RU2805157C1 - Method and device for image decoding based on intra prediction in image coding system - Google Patents

Abstract

FIELD: video encoding.

SUBSTANCE: method comprises building a list of most probable modes (MPM) for the current block based on the intra-prediction mode of the block adjacent to the current block; determining the intra-prediction mode of the current block based on the MPM list; deriving adjacent samples including left adjacent samples and top adjacent samples of the current block; deriving reference samples for predicting a target sample of the current block among neighbouring samples based on the position of the target sample and the prediction angle of the intra-prediction mode; determining an interpolation filter for the target sample based on the intra-prediction mode and the size of the current block, the interpolation filter being defined as one of a Gaussian filter and a cube filter; deriving a prediction sample of the target sample based on the interpolation filter and the reference samples; and encoding a bit stream including prediction mode information for the current block representing the intra-prediction mode of the current block among the MPM candidates in the MPM list, wherein determining an interpolation filter for the target sample based on the intra-prediction mode and the size of the current block comprises: determining whether the size of the current block is smaller than the first specific value; and, based on the current block size being smaller than the first specific value, determining whether the prediction angle of the intra-prediction mode is smaller than the second specific value, wherein the intra-prediction mode is one of the directional intra-prediction modes with angle greater than 0, wherein based on the current block size being greater than or equal to the first specific value, the interpolation filter for the target sample is output as a Gaussian filter, wherein based on the current block size being smaller than the first specific value, and the prediction angle of the intra-prediction mode being smaller than the second specific value, the interpolation filter for the target sample is output as a cube filter, and wherein based on the current block size being smaller than the first specific value, and the prediction angle of the mode of an intra-prediction being no less than the second specific value, the interpolation filter for the target sample is output as a Gaussian filter.

EFFECT: increased efficiency of video coding.

2 cl, 14 dwg, 3 tbl

Description

TECHNICAL FIELD TO WHICH THE DISCLOSURE RELATES

[1] The present disclosure relates to a video encoding method, and more specifically, to a method and apparatus for video decoding based on intra-prediction in a video encoding system.

BACKGROUND ART

[2] 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 fields. Since the image data is of high resolution and high quality, the amount of information or bits to be transmitted increases relative to the legacy image data. Therefore, when image data is transmitted using a medium such as a traditional wired/wireless broadband line, or image data is stored using an existing storage medium, transmission and storage costs increase.

[3] Accordingly, there is a need for a highly efficient image compression method for efficiently transmitting, storing and reproducing high resolution and high quality image information.

Brief description of the invention

[4] The present disclosure provides a method and apparatus for improving the efficiency of image encoding.

[5] The present disclosure also provides an intra-prediction method and apparatus for selecting an interpolation filter for a target sample in a current block.

[6] The present disclosure also provides a method and apparatus for performing intra-prediction based on an interpolation filter for a selected target sample.

[7] In one aspect, a video decoding method performed by a decoding apparatus is provided. The method includes deriving an intra-prediction mode of the current block; outputting neighboring samples including left neighboring samples and upper neighboring samples of the current block; deriving reference samples for predicting a target sample of the current block among neighboring samples based on the position of the target sample and the prediction angle of the intra-prediction mode; defining an interpolation filter for the target sample; and deriving a prediction sample of the target sample based on the interpolation filter and the reference samples.

[8] In another aspect, a decoding apparatus for performing image decoding is provided. The decoding apparatus includes: an entropy decoder obtaining prediction information for the current block; and a prediction module outputting an intra-prediction mode of the current block; outputting neighboring samples including left neighboring samples and upper neighboring samples of the current block, outputting reference samples for predicting a target sample among neighboring samples based on the position of the target sample of the current block and the prediction angle of the intra-prediction mode, determining an interpolation filter for the target sample, and outputting sample prediction of the target sample based on the interpolation filter and reference samples.

[9] In another aspect, a video encoding method performed by an encoding apparatus is provided. The method includes: determining an intra-prediction mode for the current block; outputting neighboring samples including left neighboring samples and upper neighboring samples of the current block; deriving reference samples for predicting a target sample of the current block among neighboring samples based on the position of the target sample and the prediction angle of the intra-prediction mode; defining an interpolation filter for the target sample; deriving a target sample prediction sample based on the interpolation filter and reference samples; and generating prediction information for the current block, encoding the generated prediction information, and outputting the encoded prediction information.

[10] In another aspect, a video encoding apparatus is provided. The encoding device includes: a prediction module determining an intra-prediction mode for a current block, outputting neighboring samples including left neighboring samples and top neighboring samples of the current block, outputting reference samples for predicting a target sample among neighboring samples based on the position of the target sample a current block and an intra-prediction mode prediction angle determining an interpolation filter for the target sample, and outputting a prediction sample of the target sample based on the interpolation filter and the reference samples; and an entropy encoder generating prediction information for the current block, encoding the generated prediction information, and outputting the encoded prediction information.

[11] According to the present disclosure, prediction may be performed on a target sample based on an interpolation filter selected according to current block size information, distance information to a reference sample, and/or prediction mode information, with the reference sample at a fractional sample position for the target sample can be accurately generated to improve the prediction accuracy for the current block, and the residual for the current block can be reduced to improve coding efficiency.

[12] According to the present disclosure, since the interpolation filter for the target sample can be selected based on various conditions described above, the number of bits of information regarding the selection of the interpolation filter can be reduced, thereby increasing the prediction accuracy of the current block and increasing the encoding efficiency of the current block.

BRIEF DESCRIPTION OF THE DRAWINGS

[13] FIG. 1 is a schematic diagram illustrating the configuration of a video encoding apparatus to which the present disclosure applies.

[14] FIG. 2 illustrates another example of a video encoding apparatus to which the present disclosure applies.

[15] FIG. 3 illustrates an example of a process for performing intra-prediction in an encoding apparatus.

[16] FIG. 4 is a schematic diagram illustrating the configuration of a video decoding apparatus to which the present disclosure applies.

[17] FIG. 5 illustrates another example of a video decoding apparatus to which the present disclosure applies.

[18] FIG. 6 illustrates an example of a process for performing intra-prediction in a decoding apparatus.

[19] FIG. 7 illustrates an example of left neighbor samples and top neighbor samples used for intra-prediction of the current block.

[20] FIG. 8 illustrates an example of intra-directional modes for 65 prediction directions.

[21] FIG. 9 illustrates an example in which the target sample prediction samples are derived based on integer samples adjacent to the left and right of the reference sample when the position of the reference sample located in the prediction direction of the directed intra-prediction mode is the position of a fractional sample.

[22] FIG. 10 illustrates an example of selecting an interpolation filter based on the current block size and intra-prediction mode.

[23] FIG. 11 illustrates an example of selecting an interpolation filter based on the distance between the target sample of the current block and the reference sample.

[24] FIG. 12 illustrates an example of deriving a target sample reference sample of a current block based on a plurality of interpolation filters and deriving a target sample prediction sample based on the reference sample.

[25] FIG. 13 schematically illustrates a video encoding method in accordance with the present disclosure.

[26] FIG. 14 schematically illustrates a video decoding method in accordance with the present disclosure.

