the binary tree), "c" depicts two N×2N blocks , obtained by some split flag (in this example, a vertical split of a binary tree), and "d" depicts four N×N blocks obtained by some split flag (in this example, four splits of a quaternary tree, or horizontal and vertical splits of a binary tree) . The shape of the resulting block can be determined according to the type of tree used for splitting. For example, if quaternary tree partitioning is performed, the resulting candidate blocks may be "a" and "d" blocks. If binary tree-based partitioning is performed, the resulting candidate blocks may be "a", "b", and "c" blocks. In the case of a quad tree, one split flag is supported. Block "a" can be obtained if the split flag is "0", and block "b" can be obtained if the split flag is "1". In the case of a binary tree, a plurality of split flags are supported, one of which may be a flag indicating whether a split has been performed, one of which may be a flag indicating whether a horizontal/vertical split has been performed, and one of which may be a flag indicating whether horizontal overlap is allowed. /vertical separation. If overlap is enabled, the received candidate blocks may be blocks "a", "b", "c", and "d", and if overlap is not allowed, the received candidate blocks may be blocks "a", "b", and "c". . The quaternary tree may be a basic partitioning tree. In addition, a partitioning scheme based on a tree (in this example, a binary tree) may be included in the partitioning tree. If a flag is enabled implicitly or explicitly to enable additional tree-based partitioning, multiple tree-based partitioning can be performed. Tree partitioning may be a method allowing recursive partitioning. That is, the partitioned block can be set again as the original block to perform tree partitioning, which can be determined according to partitioning settings such as a partitioning range and a valid splitting depth. This scheme may be an example of a hierarchical partitioning scheme.

[00128] FIG. 4 is a schematic diagram depicting a typed block shape.

[00129] As shown in FIG. 4, according to the type, the block to be divided can have 1 division form (in this example, "a"), 2 division forms (in this example, "b", "c", "d", "e", "f ” and “g”) and 4 division forms (in this example, “h”). The candidate group may be configured with various configurations. For example, in FIG. 5, the candidate group may be configured with a, b, c, n, or a, b-g, n, or a, n, q, etc. But this example is not limited to them, and includes the following examples, whereby various converted examples can be executed. When the flag enabling symmetric separation is activated, the supported blocks may be "a", "b", "c", and "h" in FIG. 4, and if the flag enabling asymmetric partitioning is activated, the supported blocks may be all blocks "a"-"h" in FIG. 4. In the first case, the associated information (in this example, the flag enabling symmetric separation) may be implicitly activated, and in the latter case, the associated information (in this example, the flag enabling asymmetric separation) may be explicitly generated. Typed partitioning may be a way to support single partitioning. Compared to tree partitioning, further partitioning of blocks obtained by typed partitioning may not be allowed. This may represent an example where the allowable split depth is zero (eg, unilevel split).

[00130] FIG. 5 is a schematic diagram showing various block shapes that can be obtained using the block splitting section according to the present invention.

[00131] With reference to FIG. 5, blocks a-s may be obtained according to the division setting and division mode, and may also have additional block shapes that are not shown.

[00132] As an example, for tree partitioning, asymmetric partitioning may be enabled. For example, in the case of a binary tree, asymmetric partitioning may be allowed for blocks b and c in FIG. 5 (in this example, the case of partitioning into multiple blocks), or asymmetric partitioning may be allowed for blocks b-g in FIG. 5 (in this example, the case of division into multiple blocks). If the flag allowing explicit or implicit asymmetric separation is deactivated according to the encoding/decoding settings, the resulting block candidate may be block b or c (in this example, it is assumed that overlapping horizontal and vertical separations is not allowed), and if the flag, allowing asymmetric partitioning is enabled, the resulting candidate blocks may be b, d, e (in this example, horizontal partitioning) or c, f, g (in this example, vertical partitioning). This example may correspond to the case in which the division direction is determined by the horizontal or vertical division flag, and the block shape is determined according to the flag allowing asymmetry. However, the present invention is not limited to these examples, and they can be converted to other examples.

[00133] As an example, additional tree-based partitioning can be used for tree partitioning. For example, division based on a ternary tree, a quaternary tree, an octal tree, and so on can be performed. to get n split blocks (in this example, 3, 4, and 8, n is an integer). In the case of a ternary tree, the supported blocks (in this example divided into multiple blocks) may be h-m blocks. In the case of a quad tree, the supported blocks may be n-p blocks. In the case of an octal tree, the supported block may be block q. Tree partitioning support may be implicitly determined according to the encoding/decoding setting, or associated information may be explicitly generated. Furthermore, depending on the encoding/decoding settings, partitioning may be used alone or in combination with binary tree or quaternary tree partitioning. For example, in the case of a binary tree, blocks such as blocks b and c can be obtained. If a binary tree and a ternary tree are used in combination (in this example, it is assumed that the application range of the binary tree and the application range of the ternary tree partially overlap), blocks such as blocks b, c, i, and l can be obtained. If the flag allowing additional partitioning other than the existing tree is explicitly or implicitly deactivated according to the encoding/decoding setting, the resulting candidate block may be block b or c. If this flag is enabled, the resulting candidate block may be blocks b and i, or blocks b, h, i, and j (in this example, a horizontal split), or may be blocks c and l, or blocks c, k, l, and m (in this example). example vertical separation). This example may correspond to the case in which the division direction is determined by the horizontal or vertical division flag, and the block shape is determined according to the additional division enabling flag. However, the present invention is not limited to these examples, and these examples are converted to other examples.

[00134] For example, non-rectangular division may be allowed for typed blocks. For example, separation in the form of blocks r and s is possible. When combined with the typed candidate block group described above, the supported blocks are blocks a, b, c, h, r, s, or a-h, r, s. In addition, blocks supporting n partitions (eg, in this example, n is an integer 3 other than 1, 2, and 4), such as h-m blocks, may be included in the candidate group.

[00135] The split mode may be determined according to the encoding/decoding settings.

[00136] As an example, the division mode may be determined according to the block type. For example, a tree partition may be used in the coding block and the transform block, and a typed partition may be used in the prediction block. In addition, a combination of the two types of separation may be used. For example, a division mode may be used in a prediction block in which a tree division is combined with a typed division, and this division mode differs according to at least one range applied to the block.

[00137] For example, the division mode may be determined according to the block size. For example, tree partitioning may be performed on some range (eg, a×b-c×d, if the latter is larger) between the block's maximum value and minimum value, and typed partitioning may be performed on some range (eg, e×f-g×h). In this case, the range information according to the split mode may be explicitly generated or may be implicitly determined.

[00138] As an example, the splitting mode may be determined according to the shape of the block (or block before splitting). For example, if the block is square in shape, tree splitting and typed splitting can be performed. Alternatively, if the block is rectangular, a tree division may be performed.

[00139] The separation setting may be determined according to the encoding/decoding settings.

[00140] As an example, a split setting may be determined according to a block type. For example, in tree partitioning, quaternary tree partitioning may be used for a coding block and a prediction block, and binary tree partitioning may be used for a transform block. Alternatively, the allowable split depth may be set to m in the case of a coding block, the allowable split depth may be set to n in the case of a prediction block, and the allowable split depth may be set to in the case of a transform block, where m, n, and o may be the same or different. .

[00141] For example, the split setting may be determined according to the block size. For example, quaternary tree partitioning may be performed for a range of a block (eg, a×b - c×d), and for a range (eg, e×f - g×h, in this example, c×d is assumed to be greater than than g×h) splitting based on a binary tree can be performed. In this case, the range may include all ranges from the maximum value to the minimum value of the block, and these ranges may be defined as non-overlapping with each other or as overlapping with each other. For example, the minimum value of a certain range may be equal to the maximum value of a certain range, or the minimum value of a certain range may be less than the maximum value of a certain range. In the case of overlapping ranges, the split scheme containing the larger maximum value will take precedence. That is, in a partition scheme having a certain priority, execution of a partition scheme having a lower priority can be determined according to the partition result. In this case, the range information according to the tree type may be explicitly generated or may be implicitly determined.

[00142] As another example, typed splitting for some candidate groups can be performed in certain block ranges (same as in the example), and typed splitting can be performed in certain ranges (same as in the example) for some candidate groups ( in the present example, at least one configuration is different from the previous candidate group). In this case, the range may include all ranges from the maximum value to the minimum value of the block, and the range may have settings that do not overlap with each other.

[00143] As an example, the separation setting may be determined according to the shape of the block. For example, if the block has a square shape, a division based on a quaternary tree can be performed. In addition, if the block is rectangular, binary tree-based separation may be performed.

[00144] For example, the separation setting may be determined according to encoding/decoding information (eg, slice type, color component, encoding mode, etc.). For example, splitting based on a quaternary tree (or binary tree) may be performed in some range (for example, a×b - c×d), if the slice type is I, may be performed in some range (for example, e×f - g×h ) if the slice type is P, and may be performed in some range (for example, i×j - k×l) if the slice type is B. In addition, the allowable split depth based on a quaternary tree (or binary tree) can be set to to m if the slice type is I, may be set to n if the slice type is P, and may be set to o if the slice type is B. Here, m, n, and o may or may not be the same. Some types of slices may have the same configuration as other slices (eg, P and B slices).

[00145] As another example, the allowable separation depth based on a quaternary tree (or binary tree) may be set to m if the color component is a luminance component, and may be set to n if the color component is a color difference component, and m and n may or may not be the same. In addition, the division range based on the quaternary tree (or binary tree) in the case where the color component is the luminance component (for example, a×b - c×d) may or may not be the same as the division range based on the quaternary tree. (or binary tree) in the case where the color component is a color difference component (for example, e×f - g×h).

[00146] As another example, the allowable split depth based on a quaternary tree (or binary tree) may be m if the coding mode is intra prediction, and may be n (in this example, n is assumed to be greater than m) if the coding mode is inter-picture prediction, where m and n may or may not be the same. In addition, the division range based on the quaternary tree (or binary tree) in the case where the coding mode is intra prediction may or may not be the same as the division range based on the quaternary tree (or binary tree) in the case where the coding mode is is the prediction made between images.

[00147] In the above example, information about supporting the adaptive configuration of the split candidate group according to the encoding/decoding information may be explicitly generated or implicitly determined.

[00148] With the above example, a case has been described in which the split mode and the split setting are determined according to the encoding/decoding setting. The example above shows some cases according to each element, and modification to other cases may be possible. In addition, the split mode and split setting can be determined according to the combination of the plurality of items. For example, the split mode and split setting may be determined according to type, size, shape, encoding/decoding information, and so on. block.

[00149] In addition, in the above example, elements related to the split mode, setting, etc. may be implicitly defined, or this information may be explicitly generated to determine whether the adaptive case is allowed, as in the example above.

[00150] The split depth in the split setting refers to the number of occurrences of spatial splits with respect to the original block (in this example, the split depth of the original block is 0), and as the split depth increases, splitting into smaller blocks can be performed. The setting related to depth may vary depending on the split pattern. For example, in the tree splitting mode, one common depth may be used for a binary tree based splitting depth and a ternary tree based splitting depth, and separate depths may be used depending on the tree type.

[00151] In the above example, if a separate split depth is used according to the type of tree, this split depth can be set to 0 at the split start position (in this example, the block before the split) based on the tree. The split depth can be calculated by taking as the center the position where the split starts, not based on the split range of each tree (in this example, the maximum value).

[00152] FIG. 6 is a schematic diagram for explaining tree division according to an embodiment of the present invention.

[00153] "a" represents an example of separation based on a quaternary tree and a binary tree. Specifically, the top left box "a" depicts a quaternary tree partition, the upper right box and the lower left box depict a quaternary tree and binary tree partition, and the lower right box depicts a binary tree partition. In the figure, the solid line (in this example, Quad1) represents the boundary line, which splits into a quad tree, the dashed line (in this example, Binary1) represents the boundary line, which splits into a binary tree, and the heavy solid line (in this example, Binin2 ) represents a boundary line that splits into a binary tree. The difference between the dashed line and the heavy solid line represents the difference in separation schemes.

[00154] As an example (the allowable splitting depth based on the quaternary tree of the upper left block is 3, if the current block has a size of N×N, splitting is performed until one of the horizontal or vertical reaches (N>>3), and the split information is generated up to (N>>2) This also applies to the following example: The maximum and minimum values of the quad tree are assumed to be N×N, (N>>3)×(N>>3)), with performing tree division, they can be divided into four blocks with 1/2 horizontal and vertical lengths. When partition is enabled, the value of the partition flag may be "1", and when partition is deactivated, the value of the partition flag may be "0". According to the setting, the top-left block split flag may be generated together with the top-left block b.

[00155] As an example (for the upper right block, it is assumed that the allowable split depth based on the quad tree is 0, the allowable split depth based on the binary tree is 4, the maximum and minimum values of the split based on the quad tree are N×N, (N >>2)×(N>>2), the maximum and minimum values of the binary tree are (N>>1)×(N>>1), (N>>3)×(N>>3)), in case performing quaternary tree partitioning with respect to the original block, the upper right block can be divided into four blocks with a length of 1/2 of the horizontal and vertical lengths. The split block size is (N>>1)×(N>>), which means that binary tree splitting can be performed (in this example, it is greater than the minimum quad tree splitting value, but the splitting depth is limited). That is, this example may be an example in which the quaternary tree partition and the binary tree partition cannot overlap. The binary tree split information in this example can be configured with a plurality of split flags. Some flags may be horizontal split flags (corresponding to x in x/y in this example), and some flags may be vertical split flags (corresponding to y in x/y in this example). The split flags configuration may have settings similar to quaternary tree split (eg, whether to perform activation). In this example, two flags can be reactivated. If "-" is used to generate flag information on a figure, "-" may correspond to flag processing implicitly. If additional splitting cannot be performed according to conditions such as maximum value, minimum value, and tree splitting depth, implicit flag processing may be generated. According to the setting, the upper right block split flag may be generated together with the upper left block b.

[00156] As an example (for the upper left block, it is assumed that the allowable split depth based on the quad tree is 3, the allowable split depth based on the binary tree is 2, the maximum and minimum values of the split based on the quad tree are N×N, (N >>3)×(N>>3), the maximum and minimum binary tree partitioning values are (N>>2)×(N>>2), (N>>4)×(N>>4), and in the overlapping range, splitting priority is given to quad-tree splitting), when performing quad-tree splitting on the original block, the lower left block can be divided into four blocks with a length of 1/2 of the horizontal and vertical lengths. The partitioned block size is (N>>1)×(N>>), which means that, according to the settings of this example, binary tree-based partitioning and quaternary tree-based partitioning can be performed. That is, this example may be an example in which the quaternary tree division and the binary tree division may overlap. In this case, the execution of the binary tree division can be determined according to the result of the quaternary tree division having the predetermined priority. When performing quaternary tree splitting, binary tree based splitting will not be performed. If quaternary tree splitting is not performed, binary tree based splitting may be performed. If quad-tree division is not performed even if the conditions for division according to the above settings are satisfied, quad-tree division may not be performed. The binary tree split information in this example can be configured with a plurality of split flags. Some flags may be split flags (corresponding to x in x/y in this example) and some flags may be split direction flags (corresponding to y in x/y in this example, and it can be determined whether to generate y information according to with x). Split flags can be configured similar to quad tree split. In this example, horizontal and vertical separations cannot be activated by overlapping each other. When generating flag information as "-" in the figure, "-" can have settings similar to the above example. According to the setting, the bottom-left block split flag may be generated together with the bottom-left block b.

[00157] As an example (for the lower right block, it is assumed that the allowable splitting depth based on the binary tree is 5, the maximum and minimum splitting values based on the binary tree are N×N, (N>>2)×(N>>3 )), in the case of performing binary tree-based partitioning on the original block, the lower right block can be divided into two blocks with a length of 1/2 of the horizontal and vertical lengths. The split flag setting in this example may be the same as for the lower left block. When generating flag information as "-" in the figure, "-" can have settings similar to the above example. This example shows a case in which the horizontal and vertical minimum values of a binary tree are specified differently. According to the setting, the bottom right block split flag may be generated together with the bottom right block b.

[00158] As shown in the above example, after block information (e.g., type, size, shape, position, slice type, color component, etc.) is determined, a separation mode and setting can be determined according to the block information. separation, and the separation process can be performed accordingly.

[00159] FIG. 7 is a schematic diagram for explaining tree division according to an embodiment of the present invention.

[00160] With regard to block a and block b, the thick solid line L0 represents the maximum coding block, the block divided by the thick solid line, and the other lines L1-L5 represent the divided coding block, with the number inside the block representing the position of the divided sub-block (in this example in raster scan order), the number of "-" characters represents the division depth of the corresponding block, and the number of the boundary line between blocks represents the number of divisions. For example, if a block is divided into four blocks (based on a quaternary tree in this example), it has the order UL(0)-UR(1)-DL(2)-DR(3), and if the block is divided into two blocks ( in this example based on a binary tree), it has the order L, or U(0)-R, or D(1), which can be determined at each split depth. The following example shows a case in which the resulting coding units are limited.

[00161] As an example, it is assumed that the maximum coding block a has a size of 64x64, the minimum coding block has a size of 16x16, and quaternary tree partitioning is used. In this case, since the size of blocks 2-0, 2-1 and 2-2 (in this example having a size of 16×16) is equal to the size of the minimum coding block, they may not be divided into smaller blocks such as blocks 2-3 -0, 2-3-1, 2-3-2 and 2-3-3 (in this example having a size of 8×8). In this case, since there are 16×16 blocks obtained in blocks 2-0, 2-1, 2-2, and 2-3, that is, one group of candidates, no block division information is generated.

[00162] As an example, it is assumed that the maximum coding block b has a size of 64x64, the length of the minimum coding block in the horizontal or vertical direction is 8, and the allowable split depth is 3. In this case, since the block is 1-0- 1-1 (in this example having a size of 16×16 and a split depth of 3) satisfies the minimum coding block condition, this block can be divided into smaller blocks. However, since it is equal to the allowable split depth, this block cannot be divided into blocks having a larger split depth (in this example, blocks 1-0-1-0-0 and 1-0-1-0-1). In this case, since there are 16×8 blocks obtained in blocks 1-0-1-0 and 1-0-1-1, that is, one candidate group, no block division information is generated.

[00163] As shown in the example above, quaternary tree partitioning or binary tree partitioning may be supported according to encoding/decoding. Alternatively, quaternary tree partitioning and binary tree partitioning may be supported together. For example, one or two of the above modes may be supported according to block size, split depth, and so on. If the block belongs to the first block range, quaternary tree-based partitioning may be supported, and if the block belongs to the second block range, binary tree-based partitioning may be supported. When supporting a plurality of split modes, there may be at least one setting of the maximum coding block size, the minimum coding block size, the allowable split depth, and so on, according to each mode. The ranges can be configured to overlap with each other, or they can be non-overlapping. Alternatively, it is also possible that a certain range includes another range. The above settings can be defined according to single or mixed factors such as slice type, encoding mode, and color component.

[00164] As an example, a split setting may be determined according to a slice type. In the case of slice I, the supported split setting is such that, in the case of a quaternary tree, the supported split range is 128x128 - 32x32, and in the case of a binary tree, the supported split range is 32x32 - 8x8. In the case of P/B slice, the supported split setting is such that, in the case of a quaternary tree, the supported split range is 128x128 - 32x32, and in the case of a binary tree, the supported split range is 64x64 - 8x8.

[00165] As an example, a split setting may be determined according to a coding mode. If the coding mode is intra-prediction, the supported split setting is: For a binary tree, a split range of 64×64 to 8×8 and a valid split depth of 2 can be supported. If the coding mode is inter-picture prediction, the supported split setting is: for binary tree, a split range of 32×32 - 8×8 can be supported, and a valid split depth of 3.

[00166] As an example, a separation setting may be determined according to a color component. For the luma component, the supported split range is 256x256 - 64x64 in the case of a quaternary tree, and the supported split range is 64x64 - 16x16 in the case of a binary tree. The chroma component in the case of a quaternary tree supports the same settings as the luma component (in this example, the ratio of the lengths of each block is set according to the chroma format), and the supported separation range in the case of a binary tree is 64x64 - 4x4 (in this example assumes that this range in the same luma component is 128×128 8×8, 4:2:0).

[00167] The examples describe the case in which the division settings differ according to the block type. In addition, some blocks may be combined with other blocks to carry out the separation process. For example, when combining an encoding block and a transformation block into one element, a separation process is performed to obtain an optimal block size and shape, which may be not only an optimal encoding block size and shape, but also an optimal transformation block size and shape. Alternatively, an encoding block and a transformation block may be combined into one element, a prediction block and a transformation block may be combined into one element, an encoding block, a prediction block and a transformation block may be combined into one element, and other blocks may be combined.

[00168] This document describes the case in which each block is provided with a separate separation setting, but it is also possible to combine multiple elements into one element having one separation setting.

[00169] In the above process, the generated information will be included in the encoder in the bitstream in at least one element of sequence, frame, slice, tile, etc., and the decoder will parse the associated information from the bitstream.

[00170] In the image encoding/decoding process, the input pixel value may differ from the output pixel value, and a pixel value correction process may be performed to prevent distortion caused by operating errors. The pixel value correction circuit is a process of correcting a pixel value that is out of the pixel value range to be within the pixel value range, and may be referred to as clipping.

[00171] Table 1 shows an example code for a clipping function (Clip_x) used to perform pixel value adjustment. With reference to Table 1, the input as parameters of the clipping function (Clip_x) may be the input pixel value (pixel_val), as well as the minimum value (minI) and maximum value (maxI) of the allowed range of pixel values. Here, if the description is based on bit depth (bit_depth), the minimum value (minI) may be 0, and the maximum value (maxI) may be (2bit_depth-1). When executing the clipping function (Clip_x), the pixel input value (pixel_val, parameter A) that is less than the minimum value (minI) is replaced by the minimum value (minI), and the pixel input value that is greater than the maximum value maxI (parameter C) can be replaced maximum value (maxI). Therefore, after the pixel value adjustment is completed, the output value (output) may be returned as the output pixel value (pixel_val').

[00172] A pixel value range is determined according to bit depth, but the pixel values (e.g., frames, slices, tiles, blocks, etc.) that make up an image differ according to the type and characteristics of the image, so they may not be generated in the range of all pixel values. According to an embodiment of the present invention, in order for a range of pixel values to be used in an image encoding/decoding process, reference can be made to a range of pixel values that constitute an actual image.

[00173] For example, in the method for correcting pixel values according to Table 1, the minimum value (minI) of the clipping function may be used as the minimum value for the pixel values that constitute the actual image, and the maximum value (maxI) of the clipping function may be used as the maximum value for the pixel values that make up the actual image.

[00174] Summarizing, an image encoding/decoding apparatus may provide a method for correcting pixel values based on bit depth and/or a method for correcting pixel values based on a range of pixel values constituting an image. The encoder/decoder may maintain flag information used to determine whether it is possible to support an adaptive method for adjusting pixel values. If the flag information is "1", information on selecting a pixel value correction method can be generated, and if the flag information is "0", a predetermined pixel value correction method (in this example, a bit depth basis). If the pixel value adjustment method selection information indicates that the pixel value adjustment method is based on a range of pixel values that make up an image, related image pixel value information can be included. For example, information about the minimum and maximum values of each image and the median value described below in accordance with the color components may be included. The information generated during the adjustment can be recorded in the elements of video, sequence, frame, slice, tile, block, and so on. encoder and then transmitted, and the information recorded in the decoder may be parsed to recover related information in the same element.

[00175] On the other hand, by this process and by adjusting pixel values based on bit depth or adjusting pixel values based on a range of pixel values that make up an image, a range of pixel values including the minimum and maximum values among the pixel values can be changed (defined or set), or additional information about the range of pixel values can also be changed (defined or set). For example, the maximum and minimum values among the pixel values used to compose the actual image can be changed, and the middle value used to configure the pixel values can also be changed.

[00176] That is, in the process of adjusting pixel values based on bit depth, minI may represent the minimum pixel value of the image, maxI may represent the maximum pixel value of the image, I may represent the color component, and medianI may represent the center value of the image pixels, minI may be 0 , maxI may be equal to (1<<bit_depth-1)-1, midI may be equal to 1<<(bit_depth-1), and medianI may be obtained in other forms, including the above examples, according to encoding/decoding settings. The median value is just a term used in the present invention to describe, and may refer to pixel value range information that can be changed (determined or set) according to the pixel value correction process in the image encoding/decoding process.

[00177] For example, in the process of adjusting pixel values according to a range of pixel values that make up an image, minI may be the minimum value of image pixels, maxI may be the maximum value of image pixels, and medianI may be the center value of image pixels, medianI may be an average value of the pixel values in the image, may be the value at the center when the pixels are aligned, or may be a value obtained according to the range information of the image pixel values. For at least one of minI or maxI, medianI can be obtained. That is, medianI may be a pixel value within a range of image pixel values.

[00178] In particular, medianI may be a value obtained in accordance with information about the range of values of the image pixel (in this example, minI or maxI), such as (minI+maxI)/2 or (minI+maxI)>>1 , (minI+maxI+1)/2, (minI+maxI+1)>>1, and medianI can be obtained in other forms including the above examples according to encoding/decoding settings.

[00179] An embodiment will be described below in accordance with the processing of adjusting pixel values (in this example, it is the middle value).

[00180] As an example, the base bit depth is 8 bits (0-255), the process of adjusting pixel values based on the range of pixel values that make up the image is selected (in this example, the minimum value is 10, the maximum value is 190, and the median value , obtained (by averaging) from the minimum and maximum values, is 100), and if the current block is the first block in the picture (in this example, in the frame), the adjacent blocks used for encoding/decoding (in this example, the left, bottom left, top left, top and top right) do not exist, so the reference pixel may be filled with a middle value of 100. The reference pixel may be used to perform an intra prediction process according to the prediction mode.

[00181] As an example, the base bit depth is 10 bits (0-1023), the pixel value adjustment process is selected based on the range of pixel values that make up the image (in this example, the median value is 600, and there are associated syntax elements), and , if the current block is the first block in the image (slice or tile in this example), adjacent blocks used for encoding/decoding (in this example left, bottom left, top left, top and top right) do not exist, so the reference pixel may be filled with a median value of 600. The reference pixel may be used to perform an intra prediction process according to the prediction mode.

[00182] As an example, the base bit depth is 10 bits, the process of adjusting pixel values based on the range of pixel values that make up the image is selected (in this example, the intermediate value is 112, and there are associated syntax elements), the setting of determining whether use pixels of corresponding blocks in prediction of the current block, is activated according to the coding mode of adjacent blocks (intra prediction/prediction between pictures) (in this example, if the coding mode of the corresponding blocks is intra prediction, then it can be used for the reference pixel of the current block , and if the encoding mode is inter-picture prediction, it cannot be used for the reference pixel of the current block; if the setting is deactivated, it can be used as the current block pixel regardless of the encoding mode of the corresponding blocks; a related syntax element is constrained_intra_pred_flag, which can appear in type P or B images). If the current block is located on the left side of the picture, adjacent blocks used for encoding/decoding (in this example, left, bottom left, and top left) do not exist, and if there are adjacent blocks used for encoding/decoding (in this example, right and top right), however, since the block coding mode is inter-picture prediction and is disabled by the above setting, there is no available reference pixel, and then the reference pixel can be filled with the middle value (in this example, 112). That is, since there is no available reference pixel, it can be filled with a median value from the range of image pixel values. The reference pixel may be used to perform an intra prediction process according to the prediction mode.

[00183] In the above embodiment, the prediction section contains various depicted cases associated with the median value, but it may be included in a different video encoding/decoding configuration. In addition, the present invention is not limited to the above-described embodiment, and modifications and extensions can be made in various cases.

[00184] In the present invention, the pixel value correction process can be applied to the encoding/decoding process of a prediction section, a transform section, a quantization section, an inverse quantization section, an inverse transform section, a filter section, a memory, and so on. For example, an input pixel in the pixel value correction method may be a reference sample or a prediction sample in a prediction process, and may be a reconstructed sample in a transform, quantization, inverse transform, and inverse quantization process. In addition, the pixel may be a recovered sample from the loop filtering process, or may be a pixel stored in the memory. In this case, in the process of transformation, quantization, inverse transformation, and inverse quantization, the reconstructed sample may refer to the reconstructed sample before applying the filtering in the loop. The restored sample value in the filter in the loop may refer to the restored sample value after applying the filter in the loop. The recovered sample during filtering in the deblocking filter may refer to the recovered sample after application of the decoding filter. The recovered sample in the SAO process may refer to the recovered sample after applying the SAO. The recovered sample in the ALF process may refer to the recovered sample after applying the ALF. Examples of the various cases described above have been described, but the present invention is not limited thereto and can be applied to the input, intermediate, and output steps of all encoding/decoding processes that cause the pixel value adjustment process.

[00185] The following example is described based on the assumption that clipping function Clip_Y of the Y luminance component and clipping functions Clip_Cb and Clip_Cr of the Cb and Cr color difference components are supported.

[00186] In the present invention, the prediction section can be classified into intra prediction and inter-picture prediction, and intra-picture prediction and inter-picture prediction can be defined as follows.

[00187] Intra prediction may be a technique for generating a prediction value from a region in which encoding/decoding of the current image (e.g., frame, slice, tile, etc.) is completed, and inter-picture prediction may be a technique for generating a prediction value from an image (eg, frame, slice, tile, etc.) for which encoding/decoding was completed before the current image.

[00188] In addition, intra prediction may be a technique for generating a prediction value from a region in which encoding/decoding of the current picture is completed, but some prediction methods may exclude, for example, a method for generating a prediction value from a reference picture, block matching, pattern matching and other predictions. Inter-picture prediction may be a technique for generating a prediction value from a picture in which at least one encoding/decoding is completed. An image in which at least one encoding/decoding is completed may be configured to include the current image.

[00189] One of the above definitions may be observed in accordance with the encoding/decoding settings, and the following example will be described on the assumption that the first determination is observed. In addition, although the description is made on the assumption that the prediction value is a value obtained by prediction in the spatial domain, the present invention is not limited to this.

[00190] FIG. 8 shows a block splitting method according to an embodiment of the present invention. In more detail, this figure shows an example of the size and shape of a block obtained in accordance with one or more partitioning methods, starting from the basic coding block.

[00191] In this figure, the heavy solid line represents the base coding block, the thick dashed line represents the quaternary tree partition boundary, the double solid line represents the symmetric binary tree partition boundary, the solid line represents the ternary tree partition boundary, and the thin dashed the line represents the boundary of the asymmetric partition based on the binary tree. Except for the thick solid line, each line represents a border divided according to each division method. The partition settings described below (eg, partition type, partition information, partition information configuration order, etc.) are not limited to the case of the corresponding example, and various modifications can be made.

[00192] For ease of description, description will be given assuming that separate block division settings are specified for top left, top right, bottom left, and bottom right blocks (NxN, 64x64) based on a base coding block (2Nx2N , 128×128). First of all, due to one splitting operation, four sub-blocks are obtained in the original block (split depth 0 -> 1, i.e. split depth is increased by 1), and the maximum coding block is 128x128, the minimum coding block is 8x8, and the maximum split depth is 4 due to the split setting for quad tree, which is the setting typically applied to each block.

[00193] (1 time, upper left block, A1-A6)

[00194] In this example, if single tree based partitioning (quaternary tree based in this example) is supported, the size and shape of resulting blocks may be determined by a single block splitting setting (e.g., maximum coding block, minimum coding block, split depth block etc.). In this example, if the block can be obtained according to the division (respectively, dividing the width and height into two parts), the division information required for one division operation (based on the 4M×4N block before division, the division depth is increased by 1) , is a flag indicating whether the split is performed (in this example, if the flag is 0, x is split, and if the flag is 1, o is split). The resulting candidates may be 4Mx4N and 2Mx2N.

[00195] (2 times, upper right block, A7-A11)

[00196] In this example, if multi-tree splitting (in this example, quaternary tree and binary tree based) is supported, the size and shape of the resulting blocks can be determined by multi-tree splitting settings. In this example, it is assumed that in the case of a binary tree, the maximum coding block has a size of 64x64, the minimum coding block has a length of 4, and the maximum split depth is 4.

