KR102443381B1 - Method and apparatus for processing a video signal - Google Patents

Abstract

An image decoding method according to the present invention includes the steps of determining whether a block deviating from a boundary of a current picture or a face can be used as a reference block, determining a reference block of the current block based on a result of the determination, and The method may include generating a prediction block of the current block by using the reference block.

Description

Video signal processing method and apparatus {METHOD AND APPARATUS FOR PROCESSING A VIDEO SIGNAL}

The present invention relates to a video signal processing method and apparatus.

Recently, the demand for high-resolution and high-quality images such as HD (High Definition) images and UHD (Ultra High Definition) images is increasing in various application fields. As the image data becomes higher resolution and higher quality, the amount of data increases relatively compared to the existing image data. The storage cost will increase. High-efficiency image compression techniques can be used to solve these problems that occur as image data becomes high-resolution and high-quality.

Inter-screen prediction technology that predicts pixel values included in the current picture from pictures before or after the current picture with image compression technology, intra-picture prediction technology that predicts pixel values included in the current picture using pixel information in the current picture, Various techniques exist, such as an entropy encoding technique in which a short code is assigned to a value with a high frequency of occurrence and a long code is assigned to a value with a low frequency of occurrence.

Meanwhile, as the demand for high-resolution images increases, the demand for stereoscopic image contents as a new image service is also increasing. A video compression technique for effectively providing high-resolution and ultra-high-resolution stereoscopic image content is being discussed.

An object of the present invention is to provide a method and apparatus capable of effectively encoding/decoding a 360-degree projection image in encoding/decoding a video signal.

An object of the present invention is to provide a method and apparatus capable of performing padding at an image boundary or a face boundary in encoding/decoding a video signal.

An object of the present invention is to provide a method and apparatus for performing padding in consideration of similarity between boundary samples in encoding/decoding a video signal.

An object of the present invention is to provide a method and apparatus for performing motion compensation by substituting a reference region deviating from an image or face boundary in encoding/decoding a video signal.

An object of the present invention is to provide a method and apparatus for performing motion compensation using a sample generated due to padding or interpolation in encoding/decoding a video signal.

The technical problems to be achieved in the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those of ordinary skill in the art to which the present invention belongs from the description below. will be able

A video signal decoding method and apparatus according to the present invention determine whether a block outside the boundary of a current picture or phase can be used as a reference block, and determine a reference block of the current block based on the determination result, A prediction block of the current block may be generated using the reference block.

A video signal encoding method and apparatus according to the present invention determine whether a block deviating from a boundary of a current picture or a face can be used as a reference block, and determine a reference block of the current block based on the determination result, A prediction block of the current block may be generated using the reference block.

In the method and apparatus for encoding/decoding a video signal according to the present invention, when the reference block is located adjacent to a first boundary of an image or a phase, the reference block includes a first sample adjacent to the first boundary or the first A sample generated based on at least one of second samples adjacent to a second boundary having proximity to the boundary may be included.

In the method and apparatus for encoding/decoding a video signal according to the present invention, the reference block may include a sample generated by interpolating the first sample and the second sample.

In the method and apparatus for encoding/decoding a video signal according to the present invention, the second boundary may be adjacent to the first boundary when the current picture is inversely projected into a 3D space.

In the method and apparatus for encoding/decoding a video signal according to the present invention, when the reference block is located adjacent to a first boundary of an image or a face, the reference block is a first sub-block adjacent to the first boundary or the It may include at least one of the second sub-blocks adjacent to the second boundary having adjacent to the first boundary.

In the method and apparatus for encoding/decoding a video signal according to the present invention, information indicating whether a block outside the boundary of the current picture or phase can be used as a reference block may be signaled through a bitstream.

The features briefly summarized above with respect to the invention are merely exemplary aspects of the detailed description of the invention that follows, and do not limit the scope of the invention.

According to the present invention, it is possible to effectively encode/decode a 360-degree projection image.

According to the present invention, there is an advantage that encoding/decoding efficiency at the boundary can be improved by performing padding on the image boundary or the face boundary.

According to the present invention, there is an advantage that encoding/decoding efficiency at the boundary can be improved by performing padding in consideration of the similarity between boundary samples.

According to the present invention, there is an advantage in that encoding/decoding efficiency can be improved by performing motion compensation using a reference region deviating from an image or face boundary.

The effects obtainable in the present invention are not limited to the above-mentioned effects, and other effects not mentioned may be clearly understood by those of ordinary skill in the art from the following description. will be.

1 is a block diagram illustrating an image encoding apparatus according to an embodiment of the present invention.
2 is a block diagram illustrating an image decoding apparatus according to an embodiment of the present invention.
3 is a diagram illustrating a partition mode that can be applied to a coding block when the coding block is encoded by inter prediction.
4 to 6 are diagrams illustrating a camera device for generating a panoramic image.
7 is a diagram schematically illustrating a 360 degree video encoding/decoding and rendering process.
8 shows an equirectangular projection method among 2D projection methods.
9 illustrates a cube projection method among 2D projection methods.
10 illustrates an icosahedron projection method among 2D projection methods.
11 illustrates an octahedral projection method among 2D projection methods.
12 illustrates a truncated pyramid projection transformation method among 2D projection methods.
13 is a diagram exemplified to explain transformation between face 2D coordinates and 3D coordinates.
14 is a flowchart illustrating an inter prediction method of a 2D image according to an embodiment to which the present invention is applied.
15 is a diagram illustrating a process of deriving motion information of a current block when a merge mode is applied to the current block.
16 is a diagram illustrating a process of deriving motion information of the current block when the AMVP mode is applied to the current block.
17 is a diagram illustrating a position of a reference block used to derive a prediction block of a current block.
18 is a diagram illustrating an example in which a face including a reference block is identified by a reference face index in a TPP-based 360-degree projection image.
19 is a diagram illustrating a motion vector when a current block and a reference block belong to the same phase.
20 is a diagram illustrating motion vectors when a current block and a reference block belong to different faces.
21 is a diagram illustrating an example of deforming a reference face according to a current face.
22 is a diagram illustrating a method of performing inter prediction of a current block in a 360-degree projection image according to the present invention.
23 is a diagram illustrating an example of generating a reference block based on a sample belonging to a reference phase.
24 is a diagram illustrating an example of generating a motion compensation reference face by deforming a second face adjacent to a first face including a reference point of a reference block.
25 is a diagram illustrating an example in which availability of a reference face is determined according to the proximity between the faces.
26 is a diagram illustrating a 360-degree projection image in a cube map format.
27 is a diagram illustrating a method of performing motion compensation in consideration of inter-face directionality.
28 is a diagram illustrating an example of performing motion compensation when a current face and a reference face have different directions.
29 is a view for explaining the continuity of the ERP projection image.
30 illustrates an example in which a length of a padding area is set differently according to an image boundary.
31 is a diagram illustrating an example in which padding is performed at the boundary of a face.
32 is a diagram illustrating an example of determining a sample value of a padding area between faces.
33 and 34 are diagrams illustrating examples of substituting an unavailable region among reference regions in a 360-degree projection image based on an ERP format.
35 is a diagram illustrating an encoded motion vector.

Since the present invention can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents and substitutes included in the spirit and scope of the present invention. In describing each figure, like reference numerals have been used for like elements.

Terms such as first, second, etc. may be used to describe various elements, but the elements should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, a first component may be referred to as a second component, and similarly, a second component may also be referred to as a first component. and/or includes a combination of a plurality of related listed items or any of a plurality of related listed items.

When an element is referred to as being “connected” or “connected” to another element, it is understood that it may be directly connected or connected to the other element, but other elements may exist in between. it should be On the other hand, when it is mentioned that a certain element is "directly connected" or "directly connected" to another element, it should be understood that no other element is present in the middle.

The terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. In the present application, terms such as “comprise” or “have” are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification exists, but one or more other features It is to be understood that this does not preclude the possibility of the presence or addition of numbers, steps, operations, components, parts, or combinations thereof.

Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the accompanying drawings. Hereinafter, the same reference numerals are used for the same components in the drawings, and repeated descriptions of the same components are omitted.

1 is a block diagram illustrating an image encoding apparatus according to an embodiment of the present invention.

Referring to FIG. 1 , the image encoding apparatus 100 includes a picture division unit 110 , prediction units 120 and 125 , a transform unit 130 , a quantization unit 135 , a rearrangement unit 160 , and an entropy encoding unit ( 165 ), an inverse quantization unit 140 , an inverse transform unit 145 , a filter unit 150 , and a memory 155 .

Each of the constituent units shown in FIG. 1 is independently illustrated to represent different characteristic functions in the image encoding apparatus, and does not mean that each constituent unit is composed of separate hardware or one software constituent unit. That is, each component is listed as each component for convenience of description, and at least two components of each component are combined to form one component, or one component can be divided into a plurality of components to perform a function, and each of these components Integrated embodiments and separate embodiments of the components are also included in the scope of the present invention without departing from the essence of the present invention.

In addition, some of the components are not essential components for performing essential functions in the present invention, but may be optional components for merely improving performance. The present invention can be implemented by including only essential components to implement the essence of the present invention except for components used for improving performance, and only having a structure including essential components excluding optional components used for improving performance Also included in the scope of the present invention.

The picture divider 110 may divide the input picture into at least one processing unit. In this case, the processing unit may be a prediction unit (PU), a transform unit (TU), or a coding unit (CU). The picture splitter 110 divides one picture into a combination of a plurality of coding units, prediction units, and transformation units, and combines one coding unit, prediction unit, and transformation unit based on a predetermined criterion (eg, a cost function). can be selected to encode the picture.

For example, one picture may be divided into a plurality of coding units. In order to split a coding unit in a picture, a recursive tree structure such as a quad tree structure can be used. A coding in which one image or a largest coding unit is used as a root and is divided into other coding units. A unit may be divided having as many child nodes as the number of divided coding units. A coding unit that is no longer split according to certain restrictions becomes a leaf node. That is, when it is assumed that only square splitting is possible for one coding unit, one coding unit may be split into up to four different coding units.

Hereinafter, in an embodiment of the present invention, a coding unit may be used as a unit for performing encoding or may be used as a meaning for a unit for performing decoding.

A prediction unit may be split in the form of at least one square or rectangle of the same size within one coding unit, and one prediction unit among the split prediction units within one coding unit is a prediction of another. It may be divided to have a shape and/or size different from the unit.