Description of Exemplary Embodiments

[27] The present disclosure may be modified in various forms, and specific embodiments thereof will be described and illustrated in the drawings. However, the embodiments are not intended to limit the disclosure. The terms used in the following description are used to describe specific embodiments, but are not intended to limit the disclosure. A singular expression includes a plural expression if it is clearly read differently. Terms such as “include” and “have” are intended to indicate that the features, numbers, steps, operations, elements, components, or combinations thereof used in the following description exist, and thus it is to be understood that that the possibility of the existence or addition of one or more other features, numbers, steps, operations, elements, components or combinations thereof is not excluded.

[28] However, the elements in the drawings described in the disclosure are shown independently for convenience in explaining various specific functions and do not imply that the elements are implemented by independent hardware or independent software. For example, two or more elements of such elements may be combined to form one element, or one element may be divided into multiple elements. Embodiments in which elements are combined and/or separated are within the scope of the disclosure without departing from the spirit of the disclosure.

[29] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In addition, like reference numerals are used to indicate like elements throughout the drawings, and the same descriptions of like elements will be omitted.

[30] In this specification, the term “picture” generally refers to a unit representing an image at a particular time; a cutout is a unit constituting part of a picture. A single picture may be composed of multiple clippings, and the terms “picture” and “clipping” may be used interchangeably as specific circumstances require.

[31] A pixel or pel (picture element) can mean the smallest unit that makes up one picture (or image). Additionally, “sample” can be used as a term corresponding to a pixel. A sample may generally represent a pixel or a pixel value, may represent only a pixel (pixel value) of a luma component, and may represent only a pixel (pixel value) of a chrominance component.

[32] Unit indicates the basic unit of image processing. The unit may include at least one of a specific area and information associated with the area. Optionally, the term “unit” may be used interchangeably with terms such as block, area, or the like. In a typical case, an M×N block may represent a set of samples or transform coefficients arranged in M columns and N rows.

[33] FIG. 1 briefly illustrates the structure of a video encoding apparatus to which the present disclosure applies.

[34] With reference to FIG. 1, the video encoding device 100 may include a picture splitter 105, a prediction module 110, a remainder processor 120, an adder 150, a filter 255, and a memory 160. The remainder processor 120 includes a subtractor 121, a converter 122, a quantization module 123, a rearrangement module. 124, dequantization module 125, inverse converter 126.

[35] The picture division module 105 may divide the input picture into at least one processing unit.

[36] In one example, a processing unit may be referred to as a coding unit (CU). In this case, the coding unit can be recursively split from the largest coding unit (LCU) according to a quadtree/binary tree (QTBT) structure. For example, one encoding unit may be divided into multiple encoding units of greater depth based on a quadtree structure and/or a binary tree structure. In this case, for example, the quadtree structure may be applied first, and the binary tree structure may be applied later. Alternatively, the binary tree structure may be applied first. The encoding procedure in accordance with the present disclosure may be performed based on the final encoding unit, which is not further divided. In this case, the largest encoding unit may be used as the final encoding unit based on encoding efficiency or the like depending on the characteristics of the image, or the encoding unit may be recursively divided into lower depth encoding units as necessary, and the encoding unit having an optimal size , can be used as the final encoding unit. Here, the encoding procedure may include a procedure such as prediction, transformation and reconstruction, which will be described below.

[37] In another example, a processing unit may include a coding unit (CU), a prediction unit (PU), or a transformation unit (TU). A coding unit may be split from a largest coding unit (LCU) into coding units of greater depth according to a quadtree structure. In this case, the largest coding unit can be directly used as the final coding unit based on coding efficiency or the like depending on the characteristics of the image, or the coding unit can be recursively divided into coding units of greater depth as needed, and the coding unit having an optimal size , can be used as the final encoding unit. When the smallest coding unit (SCU) is set, the coding unit may not be divided into coding units smaller than the smallest coding unit. Here, a final coding unit refers to a coding unit that is divided or broken down into a prediction unit or a transformation unit. A prediction unit is a unit that is divided from a coding unit and may be a sample prediction unit. Here, the prediction unit may be divided into sub-blocks. The transform unit may be divided from the encoding unit in accordance with the quadtree structure and may be a unit for deriving a transform coefficient and/or a unit for deriving a residual signal from the transform coefficient. Further, a coding unit may be referred to as a coding block (CB), a prediction unit may be referred to as a prediction block (PB), and a transform unit may be referred to as a transform block (TB). A prediction block or prediction unit may refer to a specific block-shaped region in a picture and may include an array of prediction samples. Also, a transform block or transform unit may refer to a specific block-shaped region in a picture and may include a transform coefficient or an array of residual samples.

[38] The prediction unit 110 may perform prediction on a target processing block (hereinafter, the current block) and may generate a predicted block including prediction samples for the current block. The unit of prediction performed in the prediction unit 110 may be a coding block, or may be a transform block, or may be a prediction block.

[39] Prediction module 110 may determine whether intra-prediction is applied or inter-prediction is applied to the current block. For example, prediction module 110 may determine whether intra-prediction or inter-prediction should be applied to the CU.

[40] In the case of intra prediction, the prediction unit 110 may output a prediction sample for the current block based on a reference sample outside the current block in the picture to which the current block belongs (hereinafter, the current picture). In this case, the prediction module 110 may output a prediction sample based on the average or interpolation of neighboring reference samples of the current block (case (i)) or may output a prediction sample based on a reference sample existing in a particular direction (prediction) as a prediction sample among neighbors reference samples of the current block (case (ii)). Case (i) may be called non-directional mode or non-angular mode, and case (ii) may be called directional mode or angular mode. In intra-prediction, prediction modes may include, by way of example, 33 directional modes and at least two non-directional modes. Omnidirectional modes may include DC mode and planar mode. The prediction module 110 may determine a prediction mode to be applied to the current block using a prediction mode applicable to a neighboring block.

[41] In the case of inter-prediction, the prediction module 110 may output a prediction sample for the current block based on the sample specified by the motion vector on the reference picture. The prediction module 110 may output a prediction sample for the current block by applying any one of a skip mode, a combining mode, and a motion vector prediction (MVP) mode. In the case of the skip mode and the merge mode, the prediction unit 110 may use the motion information of the adjacent block as the motion information of the current block. In the skip mode, unlike the merge mode, the difference (residual) between the prediction sample and the original sample is not transmitted. In the case of MVP mode, the motion vector of an adjacent block is used as a motion vector predictor, and thus is used as a motion vector predictor of the current block to output the motion vector of the current block.

[42] In the case of inter-prediction, a neighbor block may include a spatial neighbor block existing in the current picture and a temporal neighbor block existing in the reference picture. A reference picture including a temporary adjacent block may also be called a correlated picture (colPic). The motion information may include a motion vector and a reference picture index. Information such as prediction mode information and motion information can be (entropy) encoded and then output in the form of a bit stream.

[43] When the motion information of a temporary adjacent block is used in the skip mode and the merge mode, the topmost picture in the list of reference pictures can be used as the reference picture. The reference pictures included in the reference picture list may be aligned based on the picture order count (POC) difference between the current picture and the corresponding reference picture. POC corresponds to display order and may differ from encoding order.

[44] Subtractor 121 generates a residual sample, which is the difference between the original sample and the prediction sample. If skip mode is used, a residual sample may not be generated as described above.

[45] Converter 122 converts the residual samples in transform block units to generate a transform coefficient. The transformer 122 may perform the transform based on the size of the corresponding transform block and the prediction mode applied to the encoding block or prediction block spatially overlapping with the transform block. For example, residual samples can be transformed using a discrete sine transform (DST) kernel if intra-prediction is applied to a coding block or prediction block overlapping with the transform block, and the transform block is a 4x4 residual array and is transformed using the kernel discrete cosine transform (DCT) in other cases.