[00197] In this example, if more than two blocks (2 or 4 in this example) can be obtained according to the split, the split information needed for one split operation (the split depth of the quad tree is increased by 1) is the flag, indicating whether a division is performed, a flag indicating the type of division, a flag indicating the shape of the division, and a flag indicating the direction of division. The resulting candidates are 4Mx4N, 4Mx2N, 2Mx4N, 4MxN/4Mx3N, 4Mx3N/4MxN, Mx4N/3Mx4N and 3Mx4N/Mx4N.

[00198] If the ranges of quaternary tree splitting and binary tree splitting overlap (that is, in the range where quad tree splitting and binary tree splitting can be performed in the current stage), and the current block (in the pre-split state) is a block obtained by splitting based on a quad tree (parent block <if the splitting depth is 1 less than the current splitting depth> a block obtained by splitting based on a quaternary tree), splitting information can be divided into the following cases. That is, if a block supported in accordance with each division setting can be obtained by a plurality of division methods, division information can be generated by the following classification.

[00199] (1) When the quaternary tree division and the binary tree division overlap with each other, the following occurs.

[00200] In the above table, "a" is a flag indicating whether splitting is performed based on a quaternary tree. If this flag is set to 1, quadruple tree (QT) splitting is performed. If this flag is 0, then "b" is asserted, which is a flag that indicates whether splitting is performed based on a binary tree. If b is 0, no split is performed in the given block (no split), and if b is 1, binary tree based split is performed.

[00201] "c" is a flag indicating the split direction, c=0 means horizontal split (hor), and c=1 means vertical split (ver). "d" denotes a flag indicating the shape of the split, d=0 means symmetric split (SBT, symmetric binary tree), and d=1 means asymmetric split (ABT, asymmetric binary tree). Only when d is equal to 1, the information on the detailed split ratio (1/4 or 3/4) is checked in asymmetric split. If d is 0, in left/right or top/bottom blocks, the ratio of left block to top block is 1/4, and the ratio of top block to bottom block is 3/4. If d is 1, the relationship is reversed.

[00202] (2) If only binary tree-based partitioning can be performed, then the following occurs.

[00203] In the above table, separation information is expressed using flags "b" through "e" with the exception of flag "a".

[00204] FIG. 8, the case of block A7 is a case in which, in blocks (A7 to A11), prior to splitting, quad-tree based splitting can be performed (that is, even if quad-tree based splitting can be performed, splitting is performed instead of quad-tree based splitting). based on the binary tree), and thus corresponds to the case in which the split information is generated in item (1).

[00205] On the other hand, in the case of blocks A8-A11, if binary tree-based partitioning is performed without first performing quaternary tree-based partitioning in the blocks (A8-A11) prior to partitioning (i.e., quaternary tree-based partitioning in the respective blocks < A8-A11> is no longer possible), this case thus corresponds to the case in which the separation information is generated in item (2).

[00206] (3 times, lower left block, A12-A15)

[00207] In this example, multiple tree-based schema partitioning (in this example, quaternary tree, binary tree, and ternary tree) is supported, and the size and shape of the block that can be obtained can be determined by multiple block partition settings. In this example, it is assumed that in the case of a binary tree/ternary tree, the maximum coding block has a size of 64x64, the minimum coding block has a length of 4, and the maximum split depth is 4.

[00208] In this example, if more than two blocks (in this example, 2, 3, and 4) can be obtained according to the split, the split information required for one split operation is a flag indicating whether the split is performed, the flag A indicating the type of split, a flag indicating the shape of the split, and a flag indicating the direction of the split. The resulting candidates are 4Mx4N, 4Mx2N, 2Mx4N, 4MxN/4Mx2N/4MxN and Mx4N/2Mx4N/Mx4N.

[00209] If the division ranges based on quaternary tree and binary tree/ternary tree overlap with each other, and the current block is a block obtained by division based on quaternary tree, the division information can be divided into the following cases.

[00210] (1) If the quaternary tree partition and the binary tree/ternary tree partition overlap with each other, then the following occurs.

[00211] In the above table "a", a flag indicating whether splitting is performed based on a quaternary tree. If this flag is set to 1, a quad-tree split is performed. If this flag is 0, flag "b" is defined indicating whether binary tree splitting or ternary tree based splitting is performed. If b is 0, no further division is performed in this block. If b is 1, a binary tree split or a ternary tree split is performed.

[00212] "c" is a flag indicating the split direction, c=0 means horizontal split, and c=1 means vertical split. "d" is a flag indicating the type of split, d=0 means binary tree (BT) split, and d=1 means ternary tree (TT) split.

[00213] (2) If only binary tree/ternary tree splitting can be performed, then the following occurs.

[00214] In the above table, the separation information is expressed using the flags "b"-"d" with the exception of the flag "a".

[00215] FIG. 8, the case of blocks A12 and A15 corresponds to a case in which quaternary tree division can be performed in blocks A12 to A15 before division, and thus corresponds to a case in which division information is generated in item (1).

[00216] On the other hand, the case of A13 and A14 corresponds to a case in which, instead of splitting based on a quaternary tree, in blocks A13 and A14, splitting based on a ternary tree is performed before splitting, and thus corresponds to a case in which split information is generated in paragraph (2).

[00217] (4 times, lower left block, A16-A20)

[00218] In this example, multiple tree-based schema partitioning (in this example, quad tree, binary tree, and ternary tree) is supported, and the size and shape of the block that can be obtained can be determined by multiple block partition settings. In this example, it is assumed that in the case of a binary tree/ternary tree, the maximum coding block has a size of 64x64, the minimum coding block has a length of 4, and the maximum split depth is 4.

[00219] In this example, if more than two blocks (in this example, 2, 3, and 4) can be obtained according to the split, the split information required for one split operation is a flag indicating whether the split is performed, the flag , indicating the type of split, a flag indicating the shape of the split, and a flag indicating the direction of the split. Candidates obtained are 4M×4N, 4M×2N, 2M×4N, 4M×N/4M×3N, 4M×3N/4M×N, M×4N/3M×4N, 3M×4N/M×4N, 4M, if possible. ×N/4Mx2N/4MxN and Mx4N/2Mx4N/Mx4N.

[00220] If the division ranges based on quaternary tree and binary tree/ternary tree overlap with each other, and the current block is a block obtained by division based on quaternary tree, the division information can be divided into the following cases.

[00221] (1) If the quaternary tree partition and the binary tree/ternary tree partition overlap with each other, then the following occurs.

[00222] In the above table, "a" is a flag indicating whether splitting is performed based on a quaternary tree. If this flag is 1, a quad-tree split is performed. If this flag is 0, the flag "b" is defined to indicate whether splitting is performed based on a binary tree. If b is 0, no further division is performed in this block. If b is 1, a binary tree split or a ternary tree split is performed.

[00223] "c" is a flag indicating the split direction, c=0 means horizontal split, and c=1 means vertical split. "d" is a flag indicating the type of split, d=0 means binary tree split, and d=1 means ternary tree split. If d is 1, a flag "e" corresponding to the split shape is determined, if e is 0, a symmetrical split is performed, and if e is 1, an asymmetric split is performed. If e is equal to 1, information about the detailed division ratio in asymmetric division is determined, which is similar to the previous example.

[00224] (2) If only binary tree/ternary tree splitting can be performed, then the following occurs.

[00225] In the above table, the separation information is expressed using the flags "b" - "f" with the exception of the flag "a".

[00226] FIG. 8, the case of the block A20 corresponds to the case in which the quaternary tree division can be performed in the blocks A16 to A19 before the division, and thus corresponds to the case in which the division information is generated in item (1).

[00227] On the other hand, in the case of A16-A19, since instead of splitting based on a quaternary tree in blocks A16-A19, splitting based on a binary tree is performed before splitting, so this case corresponds to the case in which splitting information is generated at point (2).

[00228] Intra prediction of the prediction section according to the present invention will be described below.

[00229] FIG. 9 is an exemplary diagram showing an intra prediction mode predefined in an image encoding/decoding apparatus.

[00230] With reference to FIG. 9, 67 prediction modes are configured as groups of prediction modes for intra prediction. 65 prediction modes are directional modes (2-66) and two prediction modes are non-directional modes (DC, flat). In this case, the directional mode may be divided into tilt information (eg dy/dx) or angle information (degrees). All or part of the prediction modes described in the above examples may be included in the luminance component or chrominance component prediction candidate mode group, and other additional modes may be included in the prediction mode candidate group.

[00231] In addition, by using correlation between color spaces, a reconstructed block from a different color space that has been coded/decoded can be used to predict the current block, and a prediction mode that supports this can be enabled. For example, in the case of a chroma component, a reconstructed luminance component block corresponding to the current block can be used to generate a prediction block of the current block. That is, a prediction block may be generated based on the reconstructed block considering the correlation between color spaces.

[00232] A group of candidate prediction modes may be adaptively determined in accordance with encoding/decoding settings. To improve the prediction accuracy, the number of candidate groups can be increased, and to reduce the number of bits according to the prediction mode, the number of candidate groups can be reduced.

[00233] For example, one group may be selected from candidate group A (67 modes, 65 directional modes, and 2 non-directional modes), candidate group B (35 modes, 33 directional modes, and 2 non-directional modes), and candidate group C (18 modes, 17 directional modes and 1 non-directional mode), and this group of candidates can be selected adaptively or determined according to the size and shape of the block.

[00234] In addition, a group of candidate prediction modes may have different configurations in accordance with encoding/decoding settings. For example, as shown in FIG. 2 in FIG. 9, a group of prediction candidate modes is configured in the same way for all modes, or in FIG. 9, the candidate group may be configured such that the number of modes from 18 to 34 is greater than the number of modes from 2 to 18. Alternatively, the opposite case may also take place, and the candidate group may be configured adaptively according to the shape of the block (i.e., square, rectangular with long horizontal sides or rectangular with long vertical sides). For example, if the width of the current block is larger than the height, intra prediction modes belonging to 2-15 are not used, and intra prediction modes belonging to 67-80 may be used for replacement. On the other hand, if the width of the current block is smaller than the height, intra prediction modes belonging to 53 66 are not used, and intra prediction modes belonging to -14 to -1 can be used for replacement.

[00235] In the present description, unless otherwise indicated, it is assumed that intra prediction is performed using a predetermined group of prediction candidate modes (candidate group A) with equal intervals between modes, however, the main elements of the present invention can also be used to specify the above-described adaptive internal prediction after change.

[00236] FIG. 9, the prediction mode supported in the case of a square block or a rectangular block can be represented. In addition, the prediction mode supported in the case of a rectangular block may be a prediction mode other than the above example. For example, prediction modes with different numbers of prediction mode candidate groups or an equal number of prediction mode candidate groups but with larger block lengths are denser. Conversely, there may be a case in which the prediction modes are scattered, or the opposite case may be the case. In the present description, as shown in FIG. 9, the prediction mode is described on the assumption that the prediction mode setting (equal intervals between directional modes) is maintained regardless of the block shape, but it is also applicable in other cases.

[00237] Various methods can be used to set the index assigned to the prediction mode. In the case of the directional mode, an index assigned to each mode may be determined based on predetermined priority information according to the angle or slope information of the prediction mode. For example, the x-axis or y-axis mode (modes 18 and 50 in FIG. 9) may have higher priority, the diagonal mode (modes 2, 34, and 66) having an angular difference of 45 or -45 degrees based on horizontal or vertical mode may have the following priority, the diagonal mode having an angle difference of 22.5 or -22.5 degrees based on the diagonal mode may have the next priority, and the priority information can be set in this way (the next is 11.25 or - 11.25 degrees, etc.) or in various other ways.

[00238] Alternatively, the index may be assigned in a particular order of directions based on a predetermined prediction mode. For example, as shown in FIG. 9, an index may be assigned in a clockwise direction from a certain diagonal mode (mode 2). In the following example, description will be made based on the assumption that an index is assigned in a clockwise direction based on a predetermined prediction mode.

[00239] In addition, the non-directional prediction mode may assign index information before the directional mode, or may assign index information between directional modes, or may assign index information at the end, which may be determined according to encoding/decoding settings. In this example, the non-directional mode is assumed to have the highest index assignment priority among the prediction modes (it is assigned a low index, mode 0 is flat, and mode 1 is DC).

[00240] Although various examples of index assignment to the prediction mode have been described with the above examples, the index may be assigned with other settings, the above example is not limiting, or various converted examples may occur.

[00241] In the above example, the priority information has been described using the prediction mode index assignment example, however, the priority information is not only used to assign the prediction mode index, but can also be used in the prediction mode encoding/decoding process. For example, the priority information may be used for MPM configuration, etc., and multiple sets of priority information may be supported in the prediction mode encoding/decoding process.

[00242] The scheme for obtaining the intra prediction mode (in particular, the luminance component) of the current block will be described below.

[00243] The current block may use the default mode predefined in the image encoding/decoding device. The default mode may be directional mode or non-directional mode. For example, the directional mode may include at least one of vertical mode, horizontal mode, or diagonal mode. The non-directional mode may include at least one of planar mode or DC mode. If it is determined that the default mode is used for the current block, the intra prediction mode of the current block may be equated to the default mode.

[00244] Alternatively, the intra prediction mode of the current block may be obtained based on a plurality of MPM candidates. First, a predetermined MPM candidate can be selected from the above group of prediction candidate modes. The number of MPM candidates may be three, four, five or more. MPM candidates may be derived based on the intra prediction mode of adjacent blocks adjacent to the current block. An adjacent block may be a block adjacent to at least one of the left side, top end, top left end, bottom left end, or top right end of the current block.

[00245] In particular, the MPM candidate can be determined considering whether the intra prediction mode (candIntraPredModeA) of the block on the left side and the intra prediction mode (candIntraPredModeB) of the block on the upper end are the same, and whether candIntraPredModeA and candIntraPredModeB are non-directional modes.

[00246] For example, if candIntraPredModeA and candIntraPredModeB are the same, and candIntraPredModeA is not Omnidirectional, MPM candidates of the current block may include at least one of candIntraPredModeA, (candIntraPredModeA-n), (candIntraPredModeA+n), or Omnidirectional. Here, n may be an integer equal to 1, 2 or more. The non-directional mode may include at least one of planar mode or DC mode. For example, the MPM candidates of the current block may be defined as shown in Table 2 below. The index in Table 2 specifies, but is not limited to, the position or priority of the MPM candidates. For example, index1 may be assigned to DC mode, or index4 may be assigned.

[00247] Alternatively, if candIntraPredModeA and candIntraPredModeB are different, and candIntraPredModeA and candIntraPredModeB are not non-directional modes, the MPM candidates of the current block may include at least one of candIntraPredModeA, candIntraPredModeB, (maxAB-n), (maxAB+n), or non-directional mode. Here, maxAB is the maximum value of candIntraPredModeA and candIntraPredModeB, and n can be an integer equal to 1, 2, or more. The non-directional mode may include at least one of planar mode or DC mode. For example, the MPM candidates of the current block may be defined as shown in Table 3 below. The index in Table 3 defines, but is not limited to, the position or priority of the MPM candidates. For example, the maximum index may be assigned to the DC mode. If the difference between candIntraPredModeA and candIntraPredModeB is within a predetermined threshold range, MPM candidate 1 from Table 3 will be applied, otherwise MPM candidate 2 may be applied. Here, the threshold range may be a range greater than or equal to 2 and less than or equal to 62.

[00248] Alternatively, if candIntraPredModeA and candIntraPredModeB are different and only one of candIntraPredModeA and candIntraPredModeB is in non-directional mode, MPM candidates of the current block may include at least one of maxAB, (maxAB-n), (maxAB+n), or non-directional mode. Here, maxAB represents the maximum value of candIntraPredModeA and candIntraPredModeB, and n can be an integer equal to 1, 2, or more. The non-directional mode may include at least one of a flat mode or a DC mode. For example, the MPM candidates of the current block may be defined as shown in Table 4 below. The index in Table 4 specifies, but is not limited to, the position or priority of the MPM candidates. For example, index 0 may be assigned to DC mode, or a maximum index may be assigned.

[00249] Alternatively, if candIntraPredModeA and candIntraPredModeB are different and both of candIntraPredModeA and candIntraPredModeB are non-directional modes, the MPM candidates of the current block may include at least one of non-directional mode, vertical mode, horizontal mode, (vertical mode-m), (vertical mode+m), (horizontal mode-m) or (horizontal mode+m). Here, m may be an integer equal to 1, 2, 3, 4 or more. The non-directional mode may include at least one of the flat mode or the DC mode. For example, the MPM candidates of the current block may be defined as shown in Table 5 below. The index in Table 5 defines, but is not limited to, the position or priority of the MPM candidates. For example, index1 may be assigned to horizontal mode, or a maximum index may be assigned.

[00250] An MPM candidate of the aforementioned MPM candidate set, indicated by the MPM index, may be set as the intra prediction mode of the current block. The MPM index may be encoded and reported by the image encoding device.

[00251] As described above, the intra prediction mode can be obtained by selectively using any of the default or MPM candidate mode. The selection may be made based on a flag provided by the encoder. In this case, the flag may indicate whether the intra prediction mode of the current block is set as the default mode. If the flag has a first value, the intra prediction mode of the current block is set as the default mode, and if not, information about whether the intra prediction mode of the current block is derived from the MPM candidate, MPM index, etc. may be transmitted.

[00252] The color difference component may have the same candidate group as the luma component prediction mode candidate group, or may include a candidate group made from part of the luminance component prediction mode candidate group. In this case, the group of candidate color difference component prediction modes may have a fixed configuration or a variable (or adaptive) configuration.

[00253] (Fixed Candidate Group Configuration Versus Candidate Group Variable Configuration)

[00254] As an example of a fixed configuration, some modes in the luminance component prediction mode candidate group are configured as a group of chroma component prediction modes to perform intra prediction (for example, it is assumed that these some modes are at least one of the DC mode, planar mode, vertical mode, horizontal mode or diagonal mode < some modes are assumed to be at least one of DL, UL or UR, DL is a mode in which prediction is performed from the lower right corner to the upper right corner, UL is the mode , in which prediction is performed from the upper left corner to the lower right corner, and UR is a mode in which prediction is performed from the upper right corner to the lower left corner, and it is assumed that the specified some modes are, respectively, modes 2, 34 and 66 on Fig. 9.

[00255] As an example of a configuration variable, some modes in the luma component prediction mode candidate group are configured as a base group of chroma component prediction modes (for example, it is assumed that these some modes are DC mode, planar mode, vertical mode, horizontal mode and diagonal UR mode, and it is assumed that more selected modes are configured as the base group of candidate prediction modes in general). However, the modes included in the candidate group may not correctly reflect the characteristics of the color difference component. To improve in this regard, the configuration of the group of candidate color difference prediction modes can be changed.

[00256] For example, at least one prediction mode of a block or subblock of a luma component at a position identical to or corresponding to a block of a chroma component may be included (Example 1 described below) in the base group of candidate prediction modes, or instead of some modes, a new group of candidate prediction modes (Example 2 described below). For example, in the case where the corresponding position in the luminance component corresponding to the color difference component <according to color format> cannot be configured with one block, but is configured as a plurality of sub-blocks by dividing one block, a preset position block is displayed. In this case, the position of the predetermined block is determined at the top left, top right, bottom left, bottom right, middle, top middle, bottom middle, left middle, and right middle positions in the luminance component block corresponding to the chroma component block; if it differs to a coordinate within the image, it can be position including coordinate (0, 0) top left, position including coordinate (blk_width-1, 0) top right, position including coordinate (0, blk_height-1) bottom left, position including the coordinate (blk_width-1, blk_height-1) bottom right, position including one of the coordinates (blk_width/2-1, blk_height/2-1), (blk_width/2, blk_height/2-1), (blk_width/2-1, blk_height/2) and (blk_width/2, blk_height/2) at the center, a position including one of the coordinates (blk_width/2-1, 0) and (blk_width/2, 0) at the top middle, a position including one of the coordinates (blk_width/2-1, blk_height-1) and (blk_width/2, blk_height-1) bottom middle, a position including one of the coordinates (0, blk_height/2-1) and (0, blk_height /2) on the left in the middle, and a position that includes one of the coordinates (blk_width-1, blk_height/2-1) and (blk_width-1, blk_height/2) on the right in the middle. That is, it means a block including the position of the coordinate. The above blk_width and blk_height refer to the horizontal and vertical lengths of the luminance block, and the coordinates are not limited to the above case and may include other cases. In the following description, for the luma component prediction mode <or color mode> added to the chroma difference component prediction mode candidate group, at least one prediction mode is added according to a predetermined priority <for example, assuming top left-top right-bottom left -lower right-middle>. If two prediction modes are added, the upper left block mode and the upper right block mode are added according to the above settings. In this case, if the blocks in the top left and top right positions are configured by one block, the bottom left block mode having the next priority is added.

[00257] Alternatively, at least one prediction mode (which can determine blocks adjacent to a block at the present position) may be included in the base prediction mode candidate group, if a plurality of modes are included in the color difference component prediction candidate mode group, blocks with a predetermined priority and sub-block prediction modes with a predetermined priority within the sub-blocks may be included as a candidate set according to priority) adjacent blocks left, top, top left, top right, bottom left, and the like. or subblocks of corresponding blocks (if adjacent blocks are configured by multiple blocks), or a new group of candidate prediction modes may be configured instead of some modes.

[00258] If additional content is added to the above description, not only the prediction mode of the luminance component block or adjacent block (luma block) can be included, but also at least one mode derived from the prediction mode can be included as the prediction mode color difference component. The following examples will include an example in which the luminance component prediction mode is adopted as the chrominance component prediction mode, and a detailed description about the example of adopting the prediction mode (for example, taking the adjacent mode and the directional mode of this mode as examples if the horizontal mode 18 is luma component prediction mode, modes 17, 19, 16, etc. are equivalent to the derivative prediction mode, and if a plurality of prediction modes are configured from the chroma component as a chroma component prediction mode candidate group, the configuration priority of the candidate group can be set in order from mode luma component prediction mode derived from luminance component prediction mode) or prediction mode derived from adjacent block prediction mode as a group of candidate chrominance component prediction modes will be omitted. However, the same or modified settings described below may apply.

[00259] As an example (1), if the luminance component prediction mode matches one of the chroma component prediction mode candidate groups, the candidate group configuration is the same (the number of candidate groups does not change), and the candidate sample group configurations are different if there is no match (the number of groups of candidates is increasing).

[00260] If the candidate group configurations in the above example are the same, the prediction mode indices may be the same, or other indices may be assigned, which may be determined according to encoding/decoding settings. For example, if the indices of the chroma component prediction mode candidate group are flat 0, DC1, vertical 2, horizontal 3, and diagonal UR4, in the case where the luminance component is horizontal, the configuration of the prediction mode candidate group remains unchanged, and the indices of each prediction mode remain unchanged. , or other indices may be assigned (in this example, horizontal 0, flat 1, DC2, vertical 3, and diagonal UR4). The reset of the indices presented above may be an example of a process performed to generate fewer mode bits during prediction mode encoding/decoding (assuming: assigning fewer bits to a small index).

[00261] If the candidate group configurations in the examples above are not the same, they may be added if the prediction mode indices may remain the same, or different indices may be assigned. For example, in the case where the prediction mode candidate group index setting is the same as in the previous example, if the luminance component prediction mode is diagonal DL, the prediction mode group configuration is incremented by 1, and the existing candidate group prediction mode index remains unchanged. In addition, the index of the newly added mode may be used as the last one (in this example diagonal DL5), or other indexes may be assigned (in this example diagonal DLO, flat 1, DC2, vertical 3, horizontal 4 and diagonal UL5).

[00262] As an example (2), if the luma component prediction mode matches one of the chroma component prediction mode candidate groups, the candidate group configuration is the same (candidate group modes do not change), and the candidate group configurations are different if there is no match (at least one of the candidate group modes is replaced).

[00263] If the candidate group configurations in the above examples are the same, the prediction mode indices may be the same, or different indices may be assigned. For example, if the chroma component prediction mode group indices are flat 0, DC1, vertical 2, horizontal 3, and diagonal UR4, in the case where the luminance component prediction mode is vertical, the configuration of the prediction mode candidate group remains unchanged, and the indices of each prediction mode, or other indices may be assigned (in this example, vertical 0, horizontal 1, diagonal UL2, flat 3, and DC4; this example includes, but is not limited to, an example in which the directional mode precedes if the luminance component mode is directional, and an example in which the omnidirectional mode precedes in case the luma component mode is omnidirectional).

[00264] If the configurations of the candidate groups in the above examples are different, the mode in which the prediction mode index remains unchanged, and the change mode may be assigned a replacement mode index, or the plurality of prediction modes may be assigned indices different from the existing index. For example, if the index setting of the prediction mode candidate group is the same as in the previous example, in case the luminance component prediction mode is diagonal DL, one mode in the prediction mode group (in this example, diagonal UL) will be replaced, and the index the prediction mode of the existing group of candidates remains unchanged. The mode index to be replaced is assigned as an index (eg, diagonal DL4) of the newly added mode, or other indices may be assigned (in this example, diagonal DL0, flat 1, DC2, vertical (3), horizontal 4).

[00265] In the foregoing description, an example of performing index reset is given to assign fewer mode bits, but this is only an example according to encoding/decoding settings, and other cases may occur. If the prediction mode index does not change, binarization may be performed in which a small number of bits are assigned to a small index, or binarization may be performed in which bits are assigned regardless of the size of the index. For example, if the reset prediction mode candidate group is flat 0, DC1, vertical 2, horizontal 3, and diagonal DL4, even if diagonal DL is assigned a large index, since diagonal DL is a mode derived from the luma component, fewer mode bits are assigned than other prediction modes.

[00266] The prediction mode may be a mode supported in a manner independent of the image type, or may be a mode that determines support according to some image types (e.g., a mode supported by image type I but not supported by type P or B Images).

[00267] The content described with the examples above is limited to the present example, and additional or other modified examples are possible. In addition, the encoding/decoding settings described with the above examples may be defined implicitly, or into video, sequence, frame, slice, tile, and so on elements. related information may be explicitly included.

[00268] (Obtaining prediction values in the same color space versus obtaining prediction values in other color spaces)

[00269] In the case of the intra prediction mode described with the above example, the prediction mode is described with respect to a method (eg, extrapolation, interpolation, averaging, etc.) of obtaining data for generating a prediction block from adjacent regions within the same time, and space.

[00270] In addition, a prediction mode can be supported with respect to a data acquisition method for generating a prediction block from regions located within the same time and different spaces.

[00271] For example, an example of the content described above may be a prediction mode with respect to a data acquisition method for generating a prediction block in other color spaces using correlation between color spaces. Here, when YCbCr space is taken as an example, the correlation between color spaces may represent the correlation between Y and Cb, Y and Cr, Cb and Cr. That is, in the case of a Cb or Cr color difference component, a reconstructed luminance component block corresponding to the current block can be generated as a prediction block of the current block (the base setting of the following example is color difference versus luminance). Alternatively, a reconstructed block of some color difference components (Cb or Cr) corresponding to the current block of some color difference components (Cr or Cb) may be generated as a block of prediction of color difference components (Cr or Cb). In this case, the reconstructed block of other color spaces can be directly generated as a prediction block (i.e., without performing correction), or as a prediction block, a block obtained taking into account the correlation between colors can be generated (for example, an existing reconstructed block is corrected by the formula P=a*R+b, where a and b represent values to be corrected, R and P respectively represent values obtained in other color spaces and prediction values of the current color space).

[00272] In this example, the description is made on the assumption that the data obtained using the color space correlation is used as the prediction value of the current block, but it is also possible in the case that the data is used as a correction value for correcting the prediction value of the existing current block (for example, the residual values of other color spaces are used as correction values; that is, there are other prediction values, and these prediction values are corrected; and although these prediction values are added to obtain a prediction value as well, the description is made for a detailed distinction). The present description is intended to describe the first case, but the present invention is not limited thereto, and the correction values may be used in a similar or variable manner.

[00273] The prediction mode may be a mode supported in a manner independent of the image type, or may be a mode that determines support according to some image types (e.g., a mode supported by image type I but not supported by type P or B Images).

[00274] (Comparison with obtaining related information)

[00275] In the example above, correlation information (a, b, etc.) between color spaces may explicitly include related information, or it may be obtained implicitly. In this case, the area of comparison to obtain related information may be 1) the current chroma component block and the corresponding luminance component block, or 2) an adjacent area (e.g., left block, top block, top left block, top right block, bottom left block, etc.). .e) of the current block of the color difference component and the adjacent area of the corresponding block of the luma component. The present example is an example in which the former may be expressed explicitly and the latter may be expressed implicitly.

[00276] For example, related information is obtained by comparing at least one pixel value of each color space (in this case, the pixel value to be compared may be a pixel value obtained from one pixel in each color space, and may be a pixel value obtained from many pixels). The pixel values obtained from the weighted average filtering process or the like, that is, the number of pixels that are referenced or used to compare one pixel value of each color space, may be one pixel versus one pixel, one pixel versus a plurality of pixels or the like. In this case, the former may be the color space for generating the prediction value, and the latter may be the reference color space. The above example may be possible according to the color format, or the pixel value for one pixel of the chroma component can be compared with the pixel value for the corresponding pixel of the luma component in a color format independent way, the pixel value for one pixel of the chroma component can be compared with the pixel value , obtained by performing the filter <a-tapseparatelDfilter, bxcmasknon-separable2Dfilter, d-tapdirectionalfilter, etc.> on a set of pixels of the luminance component, and either of these two methods can be used according to the encoding/decoding settings. Examples of color difference and brightness have been described above, but converted examples such as color difference <Cb> and color difference <Cr> may also occur.

[00277] In the above example, the area of comparison, if the related information is obtained implicitly, may be the nearest line of pixels of the current block of the current color component (for example, pixels, including from p[-1, -1] to p[blk_width-1, -1], p[-1, 0] to p[-1, blk_height-1]), and the corresponding pixel lines of other color spaces, or the set of pixel lines of the current block of the current color component (for example, in the above case, the pixels included in a plurality of pixel lines including p[-2, -2] to p[blk_width-1, -2], p[-2, -1] to p[-2, blk_height-1], and corresponding pixel lines of others color spaces.

[00278] Specifically, assuming the color format is 4:2:0, the pixel value for one pixel at a predetermined position can be used to compare the pixel value for one pixel in the current color space (in this example, color difference). (in this example, selected from top left, top right, bottom left, and bottom right within 2x2) in 4 corresponding pixels (one chroma component pixel corresponds to 4 pixels within a 2x2 luma component) in other color spaces (in this bit brightness example). Alternatively, to compare a pixel value for a single pixel in chrominance space, a pixel value obtained by performing filtering on a plurality of pixels (e.g., at least two pixels in corresponding 2×2 pixels, etc.) in luma space can be used.

[00279] To summarize, parameter information can be obtained from reconstructed pixels in an adjacent area of the current block and reconstructed pixels in other respective color spaces. That is, based on the associated information, at least one parameter (e.g., a or b, a1, b1 or a2, b2, etc.) can be generated and can be used as a value that is multiplied by or added to pixels restored blocks in other color spaces (for example, a, a1, a2/b, b1, b2).

[00280] In this case, the comparison processing may be performed after determining the availability of the pixels subjected to comparison in the above example. For example, if the adjacent area is available, it can be used as a comparison pixel, and if the adjacent area is not available, it can be determined according to the encoding/decoding settings. For example, if a pixel in the adjacent area is unavailable, it may be excluded from the process of obtaining the associated color space information, or it may be included in the comparison process after filling the unavailable area, which may be determined according to encoding/decoding settings.

[00281] For example, when excluding from the process of obtaining information about the correlation between color spaces, this example may be an example corresponding to the case in which an area containing pixels of at least one color space is inaccessible. In more detail, this may be an example corresponding to the case in which the pixels of one of the two color spaces are not available, or the pixels of the two color spaces are not available, which can be determined according to the encoding/decoding settings.

[00282] Alternatively, if the process of obtaining information about the correlation between color spaces is performed after filling data for comparison in the inaccessible area (or an operation similar to the process of filling the reference pixel), various methods of filling can be used. For example, a predefined pixel value can be used for padding (e.g., median value 1 of bit_depth << (bit_depth-1), a value between the minimum and maximum values of the actual image pixel, the average of the actual image pixel, the median value, etc.) , adjacent pixels or a value obtained by performing filtering (an operation similar to the filtering of reference pixels) on adjacent pixels may be used for filling, or other methods may be used.