When a prediction unit for performing intra prediction based on a coding unit is generated, if it is not the smallest coding unit, intra prediction may be performed without splitting into a plurality of prediction units NxN.

The prediction units 120 and 125 may include an inter prediction unit 120 performing inter prediction and an intra prediction unit 125 performing intra prediction. Whether to use inter prediction or to perform intra prediction for a prediction unit may be determined, and specific information (eg, intra prediction mode, motion vector, reference picture, etc.) according to each prediction method may be determined. In this case, a processing unit in which prediction is performed and a processing unit in which a prediction method and specific content are determined may be different. For example, a prediction method and a prediction mode may be determined in a prediction unit, and prediction may be performed in a transformation unit. A residual value (residual block) between the generated prediction block and the original block may be input to the transform unit 130 . Also, prediction mode information, motion vector information, etc. used for prediction may be encoded by the entropy encoder 165 together with a residual value and transmitted to a decoder. When a specific encoding mode is used, the original block may be encoded and transmitted to the decoder without generating the prediction block through the predictors 120 and 125 .

The inter prediction unit 120 may predict a prediction unit based on information on at least one of a picture before or after a picture of the current picture, and in some cases, prediction based on information of a partial region in the current picture that has been encoded Units can also be predicted. The inter prediction unit 120 may include a reference picture interpolator, a motion prediction unit, and a motion compensator.

The reference picture interpolator may receive reference picture information from the memory 155 and generate pixel information of integer pixels or less in the reference picture. In the case of luminance pixels, a DCT-based 8-tap interpolation filter in which filter coefficients are different to generate pixel information of integer pixels or less in units of 1/4 pixels may be used. In the case of the color difference signal, a DCT-based 4-tap interpolation filter in which filter coefficients are different to generate pixel information of integer pixels or less in units of 1/8 pixels may be used.

The motion prediction unit may perform motion prediction based on the reference picture interpolated by the reference picture interpolator. As a method for calculating the motion vector, various methods such as Full search-based Block Matching Algorithm (FBMA), Three Step Search (TSS), New Three-Step Search Algorithm (NTS) may be used. The motion vector may have a motion vector value of 1/2 or 1/4 pixel unit based on the interpolated pixel. The motion prediction unit may predict the current prediction unit by using a different motion prediction method. Various methods, such as a skip method, a merge method, an AMVP (Advanced Motion Vector Prediction) method, an intra block copy method, etc., may be used as the motion prediction method.

The intra prediction unit 125 may generate a prediction unit based on reference pixel information around the current block, which is pixel information in the current picture. When the neighboring block of the current prediction unit is a block on which inter prediction is performed, and the reference pixel is a pixel on which inter prediction is performed, the reference pixel included in the block on which inter prediction is performed is a reference pixel of the block on which the intra prediction is performed. information can be used instead. That is, when the reference pixel is not available, the unavailable reference pixel information may be replaced with at least one reference pixel among the available reference pixels.

In intra prediction, the prediction mode may have a directional prediction mode in which reference pixel information is used according to a prediction direction and a non-directional mode in which directional information is not used when prediction is performed. A mode for predicting luminance information and a mode for predicting chrominance information may be different, and intra prediction mode information used for predicting luminance information or predicted luminance signal information may be utilized to predict chrominance information.

When intra prediction is performed, if the size of the prediction unit and the size of the transformation unit are the same, intra prediction of the prediction unit based on the pixel present on the left side, the pixel present on the upper left side, and the pixel present on the upper side of the prediction unit can be performed. However, when the size of the prediction unit is different from the size of the transformation unit when intra prediction is performed, intra prediction may be performed using a reference pixel based on the transformation unit. Also, intra prediction using NxN splitting may be used only for the smallest coding unit.

The intra prediction method may generate a prediction block after applying an adaptive intra smoothing (AIS) filter to a reference pixel according to a prediction mode. The type of the AIS filter applied to the reference pixel may be different. In order to perform the intra prediction method, the intra prediction mode of the current prediction unit may be predicted from the intra prediction mode of the prediction unit existing around the current prediction unit. When the prediction mode of the current prediction unit is predicted using the mode information predicted from the neighboring prediction unit, if the intra prediction mode of the current prediction unit and the neighboring prediction unit are the same, the current prediction unit and the neighboring prediction unit are used using predetermined flag information It is possible to transmit information that the prediction modes of . , and if the prediction modes of the current prediction unit and the neighboring prediction units are different from each other, entropy encoding may be performed to encode prediction mode information of the current block.

In addition, a residual block including residual information that is a difference value from the original block of the prediction unit and the prediction unit in which prediction is performed based on the prediction unit generated by the prediction units 120 and 125 may be generated. The generated residual block may be input to the transform unit 130 .

The transform unit 130 converts the original block and the residual block including residual information of the prediction units generated by the prediction units 120 and 125 to DCT (Discrete Cosine Transform), DST (Discrete Sine Transform), KLT and It can be converted using the same conversion method. Whether to apply DCT, DST, or KLT to transform the residual block may be determined based on intra prediction mode information of a prediction unit used to generate the residual block.

The quantization unit 135 may quantize values transformed in the frequency domain by the transform unit 130 . The quantization coefficient may vary according to blocks or the importance of an image. The value calculated by the quantization unit 135 may be provided to the inverse quantization unit 140 and the rearrangement unit 160 .

The reordering unit 160 may rearrange the coefficient values on the quantized residual values.

The rearranging unit 160 may change the two-dimensional block form coefficient into a one-dimensional vector form through a coefficient scanning method. For example, the reordering unit 160 may scan from DC coefficients to coefficients in a high frequency region using a zig-zag scan method and may change it into a one-dimensional vector form. A vertical scan that scans a two-dimensional block shape coefficient in a column direction and a horizontal scan that scans a two-dimensional block shape coefficient in a row direction may be used instead of a zig-zag scan according to a size of a transform unit and an intra prediction mode. That is, it may be determined whether any of the zig-zag scan, the vertical scan, and the horizontal scan is used according to the size of the transform unit and the intra prediction mode.

The entropy encoding unit 165 may perform entropy encoding based on the values calculated by the reordering unit 160 . For entropy encoding, various encoding methods such as Exponential Golomb, Context-Adaptive Variable Length Coding (CAVLC), and Context-Adaptive Binary Arithmetic Coding (CABAC) may be used.

The entropy encoding unit 165 receives the residual value coefficient information and block type information, prediction mode information, partitioning unit information, prediction unit information and transmission unit information, motion of the coding unit from the reordering unit 160 and the prediction units 120 and 125 . Various information such as vector information, reference frame information, interpolation information of a block, and filtering information may be encoded.

The entropy encoder 165 may entropy-encode the coefficient values of the coding units input from the reordering unit 160 .

The inverse quantizer 140 and the inverse transform unit 145 inversely quantize the values quantized by the quantizer 135 and inversely transform the values transformed by the transform unit 130 . The residual values generated by the inverse quantization unit 140 and the inverse transform unit 145 are combined with the prediction units predicted through the motion estimation unit, the motion compensator, and the intra prediction unit included in the prediction units 120 and 125 and restored. You can create a Reconstructed Block.

The filter unit 150 may include at least one of a deblocking filter, an offset correcting unit, and an adaptive loop filter (ALF).

The deblocking filter may remove block distortion caused by the boundary between blocks in the reconstructed picture. In order to determine whether to perform deblocking, it may be determined whether to apply the deblocking filter to the current block based on pixels included in several columns or rows included in the block. When a deblocking filter is applied to a block, a strong filter or a weak filter can be applied according to the required deblocking filtering strength. In addition, in applying the deblocking filter, horizontal filtering and vertical filtering may be concurrently processed when performing vertical filtering and horizontal filtering.

The offset corrector may correct the offset of the deblocked image with respect to the original image in units of pixels. In order to perform offset correction on a specific picture, a method of dividing pixels included in an image into a certain number of regions, determining the region to be offset and applying the offset to the region, or taking edge information of each pixel into consideration can be used to apply

Adaptive loop filtering (ALF) may be performed based on a value obtained by comparing the filtered reconstructed image and the original image. After dividing pixels included in an image into a predetermined group, one filter to be applied to the corresponding group is determined, and filtering can be performed differentially for each group. As for information related to whether to apply ALF, the luminance signal may be transmitted for each coding unit (CU), and the shape and filter coefficients of the ALF filter to be applied may vary according to each block. In addition, the ALF filter of the same type (fixed type) may be applied regardless of the characteristics of the application target block.

The memory 155 may store the reconstructed block or picture calculated through the filter unit 150 , and the stored reconstructed block or picture may be provided to the predictors 120 and 125 when inter prediction is performed.

2 is a block diagram illustrating an image decoding apparatus according to an embodiment of the present invention.

2, the image decoder 200 includes an entropy decoding unit 210, a reordering unit 215, an inverse quantization unit 220, an inverse transform unit 225, prediction units 230 and 235, and a filter unit ( 240) and a memory 245 may be included.

When an image bitstream is input from an image encoder, the input bitstream may be decoded by a procedure opposite to that of the image encoder.

The entropy decoding unit 210 may perform entropy decoding in a procedure opposite to that performed by the entropy encoding unit of the image encoder. For example, various methods such as Exponential Golomb, Context-Adaptive Variable Length Coding (CAVLC), and Context-Adaptive Binary Arithmetic Coding (CABAC) may be applied corresponding to the method performed by the image encoder.

The entropy decoder 210 may decode information related to intra prediction and inter prediction performed by the encoder.

The reordering unit 215 may perform reordering based on a method of rearranging the entropy-decoded bitstream by the entropy decoding unit 210 by the encoder. Coefficients expressed in a one-dimensional vector form may be restored and rearranged in a two-dimensional block form. The reordering unit 215 may receive information related to coefficient scanning performed by the encoder and perform reordering by performing a reverse scanning method based on the scanning order performed by the corresponding encoder.

The inverse quantization unit 220 may perform inverse quantization based on the quantization parameter provided by the encoder and the reordered coefficient values of the blocks.

The inverse transform unit 225 may perform inverse transforms, ie, inverse DCT, inverse DST, and inverse KLT, on the transforms performed by the transform unit, ie, DCT, DST, and KLT, on the quantization result performed by the image encoder. Inverse transform may be performed based on a transmission unit determined by the image encoder. The inverse transform unit 225 of the image decoder may selectively perform a transformation technique (eg, DCT, DST, KLT) according to a plurality of pieces of information such as a prediction method, a size of a current block, and a prediction direction.