[46] Quantization module 123 may quantize the transform coefficients to generate quantized transform coefficients.

[47] Rearrangement module 124 rearranges the quantized transform coefficients. The rearrangement module 124 may rearrange the quantized transform coefficients in a block form into a one-dimensional vector through a coefficient scanning method. Although rearrangement module 124 is described as a separate component, rearrangement module 124 may be part of quantization module 123.

[48] Entropy encoder 130 may perform entropy encoding on the quantized transform coefficients. Entropy coding may include a coding method such as exponential Golomb coding, context-adaptive variable-length coding (CAVLC), context-adaptive binary arithmetic coding (CABAC), or the like. The entropy encoder 130 may perform encoding jointly or separately on the information (eg, syntactic element value or the like) required for video reconstruction, in addition to the quantized transform coefficients. Entropy-encoded information can be transmitted or stored in a network abstraction layer (NAL) unit in the form of a bit stream.

[49] The inverse quantization unit 125 inversely quantizes the values (conversion coefficients) quantized by the quantization unit 123, and the inverse converter 126 inversely converts the values inversely quantized by the inverse quantization unit 125 to generate a residual sample.

[50] An adder 140 adds the residual sample with the prediction sample to reconstruct the picture. The residual sample can be summed with the prediction sample in block units to generate a reconstructed block. Although the adder 140 is described as a separate component, the adder 140 may be part of the prediction module 110. The adder 140 may be referred to as a reconstruction module or a reconstructed block generator.

[51] Filter 150 may apply deblocking filtering and/or sample-adaptive offset to the reconstructed picture. Block boundary artifacts in the reconstructed image or quantization distortion can be corrected through deblocking filtering and/or sampling-adaptive offset. Sampling-adaptive bias can be applied in sampling units after deblocking filtering is completed. Filter 150 may apply an adaptive loop filter (ALF) to the reconstructed picture. ALF can be applied to a reconstructed image to which deblocking filtering and/or sample-adaptive shifting has been applied.

[52] Memory 160 may store a reconstructed picture (decoded picture) or information necessary for encoding/decoding. Here, the reconstructed picture may be a reconstructed picture filtered by filter 150. The stored reconstructed picture may be used as a reference picture for (inter-)prediction of other pictures. For example, memory 160 may store (reference) pictures used for inter-prediction. Here, pictures used for inter-prediction may be specified in accordance with a set of reference pictures or a list of reference pictures.

[53] FIG. 2 illustrates another example of a video encoding apparatus to which the present disclosure applies.

[54] With reference to FIG. 2, the video encoding device includes an intra-prediction module, a reference smoother 200, a prediction module 210, a post filter 220, a converter 230, and a quantizer 240. Here, the intra-prediction module may include a reference smoother 200, a prediction module 210, and a post filter. 220.

[55] When intra-prediction is applied to the current block, the reference smoother 200 may perform the smoothing process on the left neighboring samples and the top neighboring samples used for intra-prediction of the current block in the picture (hereinafter referred to as the current picture) to which the current block belongs. based on the current block size, intra-prediction mode information, and sample value. Accordingly, it is possible to prevent visual artifacts regarding the prediction samples of the current block that may occur due to the difference between the sample values of the left neighboring samples and the upper neighboring samples.

[56] Prediction module 210 may (i) derive a prediction sample based on the average or interpolation of left neighbor samples and top neighbor samples of the current block or (ii) may derive prediction samples based on neighbor samples present in a particular direction (prediction) relative to the prediction samples among the left neighboring samples and the top neighboring samples. Case (i) may be called non-directional mode or non-angular mode, and case (ii) may be called directional mode or angular mode. In intra-prediction, a prediction mode may have 33 directional prediction modes and at least two non-directional modes. The omnidirectional mode may include a DC prediction mode and a Planar mode. The prediction module 210 may also determine the prediction mode applied to the current block using the prediction mode applied to the adjacent block.

[57] Post filter module 220 may selectively perform post-processing filtering to reduce discontinuity between the current block and adjacent samples in accordance with the prediction mode in which the prediction sample of the current block is output. Thereafter, the encoder can output, as a residual sample, the difference between the prediction sample and the original sample, and the converter 230 can convert the residual sample into block units to generate transform coefficients. Also, quantizer 240 may quantize the transform coefficients to generate quantized transform coefficients.

[58] FIG. 3 illustrates an example of a process for performing intra-prediction in an encoding apparatus. The encoding device may perform intra prediction to generate a prediction sample of the current block (S300). The prediction sample may be called a prediction signal or an intra-prediction signal. Specifically, the encoding device can smooth left adjacent samples and upper adjacent samples (ie, perform a smoothing process on them) used for intra-prediction of the current block based on the size of the current block, mode information and sample value (S310). Thereafter, as described above, the encoding apparatus may perform prediction in accordance with the intra-prediction mode to generate a prediction sample (S320), and perform post-processing filtering to reduce discontinuity between the current block and neighboring samples (S330). The encoding apparatus may generate, as a residual sample, the difference between the prediction sample and the original sample (S340) and convert the residual sample into block units to generate conversion coefficients. Further, the encoding device may quantize the transform coefficients to generate quantized transform coefficients (S360), and entropy encode the quantized transform coefficients to perform signaling (S370).

[59] FIG. 4 briefly illustrates the structure of a video decoding apparatus to which the present disclosure applies.

[60] With reference to FIG. 4, video decoding apparatus 400 may include an entropy decoder 410, a residual processor 420, a predictor 430, an adder 440, a filter 450, and a memory 460. The residual processor 420 may include a rearranger 421, an inverse quantizer 422, an inverter 423 .

[61] When a bitstream including video information is input, the video decoding apparatus 400 can reconstruct the video in association with the process by which the video information is processed in the video encoding apparatus.

[62] For example, the video decoding apparatus 400 may perform video decoding using a processing unit applied in the video encoding apparatus. Thus, a block of video decoding processing units may be, for example, an encoding unit and, in another example, an encoding unit, a prediction unit, or a transform unit. The coding unit may be split from the largest coding unit according to a quadtree structure and/or a binary tree structure.

[63] A prediction unit and a transformation unit may be further used in some cases, in which case, a prediction block is a block output or divided from a coding unit, and may be a sample prediction unit. Here, the prediction unit may be divided into sub-blocks. The transform unit may be decomposed from the encoding unit according to a quadtree structure, and may be a unit that outputs a transform coefficient or a unit that outputs a residual signal from the transform coefficient.

[64] Entropy decoder 410 may analyze the bit stream to output information required for video reconstruction or picture reconstruction. For example, entropy decoder 410 may decode information in the bit stream based on an encoding method such as exponential Golomb coding, CAVLC, CABAC, or the like, and may output a syntax element value required for video reconstruction and a quantized transform coefficient value with respect to the remainder.

[65] More specifically, the CABAC entropy encoding method can take a bin corresponding to each syntactic element in a bitstream, determine a context model using decoding the information of the target syntactic element and decoding the information of neighboring ones, and decoding the target blocks or the symbol/bin information decoded in the previous step , predict the probability of generating a bin according to a certain context model, and perform arithmetic decoding of the bin to generate a symbol corresponding to each value of the syntactic element. Here, the CABAC entropy decoding method can update the context model using the symbol/bin information decoded for the context model of the next symbol/bin after determining the context model.