[00283] FIG. 10 shows an example of comparing pixels from different color spaces to obtain correlation information. For ease of description, it is assumed that a 4:4:4 color format is used. In this case, the following process according to the component ratio of the color format (ie, the process including the conversion process based on the component ratio) is described.

[00284] R0 represents an example in which color space regions on both sides can be used. Since the available areas are the color space areas on both sides, pixels in the respective areas can be used in the comparison process to obtain related information.

[00285] R1 represents an example in which one of the color space regions on both sides is not available (in this example, the adjacent region of the current color space is available while the corresponding regions of the other color spaces are not available). The blank areas can be used for the comparison process after using various methods to fill the blank areas.

[00286] R2 represents an example in which one of the color space regions on both sides is not available (in this example, the adjacent region of the current color space is not available, while the corresponding regions of other color spaces are available). Since there is an inaccessible area on one side, the corresponding areas of the two color spaces cannot be used in the comparison process.

[00287] R3 represents an example in which color space regions on both sides are not available. Inaccessible areas are filled using various methods and can be used for the comparison process.

[00288] R4 represents an example in which color space regions on both sides are not available. Since there are inaccessible areas on both sides, the corresponding areas of color spaces on both sides cannot be used in the comparison process.

[00289] In addition, in contrast to FIG. 10, if adjacent areas of the current block or corresponding areas of other color spaces are not available, various adjustments can be made.

[00290] As an example, a and b can be assigned predefined values (in this example, a is 1 and b is 0). This case may involve maintaining a mode in which data in a different color space is filled with the current block's prediction block. In addition, this case may allow, when performing encoding/decoding of the prediction mode, that the setting or priority of the occurrence (or selection) probability of the mode has a setting different from the existing case (for example, the selection probability is set to be less high, or the priority is set to be less high). ; in other words, due to the low accuracy of the associated information, the accuracy of the prediction block obtained using this prediction mode may be very low, so it is estimated that the mode will not be selected as the optimal prediction mode).

[00291] For example, since there is no data to compare, the mode of filling different color space data with a prediction block of the current block may not be supported. That is, this mode may be a mode supported when there is at least one available area. In this case, when encoding/decoding is performed, the prediction mode may be configured to allow or not to allow this mode to be replaced by other modes. The first case may be a setting in which the number of prediction mode groups is kept, and the latter case is a setting in which the number of prediction mode groups is reduced.

[00292] The present invention is not limited to the above examples, and various converted examples may also take place.

[00293] In the above example, a disabled case is a case in which an area is outside the boundaries of an image (e.g., frame, slice, tile, etc.) (i.e., the current block and the area are not included in the same image), if the area has not been encoded/decoded. For example, a non-accessible case can be added according to encoding/decoding settings (eg, constrained_intra_pred_flag, etc., for example, if the type is slice P or B, the above flag is 1, and the encoding mode of the corresponding region is inter-picture prediction. ).

[00294] In the following example, the above limitation may occur when generating a prediction block of the current block using recovered data from other color spaces after obtaining related information by comparing two color spaces. That is, as described above, if it is determined that the corresponding regions of other color spaces corresponding to the current block are not available, the use of this mode may be limited, or this mode is not available.

[00295] To generate a prediction value of the current block, a parameter representing correlation information between color spaces obtained by the above process and recovered data of other color spaces corresponding to the current block can be used. In this case, the reconstructed data of other color spaces used to predict the current block may be a pixel value for a pixel at a predetermined position, or a pixel value obtained by a filtering process.

[00296] For example, in the case of 4:4:4, to generate a prediction value of one pixel in chrominance space, a pixel value for the corresponding pixel in luma space can be used. Alternatively, to generate a prediction value of one pixel in chrominance space, pixel values obtained by filtering a plurality of pixels in luma space (e.g., pixels in directions such as left, right, up, down, up left, up right, down left) can be used. and down right by centering on the corresponding pixel, if 5-tap and 7-tap filters are used as examples, it is understood that there are two pixels and three pixels specifically in the left, right, up and down directions, respectively, by centering on corresponding pixel).

[00297] For example, in the case of 4:2:0, a pixel value for one pixel at a predetermined position (selected from top left, top right, bottom left, and bottom right) of four corresponding pixels (one chroma component pixel corresponds to 2×2 luminance component pixels) in luma space. Alternatively, pixel values obtained by filtering a plurality of pixels (e.g., at least two pixels in corresponding 2x2 pixels, or pixels in directions such as left, right, up, down, up left, up right, down left, and down right, by centering on 2×2 pixels) in luma space.

[00298] Summarizing, in order to obtain a pixel value used as a pixel prediction value in the current color space, a parameter representing information can be applied to the pixel value obtained in other color spaces (by multiplying or adding, etc.) about the correlation obtained using the above process.

[00299] In the above example, some color formats, some processes for obtaining pixel values, etc. have been described, however, the present invention is not limited to them, and the same or modified examples may be used in other cases.

[00300] The described content (obtaining prediction values in one color space versus obtaining prediction values in different color spaces) can be applied to a fixed candidate group configuration versus a variable candidate group configuration. For example, if a prediction value cannot be obtained in other color spaces, a prediction value replacement mode may be included in the candidate group.

[00301] With the above example, in the case of the above prediction mode, related information (e.g., support or lack of support information, parameter information, etc.) can be included in picture, sequence, frame, slice, and tile elements.

[00302] Summarizing, the prediction mode (mode A) associated with the method (the above method is a data acquisition method for generating prediction blocks from adjacent areas within the same time and the same space according to the encoding/decoding settings) is used to configure the prediction mode candidate group, or, in addition to the prediction mode, the prediction mode candidate group may comprise a prediction mode (B mode) associated with a method (the above method is a data acquisition method for generating prediction blocks from regions located within the same time and different spaces).

[00303] In the above example, the prediction mode group candidate can be configured using only mode A or mode B only, and the prediction mode group can be configured by combining mode A and mode B. Candidate mode group configuration setting information predictions may be explicitly generated in a correlation manner, or the configuration information of a group of candidate prediction modes may be implicitly predetermined.

[00304] For example, groups of candidate prediction modes may have one configuration regardless of any encoding/decoding settings (picture types in this example) or have one configuration according to some encoding/decoding settings (e.g., mode A, mode B_1<color mode> and mode B_2<color copy mode> are used to configure a group of prediction candidate modes in picture type I, mode A and mode B_1 are used to configure a group of candidate prediction modes in picture type P, and mode A and mode B_2 are used to configure a group of candidate prediction modes in picture type B).

[00305] In the present description, a group of candidate prediction modes for the luma component is depicted in FIG. 9, and it is assumed that the group of candidate prediction modes for the color difference component is configured as the horizontal, vertical, and diagonal modes in FIG. 9 (flat, DC, color mode 1, color mode 2, color mode 3, copy color mode 1, copy color mode 2, adjacent block (left block) mode 1), adjacent block mode 2 (upper block)), but may also have other settings of various groups of candidate prediction modes.

[00306] In the image encoding method according to an embodiment of the present invention, intra prediction may be performed as follows. The intra prediction of the prediction section may comprise a step of configuring the reference pixels, a step of generating a prediction block, a step of determining a prediction mode, and a step of encoding a prediction mode. In addition, the image encoding device may be configured to include a reference pixel configuration section, a prediction block generation section, and a prediction mode coding section to implement the reference pixel configuration step, the prediction block generation step, the prediction mode determination step, and the prediction mode encoding step. Some of the steps in the above process may be omitted, or other steps may be added and their order may be changed to other orders than the above order.

[00307] In addition, in the image decoding method according to the embodiment of the present invention, intra prediction can be performed as follows. Intra prediction of the prediction section may include a prediction mode decoding step, a reference pixel configuration step, and a prediction block generation step. In addition, the image decoding apparatus may be configured to include a prediction mode decoding section for implementing the prediction mode decoding step, a reference pixel configuration section for implementing the reference pixel configuration step, and a prediction block generating section for implementing the prediction block generating step. Some of the steps described above may be omitted, or other steps may be added and their order may be changed to other orders than the above order.

[00308] At the prediction block generation step, intra prediction may be performed on an element of the current block (eg, coding block, prediction block, transform block, etc.) or intra prediction may be performed on an element of a predetermined sub-block. For this purpose, a flag may be used indicating whether the current block is divided into subblocks to perform intra prediction. This flag may be encoded and reported by the encoder. If the flag is equal to the first value, the current block is divided into a plurality of subblocks, otherwise the current block is not divided into a plurality of subblocks. In this case, the division may be an additional division performed after the division based on the above tree structure. Subblocks belonging to the current block share the same intra prediction mode, but different reference pixels may be configured for each subblock. Alternatively, the sub-blocks may use the same intra prediction mode and the same reference pixels. Alternatively, the subblocks may use the same reference pixels, but different intra prediction modes may be used for each subblock.

[00309] The separation can be performed in the vertical or horizontal direction. The split direction may be determined based on a flag reported by the encoder. For example, if the flag has a first value, splitting may be performed in the horizontal direction, otherwise splitting may be performed in the vertical direction. Alternatively, the split direction may be determined based on the size of the current block. For example, if the height of the current block is greater than a predetermined threshold, division may be performed in the horizontal direction, and if the width of the current block is greater than the predetermined threshold, division may be performed in the vertical direction. Here, the threshold may be a fixed value predetermined in the encoding/decoding device, or the threshold may be determined based on block size information (eg, maximum transform block size, maximum coding block size, etc.). The block size information may be reported in at least one sequence, frame, slice, tile, chunk, or row level of the CTU.

[00310] The number of sub-blocks may be variably determined based on size, shape, split depth, intra prediction mode, and so on. current block. For example, if the current block has a size of 4×8 or 8×4, the current block may be divided into two sub-blocks. Alternatively, if the current block has a size greater than or equal to 8×8, the current block may be divided into four sub-blocks.

[00311] In the present description, the encoder will be mainly described. In the case of a decoder, the detailed description regarding the decoder is omitted because it can be obtained from the contents of the encoder in reverse order.

[00312] FIG. 11 is a schematic diagram for explaining the configuration of reference pixels for intra prediction. The size and shape (M×N) of the current block used for prediction can be obtained from the block division section and will be described based on the assumption that intra prediction is maintained within the range of 4×4 to 128×128. In general, intra prediction can be performed using a prediction block element, however, according to the settings of the block division section, intra prediction can be performed in coding block and transform block elements. After confirming the block information, the reference pixel configuration section may configure the reference pixel used to predict the current block. In this case, the reference pixel processing may be performed by a temporary storage device (for example, <Array>, main array or auxiliary array, etc.), the reference pixel may be generated and removed during each intra block prediction process, and the size of the temporary storage may be determined according to the configuration of the reference pixel.

[00313] This example is described based on the assumption that left box, top box, top left box, top right box, and bottom left box are used to predict the current block by centering on the current block. However, it is not limited to this, and a candidate block group having other configurations may be used to predict the current block. For example, a candidate group for adjacent blocks of a reference pixel may be an example according to a raster or Z-scan, and part of the candidate group may be removed according to the scan order, or the candidate group may be configured to include other candidate block groups ( for example, additional configuration of top block, bottom block, bottom right block, etc.).

[00314] Alternatively, blocks (e.g., having corresponding coordinates according to the same coordinate or component compositional ratio in each color space) corresponding to the current block in other color spaces (e.g., if the current block belongs to Cr, other color spaces) can be used to predict the current block. spaces correspond to Y or Cb). In addition, for ease of description, the description is made on the assumption that the predetermined position (left, top, top left, top right, and bottom left) is configured as one block, but at least one block may exist at the corresponding position. That is, at a predetermined position according to the corresponding block, there may be a plurality of sub-blocks of the separation block.

[00315] Summarizing, the adjacent area of the current block can be the position of the reference pixel for intra prediction of the current block, and the position of the reference pixel according to the prediction mode can be considered the area corresponding to the current block of another color space. In addition to the above examples, the position of the reference pixel set according to the prediction mode, method, and so on can be determined. For example, if a prediction block is generated in a manner such as block matching, the position of the reference pixel can be considered to be an area in which encoding/decoding has been completed before the current block of the current image, or an area included in the search range in an area where encoding/decoding is completed (e.g. enabled on the left side, or top side, or top left side and top right side of the current block).

[00316] As shown in FIG. 11, a reference pixel for predicting the current block may be configured with adjacent pixels of the left block, top block, top left block, bottom right block, and bottom left block (Ref_L, Ref_T, Ref_TL, Ref_TR, and Ref_BL in FIG. 11). In this case, the reference pixel generally consists of adjacent block pixels closest to the current block (a in Fig. 11), but may also include other pixels (pixels in line "b" and other outer lines in Fig. 11). That is, at least one of the first pixel line "a" adjacent to the current block, the second pixel line "b" adjacent to the first pixel line, the third pixel line adjacent to the second pixel line, or the fourth pixel line adjacent with a third line of pixels. For example, according to the encoding/decoding settings, the plurality of pixel lines may include everything from the first to the fourth pixel line, or may include only the rest of the pixel lines except for the third pixel line. Alternatively, the plurality of pixel lines may include only the first pixel line and the fourth pixel line.

[00317] The current block may perform intra prediction by selectively referring to any of the plurality of pixel lines. In this case, the selection may be made based on an index (refIdx) reported by the encoder. Alternatively, any of the plurality of pixel lines may be selectively used based on the size, shape, and division type of the current block, whether the intra prediction mode is the non-directional mode, the angle of the intra prediction mode, and so on. For example, if the intra prediction mode is the planar mode or the DC mode, only the first line of pixels may be used. Alternatively, only the first line of pixels may be used if the size (width or height) of the current block is less than or equal to a predetermined threshold. Alternatively, only the first pixel line may be used if the intra prediction mode is greater than a predetermined threshold angle (or less than a predetermined threshold angle). The threshold angle may be an intra prediction mode angle corresponding to mode 2 and mode 66 of the aforementioned group of candidate prediction modes.

[00318] On the other hand, pixels adjacent to the current block can be classified as at least one layer of reference pixels, which can be classified such that the pixels closest to the current block are ref_0 (pixels having a pixel value difference equal to 1, relative to the boundary pixel of the current block; from p(-1, -1) to p(2m-1, -1), from p(-1, 0) to p(-1, 2n-1)), next adjacent pixel (the difference between the border pixel of the current block and the pixel value is 2; from p(-2, -2) to p(2m, -2), from p(-2, -1) to p(-2, 2n) ) is ref_1, and the next adjacent pixel (the difference between the border pixel of the current block and the pixel value is 3; from p(-3, -3) to p(2m+1, -3), from p(-3, -2 ) to p(-3, 2n+1)) is ref_2, etc. That is, a reference pixel can be classified into a plurality of reference pixel layers according to distances of pixels that are adjacent to a boundary pixel of the current block.

[00319] In addition, the reference pixel layers may be arranged differently for each adjacent block. For example, the reference pixel according to layer ref_0 may be used when the block adjacent to the current block and the block adjacent to the top end are used as reference blocks, and the reference pixel according to layer ref_1 may be used when the block adjacent to top end.

[00320] In this case, the reference pixel set usually referred to when performing intra prediction belongs to contiguous blocks adjacent to the current block. Adjacent blocks are located at the bottom left, left, top left, top end, and top right end and are pixels that belong to layer ref_0 (pixels closest to the boundary pixel), and pixels that belong to each other are considered to be these pixels if not otherwise specified. However, only a portion of the pixels belonging to the aforementioned contiguous blocks may be used as the reference pixel set, or pixels belonging to more than two layers may be used as the reference pixel set. In this case, the set, or layer, of reference pixels may be implicitly defined (predetermined in the encoder/decoder) or may be explicitly defined (determination information may be received from the encoder).

[00321] The present description is made on the basis that the maximum number of reference pixel layers supported is 3, but it can also have a value greater than this number, and the number of reference pixel layers and the number of reference pixel sets according to the positions of adjacent blocks, to which may be referred to (or which may be referred to as candidate reference pixel groups) may be specified differently according to size, shape, prediction mode, image type (I/P/B, in this case the image is a frame, slice, tile etc.), color component, etc. block, and related information can be included in sequence, frame, slice, tile, and so on elements.

[00322] The present invention has been described assuming that the subscript (incremented by 1 from 0) is assigned from the reference pixel layer closest to the current block, but the present invention is not limited to this. In addition, the reference pixel configuration related information described below may be generated by the above-described index setting (binarization of assigning a short bit to a sub-index in the case of selecting one of the plurality of reference pixel sets, etc.).

[00323] In addition, if there are more than two reference pixel layers supported, each reference pixel included in more than two reference pixel layers can be used for weighted averaging.

[00324] For example, a prediction block may be generated using a reference pixel obtained from a weighted sum of pixels located in layer ref_0 (nearest pixel layer) and layer ref_1 (next pixel layer) of FIG. 11. In this case, according to the prediction mode (eg, directional prediction mode), the pixels to which the weighted sum is applied in each reference pixel layer may be not only integer bin pixels but also decimal bin pixels. In addition, to obtain a prediction block, weighted values (for example, 7:1, 3:1, 2:1, and 1:1), respectively, are applied to the prediction block obtained by using the reference pixels according to the first reference pixel layer, and a prediction block obtained by using the reference pixels corresponding to the second reference pixel. In this case, if the weighted value is a prediction block according to the reference pixel layer adjacent to the current block, the weighted value may be higher.

[00325] In general, for the above example, it is possible to use the nearest pixel from an adjacent block as a reference pixel, but it is not limited to this. For example, there may be various cases (for example, the case in which ref_0 and ref_1 are selected as reference pixel layers, and ref_0 and ref_1 are used to perform weighted summation and other methods for generating prediction pixel values, etc., i.e. implicit happening).

[00326] In addition, the associated reference pixel configuration information (e.g., selection information for a layer, or set, of reference pixels, etc.) may not include predefined information (e.g., if the reference pixel layer is preset as ref_0) . For example, it may be configured as, but not limited to, ref_1, ref_2, ref_3, and so on.

[00327] Some reference pixel configurations have been described using the above examples, which can be combined with various encoding/decoding information, etc. to define the settings for intra prediction. In this case, the encoding/decoding information includes an image type, a color component, a current block shape and size, a prediction mode (prediction mode type (directional or non-directional)), prediction mode direction (vertical, horizontal, diagonal 1, diagonal 2, etc.). etc.), and so on, and intra prediction settings (in this example, reference pixel configuration settings) can be determined according to the encoding/decoding information of the adjacent block and the combination of the encoding/decoding information of the current block and the adjacent block.

[00328] FIG. 12 is a circuit diagram for explaining the range of a reference pixel for intra prediction. In more detail, the reference pixel range is determined according to the block size and shape, the prediction mode configuration (in this example, prediction mode angle information), and so on. The position indicated by the arrow in Fig. 12 is a pixel used for prediction.

[00329] With reference to FIG. 12, pixels A, A', B, B', and C refer to pixels at the lower right end of the 8x2, 2x8, 8x4, 4x8, and 8x8 blocks. To perform pixel prediction, the range of the reference pixel of each block may be determined by the pixels AT, AL, BT, BL, CT, and CL used in the top and left blocks.

[00330] For example, if the reference pixels are pixels A and A' (rectangular blocks), they are located within the range from p(0, -1) to p(9, -1), from p(-1, 0) to p(-1, 9), p(-1, -1), if the reference pixels are pixels B and B' (rectangular blocks), they are located within the range from p(0, -1) to p(11, - 1), from p(-1, 0) to p(-1, 11), p(-1, -1), and if the reference pixel is pixel C (square block), it is located within the range from p( 0, -1) to p(15, -1), from p(-1, 0) to p(-1, 15), p(-1, -1).

[00331] Range information (e.g., P(-1, -1), P(M+N-1, -1), P(-1, N+M-1), etc.) of the reference pixel, obtained by the above process can be used for an intra prediction process (eg, a reference pixel filtering process, a reference pixel generating process, etc.). In addition, support for reference pixels is not limited to the cases described above, and various other cases are also possible.

[00332] The reference pixel configuration section for intra prediction may include a reference pixel generation section, a reference pixel interpolation section, a reference pixel filtering section, and so on. and may have all or part of the configurations described above.

[00333] The reference pixel configuration section is configured to check the availability of the reference pixels in order to classify available reference pixels and unavailable reference pixels. For example, if a block at a predetermined position (or candidate block of a reference pixel) can be used, then the corresponding block can be used as a reference pixel, and if it is not available, that block cannot be used as a reference pixel.

[00334] If at least one of the following conditions is satisfied, it is determined that the reference pixel is unavailable. For example, a reference pixel is determined to be unavailable if one of the conditions is satisfied that the reference pixel is outside the frame boundary, the reference pixel does not belong to the same partition block as the current block (e.g., slice, tile, etc.), encoding/decoding is not completed and the use of the reference pixel is limited according to the encoding/decoding setting. That is, if none of the above conditions is satisfied, it is determined that the reference pixel is available.

[00335] In addition, the use of reference pixels may be limited by encoding/decoding settings. For example, the use of reference pixels may be constrained in accordance with the performance of constrained intra prediction (eg, constrained_intra_pred_flag). Limited intra prediction may be performed when error-tolerant coding/decoding needs to be performed with respect to external factors such as the communication environment if an attempt is made to prohibit the use of blocks recovered by referring to other images as reference pixels.

[00336] If constrained intra prediction is disabled (e.g., in type I frame or type P or B frame constrained_intra_pred_flag=0), all candidate reference pixel blocks are available, and if enabled (e.g., type P or B frame constrained_intra_pred_flag=1), the use of the reference pixel from the corresponding block may be determined according to the coding mode (intra-prediction or inter-picture prediction) of the reference pixel candidate block. That is, if the coding mode of a block is intra prediction, the block can be used regardless of whether limited intra prediction is enabled. In the case of inter-picture prediction, the availability (deactivated) or unavailability (activated) of a block is determined depending on the activation of restricted intra prediction.

[00337] In addition, limited intra prediction may be applied in accordance with the coding mode of the reconstructed block corresponding to the current block in a different color space. For example, the current block belongs to any Cb and Cr color difference components, and the availability of restricted intra prediction is determined according to the coding mode of the block in which encoding/decoding of the Y component of the luma corresponding to the current block is completed. The above example may be an example of a case in which a reconstructed block of a different color space is used as a reference pixel. Further, it may be an example of a case in which an encoding mode is determined independently according to a color space.

[00338] In this case, if the candidate block of the reference pixel is encoded/decoded by some prediction methods (for example, prediction by block matching or pattern matching in the current image), the use of this reference pixel can be determined according to the encoding/decoding settings .

[00339] As an example, if encoding/decoding is performed using the specified prediction method, and if the encoding mode is set to intra prediction, the corresponding block may be determined to be available. Alternatively, even if intra prediction is used, special cases are allowed in which a block is not available.

[00340] For example, if encoding/decoding is performed using the specified prediction method, and if the encoding mode is set to prediction performed between pictures, the corresponding block may be determined to be unavailable. Alternatively, even if inter-picture prediction is used, special cases are allowed in which a block is available.

[00341] That is, according to the encoding/decoding settings, it can be determined whether to make an exception for the use determination case according to the encoding mode.

[00342] Restricted intra prediction may be a setting applied to certain types of images (eg, slice/tile types P or B, etc.).

[00343] According to the availability of the reference pixel, the candidate reference pixel blocks can be classified into a case in which all candidate blocks of the reference pixels can be used, a case in which some of the reference pixels can be used, and a case in which all of the reference pixels not available. Except in the case where all candidate blocks of reference pixels may be used, in all other cases, reference pixels not available at the position of the candidate block may be filled or generated.

[00344] If a candidate reference pixel block is available, a pixel at a predetermined position of that block (in this example, the pixel is assumed to be adjacent to the current block) can be stored in the current block's reference pixel memory. In this case, the pixel data from the corresponding block position can be stored in the reference pixel memory by processes such as direct copying or filtering of the reference pixels.

[00345] If a candidate reference pixel block is not available, the pixels obtained by the reference pixel generation process may be stored in the current block's reference pixel memory.

[00346] Briefly, if the reference pixel candidate block is in the available state, the reference pixel may be configured, and if the reference pixel candidate block is in the unavailable state, the reference pixel may be generated.

[00347] The following are examples of using different methods for filling reference pixels at the positions of inaccessible blocks.

[00348] For example, an arbitrary pixel value may be used to generate a reference pixel, and there may be a pixel value belonging to a range of pixel values (e.g., a value obtained from a pixel value adjustment process based on bit depth or a pixel value adjustment process based on information about the range of image pixel values, such as the minimum value, maximum value, and median value). In more detail, this may be an example applicable when all candidate reference pixel blocks are not available.

[00349] Alternatively, the reference pixels may be generated upon completion of image encoding/decoding. More specifically, the reference pixels may be generated from at least one available block adjacent to the inaccessible block. In this case, at least one of extrapolation, interpolation, copying, or other methods may be used, and the direction of generation (or copying and extrapolation) of the reference pixels may be clockwise or counterclockwise, and may be determined according to encode/decode settings. For example, the generation direction of the reference pixels in the block may be a predetermined direction or a direction adaptively determined according to the position of the inaccessible block. Alternatively, for the area corresponding to the current block in other color spaces, the same method can be used as in the above example. The difference is that if the filling process of the adjacent reference pixels of the current block is performed in the current color space, the filling process of the M×N block corresponding to the current m×n block is performed in other color spaces. Therefore, various other methods can be used to generate the corresponding areas, including the above method (for example, extrapolation in the vertical, horizontal and diagonal directions of the surrounding pixels, planar interpolation, averaging, etc., in this case, the fill direction refers to the direction from the surrounding pixels of the block corresponding to the current block inside the block). This example may be a case in which the prediction mode used to generate a prediction block from a different color space is included in the candidate group but not excluded from the candidate group.

[00350] In addition, after the reference pixel configuration is completed, reference pixels in decimal units can be generated by the reference pixel availability confirmation process by linear interpolation of the reference pixels. Alternatively, the reference pixel interpolation process may be performed after the reference pixel filtering process is performed. Alternatively, only the filtering process for the configured reference pixels may be performed. Briefly, it may be performed before the prediction block generation process.

[00351] In this case, in landscape mode, vertical mode, some diagonal modes (e.g., diagonal bottom right, diagonal bottom left, and diagonal top right) and non-directional modes, color mode, color copy mode, etc. the interpolation process is not performed, and in the case of other modes (other diagonal modes), interpolation can be performed.

[00352] The interpolation accuracy may be determined according to a supported prediction mode candidate group (or total number of prediction modes), prediction mode configuration (e.g., prediction mode direction angle and interval between prediction modes), and so on.

[00353] A predefined filter (e.g., 2-tap linear interpolation filter) may be used to interpolate the reference pixels in the decimal element, and one of a plurality of candidate filter groups may be used (e.g., 4-tap cube filter, 4-tap Gaussian filter, 6-tap Wiener filter, 8-tap Kalman filter).

[00354] If one of the plurality of candidate filter groups is used, the filter selection information may be explicitly generated or determined implicitly and may be determined in accordance with encoding/decoding settings (e.g., interpolation precision, block size and shape, prediction mode etc.).

[00355] For example, an interpolation filter to use may be determined according to a range of block sizes, an interpolation filter to use may be determined according to interpolation accuracy, and an interpolation filter may be determined according to prediction mode characteristics (e.g., directionality information, etc.). d.).

[00356] In more detail, according to the block size range, predefined interpolation filter a can be used in some ranges A, predefined interpolation filter b can be used in some ranges B, interpolation filter c of the plurality of interpolation filters C can be used in some ranges C, an interpolation filter d of the plurality of interpolation filters D may be used in some ranges of D, in some ranges a predetermined interpolation filter may be used, and in some ranges one of the plurality of interpolation filters may be used. In this case, if the use of one interpolation filter is implicit, the use of one interpolation filter in the plurality of interpolation filters is explicit, and the block size for separating the range of block sizes may be M×N (in this example, M and N are 4, 8, 16, 32, 64, 128, etc., i.e. M and N can be the minimum or maximum value of the size range of each block).

[00357] Interpolation related information may be included in video, sequence, frame, slice, tile, block, and so on elements. The interpolation process may be a process performed in a reference pixel configuration section, or may be a process performed in a prediction block generation section.

[00358] In addition, after configuring the reference pixel, the reference pixel may be filtered to mitigate the degradation remaining after the encoding/decoding process in order to improve the prediction accuracy. In this case, the filter used may be a low pass filter. The performance of filtering may be determined according to encoding/decoding settings, if filtering is applied, fixed or adaptive filtering may be applied, and encoding/decoding settings may be set according to block size, shape, prediction mode, etc.

[00359] Fixed filtering refers to a case where a predetermined filter is applied to the reference pixel filter section, and adaptive filtering refers to a case where one of a plurality of filters is applied to the reference pixel filter section. In this case, in adaptive filtering, one of the plurality of filters may be implicitly determined according to encoding/decoding settings, or the selection information may be explicitly generated, and a 3-tap filter may be included as a candidate filter group (e.g., , [1, 2, 1]/4) or a 5-tap filter (for example, [2, 3, 6, 3, 2]).

[00360] As an example, filtering may not be applied in some settings (block range A).

[00361] As an example, filtering may not be applied at some settings (B block range, some C mode), and at some settings (B block range, some D mode), filtering with a preset filter (3-tap filter ).

[00362] For example, filtering may not be applied in some settings (E block range, some F mode), in some settings (E block range, some G mode) filtering with a pre-filter (3-tap filter) may be applied, in some settings (E block range, some H mode) filtering with a pre-filter (5-tap filter) can be applied, and in some settings (E block range, some I mode), one of a variety of filters can be applied for filtering.

[00363] As an example, in some settings (range J blocks, some mode K), filtering with a predefined filter (5-tap filter) may be applied, and filtering may additionally be applied with a predefined filter (3-tap filter) . That is, a plurality of filtering processes can be performed. More specifically, filtering can also be applied based on previous filtering results.

[00364] In the above example, the block size for dividing the range of block sizes may be M×N (in this example, M and N are 4, 8, 16, 32, 64, 128, etc., i.e., M and N can be the minimum or maximum value of each block size range). In addition, the prediction mode can be roughly classified into a directional mode, an omnidirectional mode, a color mode, a color copy mode, etc., and more specifically, mode 1 adjacent to the horizontal or vertical mode / diagonal mode (45 degree interval) / horizontal or vertical mode, mode 2 adjacent to the horizontal or vertical mode (the interval between modes is slightly longer than the above interval), etc. That is, as described above, the filtering performance and the filtering type are determined according to the classification mode.

[00365] In addition, the above examples show a case in which adaptive filtering is applied in accordance with a plurality of factors such as a block range and a prediction mode. However, multiple factors are not always required, and there may also be an example where adaptive filtering is performed in accordance with at least one factor. In addition, there may be a plurality of transformed examples, without being limited to the examples above, and information related to filtering reference pixels may be included in video, sequence, frame, slice, tile, and block elements.

[00366] Filtering may be selectively performed based on a predetermined flag. The flag may indicate performing filtering on the reference pixel to perform intra prediction. The flag may be encoded and reported by the encoder. Alternatively, the flag may be obtained from the device for decoding based on the encoding parameter of the current block. The encoding parameter may include at least one of the position/area of the reference pixel, block size, component type, whether intra prediction has been applied in subblock elements, or intra prediction mode.