The prediction units 230 and 235 may generate a prediction block based on the prediction block generation related information provided from the entropy decoding unit 210 and previously decoded block or picture information provided from the memory 245 .

As described above, when intra prediction is performed in the same manner as in the operation of the image encoder, when the size of the prediction unit and the size of the transformation unit are the same, the pixel present at the left side of the prediction unit, the pixel present at the upper left side, and the upper side exist Intra prediction is performed on the prediction unit based on the pixel can Also, intra prediction using NxN splitting may be used only for the smallest coding unit.

The prediction units 230 and 235 may include a prediction unit determiner, an inter prediction unit, and an intra prediction unit. The prediction unit determining unit receives various information such as prediction unit information input from the entropy decoder 210, prediction mode information of the intra prediction method, and motion prediction related information of the inter prediction method, and divides the prediction unit from the current coding unit, and predicts It may be determined whether the unit performs inter prediction or intra prediction. The inter prediction unit 230 uses information required for inter prediction of the current prediction unit provided from the image encoder, and predicts the current based on information included in at least one picture before or after the current picture including the current prediction unit. Inter prediction may be performed on a unit. Alternatively, inter prediction may be performed based on information on a pre-restored partial region in the current picture including the current prediction unit.

In order to perform inter prediction, a motion prediction method of a prediction unit included in a corresponding coding unit based on a coding unit is selected from among skip mode, merge mode, AMVP mode, and intra block copy mode. You can decide which way to go.

The intra prediction unit 235 may generate a prediction block based on pixel information in the current picture. When the prediction unit is a prediction unit on which intra prediction is performed, intra prediction may be performed based on intra prediction mode information of the prediction unit provided by the image encoder. The intra prediction unit 235 may include an Adaptive Intra Smoothing (AIS) filter, a reference pixel interpolator, and a DC filter. The AIS filter is a part that performs filtering on the reference pixel of the current block, and can be applied by determining whether to apply the filter according to the prediction mode of the current prediction unit. AIS filtering may be performed on the reference pixel of the current block by using the prediction mode and AIS filter information of the prediction unit provided by the image encoder. When the prediction mode of the current block is a mode in which AIS filtering is not performed, the AIS filter may not be applied.

When the prediction mode of the prediction unit is a prediction unit that performs intra prediction based on a pixel value obtained by interpolating the reference pixel, the reference pixel interpolator may interpolate the reference pixel to generate a reference pixel of a pixel unit having an integer value or less. When the prediction mode of the current prediction unit is a prediction mode that generates a prediction block without interpolating the reference pixel, the reference pixel may not be interpolated. The DC filter may generate the prediction block through filtering when the prediction mode of the current block is the DC mode.

The reconstructed block or picture may be provided to the filter unit 240 . The filter unit 240 may include a deblocking filter, an offset correction unit, and an ALF.

Information on whether a deblocking filter is applied to a corresponding block or picture and information on whether a strong filter or a weak filter is applied when the deblocking filter is applied may be provided from the video encoder. The deblocking filter of the image decoder may receive deblocking filter-related information provided from the image encoder, and the image decoder may perform deblocking filtering on the corresponding block.

The offset correction unit may perform offset correction on the reconstructed image based on the type of offset correction applied to the image during encoding, offset value information, and the like.

ALF may be applied to a coding unit based on information on whether ALF is applied, ALF coefficient information, etc. provided from the encoder. Such ALF information may be provided by being included in a specific parameter set.

The memory 245 may store the reconstructed picture or block to be used as a reference picture or reference block, and may also provide the reconstructed picture to an output unit.

As described above, hereinafter, in the embodiment of the present invention, a coding unit is used as a term for a coding unit for convenience of description, but it may also be a unit for performing decoding as well as coding.

In addition, the current block denotes an encoding/decoding target block, and depending on the encoding/decoding step, a coding tree block (or coding tree unit), a coding block (or a coding unit), a transform block (or a transform unit), or a prediction block (or prediction unit) and the like. In this specification, a 'unit' may indicate a basic unit for performing a specific encoding/decoding process, and a 'block' may indicate a sample array of a predetermined size. Unless otherwise specified, 'block' and 'unit' may be used interchangeably. For example, in the embodiments to be described later, it may be understood that the coding block (coding block) and the coding unit (coding unit) have the same meaning.

One picture may be divided into square or non-square basic blocks and encoded/decoded. In this case, the basic block may be referred to as a coding tree unit. A coding tree unit may be defined as a coding unit of the largest size allowed in a sequence or a slice. Whether the coding tree unit is square or non-square, or information related to the size of the coding tree unit may be signaled through a sequence parameter set, a picture parameter set, a slice header, or the like. A coding tree unit can be divided into smaller-sized partitions. In this case, when a partition generated by dividing a coding tree unit is referred to as a depth 1, a partition generated by dividing a partition having a depth 1 may be defined as a depth 2 . That is, a partition generated by dividing a partition having a depth k in a coding tree unit may be defined as having a depth k+1.

A partition of an arbitrary size generated as the coding tree unit is divided may be defined as a coding unit. The coding unit may be divided recursively or divided into basic units for performing prediction, quantization, transformation, or in-loop filtering, and the like. As an example, a partition of an arbitrary size generated as a coding unit is divided may be defined as a coding unit, or may be defined as a transform unit or a prediction unit that is a basic unit for performing prediction, quantization, transform, or in-loop filtering.

Alternatively, when the coding block is determined, a prediction block having the same size as the coding block or a size smaller than the coding block may be determined through prediction division of the coding block. Predictive division of a coding block may be performed by a partition mode (Part_mode) indicating a division form of the coding block. The size or shape of the prediction block may be determined according to the partition mode of the coding block. The division type of the coding block may be determined through information specifying any one of the partition candidates. In this case, the partition candidates usable by the coding block may include an asymmetric partition type (eg, nLx2N, nRx2N, 2NxnU, 2NxnD) according to the size, shape, or coding mode of the coding block. For example, a partition candidate usable by the coding block may be determined according to the coding mode of the current block. As an example, FIG. 3 is a diagram illustrating a partition mode that can be applied to a coding block when the coding block is encoded by inter prediction.

When the coding block is encoded by inter prediction, as in the example shown in FIG. 3 , any one of eight partition modes may be applied to the coding block.

On the other hand, when the coding block is encoded by intra prediction, the partition mode PART_2Nx2N or PART_NxN may be applied to the coding block.

PART_NxN may be applied when the coding block has a minimum size. Here, the minimum size of the coding block may be predefined in the encoder and the decoder. Alternatively, information about the minimum size of a coding block may be signaled through a bitstream. As an example, the minimum size of the coding block is signaled through a slice header, and accordingly, the minimum size of the coding block may be defined for each slice.

As another example, a partition candidate usable by the coding block may be determined differently according to at least one of a size or a shape of the coding block. As an example, the number or type of partition candidates usable by the coding block may be determined differently according to at least one of the size or shape of the coding block.

Alternatively, the type or number of asymmetric partition candidates among the partition candidates usable by the coding block may be limited according to the size or shape of the coding block. As an example, the number or type of asymmetric partition candidates usable by the coding block may be determined differently according to at least one of the size or shape of the coding block.

In general, the size of the prediction block may range from 64x64 to 4x4. However, when the coding block is encoded by inter prediction, the prediction block may not have a size of 4x4 in order to reduce memory bandwidth when motion compensation is performed.

Depending on the angle of view of the camera, the field of view of the video captured by the camera is limited. To overcome this, one video or one bitstream may be configured by photographing an image using a plurality of cameras and stitching the photographed image. As an example, FIGS. 4 to 6 show examples of simultaneously photographing up and down, left and right, or front and rear using a plurality of cameras. In this way, a video generated by stitching a plurality of videos may be referred to as a panoramic video. In particular, an image having 360 degrees of freedom with respect to a predetermined central axis may be referred to as a 360-degree video.

The camera structure (or camera arrangement) for acquiring 360-degree video has a circular arrangement, as in the example shown in FIG. 4 , or a one-dimensional vertical/horizontal arrangement as in the example shown in FIG. 5 ( a ). Alternatively, it may have a two-dimensional arrangement (ie, a form in which a vertical arrangement and a horizontal arrangement are mixed) as in the example shown in (b) of FIG. 5 . Alternatively, as in the example shown in FIG. 6 , a plurality of cameras may be mounted on an older device.

Embodiments to be described below will be described with a focus on 360-degree video, but it will be included in the technical scope of the present invention to apply the embodiments described later to panoramic video other than 360-degree video.

7 is a diagram schematically illustrating a 360 degree video encoding/decoding and rendering process.

In order to encode/decode a 360-degree video using the encoder/decoder of FIGS. 1 and 2 , it is necessary to convert the 360-degree video into a 2D video. That is, after 2D projection transformation of image information on a 3D space into a 2D form, encoding/decoding may be performed on the transformed image. By performing inverse projection on the encoded/decoded 2D image, it is possible to view an image having 360 degrees of freedom in the up-down, left-right, front-rear, etc.

For projection transformation of 360 degree video to 2D: Equirectangular Procection (ERP), Cube Map Projection (CMP), Icosahedral Projection (ISP), Octahedron Projection (Octahedron Projection, OHP), a truncated pyramid projection transformation (TPP), a Segment Projection (SSP), or a rotated sphere projection (RSP) may be used.

8 shows an equirectangular projection method among 2D projection methods.

Equirectangular projection is a method of projecting pixels corresponding to a sphere as a rectangle with a ratio of 2:1, and is the most widely used 2D transformation technique. In the case of using the isotonic projection method, the actual length of the sphere corresponding to the unit length on the 2D plane becomes shorter toward the pole of the sphere. For example, the coordinates of both ends of the unit length on the 2D plane may correspond to a distance difference of 20 cm near the equator of the sphere, whereas it may correspond to a difference of 5 cm in the vicinity of the pole of the sphere. Accordingly, the equirectangular projection method has a disadvantage in that the image distortion is large in the vicinity of the pole of the sphere, so that the encoding efficiency is lowered.

9 shows a cube projection method among 2D projection methods.

The cube projection method approximates 3D data to correspond to a cube, and then converts the cube into 2D projection. When projecting 3D data into a cube, one face (or plane) is configured to be adjacent to four faces. Since the continuity between each face is high, the cube projection method has an advantage in that the encoding efficiency is higher than that of the equirectangular projection method. After projection-transformation of 3D data into 2D, encoding/decoding may be performed by rearranging the 2D-projection-transformed image into a rectangular shape. Reordering the 2D projection-transformed image in a rectangular shape may be referred to as frame rearrangement or frame packing.