[66] Prediction information among the information decoded in the entropy decoder 410 can be provided to the prediction module 450, and residual values, that is, quantized transform coefficients on which entropy decoding has been performed by the entropy decoder 410, can be input to the rearrangement module 421.

[67] Rearrangement module 421 may rearrange the quantized transform coefficients into a two-dimensional block form. The rearrangement module 421 may perform rearrangement corresponding to the coefficient scan performed by the encoder. Although rearrangement module 421 is described as a separate component, rearrangement module 421 may be part of inverse quantization module 422.

[68] The inverse quantization module 422 may dequantize the quantized transform coefficients based on the (de)quantization parameter to output the transform coefficient. In this case, information for outputting a quantization parameter may be signaled from the encoding device.

[69] Inverter 423 may inversely transform the transform coefficients to output residual samples.

[70] Prediction module 430 may perform prediction on the current block and may generate a predicted block including prediction samples for the current block. A unit of prediction performed in prediction unit 430 may be an encoding block, or may be a transform block, or may be a prediction block.

[71] The prediction module 430 may determine whether intra-prediction or inter-prediction should be applied based on the prediction information. In this case, the unit for determining what will be used from the intra-prediction and inter-prediction may be different from the unit for generating the prediction sample. In addition, the unit for generating the prediction sample may also be different in inter-prediction and intra-prediction. For example, what will be applied from inter-prediction and intra-prediction can be determined in a CU unit. Additionally, for example, in inter-prediction, a prediction sample can be generated by determining a prediction mode in the PU, and in intra-prediction, a prediction sample can be generated in a TU by determining a prediction mode in the PU.

[72] In the case of intra-prediction, prediction module 430 may derive a prediction sample for the current block based on an adjacent reference sample in the current picture. Prediction module 430 may derive a prediction sample for the current block by applying a directed mode or an undirected mode based on a neighboring reference sample of the current block. In this case, the prediction mode to be applied to the current block can be determined by using the intra-prediction mode of the neighboring block.

[73] In the case of inter-prediction, the prediction unit 430 may output a prediction sample for the current block based on the sample specified in the reference picture according to the motion vector. The prediction module 430 may output a prediction sample for the current block using one of a skip mode, a merge mode, and an MVP mode. Here, motion information required for inter-prediction of the current block provided by the video encoding device, for example, motion vector and reference picture index information, can be obtained or output based on the prediction information.

[74] In the skip mode and the merge mode, the motion information of an adjacent block can be used as the motion information of the current block. Here, the neighbor block may include a spatial neighbor block and a temporal neighbor block.

[75] The prediction unit 430 may construct a merging candidate list using the motion information of available neighboring blocks, and use the information indicated by the merging index in the merging candidate list as a motion vector of the current block. The association index may be signaled by the encoder. The motion information may include a motion vector and a reference picture. When the motion information of the temporary neighboring block is used in the skip mode and the merge mode, the topmost picture in the list of reference pictures can be used as the reference picture.

[76] In the case of the skip mode, the difference (residual) between the prediction sample and the original sample is not transmitted, unlike the merge mode.

[77] In the case of MVP mode, the motion vector of the current block can be output by using the motion vector of the adjacent block as the motion vector predictor. Here, the neighbor block may include a spatial neighbor block and a temporal neighbor block.

[78] When the merging mode is applied, for example, a merging candidate list may be generated using the motion vector of the reconstructed spatial neighbor block and/or the motion vector corresponding to the Col block, which is a temporal neighbor block. The motion vector of a candidate block selected from the merging candidate list is used as the motion vector of the current block in merging mode. The above prediction information may include a merging index indicating a candidate block having a best motion vector selected from the candidate blocks included in the merging candidate list. Here, the prediction unit 430 may output the motion vector of the current block using the merging index.

[79] When the MVP (Motion Vector Prediction) mode is applied as another example, a motion vector prediction module candidate list may be generated using the motion vector of the reconstructed spatial neighbor block and/or the motion vector corresponding to the Col block, which is a temporal neighbor block. That is, the motion vector of the reconstructed spatial neighbor block and/or the motion vector corresponding to the block Col, which is a temporal neighbor block, can be used as motion vector candidates. The above prediction information may include a prediction motion vector index indicating the best motion vector selected from motion vector candidates included in the list. Here, the prediction unit 430 may select a prediction motion vector of the current block from the motion vector candidates included in the motion vector candidate list using the motion vector index. The encoder predictor may obtain a motion vector difference (MVD) between the motion vector of the current block and the motion vector predictor, encode the MVD, and output the encoded MVD in the form of a bitstream. That is, the MVD can be obtained by subtracting the motion vector prediction module from the motion vector of the current block. Here, the prediction unit 430 can obtain the motion vector included in the prediction information and output the motion vector of the current block by adding the difference of the motion vectors with the motion vector prediction unit. In addition, the prediction module can obtain or output a reference picture index indicating the reference picture from the above prediction information.

[80] Adder 440 may add the residual sample with the prediction sample to reconstruct the current block or current picture. Adder 440 may reconstruct the current picture by summing the residual sample with the prediction sample in block units. When the skip mode is applied, the remainder is not transmitted, and thus the prediction sample may become a reconstructed sample. Although the adder 440 is described as a separate component, the adder 440 may be part of the prediction module 430. The adder 440 may be referred to as a reconstruction module or a reconstructed block generator.

[81] Filter 450 may apply deblocking filtering, adaptive sampling offset, and/or ALF to the reconstructed picture. Here, adaptive sampling bias can be applied in sampling units after deblocking filtering. ALF can be applied after deblocking filtering and/or applying adaptive sampling bias.

[82] Memory 460 may store a reconstructed picture (decoded picture) or information necessary for decoding. Here, the reconstructed picture may be a reconstructed picture filtered by filter 450. For example, memory 460 may store pictures used for inter-prediction. Here, pictures used for inter-prediction may be specified in accordance with a set of reference pictures or a list of reference pictures. The reconstructed picture can be used as a reference picture for other pictures. The memory 460 may output the reconstructed pictures in output order.

[83] FIG. 5 illustrates another example of a video decoding apparatus to which the present disclosure applies.

[84] With reference to FIG. 5, the video encoding device includes an intra-prediction module, a reference smoother 500, a prediction module 810, a post filter 520, an inverse quantization module 530, and an inverse converter 540. Here, the intra-prediction module may include a reference smoother 500, a prediction module 510 and a post filter 520. The intra-prediction module may output a prediction sample for the current block by applying a directed mode or an undirected mode based on a neighboring reference sample of the current block. Here, the prediction mode to be applied to the current block can be determined using the intra-prediction mode of the neighboring block.

[85] Specifically, when intra-prediction is applied to the current block, the reference smoother 500 may perform the smoothing process on the left neighboring samples and the upper neighboring samples used for intra-prediction of the current block in the picture (hereinafter referred to as the current picture) to which the current one belongs. block, based on the current block size, prediction mode, and sample value. In this way, it is possible to prevent visual artifacts regarding the prediction samples of the current block, which may occur due to differences between the sample values of the left neighboring samples and the top neighboring samples.