[00367] For example, if the reference pixel of the current block is the first line of pixels adjacent to the current block, then the reference pixel may be filtered, otherwise the reference pixel may not be filtered. Alternatively, the reference pixel may be filtered if the number of pixels belonging to the current block is greater than the number of predetermined thresholds, otherwise the reference pixel may not be filtered. The threshold value is a value predetermined for the encoder/decoder, and may be an integer of 16, 32, 64 or more. Alternatively, the reference pixel may be filtered if the current block is greater than a predetermined threshold, otherwise the reference pixel may not be filtered. The threshold value may be represented as M×N, and it is a value predetermined for the encoder/decoder, and M and N may be integers of 8, 16, 32 or more. The threshold number or threshold size may be defined to determine the filtering of the reference pixel by one threshold number or threshold size, or a combination thereof. Alternatively, the reference pixel may be filtered if the current block is a luminance component, otherwise the reference pixel may not be filtered. Alternatively, if the current block does not perform the above-described intra prediction in subblock elements (ie, the current block is not split into multiple subblocks), the reference pixel is filtered, otherwise the reference pixel may not be filtered. Alternatively, the reference pixel may be filtered if the intra prediction mode of the current block is non-directional mode or a predetermined directional mode, otherwise the reference pixel may not be filtered. Here, the non-directional mode may be a planar mode or a DC mode. However, in the non-directional mode DC mode, it may be limited to no filtering of the reference pixel. The directional mode may refer to an intra prediction mode with reference to an integer pixel. For example, the steered mode may include at least one of the intra prediction modes corresponding to modes -14, -12, -10, -6, 2, 18, 34, 50, 66, 72, 78, and 80 shown in FIG. 9. However, the directional mode may be limited to exclude the horizontal mode and the vertical mode, respectively, for modes 18 and 50.

[00368] When filtering the reference pixel in accordance with the flag, filtering may be performed based on a filter predefined in the device for encoding/decoding. The number of filter taps can be 1, 2, 3, 4, 5 or more. The number of filter taps can be defined as variable according to the position of the reference pixel. For example, a 1-tap filter may be applied to the reference pixels corresponding to at least one side of the bottom-most, top-most, left-most, and right-most ends of the line of pixels, and a 3-tap filter may be applied to the remaining reference pixels. In addition, the strength of the filter may be determined to be variable according to the position of the reference pixel. For example, a filter strength s1 may be applied to reference pixels corresponding to at least one side of the bottommost, topmost, leftmost, and rightmost ends of the line of pixels, and a filter strength s2 (s1<s2) may be applied to the remaining reference pixels. The strength of the filter may be reported at the encoding device, or may be determined based on the encoding parameters described above. When applying an n-tap filter to a reference pixel, the filter may be applied to the current reference pixel and (n-1) surrounding reference pixels. Surrounding reference pixels may indicate pixels located at at least one of the top end, bottom end, left side, or right side of the current reference pixel. The surrounding reference pixels may belong to the same pixel line as the current reference pixel, and a portion of the surrounding reference pixels may belong to a different pixel line than the current reference pixel.

[00369] For example, if the current reference pixel is located on the left side of the current block, the surrounding reference pixels may be pixels adjacent in at least one direction of the top end or bottom end of the current reference pixel. Alternatively, if the current reference pixel is located at the top end of the current block, the surrounding reference pixels may be pixels adjacent on at least one of the left and right sides of the current reference pixel. Alternatively, if the current reference pixel is located at the upper left end of the current block, the surrounding reference pixels may be pixels adjacent at least one of the lower end or right side of the current reference pixel. The ratio between the filter coefficients can be [1:2:1], [1:3:1], or [1:4:1].

[00370] The prediction block generating section is configured to generate a prediction block in accordance with at least one prediction mode and use a reference pixel based on the intra prediction mode. In this case, the reference pixel may be used in methods such as extrapolation (directional mode) according to the prediction mode, and may be used in methods (non-directional mode) such as interpolation or averaging (DC) or copying. At the same time, as described above, the current block may use filtered reference pixels or may use unfiltered reference pixels.

[00371] FIG. 13 is a diagram showing a block adjacent to a current block for generating a prediction block.

[00372] For example, in directional mode, in a mode between horizontal mode and some diagonal modes (diagonally top right, including non-horizontal diagonal lines), reference pixels of the lower left block + left block (Ref_BL and Ref_L in Fig. 13) can be used, in horizontal mode, the left block reference pixel can be used, in the mode between horizontal and vertical modes, the left block + top left block + top block reference pixels (Ref_L, Ref_TL, and Ref_T in Fig. 13) can be used, in vertical mode, the top block reference pixel can be used block (Ref_L in Fig. 13), and in a mode between vertical mode and some diagonal modes (diagonally bottom left, including non-vertical diagonals), top block + top right block reference pixels (Ref_T and Ref_TR in Fig. 13) can be used . Alternatively, in non-directional mode, the left block and top block reference pixels (Ref_L and Ref_T in FIG. 13) or the bottom left block, left block, top left block, top block, and top right block reference pixels (Ref_BL, Ref_L, Ref_TL, Ref_T, and Ref_TR in Fig. 13). Alternatively, in the case of using the color space correlation mode (color copying mode), reconstructed blocks of other color spaces (not shown in Fig. 12, but called Ref_Col in the present description) are used as reference pixels; this means that the placed block reference is in different spaces At the same time).

[00373] The reference pixels used for intra prediction may be divided into a plurality of groups. For example, reference pixels used for intra prediction may be allocated to a first reference pixel and a second reference pixel, and the first reference pixel may be a pixel directly used to generate a prediction value of the current block, and the second reference pixel may be a pixel indirectly used to generate the prediction value of the current block. Alternatively, the first reference pixel may be a pixel used to generate prediction values of all pixels of the current block, and the second reference pixel may be a pixel used to generate prediction values of some pixels of the current block. Alternatively, the first reference pixel may be the pixel used to generate the main prediction value of the current block, and the second reference pixel may be the pixel used to generate the auxiliary prediction value of the current block. Alternatively, the first reference pixel (certainly) may be a pixel in an area at the starting point of the prediction direction of the current block, the second reference pixel may be a pixel not located at the starting point of the (desired) prediction direction of the current block.

[00374] As described in the above example, although the reference pixels can be distinguished using different definitions, there may be cases in which some definitions are not distinguished according to the prediction mode. That is, it should be noted that the definition used to distinguish the reference pixels may differ according to the prediction mode.

[00375] The reference pixel described with the above example may be the first reference pixel, and the second reference pixel may further participate in generating the prediction block. In the mode between some diagonal modes (diagonal top right, including diagonal lines other than horizontal), the top left block + top block + top right block reference pixels (Ref_TL, Ref_T and Ref_TR in Fig. 13) can be used, in horizontal mode can be used top left block + top block + top right block reference pixels (Ref_TL, Ref_T and Ref_TR in Fig. 13), in vertical mode, top left block + top block + bottom left block reference pixels (Ref_TL, Ref_T and Ref_TR in Fig. 13) can be used. 13), in vertical mode, the reference pixels of the upper left block + left block + lower left block (Ref_TL, Ref_L and Ref_BL in Fig. 13) can be used, and in a mode between vertical mode and some diagonal modes (diagonal bottom left, including diagonals, other than vertical) reference pixels of upper left block + left block + lower left block (Ref_TL, Ref_L and Ref_BL in FIG. 13). In more detail, the reference pixel may be used as the second reference pixel. In addition, the first reference pixel or the first and second reference pixels may be used to generate a prediction block in the non-directional mode and the color copy mode.

[00376] In addition, the second reference pixel can be considered to include not only pixels that have been encoded/decoded, but also pixels in the current block (prediction pixels in this example). That is, the main prediction value may be a pixel used to generate the auxiliary prediction value. The present invention will mainly describe an example in which the second reference pixel is considered to be a pixel that has been encoded/decoded, but the present invention is not limited to this, and there may also be a converted example that uses pixels that have not been encoded/decoded ( in this example, prediction pixels).

[00377] To compensate for the shortcomings of the existing prediction mode, a prediction block may be generated or corrected using a plurality of reference pixels.

[00378] For example, the directional mode is a mode used to predict the directionality of a corresponding block using some reference pixels (the first reference pixel), however, it may not accurately reflect changes in the block, which may cause a reduction in prediction accuracy. In this case, if a prediction block is generated or corrected using an additional reference pixel (second reference pixel), prediction accuracy can be improved.

[00379] To this end, the following example will describe a case in which a prediction block is generated using different reference pixels as in the above example, but the present invention is not limited to the case of the above example, and even if terms such as first and second reference pixels are not used, these terms can be derived and understood from the above definitions.

[00380] The setting for using an additional reference pixel to generate a prediction block may be explicitly specified or implicitly specified. In the explicit case, an element can contain a video, sequence, frame, slice, tile, and so on. In the following example, the case of implicit processing will be described, however, the present invention is not limited thereto, and there may be other modified (explicit or combined) examples.

[00381] The prediction block may be generated in various ways according to the prediction mode. More specifically, a prediction block may be determined according to the position of a reference pixel used in the prediction mode. In addition, the prediction method may be determined according to the position of the pixel in the block.

[00382] The case of the horizontal mode will be described below.

[00383] For example, if the left block (Ref_L in FIG. 13) is used as the reference pixel, a prediction block is generated in the horizontal direction using (eg, extrapolation) the nearest pixel (1300 in FIG. 13).

[00384] In addition, the prediction block may be generated (or corrected; the generation may include the final prediction value; the correction may not include all pixels) by using reference pixels adjacent to the current block and corresponding to the horizontal direction. In more detail, the closest pixel of the corresponding block (1310 in FIG. 13; other pixels 1320 and 1330 can also be taken into account) and information about the degree of change, or gradient, of the pixel, etc. can be used to correct the prediction value. (for example, information about the degree of change, or gradient, of the pixel values reflected in R0-T0, T0-TL, T2-TL, T2-T0, T2-T1, etc.) can be fed back to the correction process .

[00385] In this case, the pixels to be corrected may be all pixels in the current block, or may be limited to some pixels (for example, they may be defined in elements of a single pixel that do not have a particular shape or exist in a non-permanent position, as in the following example, they are defined in pixel elements of a line having a specific shape; for ease of description, the description in the following example is made assuming a pixel element with a line as an element). If the correction pixel is limited applied to some pixels, the element may be defined as at least one line corresponding to the direction of the prediction mode. For example, pixels corresponding to a-d may be included in the correction target, and furthermore, pixels corresponding to e-h may also be included in the correction target. In addition, correction information obtained from adjacent pixels of a block may be identically applied regardless of line position, or correction information may be applied differently in line elements, and as the distance from adjacent pixels increases, the applicability of correction information may decrease. (for example, according to the distance such as L1-TL and L0-TL, a larger separation value can be set).

[00386] In this case, the pixels included in the correction object may have only one setting in one image, or may be determined adaptively in accordance with different encoding/decoding elements.

[00387] As an example, in the case of adaptive determination, the pixels to be corrected may be determined according to the block size. In a block of less than 8x8, no line is corrected, only one pixel can be corrected in a block of more than 8x8 and less than 32x32, and in a block of more than 32x32, two lines of pixels can be corrected. The definition of the range of block sizes can be obtained from the previous description of the present invention.

[00388] Alternatively, the pixels to be corrected may be determined according to the shape of the block (eg, square or rectangular; in particular, a rectangle elongated in the horizontal direction or a rectangle elongated in the vertical direction). For example, in the case of an 8×4 correction block, two pixel lines (ah in FIG. 13) may be subjected, and in the case of a 4×8 correction block, one pixel line (ad in FIG. 13) may be subjected. If the shape of the horizontally stretched box is determined to be in horizontal mode in the 8×4 case, the directionality of the current box may be more dependent on the top box, and if the shape of the vertically stretched box is determined to be in horizontal mode in the case of 4 ×8, the directionality of the current block may depend on the top block to a lesser extent. In addition, an arrangement opposite to that described above is also possible.

[00389] Alternatively, the pixels to be corrected may be determined according to the prediction mode. In the horizontal or vertical mode, the correction targets may be a pixel lines, while in the other modes, the correction targets may be b pixel lines. As described above, in the case of some modes (such as DC non-directional mode, color copy mode, etc.), the pixels to be corrected are not rectangular, but the range is specified in pixel units (such as a-d, e, i, m) . This will be detailed in the following example.

[00390] In addition to the above description, adaptive settings may be applied according to additional coding/decoding elements. The above description focuses on the horizontal mode limitation. However, it does not only apply to the examples above, but the same or similar settings apply to other modes as well. In addition, the same case as in the above example is implemented according to the combination of a plurality of elements rather than a single encoding/decoding element.

[00391] In the case of the vertical mode, since it can be obtained by applying different directions to the horizontal mode prediction mode, its detailed description is omitted. In addition, the following example omits repetitions of content associated with landscape mode.

[00392] The case of the diagonal mode (diagonal upward to the right) will be described below.

[00393] For example, if the left block and the lower left block (the first reference pixels, or main reference pixels, Ref_L and Ref_BL in Fig. 13) are used as reference pixels, a prediction block is generated in the diagonal direction using the nearest pixels (1300 and 1340 in Fig. 13) of the corresponding block (for example, by extrapolation, etc.).

[00394] In addition, a prediction block may be generated (or corrected) using reference pixels (second reference pixels, or auxiliary reference pixels; Ref_T and Ref_TR in FIG. 13) adjacent to the current block located at diagonally opposite positions. In more detail, the prediction value may be corrected using the nearest pixels of the respective block (1310 and 1330 in FIG. 13, and additionally may be taken into account 1320) and may be fed back to the correction process using a weighted average minor reference pixel and a main reference pixel (e.g. , the weighted value can be obtained based on at least one of the prediction pixel, the main reference pixel, or the x-axis or y-axis distance difference between the prediction pixel and the sub-reference pixel; an example of a weighted value applicable to the main reference pixel and the sub-reference pixel, may include 15:1 - 8:8; if there are more than two auxiliary reference pixels, an example of a weighted value is such that the weighted values in the auxiliary reference pixels, for example, 14:1:1, 12:2:2, 10:3:3 , 8:4:4, are the same, or the weighted values between the auxiliary reference pixels, for example, 12:3:1, 10:4:2, 8:6:2, may differ; in this case, by centering on the corresponding prediction pixel, different weight values are determined according to the proximity of the tilt information in the current direction of the prediction mode, that is, the gradient of the current prediction mode, which is closer to the gradient of the corresponding prediction pixel and each auxiliary reference pixel, is determined).

[00395] In this case, the filtering has one setting in one picture, or the filtering is determined adaptively according to different coding/decoding elements.

[00396] As an example, in the case of adaptive determination, the pixels to be filtered (eg, the number of pixels, etc.) may be determined according to the positions of the pixels to be corrected. If the prediction mode is the diagonal mode (in this example, mode 2) and the pixel to be corrected is c, prediction is performed using L3 (in this example, the first reference pixel), and correction is performed using T3 (in this example, example, the second reference pixel). That is, this case may be a case in which one first reference pixel and one second reference pixel are used to predict one pixel.

[00397] Alternatively, if the prediction mode is diagonal mode (mode 3 in this example) and the pixel to be corrected is b, prediction is performed using L1* (or L2*, in this example, the first reference pixel) obtained by performing interpolation on the pixels of the decimal elements between L1 and L2, and the correction is performed using T2 (in this example, the second reference pixel), or the correction may be performed using T3. Alternatively, the correction may be performed using T2 and T3, or the correction may be performed using T2* (or T3*) which are obtained by performing interpolation on the bin pixels between T2 and T3, which are obtained based on the directivity of the prediction mode. That is, one first reference pixel can be used to predict one pixel (in this example, assuming L1*, it can be considered as two pixels if L1 and L2 are considered directly used pixels; or, according to the filter used to perform interpolation with respect to L1*, it can be considered as two or more pixels) and two second reference pixels (in this example, assuming T2 and T3, L1* can be considered as one pixel).

[00398] Briefly, at least one first reference pixel and at least one second reference pixel can be used to predict one pixel, which can be determined according to the prediction mode and prediction pixel position.

[00399] If the correction pixels are limited applied to some pixels, the corrected pixels may be determined in the elements of at least one horizontal line or vertical line in accordance with the direction of the intra prediction mode. For example, pixels corresponding to a, e, i, and m or pixels corresponding to a-d may be included in the correction target, and furthermore, pixels corresponding to b, f, j, and n, or pixels corresponding to e-h. In some diagonal up right mode cases, pixels in horizontal line elements may be corrected, and in some diagonal down left mode cases, pixels in vertical line elements may be corrected, which is not limited to this.

[00400] In addition to the above description, adaptive settings may be applied according to additional encoding/decoding elements. The description above is focused on limiting the mode diagonally up to the right. However, it is applicable to the above examples not only in a limited sense, and similar or similar settings are applicable to other modes. In addition, the above example is implemented according to a plurality of elements rather than a combination of a single encoding/decoding element.

[00401] In the case of the diagonal down left mode, since it can only be obtained by applying different directions to the diagonal up right prediction method, a detailed description thereof is omitted.

[00402] The case of the diagonal upward left mode will be described below.

[00403] For example, if the left block, the upper left block, and the upper block (the first reference pixels, or main reference pixels, Ref_L, RefT_L, and Ref_T in FIG. 13) are used as reference pixels, a prediction block is generated in the diagonal direction using the nearest pixels (1300, 1310, and 1320 in FIG. 13) of the corresponding block (eg, by extrapolation, etc.).

[00404] In addition, a prediction block may be generated (or corrected) using reference pixels (second reference pixels, or auxiliary reference pixels, Ref_L, Ref_TL and Ref_T in Fig. 13, at the same position as the main reference pixel) adjacent with the current block, located in positions coinciding with the diagonal lines. In more detail, pixels other than the nearest pixels of the corresponding block (pixels located on the left side 1300, pixels located on the left side, top side and top left side 1320, pixels located on the top side 1310, and etc. in Fig. 13), and they can be fed back to the correction process using the weighted average auxiliary reference pixel and the main reference pixel (for example, an example of the ratio of weighted values applied to the main reference pixel and the auxiliary reference pixel, may include 7:1-4:4, etc., if there are more than two auxiliary reference pixels, an example of a weighted value may be a case in which the weighted values in the auxiliary reference pixels are the same, for example, 14:1:1, 12:2:2, 10:3:3 and 8:4:4, or the weighted values between the auxiliary reference pixels are different, such as 12:3:1, 10:4:2 and 8:6:2; in this case, the weighted values applied differently may be determined according to whether they are adjacent to the main reference pixel) or linear extrapolation, etc.

[00405] If the correction pixels are limited applied to some pixels, the corrected pixels may be determined in horizontal line or vertical line elements adjacent to the reference pixel used in the prediction mode. In this case, horizontal and vertical lines can be taken into account at the same time, and overlap is allowed. For example, pixels corresponding to a-d and pixels (overlapping with a) corresponding to a, e, i, and m may be included in the correction target. In addition, pixels corresponding to e-h and pixels corresponding to b, f, j, and n (a, b, e, and f overlap) can be included in the correction target.

[00406] The case of non-directional mode (DC) will be described below.

[00407] For example, if at least one of the left block, top block, top left block, top right block, or bottom left block is used as a reference pixel, (e.g., average, etc.) can be used to generate a prediction block. the nearest block pixel (in this example, pixels 1300 and 1310 in FIG. 13 are assumed).

[00408] Alternatively, an adjacent pixel of the reference pixel (the second reference pixel, or minor reference pixel, of the reference pixel) can be used to generate (or correct) the prediction block; the positions of Ref_L and Ref_T in Fig. 13 in this example are the same as the main reference pixel , or are the complement of the main reference pixel, the pixel located in the next adjacent position is also included; the case is similar to the mode diagonally up to the right). In more detail, pixels at positions the same or similar to the position of the main reference pixel of the corresponding block may be used to correct the prediction value, and may be fed back to the correction process using the weighted average sub-reference pixel and the main reference pixel (for example, an example of the ratio of the weighted values applicable to the main reference pixel and the sub-reference pixel may include 15:1-8:8, etc., if there are more than two sub-reference pixels, an example of a weighted value may be a case in which the weighted values in the sub-reference pixels are the same, such as 14:1:1, 12:2:2, 10:3:3, and 8:4:4, or the weighted values between the auxiliary reference pixels are different, such as 12:3:1, 10: 4:2 and 8:6:2).

[00409] In this case, the filtering has one setting for one picture, or the filtering is determined adaptively according to different coding/decoding elements.

[00410] As an example, in the case of adaptive determination, the filter may be determined according to the block size. For pixels located at the top left, top end, and left end of the current block (in this example, it is assumed that the pixel located at the top left end filters out pixels to the left of and above that pixel, and the pixel located at the top end filters out pixels at top side of that pixel, and the pixel at the left end filters out pixels to the left of that pixel), some filtering settings can be used in a block smaller than 16x16 (in this example, filtering is applied according to the weight ratio of 8:4:4 and 12:4 ), and some filtering settings can be used in a block larger than 16x16 (in this example, filtering is applied according to the weight ratio of 10:3:3 and 14:2).

[00411] Alternatively, a filter may be defined according to the shape of the block. For example, for a 16x8 block, some filter settings can be used for pixels at the top end of the current block (in this example, it is assumed that pixels on the top left, top, and top right sides of the pixel are filtered, i.e. they can be considered as an example , in which the filtered pixels also change; the filtering weight ratio is 10:2:2:2), and some filter settings can be used for pixels at the left end of the current block (in this example, it is assumed that the pixel at the left end of this pixel is being filtered, and the filtration weight ratio is 12:4). This example assumes that it can be applied to a plurality of pixels on the top side of a block with the block shape elongated in the horizontal direction. In addition, the opposite arrangement can also be realized.

[00412] If the correction is limited to some pixels, the corrected pixels may be determined in horizontal line or vertical line elements adjacent to the reference pixel used in the prediction mode. In this case, horizontal and vertical lines can be taken into account at the same time, and overlap is allowed. For example, pixels corresponding to a-d and pixels (overlapping with a) corresponding to a, e, i, etc. may be included in the correction target. Also, pixels corresponding to e-h and pixels corresponding to b may be included in the correction target. f, j and n (a, b, e and f overlap).

[00413] In addition to the above description, adaptive settings may be implemented in accordance with additional coding/decoding elements. The above description focuses on the non-directional mode limitation. However, it is applicable to the above examples not only in a limited sense, and similar or similar settings are applicable to other modes. In addition, the same example as described above is implemented according to a combination of a plurality of elements rather than a single encoding/decoding element.

[00414] The case of the color copy mode will be described below.

[00415] For the color copy mode, the prediction block is generated in a different way than the existing prediction mode, but the reference pixels can be used in the same or similar way to generation (or correction) of the prediction block. Since the content regarding obtaining a prediction block can be obtained using the above examples and the following examples, the content regarding obtaining a prediction block is omitted.

[00416] For example, a prediction block may be generated by using (eg, copying) as a reference pixel (first reference pixel, or master reference pixel) a block corresponding to the current block in a different color space.

[00417] Alternatively, a prediction block may be generated (or corrected) using reference pixels (second reference pixels, or auxiliary reference pixels; Ref_L, Ref_T, Ref_TL, Ref_TR, and Ref_BL in FIG. 13) of a block adjacent to the current block. In more detail, the nearest pixels (assumed to be pixels 1300 and 1310 in FIG. 13 in this example) of the respective block may be used to correct the prediction value and may be fed back into the correction process using the weighted average auxiliary reference pixel and the main reference pixel (for example, an example of the ratio of weighted values applicable to the main reference pixel and the auxiliary reference pixel may include 15:1 8:8, etc., if there are more than two auxiliary reference pixels, the example of the weighted value may be the case in where the weighted values in the auxiliary reference pixels are the same, such as 14:1:1, 12:2:2, 10:3:3, and 8:4:4, or the weighted values between the auxiliary reference pixels are different, such as 12: 3:1, 10:4:2 and 8:6:2).

[00418] Alternatively, to generate (or correct) a prediction block, pixels of a block adjacent to blocks obtained in other color spaces (second reference pixels, or auxiliary reference pixels; if it is assumed that the frame in Fig. 13 is a block corresponding to current block in other color spaces, the pixels are Ref_L, Ref_T, Ref_TL, Ref_TB, Ref_BL and Ref_R, Ref_BR, Ref_B, not shown in Fig. 13). Pixels to be corrected and surrounding pixels (for example, the first reference pixel, or the first reference pixel and the second reference pixel in other color spaces, i.e. within a block, if a weighted average or the like is applied, the first reference pixel is required, the first reference pixel to be corrected and the filtered first reference pixel and a block boundary, the first reference pixel to be corrected and the filtered first reference pixel are required) may be filtered and fed back to the correction process.

[00419] In case a combination of these two cases occurs, not only pixels in adjacent blocks of the current block may be used, but pixels in prediction blocks obtained in other color spaces, and pixels to be corrected, and surrounding pixels (for example, , pixels of adjacent blocks for correction target pixels and pixels within the current block adjacent to adjacent blocks for correction of target pixels) are filtered (for example, at the correction target position, an M×N mask is used to apply the filter example; at this point, the mask filters these pixels and all or part of the pixels from top, bottom, left, right, top left, top right, bottom left, bottom left, etc.) and are fed back to the correction process.

[00420] This example illustrates a case where filtering is applied after obtaining the prediction value of the current block in other color spaces, but it is also possible to receive the filtered value in the corresponding color space before obtaining the prediction value as the prediction value of the current block. In this example, it should be noted that this case differs from the example above only in order, and the object being filtered is the same.

[00421] In this case, the filtering has only one setting in one picture, or the filtering is determined adaptively according to different coding/decoding elements.

[00422] For example, for the case of adaptive determination, the filtering setting may be determined according to the prediction mode. In more detail, adaptive filtering can be set according to the detailed color copy mode in the color copy mode. For example, in some color copying modes (in this example, if the set of related information <a and b> is obtained in the adjacent area of the current block and the adjacent area of the block corresponding to other color spaces), some <1> filter settings can be accepted, and in some In color copying modes (in this example, compared to the mode described above, if multiple sets of correlation information are obtained, ie, a1 and b1, a2 and b2), some filter <2> settings can be adopted.

[00423] In the filter settings, it can be determined whether to apply filtering. For example, according to the filter settings, filter <1> may be applied or filter <2> may not be applied. Alternatively, filter A<1> may be used, or filter B<2> may be used. Alternatively, the filtering may be applied to all pixels on the left and top sides of the current block, or the filtering may be applied to some pixels on the left and top sides.

[00424] If the pixels to be corrected are limited to some pixels, the pixels to be corrected may be determined in the elements of horizontal lines or vertical lines adjacent to the reference pixels used in the prediction mode (in this example, different from the previous example, auxiliary reference pixels ). In this case, horizontal and vertical lines can be taken into account at the same time, and overlap is allowed.

[00425] For example, pixels corresponding to a-d and pixels (overlapping with a) corresponding to a, e, i, and m may be included in the correction target. In addition, pixels corresponding to e-h and pixels corresponding to b, f, j, and n (a, b, e, and f overlap) can be included in the correction target.

[00426] Briefly, main reference pixels for generating a prediction block may be derived from other color spaces, and auxiliary reference pixels for correcting a prediction block may be derived from blocks adjacent to the current block of the current color space. In addition, they can be obtained from blocks adjacent to the corresponding blocks of other color spaces. In addition, they can be obtained from some pixels of the prediction block of the current block. That is, some pixels in the prediction block may be used to correct some pixels in the prediction block.

[00427] In addition to the above description, adaptive settings may be applied according to additional coding/decoding elements. The above description focuses on the non-directional mode limitation. However, it is applicable to the above examples in a non-limiting sense, and similar or similar settings are applicable to other modes. In addition, the same example as described above is implemented according to a combination of a plurality of elements rather than a single encoding/decoding element.

[00428] This example is an example of obtaining a prediction block of the current block using correlation between color spaces, but the block for obtaining the correlation is obtained from the adjacent area of the current block and the adjacent area of the corresponding blocks in other color spaces, and therefore filtering can be applied to the block boundary.

[00429] There are many cases in which a plurality of reference pixels can be used to generate a prediction block in accordance with the encoding/decoding settings. In more detail, it can be determined according to the encoding/decoding settings whether generation or correction of a prediction block using the second reference pixel is supported.

[00430] As an example, the use of additional pixels in the prediction process can be defined implicitly or explicitly. In an explicit case, the above information may be included in the elements of a video, sequence, frame, slice, tile, block, and so on.

[00431] By way of example, the determination as to whether to use additional pixels in the prediction process may apply to all prediction modes, or may apply to some prediction modes. In this case, some of the prediction modes may be at least one of horizontal, vertical, some diagonal modes, non-directional modes, color copy modes, or the like.

[00432] As an example, the determination as to whether to use additional pixels in the prediction process may apply to all blocks or may apply to some blocks. In this case, some blocks may be defined according to size, shape, etc. blocks, and the corresponding blocks are M×N blocks (for example, the lengths of M and N are 8, 16, 32, 64, etc.; if the blocks are square, the length is 8×8, 16×16, 32×32, 64x64, etc., if the blocks are rectangular, the rectangle might have an aspect ratio of 2:1, 4:1, etc.).

[00433] In addition, the use of additional pixels in the prediction process can be determined in accordance with some encoding/decoding settings. In this case, the encoding/decoding settings may be constrained_intra_pred_flag, and additional reference pixels may be limitedly used in the prediction process according to the flag.

[00434] For example, if the flag restricts the use of the area containing the second reference pixel (i.e., the area filled by the reference pixel filling process or the like according to the flag is assumed), the use of the second reference pixel in the prediction process may be limited. Alternatively, the second reference pixel may be used in the prediction process regardless of the flag.

[00435] In addition to the cases described using the above example, there may be various applications and modifications, such as combinations of one or more elements. In addition, although the above example only describes some cases related to the color copy mode, in addition to color copy, this example can be equally or variably used to generate or correct the prediction mode of a prediction block using a plurality of reference pixels. .

[00436] In the above example, a case has been described including one setting for generating or correcting a prediction block using a plurality of reference pixels in each prediction mode, but a plurality of settings may be used for each prediction mode. That is, a plurality of filter setting candidate groups can be configured to generate the selection information.

[00437] Briefly, the filtering execution information may be processed explicitly or implicitly, and when filtering is performed, information about the filtering selection information may be processed explicitly or implicitly. When processing information explicitly, this information can be included in video, sequence, frame, slice, tile, and block elements.

[00438] The generated prediction block can be corrected, and the prediction block correction process will be described below.

[00439] The correction process may be performed based on predetermined reference pixels and weights. In this case, the reference pixel and the weighted value may be determined according to the position of a pixel (hereinafter referred to as the current pixel) in the current block to be corrected. The reference pixel and the weighted value may be determined according to the intra prediction mode of the current block.

[00440] If the intra prediction mode of the current block is non-directional, the reference pixels ref_L and ref_T of the current pixel may belong to the first pixel line adjacent to the current block and may be located in the same horizontal/vertical line as the current pixel. The weighted value may include at least one of a first weighted value wL in the x-axis direction, a second weighted value wT in the y-axis direction, or a third weighted value wTL in the diagonal direction. The first weighted value may refer to the weighted value applied to the left reference pixel, the second weighted value may refer to the weighted value applied to the top reference pixel, and the third weighted value may refer to the weighted value applied to the top left reference pixel. Here, the first weighted value and the second weighted value may be determined based on the position information of the current pixel and a predetermined scaling factor (nScale). The scale factor may be determined based on the width W and height H of the current block. For example, the first weighted value wL[x] of the current pixel predPixel[x][y] is defined as (32>>((x<<1)>>n Scale)), and the second weighted value (wT[x]) is defined as (32>>((y<<1)>>nScale)). The third weighted value wTL[x][y] can be defined as ((wL[x]>>4)+(wT[y]>>4)). However, if the intra prediction mode is planar mode, the third weighted value may be determined to be 0. The scaling factor may be set to ((Log2(nTbW)+Log2(nTbH)-2)>>2).

[00441] If the intra prediction mode of the current block is vertical/horizontal mode, the reference pixels ref_L and ref_T of the current pixel may belong to the first pixel line adjacent to the current block and may be located in the same horizontal/vertical line as the current pixel . In vertical mode, the first weighted value wL[x] of the current pixel predPixel[x][y] is defined as (32>>((x<<1)>>n Scale)), the second weighted value wT[y] can be defined as equal to 0, and the third weighted value wTL[x][y] can be determined to be equal to the first weighted value. On the other hand, in horizontal mode, the first weighted value wL[x] of the current pixel predPixel[x][y] is defined as 0, the second weighted value wT[y] can be defined as (32>>((y<<1) >>nScale)) and the third weighted value wTL[x][y] can be determined to be equal to the second weighted value.