10 illustrates an icosahedron projection method among 2D projection methods.

The icosahedral projection method is a method of approximating 3D data so as to correspond to an icosahedron, and converting it to 2D projection. The icosahedral projection method is characterized by strong inter-face continuity. In addition, frame packing for rearranging the 2D projection-transformed image may be performed.

11 illustrates an octahedral projection method among 2D projection methods.

The octahedral projection method is a method of approximating 3D data so as to correspond to an octahedron, and converting it to 2D projection. The octahedral projection method is characterized by strong continuity between faces. In addition, frame packing for rearranging the 2D projection-transformed image may be performed.

12 illustrates a truncated pyramid projection transformation method among 2D projection methods.

The truncated pyramid projection transformation method is a method of approximating 3D data so as to correspond to a truncated pyramid, and converting it into 2D projection. Under the truncated pyramid projection transformation method, frame packing may be performed so that a face at a specific point of time has a different size from a neighboring face. For example, as in the example shown in FIG. 12 , the front face may have a larger size than the side face and the back face. In the case of using the truncated pyramid projection transformation method, there is an advantage in that image data at a specific viewpoint is large, and encoding/decoding efficiency at a specific viewpoint is higher than that of other viewpoints.

SSP divides the sphere into high-latitude regions, low-latitude regions and mid-latitude regions, mapping two high-latitude regions north and south into two circles, and mapping the mid-latitude regions to a rectangle like ERP.

RSP represents a method of mapping a sphere into the shape of two ellipses surrounding a tennis ball.

Hereinafter, in an embodiment to be described later, a 2D image constructed using 2D projection transformation will be referred to as a 360-degree projection image. In addition, in the embodiments to be described later, even if the embodiments are described based on a specific projection method, it will be said that the embodiments described below may be extended and applied to projection methods other than the described projection methods.

Each sample of the 360-degree projection image may be identified by face 2D coordinates. The face 2D coordinates may include an index f for identifying a face where the sample is located, and coordinates (m, n) indicating a sample grid in a 360-degree projected image.

Through transformation between face 2D coordinates and 3D coordinates, 2D projection transformation and image rendering may be performed. As an example, FIG. 13 is a diagram illustrating transformation between face 2D coordinates and 3D coordinates. When a 360-degree projection image is generated based on ERP, conversion between three-dimensional coordinates (x, y, z) and face 2D coordinates (f, m, n) can be performed using Equations 1 to 4 below. have.

In the 360-degree projection image, the current picture may include at least one face. In this case, the number of faces may be 1, 2, 3, 4 or more natural numbers according to the projection method. Among the face 2D coordinates, f may be set to a value equal to or smaller than the number of faces. The current picture may include at least one or more faces having the same temporal order or output order (POC).

Alternatively, the number of faces constituting the current picture may be fixed or variable. For example, the number of faces constituting the current picture may be limited so as not to exceed a predetermined threshold. Here, the threshold value may be a fixed value previously agreed upon by the encoder and the decoder. Alternatively, information about the maximum number of faces constituting one picture may be signaled through a bitstream.

The faces may be determined by dividing the current picture using at least one of a horizontal line, a vertical line, or a diagonal line according to a projection method.

Each face in the picture may be assigned an index for identifying each face. Each face may be capable of parallel processing such as a tile or a slice. Accordingly, when intra prediction or inter prediction of the current block is performed, a neighboring block belonging to a different phase from the current block may be determined to be unavailable.

Phases in which parallel processing is not allowed (or non-parallel processing regions) may be defined, or phases having interdependencies may be defined. For example, phases in which parallel processing is not allowed or phases having interdependence may be sequentially encoded/decoded instead of parallel encoded/decoded. Accordingly, even a neighboring block belonging to a different face from the current block may be determined to be usable during intra prediction or inter prediction of the current block, depending on whether parallel processing between faces is possible or a dependency, or the like.

Inter prediction in the 360-degree projection image may be performed based on motion information of the current block, such as encoding/decoding of a 2D image. As an example, FIGS. 14 to 16 are flowcharts for explaining an inter prediction method for a 2D image.

14 is a flowchart illustrating an inter prediction method of a 2D image according to an embodiment to which the present invention is applied.

Referring to FIG. 14 , motion information of the current block may be determined ( S1410 ). The motion information of the current block may include at least one of a motion vector related to the current block, a reference picture index of the current block, and an inter prediction direction of the current block.

The motion information of the current block may be obtained based on at least one of information signaled through a bitstream or motion information of a neighboring block adjacent to the current block.

15 is a diagram illustrating a process of deriving motion information of a current block when a merge mode is applied to the current block.

When the merge mode is applied to the current block, a spatial merge candidate may be derived from a spatial neighboring block of the current block ( S1510 ). The spatial neighboring block may include at least one of blocks adjacent to the upper, left, or corner of the current block (eg, at least one of upper left corner, upper right corner, or lower left corner).

Motion information of a spatial merge candidate may be set to be the same as motion information of a spatial neighboring block.

A temporal merge candidate may be derived from a temporal neighboring block of the current block (S1520). The temporal neighboring block may mean a co-located block (collocated block) included in the collocated picture. The collocated picture has a different temporal order (Picture Order Count, POC) from the current picture including the current block. The collocated picture may be determined as a picture having a predefined index in the reference picture list or may be determined by an index signaled from a bitstream. The temporal neighboring block may be determined as an arbitrary block in the block having the same position and size as the current block in the collocated picture, or a block adjacent to the block having the same position and size as the current block. For example, at least one of a block including the center coordinates of a block having the same position and size as the current block in the collocated picture or a block adjacent to a lower right boundary of the block may be determined as a temporal neighboring block.

Motion information of a temporal merge candidate may be determined based on motion information of a temporal neighboring block. As an example, the motion vector of the temporal merge candidate may be determined based on the motion vector of the temporal neighboring block. Also, the inter prediction direction of the temporal merge candidate may be set to be the same as the inter prediction direction of the temporal neighboring block. However, the reference picture index of the temporal merge candidate may have a fixed value. As an example, the reference picture index of the temporal merge candidate may be set to '0'.

Thereafter, a merge candidate list including a spatial merge candidate and a temporal merge candidate may be generated (S1530). If the number of merge candidates included in the merge candidate list is smaller than the maximum number of merge candidates, a combined merge candidate combining two or more merge candidates or a merge candidate having a (0,0) motion vector It may be included in the merge candidate list.

When the merge candidate list is generated, at least one of the merge candidates included in the merge candidate list may be specified based on the merge candidate index ( S1540 ).

The motion information of the current block may be set to be the same as the motion information of the merge candidate specified by the merge candidate index (S1550). For example, when a spatial merge candidate is selected by the merge candidate index, motion information of the current block may be set to be the same as motion information of a spatial neighboring block. Alternatively, when a temporal merge candidate is selected by the merge candidate index, motion information of the current block may be set to be the same as motion information of a temporal neighboring block.

16 is a diagram illustrating a process of deriving motion information of the current block when the AMVP mode is applied to the current block.

When the AMVP mode is applied to the current block, at least one of an inter prediction direction and a reference picture index of the current block may be decoded from the bitstream (S1610). That is, when the AMVP mode is applied, at least one of the inter prediction direction and the reference picture index of the current block may be determined based on information encoded through the bitstream.

A spatial motion vector candidate may be determined based on the motion vector of the spatial neighboring block of the current block (S1620). The spatial motion vector candidate may include at least one of a first spatial motion vector candidate derived from an upper neighboring block of the current block and a second spatial motion vector candidate derived from a left neighboring block of the current block. Here, the upper neighboring block includes at least one of blocks adjacent to the upper or upper right corner of the current block, and the left neighboring block of the current block includes at least one of blocks adjacent to the left or lower left corner of the current block. can A block adjacent to the upper left corner of the current block may be treated as an upper neighboring block, or may be treated as a left neighboring block.

When the reference picture between the current block and the spatial neighboring block is different, the spatial motion vector may be obtained by scaling the motion vector of the spatial neighboring block.

A temporal motion vector candidate may be determined based on the motion vector of the temporal neighboring block of the current block ( S1630 ). When the reference picture between the current block and the temporal neighboring block is different, the temporal motion vector may be obtained by scaling the motion vector of the temporal neighboring block.

A motion vector candidate list including a spatial motion vector candidate and a temporal motion vector candidate may be generated (S1640).

When the motion vector candidate list is generated, at least one of the motion vector candidates included in the motion vector candidate list may be specified based on the information specifying at least one of the motion vector candidate list ( S1650 ).

The motion vector candidate specified by the information may be set as the motion vector prediction value of the current block, and the motion vector difference value may be added to the motion vector prediction value to obtain the motion vector of the current block ( S1660 ). In this case, the motion vector difference value may be parsed through a bitstream.

When motion information of the current block is obtained, motion compensation may be performed on the current block based on the obtained motion information (S1420). Specifically, motion compensation may be performed on the current block based on the inter prediction direction, the reference picture index, and the motion vector of the current block.

As described with reference to FIGS. 14 to 16 , inter prediction for the 360-degree projection image may be performed in units of blocks and may be performed based on motion information of the current block. For example, when performing inter prediction on a 360-degree projection image, the prediction block of the current block to be encoded/decoded in the current picture may be derived from a region most similar to the prediction block in the reference picture. In this case, the reference block in the reference picture used to derive the prediction block of the current block may be located in the same phase as the current block or in a different phase.

17 is a diagram illustrating a position of a reference block used to derive a prediction block of a current block.

As in the example shown in FIG. 17 , the reference block in the reference picture used to derive the prediction block of the current block may exist in the same phase as the current block in the current picture (see (b)), and the current block in the current picture It may be present at a different face than (see (c)). Alternatively, the reference block may span two or more phases (see (a)).

The reference picture including the reference block may be a picture having a different temporal order or output order (POC) from the current picture.

Alternatively, the current picture may be used as a reference picture. For example, a block encoded/decoded before the current block in the current picture including the current block may be set as a reference block of the current block.

As in the illustrated example, the prediction block of the current block may be derived from a reference block included in the same phase as the current block or a reference block included in a different phase than the current block. In this case, the position of the reference block may be specified through a motion vector between the reference block from the corresponding position block corresponding to the position of the current block in the reference picture.