[86] Prediction module 510 may output a prediction sample based on the average or interpolation of left neighbor samples and top neighbor samples of the current block (case (i)) or may output a prediction sample based on a neighbor sample present in a particular direction (prediction) relative to the prediction sample among the left neighboring samples and the top neighboring samples (case (ii)). Case (i) may be called non-directional mode or non-angular mode, and case (ii) may be called directional mode or angular mode. In intra-prediction, a prediction mode may include, for example, 33 directional prediction modes and at least two non-directional modes. Omnidirectional modes may include DC prediction mode and planar mode. Prediction module 510 may determine a prediction mode to be applied to the current block using the prediction mode applied to an adjacent block.

[87] Post filter module 520 may selectively perform post-processing filtering to reduce discontinuity between the current block and adjacent samples in accordance with the prediction mode in which the prediction sample of the current block is output. Thereafter, the inverse quantizer 530 may inversely quantize the quantized transform coefficients received from the encoder, and the inverse transformer 540 may inversely transform the quantized transform coefficients to generate residual samples in block units. The decoding apparatus can reconstruct the current intra-prediction encoded block based on the residual sample and the prediction sample.

[88] FIG. 6 illustrates an example of a process for performing intra-prediction in a decoding apparatus. The decoding apparatus entropy decodes the entropy encoded information received by the bit stream to obtain quantized transform coefficients (S600). Next, the decoding apparatus may inversely quantize the quantized transform coefficients to obtain transform coefficients (S610), and inversely transform the transform coefficients to generate a residual sample in block units (S620). Next, the decoding apparatus may perform intra prediction to generate a prediction sample of the current block (S630). The prediction sample may be referred to as a prediction signal or an intra-prediction signal. Specifically, the decoding apparatus may perform a smoothing process on the left neighbor samples and the top neighbor samples used for intra-prediction of the current block based on the size of the current block, the prediction mode, and the sample value (S640). Thereafter, the decoding apparatus may perform prediction in accordance with the intra-prediction mode to generate a prediction sample as described above (S650), and perform post-processing filtering to reduce discontinuity between the current block and adjacent samples (S660). The decoding apparatus may add the prediction sample and the residual sample to generate a reconstructed sample of the current block (S670).

[89] When prediction is performed on the current block as described above, prediction may be performed based on an intra-prediction mode. For example, intra-prediction may be performed based on a neighboring sample that has already been encoded/decoded during decoding of the current block. That is, the prediction sample of the current block can be reconstructed using the left neighbor samples and the top neighbor samples of the current block. Left adjacent samples and top adjacent samples can be represented as shown in FIG. 7.

[90] FIG. 7 illustrates an example of left neighbor samples and top neighbor samples used for intra-prediction of the current block. When intra-prediction is performed on the current block, an intra-prediction mode with respect to the current block may be output, and a prediction sample with respect to the current block may be generated using at least one of the left neighbor samples and the top neighbor samples in accordance with the intra-prediction mode . The left neighbor samples and the top neighbor samples used for intra-prediction of the current block may undergo a smoothing process based on the current block size, prediction mode, and sample value. That is, filtering may be performed to reduce the difference between the sample values of the left neighbor samples and the top neighbor samples based on the current block size, prediction mode, and sample values. In this way, visual artifacts with respect to the prediction samples of the current block that may occur due to differences between the sample values of the left neighbor samples and the top neighbor samples can be prevented.

[91] Here, the intra-prediction modes may include two non-directional intra-prediction modes and 33 directed intra-prediction modes. Non-directional intra-prediction modes may include a planar intra-prediction mode and a DC intra-prediction mode, and directed intra-prediction modes may include intra-prediction modes #2-#34. The planar intra-prediction mode may be referred to as the planar mode, and the DC intra-prediction mode may be referred to as the DC mode. Intra-prediction mode #10 may indicate a horizontal intra-prediction mode or horizontal mode, intra-prediction mode #26 indicates a vertical intra-prediction mode or a vertical mode, based on which the directional intra-mode prediction direction can be expressed by an angle. In other words, the relative angle corresponding to each intra-prediction mode can be expressed with reference to the horizontal reference angle 0° corresponding to intra-prediction mode #10, and the relative angle corresponding to each intra-prediction mode can be expressed with reference to vertical reference angle 0°, corresponding to intra-prediction mode #26.

[92] Additionally, the requirement for high quality video is increased, and in order to improve the efficiency of the video codec, the number of directional intra-prediction directions may be increased to 65. That is, the intra-prediction mode may include two omnidirectional intra-prediction modes and 65 directional intra-prediction modes. intra-prediction modes. Non-directional intra-prediction modes may include a planar intra-prediction mode and a DC intra-prediction mode, and directed intra-prediction modes may include intra-prediction modes #2-#66.

[93] FIG. 8 illustrates intra-directional modes of 65 direction predictions.

[94] With reference to FIG. 8, intra-prediction modes having a horizontal direction and intra-prediction modes having a vertical direction can be classified based on the intra-prediction mode #34 having an upper left diagonal prediction direction. H and V in Fig. 8 represents the horizontal directivity and vertical directivity, respectively, and the numbers -32 to 32 represent 1/32 unit offsets at the sample grid positions. Intra-prediction modes #2-#33 are horizontally directed, and intra-prediction modes #34-#66 are vertically directed. The intra-prediction mode #18 and the intra-prediction mode #50 represent the horizontal intra-prediction mode and the vertical intra-prediction mode, respectively, based on which the prediction direction of the angular intra-prediction mode can be expressed by an angle. In other words, the relative angle corresponding to each intra-prediction mode can be expressed based on the horizontal reference angle of 0° corresponding to the intra-prediction mode #18, and the relative angle corresponding to each intra-prediction mode can be expressed based on the vertical reference 0° angle corresponding to intra-prediction mode #50.

[95] In the case where the directed intra-prediction mode is applied to the current block, the prediction sample of the target sample may be output based on a reference sample located in the prediction direction of the directed intra-prediction mode relative to the target sample on which intra-prediction is performed in the current block . That is, a reference sample located in the prediction direction can be copied and output as a prediction sample. Here, the reference sample may represent a neighboring sample located in the directional intra-prediction mode prediction direction relative to the target sample among the upper neighboring samples and the left neighboring samples of the current block. Here, when there is no reference sample in units of integer samples in the direction of the target sample based on the directional intra-prediction mode prediction, that is, when the position of the reference sample located in the direction of the directional intra-prediction mode prediction is the position of a fractional sample based on the target samples, the sample value of the reference sample can be derived through interpolation between integer samples adjacent to the left and right of the reference sample, and the prediction sample of the target sample can be derived based on the reference sample. For example, interpolation between integer samples can be performed based on the distance ratio of the reference sample and the integer samples.

[96] FIG. 9 is a view illustrating an example in which, when the position of a reference sample located in the prediction direction of the directed intra-prediction mode is a fractional sample position, the prediction sample of the target sample is derived based on the entire samples adjacent to the left and right of the reference sample.

[97] With reference to FIG. 9, the position of the fractional sample of the reference sample located in the prediction direction of the directed intra-prediction mode based on the target sample can be output as tanθ*(y+1). The tanθ value for each angle θ of each directional intra-prediction mode for computing the fractional sample position may be scaled in advance in units of integers and determined to facilitate the calculation. The tanθ values of the corresponding scaled directional intra-prediction modes can be output as shown in the following table.