[00442] If the intra prediction mode of the current block is the diagonal mode, the reference pixels ref_L and ref_T of the current pixel may belong to the first pixel line adjacent to the current block and may be located on the same diagonal line as the current pixel. Here, the diagonal line has the same angle as the intra prediction mode of the current block. The diagonal line may represent a diagonal line from the lower left end to the upper right end, or may represent a diagonal line from the upper left end to the lower right end. In this case, the first weighted value wL[x] of the current pixel predPixel[x][y] is defined as (32>>((x<<1)>>n Scale)), the second weighted value wT[y] is defined as (32 >>((y<<1)>>n Scale)), and the third weighted value wTL[x][y] can be defined as 0.

[00443] If the intra prediction mode of the current block is less than or equal to mode 10, the reference pixels ref_L and ref_T of the current pixel may belong to the first pixel line adjacent to the current block and may be located on the same diagonal line as the current pixel. Here, the diagonal line has the same angle as the intra prediction mode of the current block. In this case, the reference pixel may be limited such that only one of the left reference pixel or the top reference pixel of the current block is used. The first weighted value wL[x] of the current pixel predPixel[x][y] is defined as 0, the second weighted value wT[y] is defined as (32>>((y<<1)>>nScale)), and the third weighted the value wTL[x][y] can be defined as equal to 0.

[00444] If the intra prediction mode of the current block is greater than or equal to mode 58, the reference pixels refL and refT of the current pixel may belong to the first pixel line adjacent to the current block and may be located on the same diagonal line as the current pixel. Here, the diagonal line has the same angle as the intra prediction mode of the current block. In this case, the reference pixel may be limited such that only one of the left reference pixel or the top reference pixel of the current block is used. The first weighted value wL[x] of the current pixel predPixel[x][y] is defined as (32>>((x<<1)>>nScale)), the second weighted value wT[y] is defined as 0, and the third weighted the value wTL[x][y] can be defined as equal to 0.

[00445] Based on the determined reference pixels refL[x][y] and refT[x][y] and the weighted values wL[x], wT[y] and wTL[x][y], correction of the current pixel predPixels[x ][y] can be done with Formula 1 below.

[00446] [Formula 1]

[00447] predPixels[x][y]=clip1Cmp((refL[x][y]*wL[x]+refT[x][y]*wT[y]-p[-1][-1]* wTL[x][y]+(64-wL[x]-wT[y]+wTL[x][y])*predPixels[x][y]+32)>>6)

[00448] However, the above-described correction process can only be performed when the current block does not perform intra prediction in subblock elements. The correction process can only be performed when the reference pixel of the current block is the first line of pixels. The correction process can only be performed when the intra prediction mode of the current block matches the specific mode. Here, the specific mode may include at least one of non-directional mode, vertical mode, horizontal mode, a mode that is less than a predetermined first threshold mode, or a mode that is greater than a predetermined second threshold mode. The first threshold mode may be 8, 9, 10, 11, or 12, and the second threshold mode may be 56, 57, 58, 59, or 60.

[00449] The prediction mode determination section performs the optimal mode selection process in the plurality of prediction mode candidate groups. In order to determine the mode with the optimal encoding cost to generate the number of bits of the corresponding mode, block distortion can be generally considered (e.g., distortion of the current block and the reconstructed block; sum of absolute difference (SAD), sum of quadratic difference (SSD), etc.) and random distortion technology. The prediction block generated based on the prediction mode determined by the above process may be sent to a subtraction section and an addition section.

[00450] The prediction mode encoding section is configured to encode the prediction mode selected by the prediction mode determination section. In the prediction mode candidate group, index information corresponding to the prediction mode may be encoded, or the prediction mode may be predicted to encode information associated therewith. That is, the first case indicates a direct prediction mode encoding method without performing prediction, and the latter case indicates a prediction mode prediction method such that encoding mode prediction information and information obtained based on the prediction information are encoded. In addition, the first example is an example that can be applied to a color difference component, and the last example is an example that can be applied to a luminance component. The present invention is not limited to these examples, and other cases are also possible.

[00451] In the case of performing encoding by predictive mode prediction, the prediction value (or prediction information) of the prediction mode may be referred to as the most probable mode (MPM). In this case, a predetermined prediction mode (e.g., DC mode, planar mode, vertical mode, horizontal mode, diagonal mode, etc.) or spatially adjacent block prediction mode (e.g., left block, top block, top left block, top right block, bottom left block, etc.). In this example, the diagonal mode may represent diagonal up right, diagonal down right, and diagonal down left, which may correspond to mode 9, mode 2, and mode 66 in FIG. 9.

[00452] In addition, a mode derived from the modes included in the MPM candidate group may be configured as an MPM candidate group. For example, in the case of a directional mode included in the MPM candidate group, a mode with a mode interval difference a (e.g., a is a non-zero integer such as 1, -1, 2, or -2; in case FIG. 9 mode 10 has been added, the derived modes are mode 9, mode 11, mode 8, mode 12, etc.) is re-included (or further included) in the MPM candidate group.

[00453] The above example may correspond to the case where the MPM candidate group is executed in a plurality of modes, the MPM candidate group (or the number of MPM candidate groups) may be determined according to the encoding/decoding settings (for example, the candidate mode group prediction, image type, block size, block shape, etc.), and at least one mode may be included.

[00454] The prediction mode for configuring a group of MPM candidates may take precedence. The order of the prediction modes included in the MPM candidate group may be determined according to priority, and the configuration of the MPM candidate group can be completed if the number of MPM candidate groups is filled according to priority. In this case, the priority may in turn be a spatially adjacent block prediction mode, a predetermined prediction mode, and a mode derived from the prediction mode originally included in the MPM candidate group, but other modifications are possible.

[00455] If the prediction mode encoding of the current block is performed using the MPM, information about whether the prediction mode matches the MPM (eg, most_probable_mode_flag) may be generated.

[00456] If the prediction mode is the same as the MPM, (eg, most_probable_mode_flag=1), MPM index information (eg, mpm_idx) may be additionally generated according to the MPM configuration. For example, if the MPM is executed as a single prediction mode, no additional MPM index information is generated, and if the MPM is performed as a plurality of prediction modes, index information corresponding to the prediction mode of the current block may be generated in the MPM candidate group.

[00457] If the prediction mode is not the same as the MPM (e.g., most_probable_mode_flag=0), in the remaining prediction mode candidate group (or non-MPM candidate mode group) other than the MPM candidate group in the prediction mode candidate groups, non-MPM mode index information (eg, non_mpm_idx) corresponding to the prediction mode of the current mode may be generated, which may be an example of a case in which the non-MPM mode is configured as one group.

[00458] If a group of non-MPM candidate modes consists of a plurality of groups, information about which group the prediction mode of the current block belongs to can be generated. For example, the non-MPM mode is configured by groups A and B (assuming that A is configured with m prediction modes, B is configured with n prediction modes, the non-MPM mode is configured with m+n prediction modes, and n is greater than m; assuming that the mode from group A is a steered mode and has equal intervals, and the mode from group B is a steered mode and does not have equal intervals), if the prediction mode of the current block is the same as the prediction mode from group A (eg, non_mpm_A_flag= 1), index information corresponding to the prediction mode of the current block may be generated in candidate group A, and if there is no match (for example, non_mpm_A_flag=0), index information corresponding to the prediction mode of the current block may be generated in the remaining mode groups- prediction candidates (or group B of candidates). As in the above example, the non-MPM mode may be configured as at least one group (or set) of candidate prediction modes, and the configuration of the non-MPM modes may be determined according to the group of prediction candidate modes. . For example, if there are no more than 35 groups of candidate prediction modes, there may be one non-MPM mode, and in other cases, there are two or more non-MPM modes.

[00459] As in the above example, the purpose of supporting non-MPM modes is that the number of prediction modes is large if the non-MPM mode consists of a plurality of groups, and the number of mode bits is reduced if the prediction mode not predicted by MRM.

[00460] If prediction mode encoding (or prediction mode decoding) of the current block is performed using MPM, the binarization table applied to each prediction mode candidate group (e.g., MPM candidate group, non-MPM candidate mode group) MPM, etc.) may be separately generated, and the binarization method applied to each candidate group may also be applied separately.

[00461] Prediction-related information that is generated by the prediction mode coding section may be transmitted to the coding section and included in the bitstream.

[00462] FIG. 14 shows an example of a tree division block according to an embodiment of the present invention.

[00463] FIG. 14, i represents a quaternary tree partition, ii represents a horizontal partition in a binary tree partition, and iii represents a vertical partition in a binary tree partition. In this figure, A-C represent the original blocks (blocks before the split, for example, assuming it is an element of the coding tree), and the numbers after the text represent the number of each section after the split has been performed. In the case of a quad tree, the top left box, top right box, bottom left box, and bottom right box are assigned numbers 0-3, respectively, and in the case of a binary tree, left/top box and bottom/right box are respectively assigned numbers 0 and 1 .

[00464] With reference to FIG. 14, the splitting state or splitting information to obtain the corresponding block can be determined by text and numbers obtained in the splitting process.

[00465] For example, in image i of FIG. 14, block A00 is the upper left block (adding 0 to AO) in the four blocks obtained by performing quad tree partitioning on the top left block A0 (adding 0 to A) of the four blocks obtained after performing quad tree partitioning. in relation to the original block A.

[00466] Alternatively, in image ii of FIG. 14, block B10 is the upper block (adding 0 to B1) in two blocks obtained by performing horizontal splitting in binary tree splitting on the lower block B1 (adding 1 to B) of the two blocks obtained after performing horizontal splitting in splitting. based on the binary tree with respect to the original block B.

[00467] With the above division, the division state and division information (e.g., supported division settings <tree mode type, etc>) of each block, the block support range such as the minimum size and maximum size <detail - range) can be obtained. support according to division method>, allowable division depth <detail - support range according to division method>, division flag<detail division flag according to division method>, picture type <I/P/B>, encoding mode division state block <intra-prediction/prediction performed between pictures>, encoding/decoding settings necessary for confirming information, information, etc., and which block (parent block) the current block belongs to can be confirmed before the division step to obtain current block (child block). For example, in image i of FIG. 14, for block A31, adjacent blocks may include A30, A12, and A13, and A30 can be confirmed to belong to the same block A3 as A31 in the division step of the previous step. In the case of A12 and A13 in the previous step, that is, in the separation step, A3, A12 and A13 belong to another block A1, and it can be confirmed that A12 and A13 belong to the same block A only in the previous step.

[00468] While the previous example is for the case of a single split operation (quaternary split of a quad tree or horizontal/vertical split of a binary tree), the following examples will explain the case of performing split based on multiple trees.

[00469] FIG. 15 shows an example of a multi-tree partitioning block according to an embodiment of the present invention.

[00470] With reference to FIG. 15, the splitting state or information about the splitting performed to obtain the corresponding block can be confirmed by text and numbers obtained in the splitting process. In this example, each word does not mean the original block, but the split information is indicated by a number.

[00471] For example, in the case of block A1A1B0, the upper right block A1 is obtained by performing quaternary tree splitting on the original block, the upper right block (adding A1 to A1) is obtained by performing quaternary tree splitting on block A1, and the top block (adding B0 to A1A1) is represented by performing horizontal binary tree splitting on block A1A1.

[00472] Optionally, in the case of block A3B1C1, the lower right block A3 is obtained by performing quaternary tree splitting on the original block, the lower block (adding B1 to A3) is obtained by performing horizontal splitting while binary tree splitting on block A3 , and the right block (adding C1 to A3B1) is represented when vertical splitting is performed in the binary tree splitting process on A3B1.

[00473] In this example, with the partitioning information according to the splitting step of each block, it is possible to determine relationship information between the current block and the adjacent block, such as at which splitting step the current block and the adjacent block are identical.

[00474] FIG. 14 and 15 show some examples of determining the separation information of each block, and the separation information of each block can be confirmed using different information to determine the separation information and combined information (for example, separation flag, depth information, maximum value of depth information, range block, etc.) so that the relationship between blocks can be determined.

[00475] FIG. 16 is a circuit diagram showing various situations of block division.

[00476] In general, since there are various texture information in an image, it is difficult to perform encoding/decoding using the encoding/decoding method. For example, some regions may contain regions in which there are stronger edge components in a particular direction, and some regions may contain complex regions in which edge components do not exist. Block division plays an important role in effectively performing encoding on these areas.

[00477] The purpose of performing block division is to efficiently divide regions according to image features. However, when using only one division mode (eg, division based on quaternary tree), it may be difficult to adequately reflect the characteristics of the image to perform division.

[00478] With reference to FIG. 16, it can be confirmed that images containing different textures are split according to quaternary tree based split and binary tree based split, a to e can be a case in which only quad tree based split is supported, and f to j can represent the case in which only binary tree-based partitioning is supported.

[00479] In this example, partitions according to quaternary tree partitioning are called UL, UR, DL, and DR, and binary tree partitioning is also described on that basis.

[00480] Image a in FIG. 16 may be a texture form in which quaternary tree division achieves optimum efficiency: it may be divided into four parts by one quaternary tree division (Div_HV), and encoding/decoding is performed in the elements of each block. At the same time, when using binary tree partitioning as shown in image f of FIG. 16, unlike a quaternary tree, three splits may be required (Div_V twice, Div_H once).

[00481] Picture b of FIG. 16 is such that the texture is split into top and bottom block regions, in image b of FIG. 16, if a quaternary tree partition is applied, one partition (Div_HV) is required, and in the image g of FIG. 16, one division Div_H can also be performed in a division based on a binary tree. Assuming that the quaternary tree split flag requires 1 bit and the binary tree split flag requires more than 2 bits, the quad tree can be considered efficient in terms of flag bits, but encoding/decoding information (e.g., information for texture information expressions (texture expression information (residual signal, coding coefficient, information indicating the presence or absence of a coding coefficient)), prediction information <for example, information related to intra prediction, and information related to prediction performed between pictures and >, transformation information <eg, transformation category information, transformation division information, etc.>) is generated in the elements of each block. In this case, the performance of quaternary tree partitioning is low compared to binary tree partitioning because the texture information is an example of dividing similar areas and regenerating corresponding information.

[00482] In other cases of Fig. 16, it is also possible to determine which tree is more efficient according to the texture category. From the above examples, it can be confirmed that, in order for the effective area to be divided according to image characteristics by the above examples, it is extremely important to support the variable block size and various division modes.

[00483] A detailed analysis of quaternary tree partitioning and binary tree partitioning is performed as follows.

[00484] With reference to image b in FIG. 16 shows the blocks (in this example, the left block and the top block) adjacent to the bottom right block (the current block). It can be seen that the left block is a block having characteristics similar to the current block, and the top block is a block having characteristics different from the current block. And although the block has properties similar to some blocks (in this example, the left block), it is possible that the block was segmented based on the properties of the division based on a quaternary tree.

[00485] In this case, since the current block has characteristics similar to the left block, similar encoding/decoding information can be generated. In the case of an intra prediction mode (e.g., MPM, i.e., information predicting the current block mode from adjacent blocks in order to reduce the number of bits of the current block prediction mode), a motion information prediction mode (e.g., skip mode, merge mode) appears. , information for reducing mode bits (eg, contention mode)) to effectively avoid this situation, while the information of the left block is available for sending. That is, if the encoding information (eg, intra prediction information, inter-picture prediction information, filter information, coefficient information, etc.) of the current block is assigned to one of the left block and the top block, there may be reference is made to more precise information (in this example, to the left block).

[00486] With reference to image e in FIG. 16, a small block is assumed in the lower right corner (a block dividing the current block twice). Likewise, the current block also has similar characteristics to the top block of the left block or the top block, so encoding efficiency can be improved by referring to encoding information from the corresponding block (in this example, the top block).

[00487] Conversely, with reference to image j in FIG. 16, reference is made to the left block adjacent to the rightmost block (the block that divides the current block twice). In this example, since there is no top block, by referring only to the left block, it can be confirmed that the current block is a block having a characteristic different from the left block. It can be confirmed that a quaternary tree partition can be similar to a portion of adjacent blocks, while a binary tree partition has a high probability of obtaining different characteristics from adjacent blocks.

[00488] In more detail, if some blocks (x) and adjacent blocks (y) are the same blocks before the corresponding split, for splitting based on a quaternary tree, some blocks (x) and adjacent blocks (y) that are the same, like pre-split blocks may have similar characteristics or different characteristics, and vice versa, for a binary tree, in most cases, some blocks (x) and adjacent blocks (y) that are the same as pre-split blocks may have different characteristics. In other words, if there is a case of similar characteristics, there is no need to perform block splitting, and blocks are finalized before splitting, but in most cases, blocks undergo splitting due to different characteristics.

[00489] In addition to the example above, in image i of FIG. 16, the lower right x block and the lower left y block may have the following relationships. That is, since the image characteristics of the upper blocks are similar, the division is determined without performing the division, and since the image characteristics of the lower blocks are different, the division is determined after division into the lower left block and the lower right block.

[00490] Briefly, in the case of a quaternary tree partition, if there is an adjacent block having a parent block identical to the current block in adjacent blocks adjacent to the current block, it may have characteristics similar to or different from the current block, since the characteristic of the quaternary tree is the unconditional division of the block into 1/2 in the horizontal and vertical directions. In the case of binary tree partitioning, if there is an adjacent block having a parent block identical to the current block in adjacent blocks adjacent to the current block, horizontal or vertical separation can be performed on the adjacent block according to image characteristics, so "was subjected to separation" can mean separation due to different characteristics.

[00491] (Partition based on quaternary tree)

[00492] An adjacent block (in this example, assumed but not limited to the left block or the top block) that is the same as the block before the split, compared to the current block, may have similar or different characteristics to the current block.

[00493] In addition, adjacent blocks other than the current block and the pre-split block may have similar or different characteristics to the current block.

[00494] (Binary Tree Partition)

[00495] An adjacent block (in this example, assumed to be the left block or the top block, the maximum number of candidates is one due to the binary tree) that is the same as the block before splitting may have different characteristics compared to the current block.

[00496] In addition, adjacent blocks other than the pre-split block may have similar or different characteristics to the current block compared to the current block.

[00497] Descriptions will be made below, provided that the above assumption is the main assumption of the present invention. According to the above contents, a case (1) in which characteristics of adjacent blocks are the same or different from the current block and a case (2) in which characteristics of adjacent blocks are different from those of the current block can be distinguished.

[00498] Reference is again made to FIG. 15.

[00499] As an example (the current block is A1A2), since the block A1 before the division of the block A1A2 is different from the block before the division (original block) of the block A0, the block A0 (left block) in adjacent blocks can be classified (i.e. in the case where it is not known whether the characteristic is the same or different from the current block) as a general case.

[00500] For the adjacent block A1A0 (upper block), since the block A1 before splitting the block A1A2 is the same as the block before splitting the block A1A0, the split mode can be confirmed. In this case, since the division mode is division (A) based on a quaternary tree, it is classified as a general case.

[00501] As an example (the current block is A2B0B1), the A2B0 block before splitting the A2B0B1 block is the same as the block A2B0B1 before splitting the A2B0B1 block for the A2B0B0B0 block (upper block) in adjacent blocks, thus confirming the split mode. In this case, since the division mode is binary tree division (B), it is classified as an exceptional case.

[00502] As an example (the current block is A3B1C0), in adjacent blocks, since block A3B1 before splitting block A3B1C0 is different from block A3 before splitting block A3B0, block A3B0 (in this example, the upper right block) can be classified as common happening.

[00503] As described in the above example, adjacent blocks can be classified into a general case and an exceptional case. For the general case, it is not known whether the encoding information of the corresponding adjacent block can be used as the encoding information of the current block, and for the exception, it is determined that the encoding information of the corresponding adjacent block cannot be used as the encoding information of the current block.

[00504] Based on the above classification, a method for obtaining prediction information of the current block from adjacent blocks can be used.

[00505] Briefly, it is confirmed whether the blocks are the same (A) before the division of the current block and adjacent blocks.

[00506] If A has the same result, the division mode of the current block (if so, not only the current block but also adjacent blocks are blocks determined using the same division mode, thus confirming only the current block) is confirmed (B).

[00507] If result A is different, the process ends (ends).

[00508] If the result of B is a quaternary tree split, adjacent blocks are marked as normal and the process ends (ends).

[00509] If result B is a binary tree split, adjacent blocks are marked as exclusive and the process ends (ends).

[00510] The above example can be applied to setting a group of intra prediction modes prediction candidates (related to MPM) according to the present invention. General prediction candidate group settings (eg, candidate group configuration priorities, etc.) and exclusive prediction candidate group settings may be supported, and priorities may be shifted back or excluded from candidate groups derived from a block based on the states of the above adjacent blocks.

[00511] In this case, adjacent blocks in which these settings are applied may be limited to a spatial case (single space) or may be applicable to a case of obtaining a single image from other color spaces, such as a color copying mode. That is, the above settings can be made in consideration of the prediction state or the like. for a block obtained using the color copying mode.

[00512] In addition to the above example, the following example is an example of adaptively determining encoding/decoding settings (e.g., prediction candidate group settings, reference pixel settings, etc.) according to a relationship between blocks (in the above example, relative the relationship between the current block and other blocks is identified using partition block information, etc.).

[00513] This example (luma component) is described based on the assumption that a total of 3 candidates (left and right in this example) are selected from adjacent blocks, and they are set as a group of MPM candidates if the encoding/decoding device has 35 predefined intra prediction modes.

[00514] By adding one candidate, respectively, in the left block L0 and the upper block T0, a total of two candidates can be configured. If a candidate group cannot be configured in each block, it can be replaced or populated with DC mode, planar mode, vertical mode, horizontal mode, diagonal mode, etc. If two candidates are filled using the above process, the remaining candidate may be considered a filled number in various cases.

[00515] For example, if the candidates filled in each block are the same, the adjacent modes of the above mode are replaced in a manner that does not overlap with the modes included in the candidate group (e.g., k_mode-2, k_mode-2, k_mode+1, k_mode+ 2, etc. if the same mode is k_mode). Alternatively, if the candidates populated in each block are the same or different, the candidate group may be configured by adding planar mode, DC mode, vertical mode, horizontal mode, and diagonal mode.

[00516] Using the above process, a group (general case) of candidate intra prediction modes can be configured. The candidate group thus established belongs to the general case, and an adaptive intra prediction mode candidate group (exceptional case) can be configured in accordance with adjacent blocks.

[00517] For example, if an adjacent block is marked with an exception condition, a candidate group derived from that block may be excluded. If the left block is marked as an exception state, the candidate group can be configured by the top block and a predetermined prediction mode (for example, the mode derived from the DC block, planar block, vertical block, horizontal block, diagonal block, and top block, etc. .). The above example can be applied identically/similarly even when the MPM candidate group is configured with more than three MPM candidates.

[00518] The description will be made based on the following assumptions: this example (color difference component) contains five intra prediction modes (DC mode, planar mode, vertical mode, horizontal mode, and, in this example, color mode), and encoding/decoding is performed by configuring prediction modes that are adaptively prioritized as candidate groups (ie, encoding/decoding is performed directly without using MPM).

[00519] First, the color mode is given the highest priority (in this example, index 0; "for 0" is assigned 1 bit), while the other modes (planar, vertical, horizontal, and DC in this example) are given low priority (in in this example, respectively, 3 bits are assigned with the values "100", "101", "110", "111").

[00520] If the color mode matches one of the other prediction modes (DC, planar, vertical, and horizontal) from the candidate group, the priority giving the matching prediction mode (indices 1-4 in this example) can be assigned a predetermined prediction mode (eg, diagonal mode, etc.), and if there is no match, the candidate group configuration is terminated.

[00521] Using the above process, a candidate group for the intra prediction mode can be configured. An example implemented in this way corresponds to the general case, and an adaptive candidate group can be configured in accordance with the received color mode blocks.

[00522] In this example, if a block corresponding to the current block in other color spaces of the received color mode is configured as a single block (i.e., a non-separated state), it can be called a general case, and if it is configured as a plurality of blocks ( i.e., a split state of more than two blocks), it may be referred to as an exceptional case. As described above, there may be an example based on the following assumptions: this example is different from the example in which classification is performed according to whether the current block is the same as the parent block of an adjacent block, with a split mode or the like, and if a plurality of corresponding blocks of different color spaces are configured, there is a higher probability of characteristics different from the current block. That is, this should be understood as an example of adaptively determining encoding/decoding settings in accordance with the relationship between blocks.

[00523] If the corresponding block is marked as an exception state in accordance with the above suggestion example, the prediction mode obtained from this block may have a low priority. In this case, one of the other prediction modes (planar, DC, vertical, and horizontal) may be given high priority, and low priority may be given to other prediction modes and color modes not included in the priority.

[00524] For ease of description, the above examples are described on the basis of some assumptions, but without limitation, and the same or similar applications can be used for various embodiments of the present invention described above.

[00525] To sum up, candidate group A can be used when there are blocks marked as exceptions in adjacent blocks, and candidate group B can be used when there are blocks marked as exceptions. It should be understood that the above classification is divided into two cases, but the configuration of the intra prediction candidate group in block elements can be adaptively implemented in accordance with exception states and block positions.

[00526] In addition, the above example describes tree division in the division mode, but is not limited to it. In more detail, in the above example, it is possible to set a block obtained using at least one tree partitioning mode as a coding block, and directly perform prediction, transformation, etc. without dividing this block into prediction blocks, transformation blocks, etc.

[00527] As another example of division settings, a coding block is obtained using tree division, and at least one prediction block is obtained based on the obtained coding block.

[00528] For example, it is assumed that a coding block (2N×2N) can be obtained using tree partitioning (in this example, based on a quaternary tree), and a prediction block is obtained using typed partitioning (candidate types supported in this example are 2N×2N, 2N×N, N×2N and N×N). In this case, if a coding block (in this example, assumed to be a parent block) is divided into a plurality of prediction blocks (in this example, assumed to be sub-blocks), setting the above-described exception state or the like. can also be applied between prediction blocks.

[00529] As shown in image g of FIG. 16, if there are two prediction blocks (separated by a thin solid line) in a coding block (thick solid line), the bottom block has a different characteristic from the top block and does not belong to the coding information of the top block or has a low priority.

[00530] FIG. 17 shows an example of block division according to an embodiment of the present invention. In more detail, an example is shown in which a basic coding block (maximum coding block, 8N×8N) obtains a coding block (diagonal block, 2N×2N) using quaternary tree partitioning, and the resulting coding block is divided into at least one prediction block. (2N×2N, 2N×N, N×2N and N×N) using typed partitioning.

[00531] The setting of the intra prediction mode candidate group for the case of obtaining rectangular blocks (2N×N and N×2N) will be described below.

[00532] FIG. 18 shows various examples of intra prediction mode candidate groups with respect to setting blocks generating prediction information (the prediction blocks in this example are 2N×N blocks).

[00533] This example assumes that if there are 67 intra prediction modes (luminance component), a total of 6 candidates are selected from adjacent blocks (left, top, top left, top right, and bottom left in this example) and are given in a group MRM candidates.

[00534] With reference to FIG. 13, four candidates may be configured in the order L3-T3-B0-R0-TL, and two candidates may be configured in a predetermined mode (eg, planar and DC). If in the above configuration, the maximum number (in this example 6) is not filled, the prediction mode (for example, k_mode-2, k_mode-2, k_mode+1, k_mode+2, etc., if k_mode is available) can be enabled, obtained from the prediction modes included in the candidate group, a predetermined mode (eg, vertical, horizontal, or diagonal, etc.), and so on.

[00535] In this example, in the case of spatially contiguous blocks, the candidate group priority is the order left block-top block-bottom left block-top right block-top left block of the current block block), and the predetermined mode, in turn, assumes the order planar-DC-vertical-horizontal-diagonal mode.

[00536] With reference to image a in FIG. 18, as with setting the candidate group in the above example, the current block (2N×N, PU0) can configure the candidate group in the order l1-t3-l2-tr-t1 (because other content is repeated, it is omitted). In this example, an adjacent block of the current block may be a block that has been encoded/decoded (a coding block, i.e. a prediction block in other coding blocks).

[00537] In contrast to the above, if the position of the current block corresponds to PU1, at least one group of candidate intra prediction modes may be defined. In general, a block adjacent to the current block is more likely to have similar characteristics to the current block, and therefore is more advantageous for configuring a candidate group from the corresponding block (1). On the other hand, it may be necessary to configure a group of candidates for parallel encoding/decoding processing (2).

[00538] If the current block is PU1, when setting up a candidate group building as in case (1), for example, in image b of FIG. 18, candidate groups can be configured in the order l3-c7-bl-k-l1 (k can be obtained from c7 or tr, etc.), and in the case of setting the candidate group configuration as in case (2), for example , in the image from Fig. 18, candidate groups may be configured in the order l3-bl-k-l1 (k may be obtained from tr, etc.). The difference between these two examples is whether the group of candidate intra prediction modes for the top block is included. That is, in the first case, the intra prediction mode of the upper block is included in the candidate group in order to improve the coding/decoding efficiency of the intra prediction mode, and in the latter case, the intra prediction mode of the upper block that cannot be referred to is excluded from the candidate group for parallel processing, or the like. etc., since it has not yet been determined whether encoding/decoding is completed.

[00539] FIG. 19 shows various examples of settings of intra prediction mode candidate groups with respect to setting of blocks generating prediction information (prediction blocks in this example are N×2N blocks).

[00540] This example assumes that if there are 67 intra prediction modes (luma component), a total of 6 candidates are selected from adjacent blocks (left, top, top left, top right, and bottom left in this example) and they are given in group of MRM candidates.

[00541] With reference to FIG. 13, one candidate can be configured in the order L3-L2-L1-L0 (top block), one candidate can be formed in the order T3-T2-T1-T0 (top block), two candidates can be configured in the order B0-R0 -TL (upper left block, upper right block and lower left block), and the two candidates can be configured in a predefined mode (eg planar or DC). If the maximum number is not filled in the above configuration, the prediction mode obtained from the prediction modes included in the candidate group, the preset mode, and so on, may be turned on. In this case, the candidate group configuration priority may be the order left block-top block-planar block-block DC-bottom left block-top right block-top left block.

[00542] With reference to image a in FIG. 19, the current unit (N×2N, PU0) can configure one candidate in the order l3-l2-l1-l0, one candidate in the order t1-t0, and two candidates in the order bl-t2-tl in the same manner as group settings. candidates in the examples above. In this example, an adjacent block of the current block may be a block that has been encoded/decoded (a coding block; i.e., a prediction block in other coding blocks).

[00543] In contrast to the above, if the position of the current block corresponds to PU1, at least one group of candidate intra prediction modes may be defined. In the above examples, (1) and (2) can be configured.

[00544] If the current block is PU1, one candidate in the order c13-c9-c5-c1, one candidate in the order t3-t2, and two candidates in the order k-tr-t1 can also be configured (k can be obtained from bl or c13, etc.) when configuring the candidate group as in (1), for example, in image b of FIG. 19, and one candidate in the order t3-t2 and two candidate groups in the order k-tr-tl (k can be obtained from bl, etc.) can also be configured by configuring the candidate group as in (2), for example, in the image from Fig. 19. The difference between these two examples is whether a group of candidate intra prediction modes for the top block is included. That is, in the first case, the left block intra prediction mode is included in the candidate group to improve the coding/decoding efficiency of the intra prediction mode, and in the latter case, the left block intra prediction mode that cannot be referred to is excluded from the parallel processing candidate group, or the like. etc., since it has not yet been determined whether encoding/decoding is completed.

[00545] Thus, the configuration of the candidate group can be determined according to the configuration settings of the candidate group. In this example, candidate group configuration settings (in this example, candidate group configuration settings for parallel processing) may be implicitly specified, or related information may be explicitly included in video, sequence, frame, slice, tile, etc. elements. .

[00546] A brief description of the above content is as follows. Related information is expected to be implicitly defined or explicitly generated.

[00547] The configuration setting of the intra prediction mode candidate group is confirmed in the initial encoding/decoding stage (A).

[00548] If the acknowledgment result for A is such that reference can be made to the setting of a previous prediction block in the same coding block, the intra block prediction mode is included in the candidate group (end).

[00549] If the confirmation result for A is such that reference to the setting of the previous prediction block in the same coding block is prohibited, the intra block prediction mode is excluded from the candidate group (end).

[00550] For ease of description, the above examples are described on the basis of some assumptions, but without limitation, and the same or similar applications can be applied to various above-described embodiments of the present invention.