As another example, in order to reduce the amount of data required for encoding/decoding a motion vector, information for specifying a face including a reference block and/or a motion vector for specifying a position of a reference block within the corresponding face is used to reduce the current block motion compensation may be performed. A face including a reference block in a reference picture may be referred to as a 'reference face'.

The information for specifying the face including the reference block includes information indicating whether the reference block belongs to the same face as the current block and/or information (eg, reference face index) for identifying the face including the reference block. It may include at least one. For example, it is possible to determine whether the reference block belongs to the same phase as the current block by using the 1-bit flag. Also, the face including the reference block in the reference picture may be specified by using the reference face index.

18 is a diagram illustrating an example in which a face including a reference block is identified by a reference face index in a TPP-based 360-degree projection image.

As in the example shown in FIG. 18 , a reference face index 'mc_face_idx' (or 'ref_face_idx') for identifying a face including a reference block may be defined. The reference face index may be encoded/decoded through a bitstream.

As another example, a reference face index may be derived from a neighboring block adjacent to the current block. For example, under the merge mode, the reference face index of the current block may be derived from a merge candidate merged with the current block. On the other hand, under the AMVP mode, the face index of the current block may be encoded/decoded through the bitstream.

When the reference block spans the boundary between two faces, the reference face index may specify a face including the reference position of the reference block. Here, the reference position may include a position of a specific corner (eg, upper left sample) in the reference block or a center point of the reference block.

The position of the reference block within the face may be specified based on a vector value from the reference position of the reference face to the reference position of the reference block. Here, the reference position of the reference face may be the position of a specific corner within the face (eg, the position of the upper left reference sample) or the center point of the face.

Alternatively, the reference position of the reference face may be variably determined according to the index of the face including the current block (ie, the current face index), the reference face index, the relative position between the current face and the reference face, or the position of the current block within the face, etc. have. For example, when the current block exists at the first position in the first face, the second position corresponding to the first position in the reference face may be determined as the reference position. As another example, when the current face is located on the left side of the reference face, the reference position of the reference face is set to the upper left corner, and when the current face is located on the top of the reference face, the reference position of the reference face is set as the upper center can be A motion vector from a reference position within a face to a reference block may be referred to as a face vector.

Whether the motion vector is a face vector may be determined based on whether the current face and the reference face are the same (ie, whether the current face index and the reference face index are the same). For example, when the current face index and the reference face index are the same, the motion vector may indicate a vector from the current block to the reference block. On the other hand, when the current face index and the reference face index are different from each other, the motion vector may indicate a vector from the reference position in the reference face to the reference block.

Alternatively, information indicating whether a motion vector is a face vector may be encoded/decoded through a bitstream.

A motion vector (eg, a face vector or a non-face vector) of the current block may be encoded/decoded through a bitstream. For example, a motion vector value may be encoded/decoded through a bitstream as it is.

Alternatively, according to the inter prediction mode of the current block, a motion vector may be encoded/decoded through a bitstream, or a motion vector of the current block may be derived from a neighboring block. For example, when the inter prediction mode of the current block is the AMVP mode, the motion vector of the current block may be encoded/decoded using differential coding. Here, differential coding refers to encoding/decoding a difference between a motion vector of a current block and a motion vector predicted value through a bitstream. The motion vector prediction value may be derived from a spatial/temporal neighboring block of the current block. Alternatively, the motion vector of the current block may be derived to be the same as the spatial/temporal neighboring block of the current block. On the other hand, when the inter prediction mode of the current block is the merge mode, the motion vector of the current block may be set to be the same as the motion vector of the spatial/temporal neighboring block of the current block.

When the motion vector of the current block and the motion vector of the neighboring block are different types, the motion vector of the current block may be derived by matching the motion vector of the neighboring block to the motion vector type of the current block. For example, when the motion vector of the current block is a non-face vector, while the motion vector of the neighboring block is a face vector, the neighboring block using the vector between the reference face reference point of the neighboring block and the reference face of the neighboring block and the face vector of the neighboring block can be converted to a non-face vector of . According to the inter prediction mode of the current block, the motion vector of the current block may be derived based on the transformed non-face vector of the neighboring block.

As another example, the encoding/decoding method of the motion vector of the current block may be determined differently depending on whether the motion vector of the current block is a face vector or a non-face vector. For example, when the motion vector of the current block is a non-face vector, the motion vector of the current block is derived using the motion vector of a neighboring block, whereas when the motion vector of the current block is a face vector, the face vector value is bit as it is. Encoding/decoding may be performed through a stream.

As described through the above example, in the 360-degree projection image, motion compensation of the current block may be performed through a reference block belonging to a different phase than the current block. However, when at least one of the phase, size, or shape of the face including the current block and the face including the reference block is different, it is difficult to search for a reference block matching the prediction block of the current block in the reference face. For example, under TPP, since the front face and the right face have different sizes and shapes, it is difficult to have a similarity between the block belonging to the front face and the block belonging to the right face. Accordingly, when motion estimation or motion compensation is to be performed from a reference face having a phase, size, or shape different from the current face, transformation to match the phase, size, or shape of the current face and the reference face needs to be performed.

Hereinafter, a method for performing inter prediction according to whether the current block and the reference block belong to the same phase (or whether they belong to a phase corresponding to each other) will be described in more detail.

19 is a diagram illustrating a motion vector when a current block and a reference block belong to the same phase.

When the current block and the face containing the reference block are the same (that is, the current face index and the reference face index are the same), the motion vector is the coordinate difference between the starting point between the current block and the starting point of the reference block, as in a 2D image. can be used as a motion vector.

20 is a diagram illustrating motion vectors when a current block and a reference block belong to different faces.

If the current block and the face containing the reference block are different (that is, the current face index and the reference face index are different), and at least one of the size, shape, or phase between the current face and the reference face is different, the face to which the reference block belongs can be transformed according to the size, shape, or phase of the face to which the prediction block belongs. For example, the reference face may be transformed using at least one of phase warping, interpolation, and/or padding. As an example, FIG. 21 is a diagram illustrating an example of transforming a reference face to match a current face. If the current face and the reference face are different in size and/or shape, as in the example shown in FIG. 21 , a phase transformation, padding or interpolation is applied to the reference face to make the reference face the same size and/or shape as the current face. can be transformed to have In deforming the reference face, at least one of phase deformation, padding, and/or interpolation may be omitted, and deformation of the reference face may be performed in an order different from that illustrated in FIG. 21 .

The reference face deformed according to the current face may be referred to as a reference face for motion compensation.

The motion compensation reference phase may be interpolated with a predefined precision (eg, quarter pel or integer pel, etc.). A block most similar to the prediction block of the current block in the interpolated motion compensation reference phase may be generated as the prediction block of the current block. As in the example shown in FIG. 20 , the motion vector of the current block may indicate a coordinate difference between the start position of the current block and the start position of the reference block (ie, non-face vector encoding/decoding). Although not shown, it is also possible to set the coordinate difference between the reference position in the motion compensation reference face and the start position of the reference block as the motion vector of the current block (ie, encode/decode the face vector).

20 and 21 , it has been described that the reference face is modified according to the phase, size, or shape of the current face. Contrary to the illustration, it is also possible to derive the motion vector of the current block by transforming the current face according to the phase, size, or shape of the reference face.

As in the above example, when at least one of the phase, size, or shape between the current face and the reference face is different, at least one of the phase, size, or shape between the current face and the reference face is modified to perform inter prediction. can

22 is a diagram illustrating a method of performing inter prediction of a current block in a 360-degree projection image according to the present invention.

Referring to FIG. 22 , first, information about a reference face may be decoded from a bitstream ( S2210 ). When the information on the reference face is decoded, based on the decoded information, it may be determined whether the current block and the reference block belong to the same face ( S2220 ).

The information about the reference face may include at least one of whether the current block and the reference block belong to the same face or a reference face index.

For example, 'isSameFaceFlag' indicating whether the face to which the current block belongs and the face to which the reference block belongs correspond to each other or whether the current face index and the reference face index are the same may be signaled through the bitstream. A value of isSameFaceFlag of 1 may mean that the current face index and the reference face index have the same value, or that the face to which the current block belongs and the face to which the reference block belongs correspond to each other. On the other hand, a value of isSameFaceFlag of 0 may mean that the current face and the reference face index have different values, or that the face to which the current block belongs and the face to which the reference block belongs do not correspond to each other.

The reference face index may be signaled only when the value of isSameFaceFlag is 0. Alternatively, it is possible to omit the signaling of isSameFaceFlag and essentially signal the reference face index. When signaling of isSameFaceFlag is omitted, it is possible to determine whether the current block and the reference block belong to the same face by comparing the current face index and the reference face index.

When it is determined that the current block and the reference block are included in the same phase, a motion vector indicating a coordinate difference between the current block and the reference block in the reference face is obtained (S2230), and motion compensation is performed using the obtained motion vector. can be performed (S2240).

On the other hand, when it is determined that the current block and the reference block are included in different faces, a motion compensation reference face in which at least one of the phase, size, or shape of the reference face is transformed to match the current face may be generated ( S2250 ). When the motion vector reference face is generated, a motion vector indicating a coordinate difference between the current block and a reference block in the motion compensation reference face may be obtained, and motion compensation may be performed using the obtained motion vector.

Even when the current block and the reference block belong to different faces, if the phase, size, or shape between the current face and the reference face is the same, the process of generating the motion vector reference face may be omitted.

As another example, based on whether the reference block belongs to a specific face, it may be determined whether to modify the reference face. For example, in a TPP-based 360-degree projection image, a flag indicating whether a reference block exists in the front face may be signaled. isRefInFrontFlag indicates whether the reference block exists in the front face. If the value is 1, it indicates that the starting point of the reference block exists in the front face. If the value is 0, the starting point of the reference block is right, left, top, It may indicate that it is present on the bottom or rear face. When both the current block and the reference block belong to the front face, or when both the current block and the reference block do not belong to the front face, the process of generating the motion compensation reference face may be omitted. On the other hand, when one of the current block and the reference block belongs to the front side and the other does not belong to the front side, a motion compensation reference phase may be generated and the reference block of the current block may be specified from the generated motion compensation reference phase.