[98]

[99] Here, predModeIntra may represent each directional intra-prediction mode, and intraPredAngle may represent the prediction angle of each directional intra-prediction mode or the scaled approximate value tanθ of each directional intra-prediction mode. The approximate value of tanθ according to the predetermined intra-prediction mode can be output based on Table 1. Meanwhile, the value of tan ^-1 θ of each scaled directional intra-prediction mode can be output as shown in the following table.

[100]

[101] Here, predModeIntra may represent each directional intra-prediction mode, intraPredAngle may represent the inverse prediction angle of each directional intra-prediction mode, or the scaled approximation tan ^-1 θ of each directional intra-prediction mode. The approximate value of tan ^-1 θ can be derived according to the predefined intra-prediction mode based on Table 2.

[102] In this case, the non-directional intra-prediction mode can be applied to the current block. Non-directional intra-prediction modes may include planar intra-prediction mode and DC intra-prediction mode. The planar intra-prediction mode may be referred to as the planar mode, and the DC intra-prediction mode may be referred to as the DC mode. In DC mode, the prediction sample of the current block can be derived based on the average of the neighboring samples of the current block. Intra-prediction based on DC mode can be efficiently performed when the sample values of the current block are similar. However, when intra prediction is performed based on the DC mode, when the sample values of the current block vary, discontinuity may occur between the predicted block of the current block and neighboring samples. In a similar case, even when intra-prediction is performed based on the directed intra-prediction mode, unintentional visible delineation may occur. Planar mode was developed to overcome this problem. The planar mode indicates a prediction mode in which horizontal linear prediction and vertical linear prediction are performed based on reference samples with respect to the target sample, and the output values are then averaged to generate a prediction sample of the target sample.

[103] When prediction is performed on the current block based on the directional intra-prediction mode, if a reference sample in units of integer samples is not present in the directional intra-prediction mode prediction direction of the current block relative to the target sample of the current block as described above, that is, when the position of the reference sample located in the prediction direction of the directional intra-prediction mode relative to the target sample is the position of the fractional sample, the sample value of the reference sample can be output by interpolating the left and right integer samples of the reference sample, and the prediction sample of the target sample can be output based on the derived reference samples. Whole samples may indicate neighboring samples of the position of the whole sample located near the reference sample.

[104] In this case, interpolation between the integer samples to the left and right of the reference sample can be obtained based on one of various interpolation filters. For example, the interpolation may be performed based on an interpolation filter having a low-pass filter effect, or the interpolation may be performed based on a complex interpolation filter. An interpolation filter having a low-pass filter effect may be a linear filter or a Gaussian filter, and a complex interpolation filter may be a spline filter. A spline filter may also be called a cubic filter. The interpolation filters can be 4-tap interpolation filters. A 4-tap interpolation filter may represent a filter on which interpolation is performed on four integer samples based on four weights. Interpolation between integer samples performed based on the interpolation filter can be expressed by the following equation.

[105] [Equation 1]

[106] Here, p[x][y] may denote the prediction sample of the target sample, f[0], f[1], f[2] and f[3] may denote the filter coefficients of the interpolation filter, ref[n] may denote the nth neighboring sample, and iIdx may denote an integer position index of a fractional sample located in the prediction direction of the intra-prediction mode of the current block relative to the target sample. The integer fractional sample position index may represent an integer value excluding the remainder of the fractional sample position.

[107] Here, the filter coefficient of a cubic filter, which is one of the low-pass filters, and the filter coefficient of the Gaussian filter, which is one of the interpolation filters having a low-pass filter effect, can be output as shown in the following table.

[108]

[109] Here, the sub-pel (sub-pixel) position n/32 may represent a residual value of a position of a fractional sample located in the intra-prediction mode prediction direction of the current block relative to the target sample. Based on the position of the fractional sample of the target sample and the above Table 3, the filter coefficients of the cubic filter or the filter coefficients of the Gaussian filter can be derived.

[110] When prediction is performed on the current block based on the directional intra-prediction mode as described above, the distance between the target sample of the current block and the reference sample may increase according to the prediction angle of the directional intra-prediction mode, and as the distance increases, the prediction accuracy can deteriorate. A method for selecting a suitable interpolation filter according to the distance between a target sample and a reference sample and performing prediction based on the reference sample derived by applying the selected interpolation filter may be proposed to improve the prediction accuracy. A method for selecting a suitable interpolation filter according to the distance between the target sample and the reference sample may be as follows.

[111] For example, the interpolation filter may be selected based on the size of the current block or the intra-prediction mode of the current block. As described above, the distance between the target sample of the current block and the reference sample may be output in accordance with the prediction angle slope of the intra-prediction mode for the current block. Since the reference sample of the current block is inferred based on the left neighboring samples and the top neighboring samples of the current block, the distance between the target sample and the reference sample may increase as the position of the target sample becomes closer to the lower right end of the current block. Also, as the value of intraPredAngle of the directional intra-prediction mode defined in Table 1 increases, the slope of the prediction angle can be close to 45°. Since the slope of the prediction angle is close to 45°, the distance between the target sample and the reference sample may increase, and accordingly, as the value of intraPredAngle increases, the distance between the target sample and the reference sample may increase.

[112] In addition, the distance between the target sample and the reference sample may be inferred based on the size of the current block. That is, as the size of the current block increases, the distance between the target sample and the reference sample may increase. Therefore, the size of the current block is considered to be closely related to the prediction accuracy of the target sample.

[113] If the intraPredAngle value is greater than 0 and less than 32 as described above, the target sample can be predicted based on the reference sample of the position of the fractional sample, as shown in FIG. 9. In this case, since only the integer sample value near the fractional sample position is represented, the encoding device can predict the fractional sample location of the reference sample based on the interpolation filter, and the reference sample value of the predicted position of the fractional sample can be copied as the sample value for the prediction sample of the target sample. Thus, the accuracy of the predicted block of the current block can be determined in accordance with the accuracy of the interpolation filter.

[114] When intra-prediction is applied to the current block, the information that can be used for intra-prediction can be limited to the left neighboring samples and the top neighboring samples of the current block that have already been reconstructed during decoding of the current block, and as the distance increases between the target sample of the current block and the reference sample, the correlation between the target sample and the reference samples derived from the left neighbor samples and the top neighbor samples may be significantly reduced.

[115] Accordingly, when the distance between the target sample and the reference sample is too large, a reference sample output method based on an interpolation filter having a low-pass filter effect so that artifacts or noise of the reference sample is not propagated, so as to improve prediction accuracy and coding efficiency . On the contrary, when the distance between the target sample and the reference sample is close to each other, the correlation between the target sample and the reference sample is high, and thus, the reference sample inference method based on precise interpolation can be advantageously used, so that the similarity between the prediction sample of the target sample and the reference sample sample is supported to the greatest extent possible. That is, when the distance between the target sample and the reference sample is close to each other, the reference sample inference method based on complex interpolation filter can improve the prediction accuracy and coding efficiency.

[116] Therefore, the current block may be selected based only on the size of the current block to output a reference sample for the target sample of the current block, may be selected based on only the intra-prediction mode of the current block, or selected based on the intra-prediction mode of the current block.