[00551] FIG. 20 shows an example of block division according to an embodiment of the present invention. In more detail, the basic coding block (maximum coding block) is an example of obtaining a coding block (A×B diagonal block) by binary tree partitioning (or multi-tree partitioning), and the resulting coding block is set as a prediction block.

[00552] In this case, the setting of the motion information prediction candidate group for the case of obtaining a rectangular block (A×B, A≠B) will be described below.

[00553] FIG. 21 shows various examples of intra prediction mode candidate groups for setting blocks generating prediction information (2N×N coding block in this example).

[00554] The description is made based on the following assumptions. If the given example (luma component) contains 67 intra prediction modes, a total of 6 candidates are selected from the adjacent blocks (left, top, top left, top right, and bottom left in this example) and set to the MPM candidate group. If a plurality of non-MPM candidate mode groups are configured (in this example, A and B, the mode in which the prediction mode of the current block is more likely to be predicted in the non-MPM candidate modes belongs to A), a total of 16 candidates, and a total of 45 candidates are configured in group B.

[00555] In this case, group A contains candidates classified according to certain rules (e.g., composed of equidistant modes in directional mode) that are not included in the modes from the MPM candidate groups, or may contain candidates that are not included in the final MPM group -candidates according to the priority of the group of MPM-candidates, etc. Group B may be composed of candidates not included in group A in the MPM candidate group and the non-MPM candidate regime group.

[00556] With reference to image a of FIG. 21, the current block (CU0 or PU0 in which the size is assumed to be 2N×N and the horizontal/vertical size ratio is 2:1) can configure six candidates as a candidate group in the order l1-t3-planar-DC-l2-tr- tl-l1*-t3*-l2*-tr*-t1* - vertical-horizontal-diagonal mode. In the above example, "*" represents the mode obtained from the prediction mode of each block (eg, plus +1 mode, -1, etc.).

[00557] On the other hand, with reference to image b in FIG. 21, the current unit (CU1 or PU1 in which the size is 2N×N) can configure six candidates as a candidate group in the order 13-c7-planar-DC-bl-k-l1-l3*-c7*-bl*-k *-l1* - vertical-horizontal-diagonal mode.

[00558] In this example, in the case where setting (2) shown in FIG. 18 as shown in image b of FIG. 21, the candidate group is excluded from the upper block prediction mode candidate group according to the order 13-planar-DC-bl-k-l1-c7-l3*-bl*-k*-l1*-vertical-horizontal-diagonal-c7 *, or the priority is reduced, and thus it can be included in group A.

[00559] However, unlike the case shown in FIG. 18 is as follows: even if a block is rectangular, the block is divided into coding units and immediately set as a prediction block without further division. Therefore, as shown in FIG. 18, the configuration setting of the candidate group may not be applicable to this example (eg, setting (2)).

[00560] However, in case k in FIG. 21, since there is a position where encoding/decoding is not completed, it can be obtained from an adjacent block where encoding/decoding is completed.

[00561] FIG. 22 shows various examples of setting intra prediction mode candidate groups for blocks generating prediction information (the coding blocks in this example are N×2N blocks).

[00562] The description will be made based on the following assumptions. If this example (color difference component) contains five intra prediction modes (in this example, DC mode, planar mode, vertical mode, horizontal mode, and color copy mode), encoding/decoding is performed by configuring the prediction modes that are adaptively prioritized as groups candidates.

[00563] In this case, the priority will be described based on the assumption that the priority is determined from adjacent blocks (in this example, left, top, top left, top right and bottom left).

[00564] With reference to FIG. 13, the first class candidate can be determined in the left block L3, the upper block T3, the upper left block TL, the upper right block R0, and the lower left block B0. In this case, the mode having the value most frequently encountered in the block prediction modes can be determined as a first class candidate. If multiple modes have the most frequently occurring values, a predetermined priority is given (eg, color copy mode - planar-vertical-horizontal-DC).

[00565] This example can be considered similar to setting MPM candidates (in this example, the first bit is defined as 0 or 1; if the first bit is 1, 2 bits need to be added; the first bit can perform bypass coding or legacy coding according to coding settings /decoding, and the remaining bits can perform bypass coding) with respect to obtaining a mode (a mode is a mode estimated as a prediction mode of the current block) from adjacent blocks and thus determining a priority (i.e., determining the number of assigned bits; for example, in the case of of the first class, it is "0"; in the case of classes two to four, they are "100", "101", "110", and "111") and with respect to the prediction of the prediction mode.

[00566] With reference to image a in FIG. 22, the current block (CU0 or PU0 in which the size is assumed to be N×2N and the horizontal/vertical size ratio is 1:2) may determine the first class candidates in blocks 13,t1,t1,t2 and b1.

[00567] On the other hand, with reference to image b in FIG. 22, the current block (CU1 or PU1 in which the size is N×2N) may determine the first class candidate in blocks c13, t3, t1, tr, and k.

[00568] In this example, if the same setting as (2) shown in FIG. 19, the current CU1 may determine the first class candidate in blocks t3, t1, tr, and k. In addition, if the block corresponding to the current block in other color spaces in which the color copy mode is obtained is not configured as one block, the priority of the color copy mode is shifted back (lowered) in the preset priority so that the priority changes to planar-vertical- horizontal - DC color copying mode, etc. As described above, this example may take place based on the following assumptions. This example is different from the example in which classification is performed according to whether the current block is the same as the parent block of an adjacent block, with a split mode, or the like, however, if a plurality of corresponding blocks of different color spaces are configured, there are more high probability of characteristics different from the current block. That is, this should be understood as an example of adaptively determining encoding/decoding settings in accordance with the relationship between blocks.

[00569] However, unlike the case shown in FIG. 19 is as follows: even if the block is rectangular, this block is divided into coding units and corresponds to the case of immediate tuning as a prediction block without further division. Therefore, as shown in FIG. 19, the candidate group configuration setting may not be applicable in this example.

[00570] However, in case k in FIG. 22, since there is a position where encoding/decoding is not completed, it can be obtained from an adjacent block where encoding/decoding is completed.

[00571] For ease of description, the above examples have been described based on some assumptions, but without limitation, and the same or similar applications are possible in various embodiments of the present invention described above.

[00572] In the embodiments shown in FIG. 21 and 22, it is assumed that M×N (M≠N) blocks that can be partitioned using binary tree partitioning appear sequentially.

[00573] In the above case, the case described in FIG. 14-16 (an example of setting up a group of candidates by determining the relationship between the current block and the adjacent block, etc.). That is, in FIG. 21 and 21, adjacent blocks prior to division of CU0 and CU1 are identical to each other, and CU0 and CU1 can be obtained by horizontal division or vertical division in binary tree division.

[00574] In addition, in the embodiments shown in FIG. 18 and 19, it is assumed that after splitting a coding block, there are a plurality of rectangular prediction blocks in a coding block.

[00575] In the above case, the case described in FIG. 14-16. Then there is a contradiction with the examples presented in Fig. 18 and 19. For example, in FIG. 18 in the case of PU1, it can be determined whether or not to use the information of PU0, and in FIG. 14-16, the PU0 information is not used because the characteristics of PU1 are different from those of PU0.

[00576] With regard to the above content, the candidate group can be configured without conflict according to the setting of the initial encoding/decoding stage. The motion information prediction candidate group may be defined in accordance with various other encoding/decoding settings.

[00577] A case of setting a group of candidate prediction modes is described in the above example. In addition, it may be set to restrict the use of reference pixels to predict the current block from adjacent blocks marked with exception conditions.

[00578] For example, if a prediction block is generated by distinguishing a first reference pixel from a second reference pixel according to an embodiment of the present invention, if the second reference pixel is included in a block marked as an exception, generating a prediction block using the second reference pixel may be limited. That is, the prediction block may be generated using only the first reference pixel.

[00579] Briefly, the above example can be considered as an item in the encoding/decoding settings related to the use of the second reference pixel.

[00580] Various cases will be described herein regarding the settings of the intra prediction mode candidate group.

[00581] In this example, it is assumed that there are 67 intra prediction modes configured as 65 directional modes and 2 non-directional modes, namely planar mode and DC mode. However, it is not limited to these, and other intra prediction modes may be set. In this example, it is assumed that six candidates are included in the MPM candidate pool. However, it is not limited to this, and the MPM candidate group may also be configured with four, five, or seven candidates. In addition, there is a setting in which there is no overlapping mode in the candidate group. In addition, priority refers to the order of determining inclusion in a group of MPM candidates, but may be considered as an element for determining binarization, entropy encoding/decoding settings, and so on. for each candidate belonging to the group of MPM candidates. The following example will be described by centering on the luma component, however the same or similar or modified applications can also be made to the chroma component.

[00582] The modes included in the intra prediction mode candidate group (e.g., MPM candidate group, etc.) according to the present invention can be configured as spatially adjacent block prediction modes, predefined prediction modes, prediction modes derived from modes predictions included in the candidates, etc. In this case, a rule (eg, priority, etc.) can be defined to configure the candidate group according to the encoding/decoding settings.

[00583] The following examples describe cases related to fixed configurations of a group of candidates.

[00584] As an example (1), a fixed priority for configuring a group of candidate intra prediction modes (in this example, a group of MPM candidates) can be maintained. For example, a predefined priority may be supported. For example, if spatially adjacent block prediction modes are added to a candidate group, the predetermined priority is the order left block (L3 in FIG. 13) - top block (T3 in FIG. 13) - bottom left block (BO in FIG. 13) - upper right block (R0 in Fig. 13) - upper left block (TL in Fig. 13). If the preset prediction mode is added to the candidate group, the preset priority is the planar-DC-vertical-horizontal-diagonal mode order. In addition, a combined configuration of the above example may take place, such as the order left block-top block-planar block-block DC-bottom left block-top right block-top left block. If the number of candidates cannot be reached even when executing the candidates according to the priority, the next priority may have a mode derived from the prediction mode that has been enabled (for example, +1 and -1 of the left block mode, +1 and -1 of the top block mode etc.), vertical mode, horizontal mode, diagonal mode, etc.

[00585] In the above example, if spatially adjacent block prediction modes are added to a candidate group, the candidate modes have left block-top block-bottom left block-top right block-top left block priorities, and left block prediction modes (L3 in FIG. 12) are sequentially included in the candidate group, at the same time, if the corresponding modules do not exist, the upper block prediction mode (T3 in FIG. 13) is included in the candidate group as the next priority. Thus, the prediction mode is included in the candidate group in some order, and if the corresponding block's prediction mode is not available or overlaps with the enabled mode, the order moves to the next block.

[00586] As another example, if spatially adjacent block prediction modes are added to a candidate group, the candidate group may be configured in the order left block-top block-bottom left block-top right block-top left block. In this case, the prediction mode of the left block first considers the prediction mode of the block located in L3, but if there is an unavailable or overlapping mode there, the candidate prediction modes of the left block are populated in the order of the next left block subblocks L2, L1, and L0. Similarly, the same or similar settings apply to the top block (T3-T2-T1-T0), bottom left block (B0-B1-B2-B3), top right block (R0-R1-R2-R3) and top left block (TL). For example, if block prediction modes cannot be added to the candidate group even if executed in L3-L2-L1-L0 order, the next consecutive block can be executed.

[00587] The following examples describe cases of adaptive candidate group configurations. The candidate group adaptive configuration may be determined according to the state of the current block (eg, block size and shape), the state of an adjacent block (eg, block size and shape, prediction mode, etc.), or the relationship between the current block and the adjacent block.

[00588] As an example (3), adaptive priority may be supported to configure a group of candidate intra prediction modes (in this example, a group of MPM candidates). The priority can be set according to the frequency. That is, prediction modes that occur more frequently may have a high priority, while prediction modes that occur less frequently may have a low priority.

[00589] For example, the priority may be set according to the frequency of prediction modes of spatially adjacent blocks.

[00590] A predefined priority for the same frequency case may be supported. For example, assuming that the modes appearing twice in the left box, top box, bottom left box, top right box, and top left box, respectively (assuming one prediction mode is obtained for each block), are mode 6 and mode 31, and mode a appearing once is mode 14, in this example (assuming mode a appearing in the previous consecutive block in the order left block-top block-bottom left block-top right block-top left block is mode 6 ) mode 6 and mode 31 are included as the first candidate and the second candidate.

[00591] A planar mode and a DC mode are included as the third and fourth candidates, and mode 14 having a frequency prediction mode of 1 is included as the fifth candidate. Then mode 5 and mode 7 derived from the first candidate and the second candidate and modes 30 and 32 can be given the next priority. In addition, the mode 13 and the mode 15 derived from the fifth candidate are given the next priority, and then the vertical mode, the horizontal mode, and the diagonal mode may have the next priority.

[00592] That is, the prediction mode having a frequency of 2 or more is given priority over the planar mode and the DC mode, and the prediction mode having a frequency of 1 can be given priority after the planar mode and the DC mode, for example, the derivative mode has low priority, preset mode, etc. the above example.

[00593] Briefly, a predetermined priority (e.g., the order left block-top block-planar block-block DC-bottom left block-top right block-top left block) may be a priority given considering the statistical characteristics of the overall image , and the adaptive priority (in this example, the case in which it is included in the candidate group according to the frequency) may be an example of performing a partial modification of the predetermined priority, taking into account some characteristic features of the image (in this example, the mode appearing with a frequency of two or more is set before the fixed planar mode and DC mode, and the mode appearing at a frequency of one is set after the fixed planar mode and DC mode).

[00594] In addition, according to the frequency, not only the priority with respect to the configuration of the candidate group can be determined, but the binarization and entropy encoding/decoding settings for each candidate belonging to the MPM candidate group can also be determined based on the frequency. . As an example, frequency-based binarization of MPM candidates having m frequencies can be determined. Alternatively, candidate context information may be adaptively determined according to frequency. That is, in this example, context information setting the mode selection probability as high may be used. That is, when m is 1-4- (in this example, omnidirectional mode is included in 2 out of a total of 6 candidates, and thus the maximum frequency value is 4), the context information may be set differently.

[00595] If 0 to 0 and 1 of a binary index (index Bin; bit order when more than one bit is configured according to binarization; for example, if some MPM candidate is configured as "010", by numbers from the first binary number to third binary number can be 0, 1, and 0) is a binary number indicating that the candidate is selected as an MPM (the binary number is used to determine whether the MPM is finite, or an additional binary number needs to be confirmed to recognize the final MPM; that is, in "010" above, if the first binary number is 0, it is necessary to confirm the second binary number and the third binary number, and whether the mode is MPM final can be confirmed; if there is one binary number, it is immediately confirmed whether the MPM is final, according to with binary values 0 and 1), and 1 is a binary number indicating that the candidate is not selected as an MPM, context information with a high probability of occurrence of 0 is applied (that is, when performing binary arithmetic, the probability of occurrence of 0 can be equated to 90%, the probability of occurrence of 1 can be equated to 10%, and it is assumed that the base case is such that the probabilities of occurrence of 0 and 1 are 60% and 40%), so CABAS can be applied.

[00596] As an example (4), adaptive priority may be supported to configure a group of candidate intra prediction modes (in this example, a group of MPM candidates). For example, the priority may be set according to the directionality of the prediction modes of spatially adjacent blocks.

[00597] In this case, the directionality categories are: mode group in the upper right corner direction (2-17 in Fig. 9), horizontal mode group (18 in Fig. 9), mode group in the lower right corner direction (19-49 in Fig. 9). Fig. 9), a vertical mode group (50 in Fig. 9), a mode group in the lower left corner direction (51-66 in Fig. 9), and a non-directional mode group (planar mode or DC mode). Alternatively, they can be divided into a group of horizontally directional modes (modes 2-34 in Fig. 9), a group of vertically directional modes (modes 35-66 in Fig. 9) and a group of non-directional modes (planar mode or DC mode), and also possible various configuration examples.

[00598] For example, if the priority of the base candidate group is the order top left block-planar block-DC block-bottom left block-top right block-top left block, then the candidate group may be configured in the above order. However, after the candidate groups are configured, the priorities for the binarization and entropy encoding/decoding settings for each candidate can be determined based on the category. As an example, binarization with fewer bits can be performed for categories in a candidate MPM group containing a large number of candidates. Alternatively, the context information may be adaptively determined according to the category. That is, the context information can be determined according to the number of modes included in each category (for example, if the first category is m and the second category is n, the context information is determined according to the combination of m and n).

[00599] As an example (5), adaptive priority may be maintained in accordance with the size and shape of the current block. For example, the priority may be determined according to the size of the block, and the priority may be determined according to the shape of the block.

[00600] If the block size is 32×32 or more, it can be included in the candidate group in the order left block-top block-planar block-block DC-bottom left block-top right block-top left block, and if the block size less than 32×32, it can be included in the candidate group in the order left block-top block-bottom left block-top right block-top left block-planar block-block DC.

[00601] Alternatively, if the block is square, it may be included in the candidate group in the order left block-top block-planar block-block DC-bottom left block-top right block-top left block if the block is rectangular (elongated horizontal), it can be included in the candidate group in the order top block-top right block-top left block-planar block-block DC-left block-bottom left block, and if the block is rectangular (vertically elongated), it can be included in the candidate group in the order left block - bottom left block - top left block - planar block - DC block - top right block. This example can be understood as the case in which blocks adjacent to longer blocks are of higher order.

[00602] As an example (6), adaptive priority may be maintained in accordance with the relationship between the current block and adjacent blocks.

[00603] With reference to image a in FIG. 23 shows an example of generating a prediction block according to the prediction mode of an adjacent block (in this example, a sub-block of the left block).

[00604] If the prediction mode (used, for ease of description, only with reference to directional content) of the left block (in this example, the upper sub-block of the left block) is the directional mode existing among the modes 51-66 in FIG. 9 (right tilt mode in vertical mode), and if a left block prediction block is generated, the reference corresponds to a diagonal part (TL or T part) in the figure. The right region of the left block may essentially be a region having a high correlation with the left region of the left block, however, since encoding/decoding has not been performed so far, a prediction block may be generated using the reference pixels of the lower region T where encoding/decoding has been completed .

[00605] Even if the prediction is performed from a part of the lower area of the block T as described above, the final prediction mode is defined as the mode existing among the modes 51-66, and this means that some areas 2400 of the current block also have the same or similar directivity (or edges, etc.) like the above prediction mode.

[00606] That is, it may mean that the probability of selecting the prediction mode of the left block as the prediction mode of the current block is high. Alternatively, this may mean that the probability of selecting modes 51-66 is higher.

[00607] With reference to image b in FIG. 23, when the corresponding description is derived from the above example, when the upper block prediction modes exist in the modes 2 to 17 in FIG. 9 (down slope mode in horizontal mode), this may mean that some areas of the current block also have the same directivity as the corresponding prediction modes.

[00608] That is, it may mean that the probability of selecting the prediction mode of the upper block as the prediction mode of the current block is higher. Alternatively, this may mean that modes 2-17 are more likely to be selected.

[00609] As described above, if it is possible to determine the prediction mode of the current block through the prediction modes of adjacent blocks, the priority may be determined adaptively.

[00610] For example, if the candidate base group is configured with a priority in the order left block-top block-planar block-block DC-bottom left block-top right block-top left block, the priority does not change in the case identical to image a in FIG. 23, and may change in the order top block-left block-planar block-DC block-bottom left block-top right block-top left block in case b in FIG. 23.

[00611] Alternatively, for case a in FIG. 23, the order left block-(left+1) block-(left-1) block-top block-planar block-block DC-bottom left block-top right block-bottom left block can take place, and for case b in FIG. 23, it may change in the order top block-(top-1) block-(top+1) block-left block-planar block-block DC-bottom left block-top right-bottom left block.

[00612] With reference to image a in FIG. 24 shows an example of generating a prediction block according to the prediction mode of an adjacent block (in this example, the left block).

[00613] In the case where the left block prediction mode is the mode having the directionality existing among the modes 51-66 in FIG. 9, its related pixels for generating the left block prediction block correspond to the diagonal portions TL and T in the figure. The right region of the left block may be a region having a high correlation with the left region of the left block, however, since encoding/decoding has not yet been performed, a prediction block may be generated using the reference pixels of the lower region of the block T where encoding/decoding has been completed.

[00614] Even if the prediction is performed from the lower region of the block T as described above, the final prediction mode is defined as the mode existing among the modes 51-66, and this means that some regions of the current block also have the same or similar directivity (or edges, etc.) like the above prediction mode.

[00615] That is, it may mean that the probability of selecting the left block prediction mode is high. Alternatively, this may mean that the probability of selecting modes 51-66 is high.

[00616] With reference to image b in FIG. 24, when the corresponding description is derived from the above example, when the upper block prediction modes exist in the modes 2 to 17 in FIG. 9, this may mean that some regions of the current block also have the same directivity as the corresponding prediction modes.

[00617] That is, it may mean that the probability of selecting the prediction mode of the upper block as the prediction mode of the current block is high. Alternatively, this may mean that the probability of selecting modes 2-17 is high.

[00618] As described above, if it is possible to determine the prediction mode of the current block through the prediction modes of adjacent blocks, the priority may be determined adaptively.

[00619] For example, if the candidate base group is configured with a priority in the order left block-top block-planar block-block DC-bottom left block-top right block-top left block, the priority does not change for case a in FIG. 24 and may change in the order top block-left block-planar block-DC block-bottom left block-top right block-top left block for case b in FIG. 23. In this case, as in case a in FIG. 24, the priority is the same as before, but instead, binarization and entropy encoding/decoding settings for the left block prediction candidate modes in the MPM candidate group can be adaptively determined. As an example, binarization (with shorter bit assignment) of the prediction candidate modes for the left block can be determined. Alternatively, the context information may be adaptively determined. That is, in this example, context information setting the probability of selecting a candidate as high can be used.

[00620] If 0 to 0 and 1 of some binary index is a binary number indicating that the candidate is selected as the MPM (in the case of determining whether it is the final MPM, using one binary number or confirming an additional binary number to recognize the final MPM ), and 1 is a binary number indicating that the candidate is not selected as an MPM, CABAC can be applied by applying contextual information that the probability of occurrence of 0 is high.

[00621] Alternatively, for case a in FIG. 24 can be left block-(left+1) block-(left-1) block-(left+2) block-(left-2) block-top block-planar block-block DC-bottom left block-top right block-bottom left block, and for case b in FIG. 24 it can change in the order top-block-(top-1) block-(top+1) block-(top-2) block-(top+2) block-left block-planar block-DC block-bottom-left block- top right-bottom left block.

[00622] With reference to image a in FIG. 25 shows an example of generating a prediction block according to the prediction mode of an adjacent block (in this example, the left block).

[00623] A case is considered in which the left block and lower left block prediction modes have a directionality existing among the modes 51-66 in FIG. 9. Left block and bottom left block may or may not belong to the same block. In this example, for ease of description, it is assumed that the left block and the lower left block belong to the same block (ie, the same mode) for which the division has been completed.

[00624] The pixels referred to in generating the prediction block (left+lower left) blocks correspond to the diagonal portions TL, T and TR in the figure. Even if prediction is performed from the lower region of the T block and the TR block as described in the above example, the end prediction mode is defined as the mode existing among the modes 51-66, which means that the entire area of the current block also has the same or similar directionality. , like the above prediction mode.

[00625] That is, this may mean that the probability of selecting the prediction mode of the left block or the lower left block (in this example, the prediction modes of these two blocks are the same) as the prediction mode of the current block is (very) high. Alternatively, this may mean that the probability of selecting modes 51-66 is high (in this example, the prediction modes of the two blocks are different).

[00626] In image b of FIG. 25, if the (upper+upper right) block prediction modes exist among the modes 2-17 in FIG. 9, this may mean that the entire area of the current block also has the same directivity as the respective prediction modes.

[00627] That is, it may mean that the probability of selecting the prediction mode of the top block or the top right block (in this example, the prediction modes of these two blocks are the same) as the prediction mode of the current block is (very) high. Alternatively, this may mean that the probability of selecting modes 2-17 is high (in this example, the prediction modes of the two blocks are different).

[00628] As described above, if it is possible to determine the prediction mode of the current block through the prediction modes of adjacent blocks, the priority may be determined adaptively.

[00629] For example, if the candidate base group is configured with priority in the order left block-top block-planar block-block DC-bottom left block-top right block-top left block, it may change in the order left block-bottom left block- planar block-block DC-top block-top right block-top left block for case a in FIG. 25. The priority may change in the order top block-top right block-planar block-DC block-left block-bottom left block-top left block for case b in FIG. 25. That is, the priority of the associated block can be pushed forward.

[00630] Alternatively, for case a in FIG. 25 priority can change in the order left block-bottom left block-(left+1) block-(left-1) block-bottom left+1) block-(bottom left-1) block-planar block-DC block-top block -upper right block-upper left block, and for case b in FIG. 25 priority can change in the order top block-top right block-(top-1) block-(top+1) block-(top right-1) block-(top right+1) block-planar block-block DC-bottom left block - top left block.

[00631] In addition, in the case of FIG. 25, based on the category described in the fourth example, the setting of binarization and entropy encoding/decoding of each candidate belonging to the MPM candidate group can be determined. As an example, binarization (ie, assignment of shorter bits) can be defined for MPM candidates (modes a and b of FIG. 24). Alternatively, context information may be adaptively determined according to the category. That is, in this example, the high probability setting context information of the category to which the mode belongs (modes a and b of FIG. 25) can be used.

[00632] If 0 to 0 and 1 of some binary index is a binary number indicating that a mode belonging to a category other than the category described above is selected as MPM, and 1 is a binary number indicating that a mode belonging to to the specified category, CABAS can be applied by applying contextual information indicating a high probability of occurrence of 1.

[00633] To confirm the above content, it is necessary to confirm whether a block located in a predetermined direction (left, top, top left, top right, and bottom left) is configured as a single (i.e., unsplit) block. Therefore, if a block is configured as a single block, then the case with the prediction mode of this block is executed, and if it is configured as a split block, then part or all of the prediction modes of each block are confirmed to execute this case. However, the complexity of this case can be significantly increased.

[00634] To this end, this process can be configured simply. It is assumed that the range of the current block is (a, b) (a+M-1, b+N-1). The left block contains pixels centered on the current block and belonging to the range (a-1, b) (a-1, b+N-1). The directionality of the left block prediction mode can be determined by confirming the prediction modes of some, but not all, pixels.

[00635] For example, the directionality of the prediction block is determined using the upper right (based on the current upper left block) coordinate prediction mode of the left block (a-1, b) and the lower right (based on the current lower left) side coordinate prediction mode left block (a-1, b+N-1). In this case, the left block may be configured as a single block and may be configured as a division block. Even if it is a single block or a division block, the prediction modes including coordinates may be the same. However, even if the prediction modes are not the same, if they are the prediction modes belonging to the modes 51 to 66 in FIG. 9 as described with the image a in FIG. 24, the prediction mode, the coding of these prediction modes may be given a high priority.

[00636] The above example assumes a left block case, however other directions apply similarly. In addition to the left block, in order to confirm the prediction mode of the lower left block, it is possible to confirm the prediction mode belonging to the coordinates of the lower left block, for example, (a-1, b+N).

[00637] Briefly, this can be used to refer to the configuration of a group of candidate intra prediction modes by confirming the prediction modes with respect to some coordinates of a block located in a predetermined direction, and confirming the prediction modes or directionality of the prediction modes in that direction. In this case, the number of coordinates may be an integer equal to 1 or more, such as 1, 2, and 3.

[00638] Summarizing, if a mode is confirmed for which a high correlation with the current block is determined, that mode can be given a high priority in the candidate group. Alternatively, a high priority in the candidate group may be assigned to prediction modes derived from a given mode (eg, adjacent prediction modes such as +1 or -1). That is, this mode may prioritize all or part of the prediction mode or predefined prediction modes (DC mode, planar mode, horizontal mode, vertical mode, diagonal mode, etc.) of other spatially contiguous blocks.

[00639] In contrast to the case described in the above example, if it cannot be determined whether there is a high correlation with the current block, a baseline-defined candidate group can be configured. That is, considering the configuration method of the candidate group of the high-correlation process, one of the plurality of configuration methods of the prediction mode candidate group or configured candidate groups can be selected based on state information (size, shape, block position, prediction mode, etc.) adjacent blocks.

[00640] FIG. 26 is a schematic diagram of the relationship between the current block and the adjacent block.

[00641] In the image a of Fig. 26 shows the encoding order according to the division block.

[00642] When performing quaternary tree partitioning, encoding is performed in the order top left block (0)-top right block (1) - bottom left block (2) - bottom right block (3). If sub-blocks are further divided based on a quaternary tree (depth 1), encoding is performed in the above order in each sub-block (depth 2), and encoding is performed before the depth information. For a top left box split at depth 1, there is an order top left box (0-0)-top right box (0-1)-bottom left box (0-2)-bottom right box (0-3) at depth 2 For a bottom right block split at depth 1, there is an order of top left block (3-0), top right block (3-1), bottom left block (3-2), and bottom right block (3-3) at depth 2. That is, encoding is performed in the x-y-z order represented by the depth information.

[00643] When performing binary tree partitioning, encoding is performed in the order left block(0)-right block(1), or top block(0)-bottom block(1) in the case of horizontal partitioning. If subblocks are further divided based on a binary tree (depth 1), encoding is performed in the above order in each subblock (depth 2), and encoding is performed before the depth information.

[00644] In the image a of Fig. 26, in the case where the block 3 is the same as in the image a in FIG. 25, and block 16 is the same as in image a in FIG. 23, block 20 corresponds to the same case as in image a in FIG. 24.

[00645] For example, block 3 is a case in which the left block is configured as one undivided block from the lower left block and has a prediction mode (one mode) in a specific direction or has prediction modes (more than one mode) in a specific direction, even if the left block is separated from the bottom left block.

[00646] Block 16 is a case in which the left block is configured as a plurality of blocks after division, and at least a portion of the subblocks has a prediction mode (one mode) in a specific direction.

[00647] Block 20 is a case in which the left block is configured as one block and has a prediction mode (one mode) in a specific direction, or the left block is configured as a plurality of sub-blocks after division and all sub-blocks have prediction modes (more than one mode) in a specific direction.

[00648] An example of a relationship between a current block and an adjacent block is depicted with image b in FIG. 26. If this relationship is defined as in image b of FIG. 26, it can be taken into account when configuring a group of candidate prediction modes.

[00649] Using the above examples, some examples of the relationship between the current block and the adjacent block are checked, however, various other cases are possible, and the configuration of the group of candidate intra prediction modes can be set according to the specific case.

[00650] In addition, in the examples above, various cases are tested with respect to the configuration of the intra prediction mode candidate group. The candidate group may be configured to apply one of the settings described in each embodiment, the settings described in each embodiment may be applied the same or similarly in other embodiments, and multiple settings may be applied to configure the candidate group. In addition, a candidate group configuration can be created by combining the above examples with additional settings.

[00651] In this example, assuming an example in which a non-MPM candidate mode group is configured as a plurality of candidate groups (group A and group B), the MPM candidate group prediction mode is not selected, and encoding/decoding is performed with using a candidate mode group (group A) in which the next of the non-MPM candidate mode group prediction modes is evaluated as the current block prediction mode.

[00652] In this example, it is assumed that 16 candidates are included in candidate group A and 45 candidates are included in candidate group B.

[00653] When configured with respect to the configuration of candidate group A, various methods may be used, and a prediction mode not included according to the MPM candidate group priority may be included in candidate group A.

[00654] Alternatively, Candidate Group A may be configured according to a predefined category regardless of the above example. For example, in FIG. 9, in the candidate group, prediction modes can be configured with some intervals. That is, blocks 2, 6, 10, 14, 18, etc. can be included in group A. in fig. 8. In this case, for group B, blocks 3, 4, 5, 7, 8.9, 11, etc. can be included in group B. Of course, the modes included in the group of MPM candidates are excluded from this setting.

[00655] The present invention is not limited to the above examples, and various modifications and other settings are also possible.

[00656] Briefly, the MPM candidate group is configured first, and then the candidate group A is configured. In addition, the remaining prediction modes may be configured as candidate group B.

[00657] A truncated unary binarization may be used for the MPM candidate mode, a fixed length binarization may be used for the candidate group A mode, and a truncated unary binarization may be used for the B candidate mode , although various other binarization methods can be used.