Motion compensation of the current block in the 360-degree projection image may be limited to use only a reference block belonging to the same phase as the current block. Motion estimation and motion compensation for the current block may be performed on a reference block belonging to the same phase as the current block. For example, as in the example shown in (c) of FIG. 17 , motion compensation such as when the current block and the reference block belong to different phases may not be performed. The face to which the reference block belongs may be determined based on the position of the reference point of the reference block. Here, the reference point of the reference block may be a corner sample or a center point of the reference block. For example, even when the reference block is located across the boundary between two faces, if the reference point of the reference block belongs to the same face as the current face, it may be determined that the reference block belongs to the same face as the current block.

Whether motion compensation can be performed using a reference block belonging to a different face from the current block may be adaptively determined based on a projection method, a size/shape of a face, a size difference between faces, and the like. Alternatively, information (eg, a flag) indicating whether motion compensation can be performed using a reference block belonging to a different phase than the current block may be signaled through a bitstream.

Motion compensation of the current block may be performed based on a reference block generated by interpolation, padding, or phase transformation of pixels belonging to a reference face corresponding to the current face. For example, when the reference block spans two or more faces, and the reference point of the reference block belongs to the reference face corresponding to the current face, the reference block is located in the reference face (hereinafter, referred to as the first face) corresponding to the current face. It may include a first region to which it belongs and a second region belonging to a reference face that is out of the current face (hereinafter referred to as a second face).

In this case, a pixel of the second area is generated by copying or interpolating a sample included in the first face, or a predetermined filter is applied to the sample included in the first face and/or pixels of the second face to generate a pixel of the second area. You can also create pixels. The predetermined filter may mean a weight filter, an average filter, or an interpolation filter. The pixel area to which the filter is applied may be an entire area belonging to the first phase and/or the second phase, or a partial area. Here, the partial region may be the first region and the second region, or a region having a size/shape predefined in the encoder/decoder. The filter may be applied to one or more pixels adjacent to the boundary between the first face and the second face.

23 is a diagram illustrating an example of generating a reference block based on a sample belonging to a reference phase.

As in the example shown in FIG. 23 , the sample included in the boundary of the reference face (first face) corresponding to the current face is padded (or copied) and/or interpolated, or the sample included in the first face and the sample included in the first face; Motion compensation may be performed on the current block based on a reference block generated by applying a filter between samples included in the second phase adjacent to the first phase.

In the example shown in FIG. 23 , a padding region is generated by padding a sample included in the front face to which the reference point of the reference block belongs, and motion compensation is performed on the current block by using the sample included in the padding region. became

Alternatively, a motion reference compensation reference phase obtained by phase-transforming all or part of the second phase may be generated using the value of the first phase, and then motion compensation may be performed on the current block using the generated motion compensation reference phase. have.

24 is a diagram illustrating an example of generating a motion compensation reference face by deforming a second face adjacent to a first face including a reference point of a reference block.

24, by performing at least one of transformation, interpolation, or padding on all or part of the second face that includes a partial region of the reference block but does not include the reference point of the reference block, A motion compensation reference face can be created. Accordingly, motion compensation may be performed on the current block using a sample belonging to the motion compensation reference phase.

Information indicating whether to use a reference block generated based on a sample value belonging to a reference face corresponding to the current phase for motion compensation may be encoded/decoded through a bitstream. The information is a 1-bit flag. For example, a flag having a value of 0 indicates that a reference block generated based on a sample value belonging to a reference phase corresponding to the current phase is not used for motion compensation, and the value of the flag A value of 1 may indicate that a reference block generated based on a sample value belonging to a reference phase corresponding to the current phase can be used for motion compensation of the current block.

Depending on the projection method of the 3D data, the size/shape between the faces may be different. For example, under the TPP projection method, the front face may be larger than the remaining face. A face with a small size has a relatively small amount of information compared to a face with a large size. Accordingly, it is possible to increase the precision of the motion vector in the face with a small size, thereby increasing the encoding efficiency. That is, the motion vector precision may be adaptively determined according to the size/shape of the reference face including the reference block.

For example, in the TPP-based 360-degree projection image, when the reference block belongs to the front face, motion compensation is performed using a quarter pel, whereas the reference block is smaller in size than the front face, right face, left In the case of belonging to the face, the upper face, or the lower face, motion compensation may be performed using an octopel (1/8pel).

Conversely, it is also possible to use a smaller motion vector precision as the size of the reference face is larger, and use a larger motion vector precision as the size of the reference face is smaller.

The reference block of the current block may be derived from a face having adjacency to the current face. That is, a face that is not adjacent to the current face cannot be used as a reference face, and blocks belonging to the corresponding face cannot be used as reference blocks of the current block. Whether the current face and the predetermined face have adjacency is determined whether the current face and the reference face are spatially adjacent to each other, or whether the difference between the index of the current face and the index of the reference face is greater than a predefined threshold. can be determined on the basis of Here, whether the current face and the reference face are spatially adjacent to each other may be determined based on a 3D space or a 2D plane. Alternatively, the adjacency between the faces in the encoder and the decoder may be predefined.

For example, in 3D space, a face that is not spatially adjacent to the current face cannot be used as a reference face. For example, in a hexahedron, a face corresponding to the opposite face of the current face is not spatially adjacent to the current face, so it may be set not to be used as a reference face.

25 is a diagram illustrating an example in which availability of a reference face is determined according to the proximity between the faces.

As in the example shown in FIG. 25 , when the current block is included in the front face, the back face located opposite the front face in 3D space cannot be used as a reference face of the current block. Similarly, in performing motion compensation of blocks included in the back face, the front face cannot be used as a reference face.

This can also be applied between the Left and Right faces and between the Top and Bottom faces.

If the reference block is included in a face that is not adjacent to the current face, motion compensation may be performed by using a predetermined location area in a picture different from the current picture instead. Here, the predetermined location area may mean a block at the same location as the current block or a block adjacent in a specific direction from the block at the same location.

As another example, the availability of the motion vector, the merge candidate, or the motion vector candidate may be determined according to the position of the reference block indicated by the motion vector of the current block or the merge candidate. For example, when the motion vector of the current block or the reference block indicated by the merge candidate of the current block is located at a face not adjacent to the current face, the corresponding motion vector or the corresponding merge candidate may be determined as unavailable. Unavailable merge candidates may be excluded from the merge candidate list.

When a 360-degree image is projected on a 2D plane, faces may have different directions according to a projection transformation format. For example, when a 360-degree image approximated as a polyhedron is developed on a 2D plane, frame packing may be performed by deforming or rotating faces in order to increase encoding/decoding efficiency.

26 is a diagram illustrating a 360-degree projection image in a cube map format.

In the example shown in FIG. 26 , the faces (faces 0, 1, 2) included in the lower row are rotated 90 degrees clockwise compared to the faces (faces 3, 4, and 5) included in the upper row. was shown As in the example shown in FIG. 26 , in consideration of spatial continuity between the faces, as the faces are rotated and disposed, the faces have different directions from each other.

When the reference face has a different direction from the current face, motion compensation for the current block may be performed by rotating the reference face or the reference block according to the directionality of the reference face. A method of performing motion compensation in consideration of inter-face directionality will be described in more detail with reference to FIG. 27 .

27 is a diagram illustrating a method of performing motion compensation in consideration of inter-face directionality.

First, the reference phase or reference block of the current block may be specified (S2710). The reference face of the current block may be specified by a reference face index or a reference block specified by a motion vector of the current block.

When the reference face or the reference block of the current block is determined, it may be determined whether the current face and the reference face have the same direction (S2720). Whether the current face and the reference face have the same direction may be determined based on rotation information related to rotation of the face. The rotation information may include information about whether to rotate, a rotation direction, a rotation angle, or an order/position of allocating a sample of a reference block to a prediction block. The rotation information may be encoded in an encoder and signaled through a bitstream, or may be derived based on inter-phase directionality.

Alternatively, whether or not the face is rotated or the angle of rotation may be predefined in the encoder and decoder according to the position of the face.

When the current face and the reference face have the same direction, a prediction block for the current block may be generated by using the reference block included in the reference face ( S2730 ).

On the other hand, when the current face and the reference face have different directions, the reference face or the reference block included in the reference face may be rotated to have the same direction as the current face ( S2740 ). In addition, a prediction block for the current block may be generated using the reference block included in the rotated reference face or the rotated reference block ( S2750 ).

28 is a diagram illustrating an example of performing motion compensation when a current face and a reference face have different directions.

When the reference face is specified by the reference face index or the motion vector of the current block, it may be determined whether the reference face has the same direction as the current face. For example, in the example shown in FIG. 28 , the reference face (phase 1) is rotated by 90 degrees relative to the current face (phase 4), so it may be determined that the current face and the reference face have different directions.

When the current face and the reference face have different directions, the reference face or the reference block included in the reference face may be rotated to have the same direction as the current face. For example, after rotating the reference face according to the direction of the current face, the reference block may be obtained from the rotated reference face. When the reference block is obtained, a prediction block of the current block may be generated using the obtained reference block.

Unlike the example shown in FIG. 28 , after obtaining the comprehension reference block in the motion vector, the obtained reference block may be rotated according to the direction of the current face. In this case, a prediction block of the current block may be generated using the rotated reference block.

In the example described with reference to FIGS. 27 and 28 , it is exemplified that the reference face or reference block is rotated according to the direction of the current block. Unlike the illustrated example, it is also possible to perform motion compensation by rotating a reference picture or rotating a current face or a current block according to the direction of the reference face. Alternatively, it is also possible to rotate both the current face and the reference face according to a predefined direction.

When the reference block of the current block spans the picture boundary, motion compensation may be performed by performing padding outside the picture boundary in which the pixel value does not exist. Padding may be performed by copying samples adjacent to a picture boundary or interpolating samples adjacent to a picture boundary. Alternatively, padding may be performed by setting an area in which a pixel value outside a picture boundary does not exist to a predefined value. For example, padding may be performed to set a pixel value outside the picture boundary to an intermediate value (eg, 128 in the case of an 8-bit image) of an available pixel value range.

Padding may be performed on a picture boundary or a face boundary even in a 360-degree projection image. In this case, padding at the picture boundary or the face boundary may be performed in consideration of continuity of a 360-degree image. By back-projecting the 360-degree projection image to a sphere or polyhedron, continuity between picture boundaries or continuity between faces may be determined. Specifically, the boundary of the 360-degree projection image or the boundary of the face in the 360-degree projection image may be performed using samples adjacent to the boundary adjacent to the corresponding boundary in 3D space.

29 is a view for explaining the continuity of the ERP projection image.