[117] For example, when the size of the current block is a size of 4×4 and intra-prediction is performed on the current block, the correlation of the current block having a size of 4×4 with the neighboring samples of the current block may be very high, and thus the reference sample can be inferred based on a complex interpolation filter regardless of the intra-prediction mode. Alternatively, if the intraPredAngle value derived from the intra-prediction mode of the current block is 11 or more regardless of the size of the current block, the distance between the target sample and the reference sample may increase, and thus the reference sample may be output based on an interpolation filter having interpolation filter effect. Alternatively, when the size of the current block is smaller than the specific size, and the intraPredAngle value of the intra-prediction mode of the current block is smaller than the specific value, a reference sample of the target sample may be output based on a complex interpolation filter, and in other cases, a reference sample of the target sample may be output based on an interpolation filter having a low-pass filter effect.

[118] In addition, when a mode such as the most likely mode (MPM) is applied to the current block to infer the intra-prediction mode of the current block based on the intra-prediction mode of a neighboring block of the current block, and the intra-prediction mode of the current block represents is a directional intra-prediction mode, not a planar mode or a DC mode, the interpolation filter used in an adjacent block selected through the MPM mode can be output as the interpolation filter of the current block. Here, in the case where the MPM mode is applied to the current block, the encoding device may determine the MPM list based on the intra-prediction mode regarding the left or upper neighboring blocks of the current block, and determine the intra-prediction mode based on the MPM list.

[119] In addition, when the interpolation filter is selected based on the intra-prediction mode of the current block, the intra-prediction mode reference, that is, the reference for determining whether an interpolation filter having a low-pass filter effect is used or whether a complex interpolation filter is used, can change according to the size or shape of the current block.

[120] In the case where the current block is a square block, the width and height of the block are equal, that is, the size of the current block is N×N, and thus the size of the reference block when selecting the interpolation filter may be N for the directional mode intra-predictions of any direction. Here, in the case where the current block has a non-square shape, that is, in the case where the size of the current block is M×N, if the mode selected as the prediction mode of the current block is the directed intra-prediction mode, and this mode is a vertical directional prediction mode, the size of the reference block for selecting the interpolation filter may be represented by M. Here, the vertical directional prediction mode may indicate intra-prediction modes #34 to #66, when the intra-prediction mode includes 65 directional intra-prediction modes and two non-directional intra-prediction modes. Similarly, when the size of the current block is M×N, the mode selected as the prediction mode of the current block is a directional mode, and the mode is a horizontal directional prediction mode, the size of the current block serving as a reference for selecting the interpolation filter may represent is N. Here, the horizontal directional prediction mode may represent intra-prediction modes #2 to #33, where the intra-prediction mode includes 65 directional intra-prediction modes and two non-directional intra-prediction modes.

[121] Alternatively, if the current block is a non-square block having size M×N, and the prediction mode of the current block is a vertical directional prediction mode, the interpolation filter of the current block can be selected based on N, and similarly if the current block is a non-square block having a size of M×N, and the prediction mode of the current block is a horizontal directional prediction mode, the interpolation filter of the current block can be selected based on M. However, in the specific example that will be described below, when the vertical directional prediction mode is applied to the current block having size M×N, the size of the current block serving as a reference for selecting the interpolation filter can be represented by M. Similarly, when the horizontal directional prediction mode is applied to the current block having size M×N, the size of the current block serving as the reference to select an interpolation filter, may be represented by N. Specifically, for example, if the block size value is less than or equal to 8, a complex interpolation filter may be selected, and a block reference sample may be output based on the complex interpolation filter, and in this case, if the size of the current block is 8×4 and the intra-prediction mode of the current block is one of the intra-prediction modes having a vertical direction, a complex interpolation filter can be selected as an interpolation filter for the current block, and a reference sample of the current block can be output based on the complex interpolation filter .

[122] Also, if the value intraPredAngle of the intra prediction mode of the current block is less than or equal to 11, a complex interpolation filter is selected, and a reference sample can be output based on the complex interpolation filter. If the intraPredAngle value of the intra prediction mode of the current block is greater than 11, an interpolation filter having a low-pass filter effect is selected, and a reference sample can be output based on the interpolation filter having a low-pass filter effect.

[123] In the case where the size value of the current block is equal to or greater than 16, if the intraPredAngle value of the intra-prediction mode of the current block is less than or equal to 5, a complex interpolation filter is selected, and a reference sample can be output based on the complex interpolation filter. If the value intraPredAngle of the intra prediction value of the current block is greater than 5, an interpolation filter having a low-pass filter effect is selected, and a reference sample can be output based on the interpolation filter having a low-pass filter effect.

[124] FIG. 10 illustrates an example of selecting an interpolation filter based on the current block size and intra-prediction mode. The encoding device/decoding device can output an intra-prediction mode for the current block and can determine whether the intra-prediction mode is a directed intra-prediction mode (S1000). The directional intra-prediction mode may be referred to as angular prediction. If the intra-prediction mode is a non-directional intra-prediction mode, the encoding/decoding device may perform intra-prediction of the current block based on the intra-prediction mode.

[125] When the intra-prediction mode is a directed intra-prediction mode, the encoding device/decoding device can determine whether the size of the current block is smaller than the first threshold (S1010). When the current block is a non-square block having a size of M×N, and the mode selected as the intra-prediction mode of the current block includes an intra-prediction mode having a vertical direction, that is, when the intra-prediction mode includes 65 directed intra-prediction modes and two non-directional intra-prediction modes, if the intra-prediction mode of the current block is one of the intra-prediction modes #34-#66, the support for selecting the interpolation filter can be represented by the width of the current block, i.e. ., M. Similarly, in the case where the current block is a non-square block having size M×N, and the mode selected as the intra-prediction mode of the current block is an intra-prediction mode having a horizontal orientation, that is, When the intra-prediction mode includes 65 directed intra-prediction modes and two non-directional intra-prediction modes, if the intra-prediction mode of the current block is one of the intra-prediction modes #2-#33, the reference for selecting the interpolation filter can be is represented by the height of the current block, i.e., N. Alternatively, when only a block having a square shape is considered, and the size of the current block is N×N, the size value of the current block can be represented by N. The first threshold can be set to 4. 8, 16, 32 or the like.

[126] If the size of the current block is not less than the value of the first threshold, that is, if the size of the current block is greater than or equal to the value of the first threshold, the encoding device/decoding device may select a Gaussian filter as the interpolation filter of the current block and output a reference sample of the target sample in the current block based on a Gaussian filter (S1020). Here, the Gaussian filter is one of the interpolation filters having a low-pass filter effect, and a reference sample of the target sample may be output based on the interpolation filter having a low-pass filter effect other than the Gaussian filter. For example, if the size of the current block is greater than or equal to the first threshold, the encoding device/decoding device may select the interpolation filter of the current block as a linear filter and output a reference sample of the target sample in the current block based on the linear filter. Here, the reference sample may indicate a neighboring sample located in the directional intra-prediction mode prediction direction of the current block based on the target sample.

[127] If the size of the current block is not smaller than the value of the first threshold, it can be determined whether the intraPredAngle of the directed intra-prediction mode of the current block is smaller than the value of the second threshold (S1030). intraPredAngle may indicate the prediction angle of the directional intra-prediction mode. For example, the second threshold could be set to 11.

[128] If the intraPredAngle of the intra-prediction directional mode of the current block is not less than the value of the second threshold, the encoding device/decoding device may select a Gaussian filter as an interpolation filter of the current block and output a reference sample of the target sample in the current block based on the Gaussian filter (S1020 ). Here, the Gaussian filter is one of the interpolation filters having a low-pass filter effect as described above, and a reference sample of the target sample may be output based on the interpolation filter having a low-pass filter effect other than the Gaussian filter.