[00658] A mode in the MPM candidate group may perform encoding/decoding on all or part of the binary numbers using the legacy encoding section, and a mode representing a non-MPM candidate mode may perform encoding/decoding on all binary numbers using the bypass coding section. Alternatively, a mode from candidate group A in a non-MPM candidate mode group may perform encoding/decoding on a portion of binary numbers with a conventional encoding section, but is not limited to this, and various cases may be used.

[00659] The above is a description of the basic settings, and in the present description, many converted embodiments can take place.

[00660] A case in which a prediction block is generated using a plurality of reference pixels will be described below.

[00661] FIG. 27 shows an example of generating a prediction block using a plurality of reference pixel layers.

[00662] As the prediction value x pixels of the current block, a pixel value generated by applying a weighted average to a of ref_1 and b of ref_2 can be used.

[00663] Generating a prediction block using multiple reference pixel layers sometimes results in better prediction results. That is, it means that when generating pixel prediction values of the current block, a plurality of pixels can lead to more accurate prediction results.

[00664] However, ref_1 and ref_2 may be applied in the case where the current block has a high correlation with the edge direction or the like, and in some cases, using multiple reference pixel layers may lead to poor prediction results. For example, if the current block has a high correlation in the vertical direction, and if the top block has a high correlation in the diagonal direction, using multiple reference pixel layers may not achieve a good effect.

[00665] In this case, it is preferable to explicitly determine whether to use an existing reference pixel layer or to use multiple reference pixel layers.

[00666] Alternatively, using a plurality of reference pixel layers to generate a prediction block by confirming whether there is a high correlation between the reference pixel layers can be implicitly determined.

[00667] In the above example, it is shown that in intra prediction, division is performed in elements of predetermined sub-blocks, and prediction and transformation based on sub-blocks, etc. are performed. In this case, the sub-blocks belonging to the current block can share the intra prediction mode. In this case, by applying multiple reference pixel layers to sub-blocks in a block, better results can be achieved.

[00668] FIG. 28 shows an example of generating sub-blocks in the current block using multiple reference pixel layers. The current block is a 4M×4N block, the sub-block is 4M×N, and forward prediction is performed (top→down or left→right as a reference point, backward prediction is bottom→up, right→down), while the prediction mode using reference pixels whose prediction directions are top left block, top block, and top right block are k (#66 in FIG. 9).

[00669] With reference to FIG. 28, the 4M×N subblock containing x can perform intra prediction using the first previous subblock that has been encoded as the reference pixel layer. Alternatively, the subblock containing x may perform intra prediction using as the reference pixel layer the second previous subblock that has been encoded. Alternatively, the subblock containing x may perform intra prediction using the first and second previous subblocks as reference pixel layers (if a weighted value is applied to each reference pixel layer, the weighted value applied to the subblock containing x is greater than or equal to the weighted value of the other subblocks).

[00670] In general, the unconditional use of reference pixel layers does not always lead to the correct result if multiple reference pixel layers are used on a coding block basis. However, in the above example, the previous sub-block of the target sub-block used for encoding has the same direction as the target sub-block, thus leading to a better result. Therefore, to explicitly determine the execution or selection of one of the reference pixel layers, multiple reference pixel layer settings can be explicitly applied to sub-blocks in the current block. Alternatively, a plurality of reference pixel layers may be implicitly applied to first subblocks of the target subblock to generate a prediction value of the target subblock.

[00671] If a plurality of sub-blocks are respectively set to use different reference pixel layers with respect to a sub-block element, and it is desired to generate a prediction block using the above settings, different settings can be performed according to the positions in the current sub-block block.

[00672] For example, it is assumed that four horizontally divided sub-blocks are supported in the current block. Then the first sub-block may perform prediction and encoding using the nearest reference pixel of the current block, and the second sub-block may perform prediction and encoding using the nearest reference pixel of the current block and the data of the first sub-block.

[00673] In addition, the third subblock may perform prediction and encoding using the data of the first and second subblocks, and the fourth subblock may perform prediction and encoding using the data of the second and third subblocks.

[00674] The reference pixel filtering and reference pixel interpolation applicable to each reference pixel layer (or subblock) may use the same filter or may use different filters as described above. In this case, the filters differ in the number of filter taps and filter coefficients. In addition, filtering can also be applied in the same way or in different ways. Alternatively, the reference pixel filtering and the reference pixel interpolation are not performed separately, but may be implemented in the same filter. For example, when filtering the reference pixels using a 6-tap interpolation filter instead of a 3-tap low-pass filter and a 4-tap interpolation filter, filtering may not be performed twice.

[00675] As described above, various reference pixel configurations can be set, and then prediction can be performed using multiple reference pixel layers. Various other combinations may be described in various other embodiments of the present invention.

[00676] An example of generating a prediction block will be described below.

[00677] It was previously described that reference pixels for intra prediction can be separated into a variety of concepts. Instead of using a plurality of reference pixel layers, a prediction block may be generated or corrected using the first reference pixel and the second reference pixel according to the prediction mode.

[00678] Examples with different application methods according to the prediction mode are described. However, this example can be applied to the case where the current block is divided into a plurality of sub-blocks. That is, compared to using a reference pixel as the first reference pixel for an adjacent block of the current block, using a pixel located at a position opposite to the prediction mode direction or a position corresponding to the prediction mode direction as the second reference pixel if the current block is divided into a plurality of sub-blocks, the top block may be the first reference pixel. That is, in the present invention, the position of the first reference pixel can be changed in an example of generating a prediction block using a plurality of reference pixels.

[00679] Alternatively, the above method may not be used when the current block is divided into a plurality of subblocks and converted. The reason is that the use of the second reference pixel may result in low efficiency in arranging nearest pixels including previously completed encoding.

[00680] That is, performing encoding by dividing the current block (transformation block) into subblocks requires confirmation. Therefore, in the present invention, a prediction block can be generated or corrected using the first reference pixel and the second reference pixel according to the prediction mode if encoding is performed by prediction, transformation, etc. at the original block size without splitting. If encoding is performed after splitting, generating or correcting a prediction block using the second reference pixel is prohibited.

[00681] If, when performing encoding, the second reference pixel is used after division according to the subdivision/decoding settings, the corresponding process can be performed by correcting the position of the first reference pixel.

[00682] FIG. 29 is a schematic diagram for explaining the configuration of reference pixels for intra prediction. The size and shape (M×N) of the current block for performing prediction can be obtained from the block division section and described based on the following assumptions: a range of 4×4 to 256×256 is supported for performing intra prediction. In general, intra prediction may be performed on elements of a prediction block, however, intra prediction may be performed on elements of coding blocks and transform blocks according to the settings of the block division section. After confirming the block information, the reference pixel configuration section may configure the reference pixel used to predict the current block. In this case, the reference pixel processing may be performed by a temporary storage device (for example, <Array>, main array or auxiliary array, etc.), the reference pixel may be generated and removed during each intra block prediction process, and the size of the temporary storage may be determined according to the configuration of the reference pixel.

[00683] This example assumes that left box, top box, top left box, top right box, and bottom left box are used to predict the current block by centering on the current block. However, it is not limited to this, and a candidate block group having other configurations may be used to predict the current block. For example, a candidate group for adjacent blocks of a reference pixel may be an example following a raster or Z-scan, and part of the candidate group may be removed according to the scan order or configured to include other candidate block groups (e.g., additional top block configuration , bottom block, bottom right block, etc.). That is, the positions of adjacent blocks suitable as references may change according to the scan order or the like.

[00684] Alternatively, blocks (e.g., having corresponding coordinates according to the same coordinate or component compositional ratio in each color space) corresponding to the current block in other color spaces (e.g., if the current block belongs to Cr, other color spaces) can be used to predict the current block. spaces correspond to Y or Cb). In addition, for ease of description, the description is made on the assumption that one block is configured in a predetermined position (left, top, top left, top right, and bottom left), but at least one block may exist in the corresponding position. That is, at a predetermined position, there may be a plurality of sub-blocks obtained by dividing the corresponding block.

[00685] Briefly, the adjacent area of the current block may be the position of the reference pixel for intra prediction of the current block, and the position of the reference pixel according to the prediction mode may be further considered the area corresponding to the current block in a different color space. In addition to the above examples, the position of the reference pixel set according to the prediction mode, method, and so on can be determined. For example, if a prediction block is generated in a manner such as block matching, the position of the reference pixel can be considered to be an area in which encoding/decoding has been completed before the current block of the current image, or an area included in the search range in an area where encoding/decoding is completed (e.g. , including the left side, or top side, or top left side and top right side of the current block).

[00686] As shown in FIG. 29, a reference pixel for prediction of the current block (M×N) may be configured with adjacent pixels of the left block, top block, top left block, bottom right block, and bottom left block (Ref_L, Ref_T, Ref_TL, Ref_TR, and Ref_BL in FIG. 7).

[00687] In this case, adjacent pixels can be classified by at least one reference pixel layer: nearest pixel is ref_0 (pixel value difference is 1, from p(-1, -1) to p(2M-1, -1) , from p(-1, 0) to p(-1, 2N-1)), the next adjacent pixel (the difference in pixel values is 2, from p(-2, -2) to p(2M, -2), from p(-2, -1) to p(-2, 2N)) can be ref_1, and the next adjacent pixel (pixel value difference is 3, from p(-3, -3) to p(2M+1, -3 ), p(-3, -2) to p(-3, 2N+1)) may be ref_2, and so on. In more detail, ref_k represents a set of pixel values in which the difference of pixel values in adjacent blocks (in this example, Ref_TL, Ref_T, Ref_TR, Ref_L and Ref_BL) is equal to k+1. The description (or main description) regarding the use of the reference pixel set ref_k in this document is made based on the following assumption. Pixels having a pixel value difference of k+1 in Ref_TL, Ref_T, Ref_TR, Ref_L, and Ref_BL are used as reference pixels and are used for a prediction block, but the present invention is not limited to this. This means that in the set of reference pixels for predicting the current block, adjacent blocks may be the same or different.

[00688] For example, ref_0 (p(-1, -1) to p(2M-1, -1)) can also be used in Ref_TL, Ref_T and Ref_TR, ref_1 (p(-2, 0) to p(-2, N-1)) in Ref_L and used ref_2 (p(-3, N) to p(-3, 2N+1)) in Ref_BL.

[00689] In addition, one reference pixel set refers to the Ref_TL, Ref_T, Ref_TB, Ref_L, and Ref_BL reference pixel set (i.e., obtained from an adjacent block used for intra prediction of the current block), but in the following example, the reference pixel set may be configured by some combination of reference pixels (e.g., Ref_T+Ref_TR, Ref_L+Ref_BL, Ref_T+Ref_TL+Ref_L, etc., or there may be a case in which a block at each position is configured by sub-blocks, i.e., Ref_T0+Ref_T1 , Ref_L0+Ref_TL+Ref_T, etc.).

[00690] The present invention has been described on the assumption that there are no more than three reference pixel sets, however, it is not limited to this, and more reference pixel sets (k is 3 or more) may be used. In this case, the supported reference pixel set (or the maximum number of sets, referred to as the candidate reference pixel group in the following examples) may be determined according to encoding/decoding settings (e.g., size, block shape, prediction mode, image type, color component, and etc.).

[00691] In the present description, a case is assumed in which a subscript (incremented by 1 from 0) is assigned from the nearest set of reference pixels, but the present invention is not limited to this. In addition, the reference pixel configuration related information described below can be generated by the above-described index settings (binarization of assigning a short bit to a small index in the case of selecting one of the plurality of reference pixel sets, etc.), but the present invention is not limited to this.

[00692] Usually, the nearest pixel of an adjacent block (in this example, the layer ref_0 of the reference pixels) can be configured as a reference pixel, however, other pixels can be configured as reference pixels according to the encoding/decoding settings. If other reference pixel layers or at least one reference pixel layer is used, the prediction accuracy can be improved, thereby improving the coding efficiency. In this case, there may be various settings associated with setting the reference pixel layer.

[00693] For example, the reference pixel set used for intra prediction may be implicitly determined, or information associated with the reference pixel set used may be explicitly generated. In more detail, at least one preset pixel set may be configured as reference pixels, or at least one reference pixel set determined by reference pixel set selection information may be configured as reference pixels. With the above process, a set of reference pixels (eg, ref_0, ref_1, ref_2, ref_0+ref_1, ref_0+ref_2, ref_1+ref_2, ref_0+ref_1+ref_2) can be configured as reference pixels for the current block.

[00694] In addition, it is possible to distinguish between a case in which the number of reference pixel sets configured as reference pixels is 1 and a case in which the number of reference pixel sets configured as reference pixels is two or more. The first case represents a case in which pixels in one set of reference pixels are used to generate a prediction block, while the latter case represents a case in which at least one pixel is used to generate a prediction block (for example, if two or more are set as reference pixels). sets of reference pixels, a value obtained from the median, weighted average, etc. of pixel values (this value may also be a value derived from one or more pixels in each reference pixel set), in each reference pixel set may be a value derived from one or more pixels).

[00695] For example, a prediction block may be generated using a reference pixel obtained by weighted summing of the pixels in the first reference pixel set (ref_0 in this example) and the pixels in the second reference pixel set (ref_1 in this example). In this case, according to the prediction mode (eg, directional prediction mode), the pixels to which the weighted sum is applied in each reference pixel layer may be not only integer bin pixels but also decimal bin pixels. Alternatively, weighted values (eg, 7:1, 3:1, 2:1, 1:1, etc.) may be respectively applied to a prediction block obtained by using pixels in the first set of reference pixels as reference pixels, and a prediction block obtained by using the pixels in the second set of reference pixels to obtain a prediction block.

[00696] Assuming the case of explicitly generated reference pixel configuration information, reference information enabling adaptive reference pixel configuration (in this example, adaptive_intra_ref_pixel_enabled_flag) may appear in video, sequence, frame, slice, tile, etc. d.

[00697] If the indication information means that adaptive reference pixel configuration is enabled (in this example, adaptive_intra_ref_sample_enabled_flag=1), adaptive reference pixel configuration information (in this example, adaptive_intra_ref_sample_flag) may appear in frame, slice, tile, block, etc. .d.

[00698] If the configuration information means that the adaptive reference pixel configuration is enabled (in this example, adaptive_intra_ref_sample_flag=1), the associated reference pixel configuration information (e.g., reference pixel set selection information, etc., in this example, intra_ref_idx ) can appear in frame, slice, tile, block, etc. elements.

[00699] In this case, if the reference pixel adaptive configuration is not enabled or the configuration is not the reference pixel adaptive configuration, the reference pixel may be configured according to a predetermined setting. In general, this can be exemplified, but not limited to, a case in which the nearest pixel of an adjacent block is configured as a reference pixel, various cases may occur (for example, a case in which ref_0 and ref_1 are configured as reference pixels, and the prediction pixel value is generated by performing a summation of weighted values or the like with respect to ref_0 and ref_1, etc., i.e. the implicit case).

[00700] Furthermore, in addition to the above-described predetermined case (e.g., ref_0 case), related reference pixel configuration information (e.g., reference pixel set selection information, etc.) can be configured by, but not limited to, a group candidates (for example, ref_1, ref_2, ref_3, etc.).

[00701] In the above example, the description is made based on the hierarchical structure, starting from the top layer element, but is not limited to this. In addition, in some cases, related information may be implicitly defined. For example, as described above, in some cases, related information may appear according to encoding/decoding settings, in some cases, related information is implicitly determined, and in some cases, a combined case may occur.

[00702] Some reference pixel configurations are described using the above examples, which can be combined with various encoding/decoding information, etc. to define the settings for intra prediction. In this case, the encoding/decoding information contains the type (I/P/B) of the image, the color component, the shape and size of the current block, the prediction mode (prediction mode type (directional or non-directional)), the direction of the prediction mode (vertical, horizontal, diagonal 1, diagonal 2, etc.), etc., and intra prediction settings (in this example, reference pixel configuration settings) can be determined according to the encoding/decoding information of the adjacent block and the combination of encoding/decoding information current block and adjacent block.

[00703] FIG. 30 is a schematic diagram with respect to the configuration of reference pixels.

[00704] In the image a in Fig. 30 shows a case in which a reference pixel set is configured as a reference pixel, and the subsequent intra prediction process (reference pixel generation, reference pixel filtering, reference pixel interpolation, prediction block generation, post-processing filtering, etc.; some intra prediction processes may be configured adaptively according to the configuration of the reference pixel set) may be performed after configuring ref_0 as a reference pixel. For example, reference pixel set information is not generated for one reference pixel set, and an intra prediction process may be performed according to the non-directional mode.

[00705] On the image b in Fig. 30 shows a case in which at most two sets of reference pixels are set as reference pixels, and the subsequent intra prediction process can be performed after ref_0+ref_1 ("and" in the figure) are configured as reference pixels. For example, although there are multiple reference pixel sets, but no reference pixel set information is generated, the intra prediction process may be performed according to some prediction mode (in this example, top left to bottom left, or bottom left). to the top right corner).

[00706] In the image with in Fig. 30 shows a case in which at most one set of reference pixels is set as a reference pixel, and a subsequent intra prediction process can be performed after one of ref_0, ref_1, and ref_2 (or in the figure) is configured as reference pixels. For example, reference pixel set information is generated by a plurality of reference pixel sets, and intra prediction processing may be performed according to some prediction modes (upper left to lower right in this example).

[00707] Briefly, in images b and c of FIG. 30 shows cases in which two and three reference pixel sets are supported respectively (in other words, two and three candidate reference pixel groups respectively), image b shows at most two reference pixel sets configured by reference pixels (e.g., derivatives of from the weighted average of the pixel values from each set of reference pixels, etc.), and the image c represents at most one set of reference pixels configured by the reference pixels (for example, derived from the pixel values of the selected set of reference pixels from the three candidate groups).

[00708] FIG. 31 is a schematic diagram of a configuration of reference pixels.

[00709] In the image a in Fig. 31 shows a case in which the candidate reference pixel group is one set of reference pixels when the block size is 64×64 or more. On image b in Fig. 31 shows a case in which the candidate reference pixel groups are two sets of reference pixels (which means at most two sets of reference pixels for the reference pixels; that is, it is clear that both cases are possible, which are shown in images b and c in Fig. 30; the same case is shown below) when the block size is greater than 16×16 and less than 64×64. In the image from Fig. 31 shows a case in which the candidate reference pixel groups are three reference pixel sets (which means at most three reference pixel sets for the reference pixels) when the block size is less than 16×16. In the present example, since the current block has a larger size, the adjacent block size is more likely to be smaller, which may be the result of splitting due to different image characteristics. Therefore, in order to avoid performing prediction from pixels far away from the pixel value of the current block, it is assumed that since the block has a larger size, the number of reference pixel set candidates decreases, but the opposite case can also be assumed. The present invention is not limited to this. Various examples of changes can be implemented.

[00710] FIG. 32 is a schematic diagram with respect to the configuration of reference pixels.

[00711] FIG. 32 shows a case in which, when the current block is rectangular (horizontally elongated in this example), the candidate reference pixel groups are a plurality of reference pixel sets on a longer side, and the candidate reference pixel group is one reference pixel set on a shorter side. side. In this example, it is assumed that one set of reference pixels is configured on the shorter side, however, it is understood that there are fewer configurations than on the longer side, and there may also be a large number of configurations otherwise. This example is intended to prevent, but not limit, the case in which, when a block is rectangular, a reference pixel with a small length is often further away from the current block, thereby degrading prediction accuracy.

[00712] For example, in the case of a directional mode predicted using reference pixels on the upper and upper right sides of the block, prediction can be performed using multiple sets of reference pixels (both cases are possible, which are shown in images b and c of Fig. 30) , and in the case of a directional mode predicted using the reference pixels of the left and lower left sides of the block, the prediction can be performed using a single set of reference pixels. That is, there may be a case in which a set of reference pixels is determined adaptively according to adjacent blocks in one block.

[00713] As an example of a case in which one of the plurality of reference pixel sets is configured as reference pixels, when each reference pixel set is the same or similar, generating reference pixel configuration information (e.g., reference pixel selection information, resolution information adaptive reference pixels, etc.) may be a case in which the information appears repeatedly. For example, if it is determined that the distribution characteristics of the reference pixels (for example, they are determined by the average value and variance of each set of reference pixels; as an example, they can be determined by comparing the value obtained in the above process with the threshold value <Threshold>; that is, if it is determined that the same or nearly the same prediction block is generated, no matter which reference pixel set is used, related information may be omitted; in this example, this may be limited to the comparison made for the last selected prediction mode) of each reference pixel set are the same or similar, the reference pixel configuration information may be omitted. In this case, the reference pixels may be configured by a predetermined method (for example, the closest set of reference pixels). In aspects of determining whether to receive the configuration information of the reference pixels by receiving intra prediction information (for example, intra prediction mode information), the decoder may be the same as the encoder, and based on this determination, it may be determined whether to receive information about the configuration of the reference pixels.

[00714] While various examples show a case in which a plurality of reference pixel sets are configured as reference pixels, it is not limited thereto. There may be various converted examples that can be combined with other, additional configurations.

[00715] The reference pixel configuration section for intra prediction may include a reference pixel generation section, a reference pixel interpolation section, a reference pixel filtering section, and so on. and may have all or part of the configurations described above.

[00716] The reference pixel configuration section is configured to confirm the availability of the reference pixels in order to classify available reference pixels and unavailable reference pixels. For example, if a block at a predetermined position (or candidate block of a reference pixel) is available, then the corresponding block can be used as a reference pixel, and if it is not available, that block cannot be used as a reference pixel.

[00717] If at least one of the following conditions is satisfied, it is determined that the reference pixel is unavailable. For example, a reference pixel is determined to be unavailable if one of the conditions is met that the reference pixel is outside the frame boundary, does not belong to the same partition block as the current block (e.g., slice, tile, etc.), encoding /decoding is not completed and use is limited according to encoding/decoding. That is, if none of the above conditions is satisfied, it is determined that the reference pixel is available.

[00718] In addition, the use of reference pixels may be limited by encoding/decoding settings. For example, the use of reference pixels may be constrained in accordance with the performance of constrained intra prediction (eg, constrained_intra_pred_flag). Limited intra prediction may be performed when error-tolerant encoding/decoding needs to be performed with respect to external factors such as the communication environment if an attempt is made to prohibit the use of blocks recovered by referring to other images as reference pixels.

[00719] According to the availability of the reference pixel, the candidate reference pixel blocks can be classified into a case in which all of the candidate reference pixel blocks can be used, a case in which some of the reference pixels can be used, and a case in which all of the reference pixels not available. Except for the case in which all candidate blocks of reference pixels may be used, in other cases, reference pixels not available at the position of the candidate block may be filled or generated.

[00720] If a candidate reference pixel block can be used, a pixel at a predetermined block position (in this example, the pixel is assumed to be adjacent to the current block) may be included in the current block's reference pixel memory.

[00721] If a candidate reference pixel block cannot be used, the pixels obtained by the reference pixel generation process may be included in the reference pixel memory of the current block.

[00722] Briefly, if the reference pixel candidate block is in the available state, the reference pixel may be configured, and if the reference pixel candidate block is in the unavailable state, the reference pixel may be generated.

[00723] In the reference pixel filtering section, a low-pass filter (such as a 3-tap filter and a 5-tap filter, such as [1, 2, 1]/4 and [2, 3, 6, 3, 2]/ 16) to perform anti-aliasing as the main purpose, but other types of filters (such as a high pass filter) may be used according to the purpose of the filters (such as sharpening, etc.). In the present description, it is emphasized that filtering is performed for the purpose of smoothing to reduce degradation in quality during the encoding/decoding process.

[00724] The filtering performance of the reference pixels can be determined according to the encoding/decoding settings. That is, filtering may or may not be applied to the reference pixels. However, such a uniform filtering operation may have the disadvantage that some image characteristics may not be reflected, and it may be advantageous to improve the encoding performance by performing filtering based on information such as encoding/decoding information reflecting some image characteristics.

[00725] Image characteristics can be determined not only according to the image type (I/P/B), color component (Y/Cb/Cr), quantization parameter, encoding/decoding information of the current block (e.g., size, shape, division information, prediction mode, etc.), but also according to encoding/decoding information of adjacent blocks, a combination of encoding/decoding information of the current block and adjacent blocks, etc. In addition, determination (for example, determining whether the area of the reference pixel is a flat area or a discontinuous area, etc.) can be performed according to distribution characteristics of the reference pixels (for example, variance, standard deviation, etc. of the area of the reference pixels). Thus, if the information about the characteristics of the image can be confirmed, the application of filtering of the reference pixels can be set in advance.

[00726] Alternatively, the filtering of the reference pixels may be explicitly specified. For example, information about whether to apply filtering may be generated. In this case, filter selection information is not generated if there is one filter, or may be generated if there are multiple candidate filter groups.

[00727] Although the above examples have described implicit and explicit settings for filtering reference pixels, there may also be combined cases, for example, in some cases, the use of settings is explicit, and in other cases, the use of settings implicitly. In this case, "implicit settings" means that reference pixel filter information (eg, whether to apply filtering and filter category information) can be obtained in the decoder.

[00728] The reference pixel interpolation section is configured to generate reference pixels in the decimal element by linearly interpolating the reference pixels. Although the process is described in the present description as part of the reference pixel configuration section, it can be understood as a component capable of being included in the prediction block generation section and execution before generating the prediction block.

[00729] In addition, although the reference pixel interpolation section is assumed to be a separate component from the previous reference pixel filtering section, they can be configured to be combined into one component. Also, there may be a configuration in which distortion is generated in the reference pixel by applying a plurality of filters in the reference pixel interpolation section and the reference pixel filtering section.

[00730] In some prediction modes (e.g., horizontal, vertical, some diagonal modes <45 degree angle modes such as diagonal bottom right, diagonal bottom left, and diagonal top right>, non-directional modes, color modes, color copy modes, and etc.; that is, in modes where decimal interpolation when generating a prediction block is not required), the reference pixel interpolation process may not be performed, and it may be performed in other prediction modes (a mode where decimal interpolation when generating a prediction block is required).

[00731] Interpolation accuracy (for example, pixel units such as 1, 1/2, 1/4, 1/8, 1/16, 1/32, and 1/64) can be determined in accordance with the prediction mode (or directivity prediction mode). For example, in a 45-degree prediction mode, no interpolation process is required, and in a 22.5-degree or 67.5-degree prediction mode, 1/2 pixel interpolation is required. As described above, according to the prediction mode, at least one interpolation accuracy and a maximum interpolation accuracy can be determined.

[00732] For interpolation of the reference pixels, a predefined filter (e.g., a 2-tap linear interpolation filter, i.e., only this interpolation filter is used in any case), and one of a plurality of groups of candidate interpolation filters (e.g., 4-tap cube filter, 4-tap Gaussian filter, 6-tap Wiener filter, and 8-tap Kalman filter; that is, it is determined that an interpolation filter is used, applied according to the encoding/decoding settings). In this case, the interpolation filter can be distinguished by differences in the number of taps (i.e., the number of pixels to which filtering is applied), filter coefficients, and so on.

[00733] The interpolation may include execution examples by stage, eg, in order from low precision to high accuracy (eg, 1/2→1/4-1/8), as well as uniform execution examples. In the first case, interpolation refers to performing interpolation (i.e., it means that not only integer elements, but also decimal elements are used for interpolation) based on pixels of integer elements and pixels of decimal elements (in this example, with reference to pixels, for which has previously been interpolated based on a lower precision than the pixel currently to be interpolated). In the latter case, interpolation means that the interpolation is performed based on integer element pixels (i.e., it means that only integer element pixels are used for interpolation).

[00734] If one of the plurality of candidate filter section groups is used, the filter selection information may be explicitly generated or determined implicitly and may be determined in accordance with encoding/decoding settings (e.g., interpolation precision, block size and shape, mode predictions, etc.). In this case, the explicitly generated element corresponds to a video, sequence, frame, slice, tile, block, and so on.

[00735] The prediction block generating section is configured to generate a prediction block in accordance with at least one prediction mode, and is configured to use a reference pixel based on the prediction mode. In this case, the reference pixel may be used in methods (directional mode) such as extrapolation according to the prediction mode, and may be used in methods (non-directional mode) such as interpolation or averaging (DC) or copying.

[00736] In addition, the prediction block may be generated by copying the reference pixels. When copying the reference pixels, more than one prediction pixel may be generated by copying one reference pixel, or more than one prediction pixel may be generated by copying more than one reference pixel, and the number of copied reference pixels is less than or equal to the number of copied prediction pixels.

[00737] In addition, although the prediction block as a whole consists of one prediction block obtained by the prediction block generation process, the following example can be included: the final prediction block is obtained by obtaining a plurality of prediction blocks and applying to this plurality of prediction blocks weighted summation or the like. In this case, the expression "multiple prediction blocks" may refer to prediction blocks obtained from a set of reference pixels, and in other cases may refer to the same prediction block.

[00738] The prediction mode determination section performs the optimal mode selection process in the plurality of prediction mode candidate groups. In order to determine the mode with the optimal encoding cost to generate the number of bits of the corresponding mode, block distortion technology (eg, current block and reconstructed block distortion; SAD, SSD, etc.) and random distortion technology can be generally considered. The prediction block generated based on the prediction mode determined by the above process may be sent to a subtraction section and an addition section.

[00739] The prediction mode encoding section is configured to encode the prediction mode selected by the prediction mode determination section. In the prediction mode candidate group, index information corresponding to the prediction mode may be encoded, or the prediction mode may be predicted to encode information associated therewith. That is, the first case represents a direct prediction mode encoding method without performing prediction, and the latter case represents a prediction mode prediction method such that encoding mode prediction information and information obtained based on the prediction information are encoded. In addition, the first example is an example that can be applied to a color difference component, and the last example is an example that can be applied to a luminance component. They are not limited to this and can also be used in other cases.

[00740] In the case of performing prediction and encoding with respect to the prediction mode, the prediction value (or prediction information) of the prediction mode may be referred to as MPM. In this case, a predetermined prediction mode (e.g., DC mode, planar mode, vertical mode, horizontal mode, diagonal mode, etc.) or spatially adjacent block prediction mode (e.g., left block, top block, top left box, top right box, bottom left box, etc.). In this example, the diagonal mode may represent the diagonal up right, diagonal down right, and diagonal down left modes, which may correspond to mode 9, mode 34, and mode 66 in FIG. 9.

[00741] In addition, a mode derived from the modes included in the MPM candidate group may be configured as an MPM candidate group. For example, in the case of a directional mode included in the MPM candidate group, a mode with a mode interval difference a (e.g., a is a non-zero integer such as 1, -1, 2, or -2; in case FIG. 9 was mode 10 enabled, derived modes are mode 9, mode 11, mode 8, mode 12, etc.) is re-included (or further included) in the MPM candidate group.

[00742] The above example may correspond to the case in which the MPM candidate group is configured in a plurality of modes, wherein the configuration of the MPM candidate group (for example, the number of MPM candidate groups and the configuration priority) can be determined in accordance with the encoding/decoding settings (eg, a group of candidate prediction modes, an image type, a block size, a block shape, etc.) may be configured to include at least one mode.

[00743] A prediction mode for configuring a group of MPM candidates may take precedence. The order of the prediction modes included in the MPM candidate group may be determined according to the priority, and the configuration of the MPM candidate group can be completed if the number of MPM candidate groups is filled in accordance with the priority. In this case, the priority may be determined in the order of a spatially adjacent block prediction mode, a predetermined prediction mode, and a mode derived from the prediction mode originally included in the MPM candidate group, but other modifications are possible.

[00744] For example, in spatially contiguous blocks, the block prediction modes may be included in the candidate group in the order left-top-bottom-left-top-right-top-left block, and so on. In the predetermined prediction modes, the prediction modes may be included in the candidate group in the order DC-planar-vertical-horizontal mode, etc., and the prediction mode obtained by adding +1, -1, or t to the included mode is included in the candidate group. .P. Alternatively, the prediction modes may be included in the left-top-DC-planar-bottom-left-top-right-top-left-left+1-left-1-top+1 priority candidate group, and so on.

[00745] If the encoding of the prediction mode of the current block is performed using the MPM, information about whether the prediction mode matches the MPM (eg, most_probable_mode_flag) may be generated.