When ERP is used, a two-dimensional 360-degree projection image can be obtained by unfolding a 360-degree image approximated by a sphere into a rectangle with a ratio of 2:1. When a rectangular 360-degree projection image is back-projected onto a sphere, the left boundary of the 360-degree projection image has continuity with the right boundary. For example, in the example shown in FIG. 24 , pixels A, B, and C outside the left boundary line can be expected to have similar values as pixels A', B' and C' inside the right boundary line, and outside the right boundary line. It can be expected that pixels D, E, and F of D', E', and F' have similar values as pixels D', E' and F' inside the left boundary line.

Also, the upper boundary on the left side has continuity with the upper boundary on the right side based on the center line in the vertical direction dividing the 360-degree projection image into two halves. For example, in the example shown in FIG. 29 , pixels G and H outside the upper-left boundary may be predicted to be similar to pixels G' and H' inside the upper-right boundary, and pixels I and J outside the upper-right boundary can be predicted to be similar to the inner pixels I' and J' of the upper-left boundary.

Similarly, based on the vertical center line dividing the 360-degree projection image into two, the upper boundary on the left has continuity with the upper boundary on the right. For example, in the example shown in FIG. 29 , pixels K and L outside the lower-left boundary may be predicted to be similar to pixels K' and L' inside the lower-right boundary, and pixels M and N outside the lower-right boundary. can be predicted to be similar to the inner pixels M' and N' of the lower-left boundary.

In consideration of continuity in 3D space, padding may be performed at the boundary of the 360-degree projection image or the boundary between faces. Specifically, in a boundary on which padding is to be performed, padding may be performed using samples included inside a boundary having continuity with the corresponding boundary. For example, in the example shown in FIG. 29 , in the left boundary of the 360-degree projection image, padding is performed using samples adjacent to the right boundary, and in the right boundary of the 360-degree projection image, padding is performed using samples adjacent to the left boundary. can be performed. That is, at positions A, B, and C of the left boundary, padding may be performed using samples at positions A', B', and C' included inside the right boundary, and positions D, E, and F of the right boundary In , padding may be performed using samples of positions D', E', and F' included inside the left boundary.

Also, when the upper boundary is bisected, padding may be performed at the upper left boundary using samples adjacent to the upper right boundary, and padding may be performed at the upper right boundary using samples adjacent to the upper left boundary. That is, at positions G and H of the upper-left boundary, padding is performed using samples at positions G' and H' included inside the upper-right boundary, and at positions I and J of the upper-right boundary, the upper-left boundary Padding may be performed using samples at positions I' and J' included inside of .

Similarly, when the lower boundary is bisected, at the lower left boundary, padding may be performed using samples adjacent to the lower right boundary, and at the lower right boundary, padding may be performed using samples adjacent to the lower left boundary. That is, at positions K and L of the lower-left boundary, padding is performed using samples at positions K' and L' included inside the upper-right boundary, and at positions M and N of the upper-right boundary, the upper-left boundary Padding may be performed using samples at positions M' and N' included inside of .

An area in which padding is performed may be referred to as a padding area, and the number of sample lines generated by padding may be referred to as a length of the padding area. As an example, in FIG. 29 , k sample lines are generated in the padding area, and it may be considered that the length of the padding area is k.

The length of the padding area may be set differently for each horizontal or vertical direction, or set differently for each face boundary. In particular, in the case of using the ERP projection transformation, in a 360-degree projection image, the closer to the upper end or lower end (ie, closer to the pole of the sphere), the shorter the actual length of the sphere corresponding to the unit length. Accordingly, the closer the 360 degree projection image is to the upper end or lower end, the more distortion will occur. To prevent this, a method of performing padding using more pixels as it approaches the region where distortion occurs, or smoothing the boundary of an image through a smoothing filter may be considered.

30 illustrates an example in which a length of a padding area is set differently according to an image boundary.

In the example shown in FIG. 30 , the length of the arrow indicates the length of the padding area.

As in the example shown in FIG. 30 , the length of the padding area performed in the horizontal direction and the length of the padding area performed in the vertical direction may be set to be different. For example, if k columns of samples are generated through padding in the horizontal direction, padding may be performed so that 2k rows of samples are generated in the vertical direction.

As another example, although padding is performed to have the same length in both the vertical direction and the horizontal direction, the length of the padding region may be extended laterally through interpolation in at least one of the vertical direction and the horizontal direction. For example, k sample lines may be generated in the vertical and horizontal directions, and k sample lines may be additionally generated in the vertical direction through interpolation or the like. That is, after generating k sample lines in both the horizontal and vertical directions (see FIG. 29), k sample lines are additionally generated in the vertical direction, so that the length in the vertical direction may be 2k (see FIG. 30). .

Interpolation may be performed using at least one of a sample included inside the boundary of the image and a sample included outside the boundary of the image. For example, an additional padding region may be created by copying samples adjacent to the inner side of the lower boundary outside the padding region adjacent to the upper boundary, and then interpolating the copied samples and samples included in the padding region adjacent to the upper boundary. . The interpolation filter may include at least one of a vertical filter and a horizontal filter. According to the position of the generated pixel, one of the filter in the vertical direction and the filter in the horizontal direction may be selectively used. Alternatively, a sample included in the additional padding area may be generated by simultaneously using the vertical filter and the horizontal filter.

As described above, the horizontal length n of the padding area and the vertical length m of the padding area may have the same value or different values. For example, n and m are natural numbers greater than or equal to 0, and may have the same value, or any one of m and n may have a smaller value than the other. In this case, m and n may be encoded in an encoder and signaled through a bitstream. Alternatively, the length n in the horizontal direction and the length m in the vertical direction may be predefined in the encoder and the decoder according to the projection transformation method.

The sample value of the padding area may be generated by copying samples located inside the image. For example, in the examples shown in FIGS. 29 and 30 , the padding area positioned at the left boundary of the image may be generated by copying a sample adjacent to the right boundary of the image.

As another example, the sample value of the padding region may be determined using at least one sample included inside a boundary to be padded and a sample included outside the corresponding boundary. For example, after copying samples that are spatially continuous with a boundary to be padded outside the boundary, a weighted average or average between the copied samples and samples included inside the boundary is used to obtain a sample in the padding area. value can be determined.

For example, in the examples shown in FIGS. 29 and 30 , the padding area positioned at the left boundary of the image is a weighted average of at least one sample inside the left boundary of the image and at least one sample inside the right boundary of the image. Or it may be generated by averaging. In this case, the weight applied to each sample may be determined based on the distance from the image boundary. For example, the closer to the left boundary of the image, the greater the weight given to the samples located inside the left boundary of the image, and the farther away from the left boundary of the image, the more samples located outside the left boundary of the image (that is, the The weight given to samples located inside the right boundary) may be increased.

When the 360-degree projection image includes a plurality of faces, frame packing may be performed by adding a padding area to the face boundary. For example, a 360-degree projection image in which a padding area is added to a boundary area between faces may be obtained.

31 is a diagram illustrating an example in which padding is performed at the boundary of a face.

For convenience of explanation, it is assumed that the 360-degree projection image is projection-converted based on the ISP. In addition, based on the drawing shown in (a) of FIG. 31, the upper face and the lower face will be divided. For example, the upper face may represent any one of faces 1, 2, 3, and 4, and the lower face may represent any one of faces 5, 6, 7, and 8.

For a predetermined face, a padding area in a shape surrounding the predetermined face may be set. As an example, as in the example shown in (a) of FIG. 31 , a padding region including m samples may be generated for a triangular-shaped face.

As a result of performing frame packing by adding a padding region, as in the example shown in FIG. 31B , a 360-degree projection image in which a padding region is added between the boundaries and faces of the image may be obtained.

Unlike the example shown in FIG. 31B , frame packing may be performed by adding a padding region only at the boundary of an image or by adding a padding region only between faces. Alternatively, in consideration of inter-face continuity, frame packing may be performed by adding a padding region only between faces in which image discontinuity occurs.

The length of the padding area between the faces may be set identically or may be set differently according to positions. For example, the length (ie, the length in the horizontal direction) n of the padding area located on the left or right side of the predetermined face and the length m in the horizontal direction of the padding area located on the top or the bottom of the predetermined face may have the same value or different values. It may have a value. For example, n and m are natural numbers greater than or equal to 0, and may have the same value, or any one of m and n may have a smaller value than the other. In this case, m and n may be encoded in an encoder and signaled through a bitstream. Alternatively, the length n in the horizontal direction and the length m in the vertical direction may be predefined in the encoder and the decoder according to the projection transformation method, the position of the face, the size of the face, or the shape of the face.

The sample value of the padding area may be determined based on a sample included in a predetermined face or a sample included in the predetermined face and a sample included in a face adjacent to the predetermined face.

For example, the sample value of the padding area adjacent to the boundary of a predetermined face may be generated by copying a sample included in the corresponding face or interpolating samples included in the corresponding face. For example, in the example shown in (a) of FIG. 31 , the upper extension region U of the upper face may be generated by copying samples adjacent to the border of the upper face or interpolating a predetermined number of samples adjacent to the border of the upper face. can Similarly, the lower extension region D of the lower face may be generated by copying samples adjacent to the border of the lower face or interpolating a predetermined number of samples adjacent to the border of the lower face.

Alternatively, the sample value of the padding area adjacent to the boundary of the predetermined face may be generated using sample values of the face spatially adjacent to the corresponding face. Here, the inter-face adjacency can be determined by whether or not there is continuity between faces when the 360-degree projection image is back-projected onto the 3D space. Specifically, the sample value of the padding area adjacent to the boundary of a predetermined face is generated by copying a sample included in a face spatially adjacent to the corresponding face, or a sample included in the corresponding face and a sample included in a face spatially adjacent to the corresponding face. It can be generated by interpolating samples. For example, a left part of the upper extension region of the second face may be generated based on samples included in the first face, and a right part may be generated based on the samples included in the third face.

32 is a diagram illustrating an example of determining a sample value of a padding area between faces.

The padding area between the first face and the second face may be obtained by averaging at least one sample included in the first face and at least one sample included in the second face. Specifically, the padding area between the upper face and the lower face may be obtained by averaging the upper extension area U and the lower extension area D by weight.

The weight w may be determined based on information encoded and signaled by the encoder. Alternatively, the weight w may be variably determined according to the position of the sample in the padding area. For example, the weight w may be determined based on a distance from the position of the sample in the padding region to the first face and the distance from the position of the sample in the padding region to the second face.

Equations 5 and 6 are diagrams illustrating an example in which the weight w is variably determined according to the position of the sample. When padding is performed between the upper face and the lower face, in an area close to the upper face, sample values of the padding area are generated based on Equation 5, and in an area close to the lower face, the sample values of the padding area are generated based on Equation 6 Sample values may be generated.

The filter for weight calculation may have a vertical direction, a horizontal direction, or a predetermined angle. When the weighting filter has a predetermined angle, a sample included in the first face and a sample included in the second face positioned on a predetermined angular line from the sample in the padding area may be used to determine a sample value of the corresponding sample. .

As another example, at least a portion of the padding area may be generated using only samples included in any one of the first face and the second face. For example, when any one of the samples included in the first face and the samples included in the second face is not available, padding may be performed using only available samples. Alternatively, padding may be performed by replacing an unavailable sample with a neighboring available sample.

When motion compensation of the current block is performed, a block generated by padding or interpolating the reference spare block may be used as the reference block of the current block after inducing the area specified by the motion vector as the reference spare block.

The reference region of the current block may be defined as a region referenced for intra prediction or inter prediction of the current block. Depending on the location of the reference region, the availability of the reference region may be determined. For example, determining availability of a reference region based on whether the reference region spans a face boundary, whether the reference region spans an image boundary, or whether the reference region exists in an area between faces (ie, a padding region) can do. Here, the reference region spanning the face boundary means that the reference region spans the boundary of the 360-degree projection image at the same time that the reference region is the face boundary, the reference region spans two or more faces, or the reference region spans the face boundary and the padding region If it spans over , etc. can be shown.

Depending on the location of the reference area, the entire reference area may be treated as available or the entire reference area may be treated as unavailable. For example, when the reference region spans the face boundary, the entire reference region may be treated as unavailable.

As another example, the reference area may be divided into an available area and an unavailable area according to the location of the reference area. For example, when the reference region spans the face boundary, an area located inside the predetermined face may be determined to be available, and an area located outside the predetermined face may be determined to be unavailable.

In inter prediction of the current block, the unavailable area may not be used or the unavailable area may be replaced with an available area. For example, a sample value generated by padding or interpolating a sample included in the available region may be used to replace the non-available region, and motion compensation of the current block may be performed using the replaced region.

Whether to use the unavailable region for motion compensation of the current block may be determined based on information signaled through a bitstream. The information may be a 1-bit flag. For example, according to the value of the flag, motion compensation of the current block is performed by excluding the unavailable area, or the unavailable area is replaced based on the available area, and then motion compensation of the current block is performed using the replaced unavailable area. can be performed.

When motion compensation is performed using the unavailable area, the unavailable area may be replaced with a sample value generated by padding or interpolating samples located inside a predetermined face. For example, the non-available region may be replaced with a sample value generated by padding or interpolating a sample included in the available region.

As another example, the non-available area may be replaced with an area located at a boundary between a predetermined face and a different face. For example, when the reference region spans the face boundary, an area located inside the predetermined face may be determined to be available, and an area located outside the predetermined face may be determined to be unavailable. In this case, the non-available region may be replaced with a region located inside a face different from the predetermined face, or may be replaced with a sample located inside a face different from the predetermined face or a sample value generated by interpolation. Here, the replacement area of the non-available area or the face including the replacement sample may be determined in consideration of proximity to a predetermined face. The adjacency may be determined in consideration of whether the predetermined face is adjacent to the predetermined face in 2D space, or whether the predetermined face is adjacent to the predetermined face when the 360-degree projection image is inversely projected on the 3D space. Alternatively, the replacement area or replacement sample may be determined from a face located on the same vertical/horizontal line as the available area.

As another example, the unavailable region may be replaced with a sample value obtained based on interpolation or a weighting operation between a sample included in the available region and a sample included in a region belonging to a different pace from the available region.

In consideration of the position of the reference region in the 360-degree projection image, the position of the reference region in the face, the index of the face to which the reference region belongs, the position or decoding order, the frame packing type, etc., the above-described processing method of the unusable image may be selectively used. may be

33 and 34 are diagrams illustrating examples of substituting an unavailable region among reference regions in a 360-degree projection image based on an ERP format.

When the reference region specified by the motion vector of the current block spans the boundary of the 360-degree projection image, the region outside the boundary of the projection image may be replaced with a replacement region. Specifically, the prediction of the current block may be performed using a reference block including a first sub-block Ref 1 in the reference image and a second sub-block Ref 2 outside the reference image.

In this case, the region outside the boundary of the projection image may be replaced with a region adjacent to the boundary having continuity with the boundary over which the reference region spans in 3D space.

Specifically, as in the example shown in FIG. 33 , when the reference region spans the left boundary of the 360-degree image, the area outside the left boundary is a boundary having adjacency to the left boundary in 3D space (ie, the right boundary). may be replaced by an area adjacent to In this case, the area outside the left boundary may be replaced with a copy of the area adjacent to the right boundary, or may be replaced with a sample value generated by padding or interpolating a sample included in the area adjacent to the right boundary.

As in the example shown in FIG. 34 , when the reference region spans the upper left boundary of the 360-degree image, the area outside the upper left boundary is defined as a boundary having adjacency to the upper left boundary in 3D space (ie, upper right boundary). ) can be replaced with an adjacent area. In this case, the area outside the upper left boundary may be replaced with a copy of the area adjacent to the upper right boundary, or may be replaced with a sample value generated by padding or interpolating a sample included in the area adjacent to the upper right boundary.

Although not shown, when the reference region spans the right border, the region outside the right border may be replaced with a region adjacent to the left border, and when the reference region spans the upper right border, the upper right outer region is replaced with the upper left It can be replaced by an area adjacent to the boundary.

As in the above-described example, the reference region of the current block may include a first sub-block adjacent to the first boundary and a second sub-block adjacent to the second boundary having continuity with the first boundary.

The encoder may selectively encode the smaller information amount among the first motion vector MVRef0 up to the first subblock and the second motion vector MVRef1 up to the second subblock. For example, when the amount of information of the second motion vector is smaller than that of the first motion vector, the second motion vector may be encoded.

35 is a diagram illustrating an encoded motion vector.

Referring to FIG. 35 , it can be confirmed that the first motion vector from the current block to the first sub-block has a smaller value than the second motion vector from the current block to the second sub-block. Accordingly, information about the first motion vector for specifying the position of the first sub-block with respect to the current block may be encoded.

The motion vector of the current block may be derived based on spatial/temporal neighboring blocks of the current block. In this case, the derived motion vector may be directly set as the motion vector of the current block, or the motion vector of the current block may be determined by resizing the derived motion vector. Here, the resizing may represent at least one of an addition, a subtraction, or a scaling operation based on a predetermined parameter. In addition, the predetermined parameter may be determined based on at least one of a size (eg, width and/or height) of a projection image and/or a face or a motion vector precision (eg, integer pels, half pels, quarter pels, etc.). can

Information indicating whether to resize the derived motion vector may be encoded and transmitted. The information may be a 1-bit flag, and the information may be signaled in at least one unit of a picture, a slice, a tile, a face, or a block.

The resizing of the motion vector may be selectively performed based on at least one of a motion vector derivation method (eg, merge mode or AMVP mode) or motion selector precision. The resizing of the motion vector may be applied to both the x-axis component and the y-axis component of the motion vector, or may be selectively applied to only one of the x-axis component and the y-axis component. When the motion vector is resized, the resized motion vector may be used to derive a motion vector of a neighboring block.

In the above example, it is assumed that a picture composed of a plurality of faces can be used as a reference picture. As another example, each face may be used as a reference picture, or a set of a predetermined number of faces may be used as a reference picture. Alternatively, in a 360-degree projection image based on TPP or the like, only the front face may be used as a reference picture, or the front face may be used as a reference picture, and a set of other faces may be used as a reference picture.

Although the above-described embodiment has been described based on a series of steps or a flowchart, this does not limit the time-series order of the invention, and may be performed simultaneously or in a different order, if necessary. In addition, each of the components (eg, unit, module, etc.) constituting the block diagram in the above-described embodiment may be implemented as a hardware device or software, or a plurality of components may be combined to form one hardware device or software. may be implemented. The above-described embodiment may be implemented in the form of program instructions that can be executed through various computer components and recorded in a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. Examples of the computer-readable recording medium include a hard disk, a magnetic medium such as a floppy disk and a magnetic tape, an optical recording medium such as a CD-ROM and DVD, and a magneto-optical medium such as a floppy disk. media), and hardware devices specially configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. The hardware device may be configured to operate as one or more software modules to perform processing according to the present invention, and vice versa.

Claims (12)

determining whether a padding area formed around a face boundary within a reference frame can be referenced during inter prediction;
obtaining motion information of a current block within a current frame;
determining a reference block of the current block in the reference frame based on the motion information; and
Comprising the step of obtaining a prediction block of the current block by using the reference block,
When it is determined that the padding area is unreferenceable, data included in the padding area is unavailable to obtain the prediction block,
Whether the padding region can be referred to during the inter prediction is determined based on a 1-bit flag parsed from a bitstream. The method of claim 1,
The padding area includes a horizontal padding area extending in a horizontal direction of the face and a vertical padding area extending in a vertical direction of the face,
Width information indicating a width of the horizontal padding region and height information indicating a height of the vertical padding region are signaled through the bitstream, respectively. The method of claim 1,
The padding area is generated based on samples belonging to an adjacent face having adjacency in 3D space to the face when reconstructed. The method of claim 1,
The padding area is characterized in that it is generated by interpolating an inner sample included in the padding area and an outer sample outside the padding area. The method of claim 1,
When it is determined that the padding area is unreferenceable and data of the padding area is included in the reference block specified based on the motion information, the data of the padding area in the reference block is replaced with available data. video decoding method. 6. The method of claim 5,
The available data is an image decoding method, characterized in that generated based on the reconstructed sample within the phase. determining whether a padding area formed around a face boundary within a reference frame can be referenced during inter prediction;
obtaining motion information of a current block within a current frame;
determining a reference block of the current block in the reference frame based on the motion information; and
Comprising the step of obtaining a prediction block of the current block by using the reference block,
When it is determined that the padding area is unreferenceable, data included in the padding area is unavailable to obtain the prediction block,
A one-bit flag indicating whether the padding region can be referred to during the inter prediction is explicitly encoded in a bitstream. delete delete delete delete delete

Images