[00746] If the prediction mode is the same as the MPM (eg, most_probable_mode_flag=1), MPM index information (eg, mpm_idx) may be additionally generated according to the MPM configuration. For example, if the MPM is executed as a single prediction mode, no additional MPM index information is generated, and if the MPM is performed as a plurality of prediction modes, index information corresponding to the prediction mode of the current block may be generated in the MPM candidate group.

[00747] If the prediction mode does not match the MPM (e.g., most_probable_mode_flag=0), in the remaining prediction mode candidate group (or non-MPM candidate mode group) other than the MPM candidate group in the prediction mode candidate groups, non-MPM mode index information (eg, non_mpm_idx) corresponding to the prediction mode of the current mode may be generated, which may be an example of a case in which the non-MPM mode is configured as one group.

[00748] If a group of non-MPM candidate modes consists of a plurality of groups, information about which group the prediction mode of the current block belongs to can be generated. For example, the non-MPM mode consists of groups A and B, if the prediction mode of the current block is the same as the prediction mode of group A (for example, non_mpm_A_flag=1), index information corresponding to the prediction mode of the current block may be generated in group A candidates, and if there is no match (eg, non_mpm_A_flag=0), index information corresponding to the prediction mode of the current block may be generated in the remaining prediction mode candidate groups (or candidate group B). As in the above example, the non-MPM mode may be configured as at least one group (or set) of candidate prediction modes, and the configuration of the non-MPM modes may be determined according to the group of prediction candidate modes. . For example, if there are no more than 35 groups of candidate prediction modes, there may be one non-MPM mode, and in other cases, there are two or more non-MPM modes.

[00749] In this case, group A may configure as a candidate group a mode different from the MPM candidate group having a higher probability of matching with the prediction mode of the current block. For example, the next sequential prediction mode not included in the MPM candidate group may be included in group A, or a prediction mode having directionality from a predetermined interval may be included in group A.

[00750] If the non-MPM mode consists of a plurality of groups, as in the above example, the number of prediction modes is large, and the prediction mode is not predicted by MPM, the non-MPM mode is supported to reduce the number of mode bits.

[00751] If prediction mode encoding (or prediction mode decoding) of the current block is performed using MPM, the binarization table applied to each prediction mode candidate group (e.g., MPM candidate group, non-MPM candidate mode group) MPM, etc.) may be separately generated, and the binarization method applied to each candidate group may also be applied separately.

[00752] In the above example, terms such as "MPM candidate group" and "Non-MPM candidate mode group" are just some of the terms used in the present invention, and it is not limited to them. In particular, these terms represent information about the category to which the prediction mode belongs after dividing the prediction modes in the current frame into a plurality of categories, and information about the mode within the category. As other expressions, they can also be used for terms such as "major group of MPM candidates" and "auxiliary group of MPM candidates".

[00753] Prediction-related information that is generated by the prediction mode coding section may be transmitted to the coding section and included in the bitstream.

[00754] If more than one reference pixel set is configured as a candidate reference pixel group, a case in which one reference pixel set is used and a case in which more than two reference pixel sets are used can be distinguished below.

[00755] (1) Using one set of reference pixels.

[00756] In this example, it is assumed that when a plurality of reference pixel sets are set as candidate groups, one reference pixel set (one of ref_0, ref_1, and ref_2 in FIG. 29) is selected for intra prediction.

[00757] If intra prediction is performed using one set of reference pixels, the following situation may occur.

[00758] The prediction block P may be generated using ref* obtained using the reference pixel set ref and using a prediction pre-processing procedure such as generating the reference pixels, filtering the reference pixels, and interpolating the reference pixels.

[00759] For example, pixels located at ref are 5-tap low-pass filtered by the reference pixel filtering section, and pixels located at ref' (i.e., pixels after reference pixel filtering) are 6-tap low-pass filtered. linear interpolation filtering using the reference pixel interpolation section. In addition, the prediction block may be generated using ref* obtained in the above process.

[00760] The prediction pre-processing procedure using a single set of reference pixels is described below.

[00761] For example, when using the reference pixel set ref_2 in FIG. 29, if Ref_TL and Ref_TR positions can be used as reference pixels, and Ref_T position cannot be used as a reference pixel, pixels at positions P(0, -3) to P(n-1, -3) Ref_T can be generated by adjacent pixels at positions P(-1, -3) to P(n, -3) of the available reference pixels and are configured as reference pixels.

[00762] After configuring the reference pixels using the above process, components such as the reference pixel filtering section and the reference pixel interpolation section may be performed before generating a prediction block.

[00763] In the case of the reference pixel filtering section, the filtering can be applied to adjacent pixels centered on the target pixel, and generally, the larger the number of pixels that are applied to the filter, the more accurate the filtering effect (e.g., smoothing, sharpening, etc.) .) can be achieved.

[00764] In the case of the reference pixel interpolation section, filtering can be applied in adjacent pixels centered on pixels to be interpolated, and the larger the number of pixels that are applied to the filter, the more accurate filtering effect can be generated.

[00765] Both of these steps serve as a process for performing filtering, and depending on situations, unnecessary filtering effects may be generated. That is, when applying filtering in each step, in case of adverse effects when applying filtering in another step, the desired effect can also occur.

[00766] For example, in the process of interpolating reference pixels, interpolation target decimal pixels may apply filtering in adjacent integer pixels, if this example assumes that the previous step applies reference pixel filtering (i.e., applied to pixels of integer elements), a quantization error of each integer pixel can be reduced by filtering the reference pixels, or it is also possible to create a case in which the quantization error, on the contrary, is further included in the interpolated pixel by using this pixel in the reference pixel interpolation process.

[00767] (S1. Smoothing Ref)

[00768] m*=v0x1'+v1xm'+v2xn'

[00769] m*=m+k1

[00770] (m - original pixel, m* - filtered pixel, m' - pixel before filtering, 1 and n' - adjacent pixels before filtering, v - filter coefficient, and k1 - quantization error)

[00771] (S2. Interpolation process)

[00772] Z*=(w0xa'+w1xb'+w2xc'+w3xd,+w4xe,+w5xf)

[00773] z*=z+k2

[00774] (z is the original pixel at the specified position, z* is the interpolated pixel; a'-f is the adjacent <filtering failed> pixels, w is the filter factor, and k2 is the quantization error)

[00775] (S3. Smoothing Ref+interpolation process)

[00776] Z**=(w0xa*+w1xb*+w2xc*+w3xd*+w4xe*+w5xf*)

[00777] z**=z+k3

[00778] (z is the original pixel at the specified position and z** is the interpolated pixel; a*-f* are adjacent filtered pixels, w is the filter coefficient, and k3 is the quantization error)

[00779] In the above equations, the input pixel value to the reference pixel filtering process S1 is the adjacent pixel of the adjacent block for which encoding/decoding is completed, and the output value obtained by the above process is the pixel having a high probability of being close to the original pixel for reducing the quantization error introduced during encoding/decoding.

[00780] In addition, the input pixel value to the reference pixel interpolation process S2 is the adjacent pixel of the adjacent block for which encoding/decoding is completed, and the output value obtained by the above process is a pixel having a high probability of being close to the original pixel value in specified pixel position. With this process, weighted summation is applied to adjacent pixels of adjacent blocks including quantization error to obtain interpolated pixels (i.e., assuming no filtering is applied to reduce quantization error, quantization error can also be reduced by this process), and thus it is possible to obtain pixels that reduce the quantization error and are close to the original pixel. That is, not only decimal element pixels may be generated, but the effect of a low-pass filtering section may also take place.

[00781] On the other hand, in the case of S3, where the processes are performed sequentially, it is necessary to consider whether the effect that is desired to be obtained in the processes can be maintained. That is, it is due to the fact that the filtering performance (performing only one or two filters, etc.), the type of filter (using an improved filter or a slightly improved filter; that is, even if a satisfactory effect is generated in the process, this effect is not achieved in other processes or, conversely, an unsatisfactory effect is obtained), adding a filter tap, etc. do not necessarily guarantee a satisfactory filtering effect.

[00782] (2) Using more reference pixel sets.

[00783] In this example, more than two reference pixel sets (more than two of ref_1, ref_2, and ref_3 in FIG. 29) are selected for intra prediction when multiple reference pixel sets are configured as candidate groups.

[00784] If intra prediction is performed using multiple sets of reference pixels, the following situation may occur.

[00785]<1>As shown in image a of FIG. 33, the first prediction block may be generated by the reference pixel set ref_A. That is, the prediction block PA may be generated using ref_A* obtained using ref_A and using a prediction pre-processing procedure such as reference pixel generation, reference pixel filtering, and reference pixel interpolation. In addition, the prediction PB block may be generated using ref_B* obtained by the prediction pre-processing procedure with respect to ref_B.

[00786] The final prediction block P may be generated by the obtained plurality of prediction blocks.

[00787] For example, pixels in ref_A are 3-tap low-pass filtered by the reference pixel filtering section, and pixels in ref_A (i.e., pixels after the reference pixel is filtered) are 4-tap cube filtered by the reference pixels. In addition, 5-tap low-pass filtering is applied to pixels located in ref_B, and 6-tap linear interpolation filtering can be applied to pixels located in ref_B' (ie, pixels after reference pixel filtering).

[00788] In addition, each prediction block is generated using ref_A* and ref_B* obtained using the above process, and then the prediction block is generated by applying a weighted average (for example, applying a high weighted value, for example, 3:1, to the prediction block , adjacent to the current block), etc.

[00789]<2>As shown in image b of FIG. 33, a prediction block may be generated by the reference pixel sets ref_A and ref_B. In this case, the prediction block P may be generated using ref_A* obtained using ref_A and ref_B and using a prediction preprocessing procedure such as reference pixel generation, reference pixel filtering, and reference pixel interpolation.

[00790] For example, the pixels located in ref_A' can be 3-tap low-pass filtered by the reference pixel filtering section, and the pixels located in ref_A' can be 4-tap cubic filtered by the reference pixel interpolation section. In addition, pixels located in ref_B can be applied 5-tap low-pass filtering, and pixels located in ref_B' can be applied 2-tap linear interpolation filtering.

[00791] In addition, a prediction block may be generated after obtaining ref* using ref_A* and ref_B* obtained in the above process. That is, after obtaining the pixel values corresponding to the directivity of the intra prediction mode in each set of reference pixels (for example, pa in ref_A* and pb in ref_B*) according to the prediction mode, a weighted average (for example, pa×A+pb ×B) to generate a prediction block.

[00792] To generate a prediction block using the methods <1> and <2> (<1> is a method of applying a weighted average between prediction blocks after generating these prediction blocks in each set of reference pixels, <2> is a method of applying a weighted average between pixels in the process of generating a prediction block; that is, the same or similar results are obtained, but the process may differ) more than two sets of pixels may be used.

[00793] The prediction pre-processing procedure when using more than two sets of reference pixels is described below.

[00794] For example, when using the reference pixel sets ref_1 and ref_2 in FIG. 29, if Ref_TL and Ref_TR positions can be used as reference pixels, and Ref_T position cannot be used as reference pixels, pixels P(0, -1) to P(n-1, -1) at Ref_T position are generated at ref_1 by adjacent pixels in P(-1,-1) and P(n,-1) of the available reference pixels of the reference pixel set ref_1, and pixels from P(0,-2) to P(n-1,-2) at position Ref_T may be generated in ref_2 by adjacent pixels in P(-1,-2) and P(n,-2) from the available reference pixels of the reference pixel set ref_2.

[00795] Alternatively, the prediction block may be generated using the available pixels in ref_1 and ref_2 at the unavailable pixel positions in ref_1, and the prediction block may also be generated using the available pixels in ref_2 and ref_1, as described above, at the unavailable pixel positions in ref_2.

[00796] In the case of the reference pixel filtering section, filtering may be applied to adjacent pixels centered on the target pixel, and in the case of the reference pixel interpolation section, filtering may be applied to adjacent pixels centered on the pixel to be interpolated.

[00797] In this example, as also mentioned in the case of using one set of reference pixels in (1) to generate a prediction block, filtering applications create unwanted effects.

[00798] Following the procedure in (1), the following description is made under the assumption <1> in (2).

[00799] (S4. Intra prediction process)

[00800] P=v0×P1+v1+P2

[00801] The above formula represents the process of obtaining a final prediction block by performing a weighted summation on the prediction blocks obtained using each set of reference pixels. That is, it is necessary to take into account the part related to obtaining prediction blocks using the values that are obtained by the S3 process in each set of reference pixels and applying the weighted summation of the prediction blocks, whereby it is possible to improve the accuracy of the prediction block.

[00802] For the problems to be considered by (1) and (2), the method may be a solution in which one of the plurality of candidate groups is explicitly selected (for example, as the candidate group associated with the filtering of the reference pixels, one of the following is selected: <no filtering, or 3-tap filtering, or 5-tap filtering>, one of <4-tap filtering or 6-tap filtering> is selected as the candidate group associated with the interpolation of the reference pixels>), and the best way to determine implicitly is described below.

[00803] An intra prediction process in which a prediction block is generated using one set of reference pixels or using more than two sets of reference pixels is described by the following embodiments.

[00804] (3) Prediction pre-processing procedure when using one set of reference pixels

[00805] As shown in image a of FIG. 34, the predicted reference pixel may require temporary storage, which may include a main array or an auxiliary array. In this example, it is assumed that the reference pixels are given by the main array, as shown in the figure.

[00806] As shown in image a of FIG. 34, ref may be filled with pixels from the set of reference pixels used for prediction. In this case, to configure ref, the reference pixels may be passed through a reference pixel filtering section after the reference pixels are configured or generated, for example, by a reference pixel generation section. The reference pixels obtained by the reference pixel filtering section are represented as ref'. Then, the reference pixels obtained by performing the reference pixel interpolation process with respect to ref' are represented as ref*, and a prediction block may be generated using ref*.

[00807] *Integer element pixel consists of ref', decimal element pixel consists of ref* obtained by performing interpolation with respect to ref (image b in Fig. 12).

[00808] As described in the above example, for ref* (the integer element pixel of ref* is the same as ref'), the reference pixel filtering process and the reference pixel interpolation process are performed sequentially, so that when applying the integer element pixel obtained by With the help of these processes, in the interpolation of the decimal element pixels due to repeated filtering, a reference pixel with less high accuracy can be obtained. Therefore, the reference pixel filtering process and the reference pixel interpolation process need to be performed separately.

[00809] For example, the filtering of the reference pixels may be omitted, and only interpolation of the reference pixels may be performed. In this case, the integer element pixel used for prediction may be an existing adjacent pixel to which no reference pixel filtering has been applied, and the decimal element pixel may be a pixel obtained by applying interpolation to existing adjacent integer element pixels.

[00810] *An integer element pixel consists of ref', a decimal element pixel consists of ref* obtained by performing interpolation using ref (image c in Fig. 34).

[00811] Alternatively, if interpolation is performed on decimal element pixels, ref can be used as a reference pixel instead of ref'. In this case, the integer element pixel used for prediction may be the pixel to which no reference pixel filtering has been applied, and the decimal element pixel may be the pixel to which interpolation is applied by using existing adjacent pixels as pixels prior to performing the reference pixel filtering.

[00812] *Integer element pixel consists of ref', decimal element pixel consists of ref* obtained by performing interpolation using ref (image d in FIG. 34).

[00813] The number of filters applied to the pixels at each position in ref is depicted by the images b d in FIG. 34. In images c and d in FIG. 34 shows examples of configuring a reference pixel by applying no more than one filter.

[00814] (4) Prediction pre-processing procedure when using more than two sets of reference pixels

[00815] For ease of description, assuming the case <2> in (2) of the preceding example, two sets of reference pixels are used to generate a prediction block.

[00816] After configuring ref_A, ref_A' can be obtained by the reference pixel filtering section, and after performing the reference pixel interpolation process with respect to ref_A', ref_A* can be obtained. Further, in the case of ref_B, ref_B* can be obtained by ref_B', and ref_A* and ref_B* (ie, ref* in image b of FIG. 34) are used to generate a prediction block. In this case, the prediction block may be generated using integer element pixels or decimal element pixels in ref_A* and ref_B* according to the prediction mode.

[00817] As described above, for ref_A* and ref_B*, the reference pixel filtering process and the reference pixel interpolation process are performed sequentially, so that by applying the integer element pixel obtained by these processes to the decimal element pixel pixel interpolation due to re-filtering, a reference pixel with less high precision. In addition, since pixels located in more than two reference pixel sets may be further subjected to weighted summation filtering or the like, it is possible to filter the pixels too many times for final prediction.

[00818] Various combinations of reference pixels for prediction pixels are shown below. For example, prediction pixels may be generated using two integer element pixels in ref_A and ref_B according to the directivity of the prediction mode (e.g., vertical mode, horizontal mode, etc.), prediction pixels may be generated using one integer element pixel and one pixel decimal element (eg, directional mode, etc.), and prediction pixels may be generated using two decimal element pixels (eg, directional mode, etc.). This can be confirmed by images b-d in FIG. 35.

[00819] In the above case, as shown in image b of FIG. 34, assuming that applying filtering to a single reference pixel set is considered as a case in which a weighted average is applied with other reference pixel sets, the number of filters applied in image b in FIG. 35 may be 2, the number of filters applied in image c in FIG. 35 may be 3, and the number of filters applied in image d in FIG. 35 may be 4.

[00820] For the above case, the following example can be applied.

[00821] For example, in each set of reference pixels, the filtering of the reference pixels may be omitted, and only the reference pixel interpolation process may be performed. In this case, the integer element used for prediction may be an existing adjacent pixel to which no reference pixel filtering has been applied, and the decimal element pixel may be a pixel obtained by applying interpolation to existing adjacent integer element pixels.

[00822] In particular, in the case of reference pixel filtering, applying filtering to pixels similar to the target pixel can more effectively reduce quantization error. In this example, prediction is performed using more than two reference pixel sets, which means that directionality of the current block also exists at the boundary of the block containing the reference pixel (i.e., more than two reference pixel sets adjacent to the current block). Therefore, as before, filtering is performed using only the nearest pixel, regardless of the directionality of the current block prediction mode, which is similar to filtering potentially different pixels based on the values of the target pixel and the source pixel.

[00823] Alternatively, the reference pixel filtering and reference pixel interpolation processes may be performed for each set of reference pixels. However, as shown in image b of FIG. 35, if the weighted average is only applied to integer element pixels, pixels to which the reference pixel filtering is applied can be used, and as shown in the image c of FIG. 35, if the weighted average is applied to integer element pixels and decimal element pixels, the integer element pixels used may use pixels to which reference pixel filtering has not been applied.

[00824] With the above examples, the prediction pre-processing procedure is shown for a case in which one set of reference pixels is used and in a case in which more than two sets of reference pixels are used.

[00825] In the above examples, although some cases are described in the case where it is assumed that adverse effects occur due to the repeated use of the filter, it cannot determine only one characteristic of the image, and there may be other options.

[00826] For example, for cases (3) and (4), it can be used either in a fixed manner (for example, regardless of a state such as the shape and size of the current block and the prediction mode), or adaptively (for example, determined according to the state of the current block).

[00827] In the case of adaptive use of (3) and (4), the settings in (3) and (4) may be applied in whole or in part, and in some cases the reverse example may also be modified or applied.

[00828] As an example, in some modes (eg, vertical, horizontal, diagonal 45 degrees, 135 degrees, etc.) of case (3), 5-tap low-pass filtering can be performed. In this example, since the interpolation process in decimals is not required, fine filtering can be performed to reduce quantization error (assuming that filtering is more accurate with more taps in this example).

[00829] Alternatively, in some modes (eg, directional mode, etc.), 3-tap low-pass filtering and 6-tap Gaussian interpolation filtering can be performed in (3). In this example, a decimal element interpolation process is required, and simple filtering can be performed by applying filtering to integer element pixels (for example, assuming that in the following cases, filtering is more simple, that is, the types of filters are different, the number of pixels to which the filter is applied, is reduced and the number of taps is small) to reduce quantization errors of multiple filterings (in this example, reference pixel filtering is applied to integer element pixels, and filtering is applied when these pixels are used to interpolate decimal element pixels). That is, a case like b in FIG. 34 (although described as an existing case when describing image b in FIG. 12, this example describes that case b can be applied instead of cases c and d in FIG. 34).

[00830] As an example, in (3), in the case of a block (for example, 32×32) that is larger than a predetermined size, 5-tap low-pass filtering may be performed and 4-tap cubic interpolation filtering may be performed, but the pixels applied when interpolated, may include pixels before low-pass filtering (i.e., integer element pixels) as input pixels. In addition, in the case of a block that is smaller than the predetermined size, the same reference pixel filtering and reference pixel interpolation processes that are identical to the above process are performed, however, the pixels applied in the interpolation may include pixels after low-pass filtering as input pixels.

[00831] As an example, in some modes (e.g., vertical, horizontal, etc.) in (4), 5-tap low-pass filtering can be performed on each set of reference pixels (i.e., using a more accurate filter than when using a 3-tap filter in other directional modes or using other filter types even if they contain the same number of taps). In this example, since no interpolation is required, filtering can be performed to reduce the quantization error. Alternatively, 5-tap low-pass filtering may be performed on the reference pixel set adjacent to the current block, and 3-tap low-pass filtering may be performed on the next adjacent reference pixel set. In this example, a set of reference pixels can be fine filtered (in this example, the weighted values are assumed to be added to the pixels near the current block). The reference pixel set assigns higher weighted values to more than two reference pixel sets. Alternatively, 5-tap low-pass filtering is performed on the reference pixel set adjacent to the current block, and no low-pass filtering is performed on the next adjacent reference pixel set. It will be understood that in this example, by centering on pixels in the higher weighted reference pixel set, low-pass filtering is applied to pixels adjacent to the reference pixel set to which the pixel belongs and to pixels adjacent to a pixel in other sets. reference pixels (i.e., if the case according to the prior art is one-dimensional filtering, in this example it is clear that two-dimensional filtering takes place).

[00832] In addition, in some modes (for example, directional mode, etc.) in (4), 3-tap low-pass filtering is performed on each reference pixel set, 8-tap cubic interpolation filtering is performed on the reference pixel set adjacent to the current block, and 2-tap linear interpolation filtering may be performed on the next adjacent set of reference pixels. Alternatively, instead of performing low-pass filtering on each reference pixel set, 6-tap Gaussian interpolation filtering is performed on the reference pixel set adjacent to the current block, and 4-tap Gaussian interpolation filtering may be performed on the next adjacent reference pixel set. Alternatively, 5-tap low-pass filtering may be performed on the reference pixel set adjacent to the current block, 3-tap low-pass filtering may be performed on the next adjacent reference pixel set, and 4-tap cubic interpolation filtering may be performed on each reference pixel set. pixels. Alternatively, 3-tap low-pass filtering is performed on the reference pixel set A adjacent to the current block, no low-pass filtering is performed on the next adjacent reference pixel set B, 8-tap Gaussian interpolation filtering is performed on the reference pixel set A adjacent to the current block, and 4-tap cubic interpolation filtering is performed on the next adjacent set B of reference pixels.

[00833] As an example, in (4), in the case of a block having a predetermined or larger size (for example, 64×64), 4-tap cubic interpolation filtering can be performed without performing low-pass filtering on each set of reference pixels. In addition, in a block that is smaller than a predetermined size and larger than a predetermined or larger size (for example, 16×16-64×64), 3-tap low-pass filtering as well as 4-tap Gaussian interpolation filtering are performed for each set of reference pixels, but when the weighted average pixels consist of only a set of integer element pixels, pixels after low-pass filtering are used, and when the weighted average pixels consist of a combination of integer elements and decimal elements, integer element pixels can use pixels before low-pass filtering. In addition, in a block that is smaller than the predetermined size (16×16), 5-tap low-pass filtering and 6-tap cubic interpolation filtering are performed for each set of reference pixels, and the pixels to be weighted average, regardless of whether they are integer elements or decimal elements can use pixels after low-pass filtering.

[00834] According to the above example, various cases are shown in which one reference pixel set is used, and more than two reference pixel sets are used, and so on.

[00835] Explicit processing may be performed on the prediction pre-processing procedure. That is, it is possible to generate one kind of selection information about using one set of reference pixels or using more reference pixels, it is also possible to generate another kind of selection information about whether filtering settings of the reference pixels are not performed, or filtering is performed using filter A or filter B, and it is possible to generate another kind of selection information regarding the use of filter A or filter B for reference pixel interpolation settings.

[00836] Alternatively, implicit processing may be performed on the prediction pre-processing procedure. That is, the number of reference pixel sets, the reference pixel filtering settings, and the reference pixel interpolation settings described above can be implicitly determined according to encoding information (states) such as block size, block shape, and prediction mode. In this case, if the number of reference pixel sets is determined to be two or more, it may be implicitly determined that the same reference pixel filtering setting or reference pixel interpolation or other setting is performed on the plurality of reference pixel sets.

[00837] Alternatively, mixed processing may be performed on the prediction pre-processing procedure. That is, implicit processing may be performed in some cases, and explicit processing may be performed in other cases. In this case, according to the encoding information, a mixed processing reference can be determined. That is, the implicit processing or the explicit processing may be determined according to a combination of at least one or more of the determined block size or shape or prediction mode.

[00838] In the present description, the case of implicit processing or mixed processing will be described. For both processes, the intra prediction setting can be configured based on coding information (e.g., prediction preprocessing, etc., which has been described above as having an adaptive configuration only in prediction preprocessing, but which also has an adaptive configuration in other processes. intra prediction mode <an intra prediction mode candidate group is configured in the following content, that is, determined from coding information such as the number of intra prediction modes and the prediction direction>).

[00839] The specified content is arranged as follows.

[00840] (Encoder)

[00841]<1> The size and shape of the current block are confirmed

[00842] - Setting up an intra prediction mode candidate group (i.e. determining the number of modes, angles, etc., which can be omitted if there is one candidate group)

[00843]<2> The size, shape, and prediction mode of the current block are confirmed (at this point, the prediction mode is determined by the intra prediction mode candidate group determined in the above process)

[00844] - Setting the reference pixel set (i.e. using one or more than two sets)

[00845] - Reference pixel filtering setting (if there are more than two reference pixel sets, a filtering setting can be defined in each reference pixel set)

[00846] - Reference pixel interpolation setting (similarly, if there are more than two sets of reference pixels, interpolation settings can be defined in each set)

[00847]<3> Encoding information related to intra prediction (if selection information is generated due to the presence of multiple candidate groups in the prediction mode and settings of the above process, this information is also included in the intra prediction information).

[00848] (Decoding)

[00849]<1> The size and shape of the current block are confirmed (in the previous step, the block information is parsed by the block dividing section and related information is confirmed)

[00850] - Intra prediction mode candidate group setting (prediction mode information is parsed and reconstructed according to the candidate group)

[00851]<2> The size, shape and prediction mode of the current block are confirmed

[00852] - Adjustment of reference pixels

[00853] - Reference pixel filtering setting

[00854] - Reference pixel interpolation setting

[00855]<3> Decoding information related to intra prediction (prediction mode and other optional information; it can be considered that analysis of this information is performed at this stage).

[00856] In the encoder, a group of candidate intra prediction modes may be confirmed in accordance with the state of the current block. That is, it is determined which candidate group is used, and the determined candidate group contains a specific number of modes and specific mode configurations. Confirmation is trivial if all conditions of the intra prediction mode candidate groups are the same, but confirmation is required if more than two intra prediction mode candidate groups are used. That is, which candidate group is used can be known according to the state of the current block, or which candidate group is used can be explicitly known by information.

[00857] Numbers of prediction modes, etc. are assigned according to the specified process, but if the groups of candidate prediction modes are different, directions or angles, etc. of each prediction mode are different according to the candidate groups, and therefore a process of unification of the candidate groups may be required. That is, #50 when using the prediction mode in FIG. 9 may be the vertical mode, however, #50 in the other candidate prediction mode group is not necessarily the vertical mode, so confirmation is required.

[00858] Therefore, this document describes that operations, settings, etc. vary depending on the type of prediction mode.

[00859] For example, when the conditional clause "if (intramode=Vertical)" is used to determine whether intra-prediction processing (e.g., reference pixel interpolation, reference pixel filtering, etc.) is used if "intramode" is #50, and the prediction mode candidate group is used, it can be directly determined to perform intra prediction processing, but since it is necessary to know the mode type in the intra prediction mode candidate group (directed mode or non-directional mode; if it is a directional mode, it is determined whether it belongs to vertical or horizontal mode, and if it is not a directional mode, it is determined whether there is a directional mode for generating a prediction block using only integer pixels, etc.), it can be entered into a conditional sentence after confirming the direction of the above mode in the mode group - prediction candidates. That is, if (intramode<x>=Vertical), it should be changed to such a process that mode x is vertical in the mode x intra-prediction candidate mode group.

[00860] In addition, to confirm the intra prediction settings, it is necessary to confirm the state of the current block. In this case, the state of the current block may be determined according to at least one or more of size, shape, prediction mode, and so on. current block. Reference pixel set, reference pixel filtering, reference pixel interpolation, etc. are set according to the state of the current block. That is, the number of reference pixel sets, which reference pixel set to use, whether to apply reference pixel filtering, and which filter to use can be determined. If the associated information is explicitly specified, a flag may be generated. If the related information is implicitly defined, it can be determined according to the encoding information (the above-described state of the current block or type <I/P/B> of the current picture, the current picture <Y/Cb/Cr>, etc., which may affect the intra-prediction settings).

[00861] Although the present invention has been described with a focus on the reference pixel set, reference pixel filtering, and reference pixel interpolation, other intra prediction-related settings may also be included. For example, in the prediction process, as before, according to the encoding information, it can also be determined that a prediction block is generated using (extrapolated, averaged, and interpolated) reference pixels according to the prediction mode, or to generate a prediction block, etc. d. different reference pixels are used than the reference pixels used according to the existing directionality (for example, in order to correct the corresponding prediction pixels according to the position of the prediction pixel, the correction value is generated by additionally using other reference pixels and then added to the existing prediction value, etc. .d.), etc. In this case, if the related information is explicitly defined, it can also be known by the related information about the prediction method used, otherwise, in the case of implicit definition of the related information, which prediction method to use can be determined by the state of the current block or t .P.

[00862] In the decoder, the description associated with it can be obtained in reverse order relative to the contents of the encoder. In the case of an explicit part, the related information is parsed. In the case of the implicit part, the related information is determined according to the encoding information, etc.

[00863] In the above description, the reference pixel filtering setting and the reference pixel interpolation setting may be determined in view of the setting change content with the number of reference pixel sets used in the present description.

INDUSTRIAL APPLICABILITY

[00864] The present invention can be applied to encoding/decoding an image signal.

Claims (18)

1. The method of intra prediction, including: obtaining an intra prediction mode of the current block; determining a pixel line for intra prediction of the current block among the plurality of pixel lines; selectively performing filtering on the first reference pixel of the determined pixel line based on the first flag indicating whether the first reference pixel has been filtered for intra prediction; And performing intra prediction of the current block based on the intra prediction mode and the determined pixel line, wherein the first flag is received from the device for decoding based on the encoding parameter of the current block, and the encoding parameter contains whether intra-prediction has been applied in the sub-block elements. 2. The method of intra prediction according to claim 1, characterized in that the encoding parameter further comprises a block size, a color component type, or an intra prediction mode. 3. The method of intra prediction according to claim 1, characterized in that it further includes: performing correction on the prediction pixel of the current block according to the intra prediction. 4. The intra prediction method of claim 3, wherein the step of performing the correction further comprises: determining at least one of the second reference pixel or a weighted value for correction based on the position of the prediction pixel of the current block. 5. The method of intra prediction according to claim 4, characterized in that the correction is selectively performed considering at least one of a pixel line position of the current block, an intra prediction mode of the current block, or whether intra prediction has been performed on subblock elements of the current block. 6. The method of intra prediction according to claim 1, characterized in that intra prediction is performed on subblock elements of the current block, and the sub-blocks are determined based on at least one of the second flag indicating whether to perform division of the division direction information or division number information. 7. The intra prediction method of claim 1, further comprising: obtaining an intra prediction mode based on a predetermined default mode or a set of most likely candidate modes (MPM).

Images