[129] If the intraPredAngle of the directional intra prediction mode of the current block is less than the second threshold value, the encoding device/decoding device may select the cube filter as the interpolation filter of the current block and output a reference sample of the target sample in the current block based on the cube filter (S1040) . Here, the cubic filter may be one of the complex interpolation filters as described above, and the reference sample of the target sample may be derived based on the complex interpolation filter other than the cubic filter. Additionally, a cubic filter may be called a spline filter.

[130] The encoding device/decoding device may generate a target sample prediction sample based on the derived reference sample of the target sample (S1050). The encoder/decoder may generate the prediction sample by copying the reference sample. The reference sample can be copied and used as a prediction sample, and thus, the prediction sample can be indicated to be generated based on the interpolation filter.

[131] Here, in addition to the above examples, a method for dividing a current block into arbitrary regions and selecting an interpolation filter for each region may be proposed as a method for selecting an interpolation filter.

[132] For example, if the size of the current block is greater than or equal to a particular size, the current block may be divided into multiple regions, and the interpolation filter of each region may be selected based on the distance between each region and adjacent samples of the current block. The size of the divided areas of the current block may be a fixed value previously determined (ie, set previously) between the encoding device and the decoding device, or may be derived based on the size of the current block, the intra-prediction mode and the like. For example, in the case where the intra prediction mode of the current block is an intra prediction mode having a vertical direction, and the mode number for the intra prediction mode is greater than #34, the size of the divided areas of the current block may be output as 4×4. In other words, in the case where the intra prediction mode of the current block is one of the intra prediction modes #35 to #66, the size of the divided areas of the current block may be output as 4x4. In the case where the current block is a block having a size of 16x16, the current block can be divided into regions having a size of 4x4, the above-described complex interpolation filter can be selected as the interpolation filter for regions #0 to #7 in raster order scanning, and an interpolation filter having a low-pass filter effect can be selected as interpolation filters for other areas. Here, the 16 area numbers having a size of 4x4 according to the raster scanning order may be sequentially output in order from the top line to the bottom line, and sequentially output in the order from left to right in each line. That is, the areas included in the first row among the 16 areas having a size of 4x4 of the current block can be represented as area #0, area #1, area #2 and area #3 in order from left to right, areas included in the second line can be represented as area #4, area #5, area #6 and area #7 in order from left to right, areas included in the third line can be represented as area #8, area #9, area #10 and area #11 is in order from left to right, and the areas included in the fourth row can be represented as area #12, area #13, area #14 and area #15 in order from left to right. This may signal the sizes of the divided regions of the current block and information indicating the interpolation filter for each of the regions. In this case, the decoding apparatus may divide the current block into a plurality of regions based on the information and select an interpolation filter of each region.

[133] Alternatively, the interpolation filter may be selected based on the distance between the target sample of the current block and the reference sample. That is, the interpolation filter may be selected based on whether the distance between the target sample and the reference sample is greater than or equal to a particular threshold value. Here, the reference sample may indicate a neighboring sample located in the intra-prediction mode prediction direction of the current block based on the target sample.

[134] For example, in the case where the size of the current block is N×N, if the distance between the target sample of the current block and the reference sample is N/2 or more, the reference sample may be output based on an interpolation filter having a low-pass filter effect, and the reference sample may be derived from a complex interpolation filter in other cases. The specific threshold value for selecting the interpolation filter may be output based on the size of the current block, as described above, or may be output based on the intra-prediction mode of the current block, depending on whether the current block is a square block/non-square block, or the like. Alternatively, information about the specific threshold value may be transmitted from the encoding device, and the decoding device may output the specific threshold value of the current block based on the information about the received specific threshold value.

[135] FIG. 11 illustrates an example of selecting an interpolation filter based on the distance between the target sample of the current block and the reference sample. The encoding device/decoding device may output an intra-prediction mode for the current block and determine whether the intra-prediction mode is a directed intra-prediction mode (S1100). The directional intra-prediction mode may be called angular prediction. If the intra-prediction mode is a non-directional intra-prediction mode, the encoding/decoding device may perform intra-prediction of the current block based on the intra-prediction mode.

[136] If the intra-prediction mode is a directed intra-prediction mode, the encoding device/decoding device can determine whether the distance between the target sample of the current block and the reference sample is less than a threshold (S1110). The reference sample may represent a neighboring sample located in the intra-prediction mode prediction direction of the current block based on the target sample. Also, the threshold value may be derived based on the size of the current block, the intra-prediction mode of the current block, whether the block is a square/non-square block, and the like, as described above. In addition, the threshold value information may be signaled, and the threshold value of the current block may be output based on the signaled threshold value information.

[137] If the distance between the target sample of the current block and the reference sample is not less than the threshold value, that is, if the distance between the target sample and the reference sample is greater than or equal to the threshold value, the encoding device/decoding device may select a Gaussian filter as an interpolation filter the current block and output a reference sample of the target sample in the current block based on the Gaussian filter (S1120). Here, the Gaussian filter is one of the interpolation filters having a low-pass filter effect, and a reference sample of the target sample may be output based on the interpolation filter having a low-pass filter effect other than the Gaussian filter. For example, if the size of the current block is greater than or equal to the threshold value, the encoding device/decoding device may select the interpolation filter of the current block as a linear filter and output a reference sample of the target sample in the current block based on the linear filter.

[138] If the distance between the target sample of the current block and the reference sample is less than the threshold value, the encoding device/decoding device may select the cube filter as the interpolation filter of the current block and output the reference sample based on the cube filter (S1130). Here, the cubic filter may be one of the complex interpolation filters as described above, and the reference sample of the target sample may be derived based on the complex interpolation filter other than the cubic filter. Additionally, a cubic filter may be called a spline filter.

[139] The encoding device/decoding device may generate a target sample prediction sample based on the derived reference sample of the target sample (S1140). The encoder/decoder may generate the prediction sample by copying the reference sample. The reference sample may be copied and used as a prediction sample, and thus, the prediction sample may be indicated to be generated based on the interpolation filter.

[140] Also, as described above, one of the interpolation filters may be selected to output a reference sample of the target sample, but a plurality of interpolation filters may also be used to output the reference sample

[141] For example, when the reference sample position of the target sample of the current block is a fractional sample position, that is, when the reference sample of the target sample is a fractional sample, the first reference sample may be generated based on the first interpolation filter, which is a complex interpolation filter, the second reference sample may be generated based on the second interpolation filter, which is an interpolation filter having the above-described low-pass filter effect, and the third reference sample may be generated based on the third interpolation filter, which is an interpolation filter different from the first interpolation filter and the second interpolation filter. filter. When the first reference sample, the second reference sample, and the third reference sample are generated, a target sample prediction sample may be generated based on the first reference sample, the second reference sample, and the third reference sample. For example, the average of the first reference sample and the second reference sample may be output as the prediction sample of the target sample, the average of the first reference sample and the third reference sample may be output as the prediction sample of the target sample, the average of the second reference sample and the third reference sample may be output as the prediction sample target sample, or the average of the first reference sample, the second reference sample, and the third reference sample may be output as a prediction sample of the target sample.

[142] Alternatively, the prediction sample of the target sample may be derived through a weighted average of the first reference sample and the second reference sample, that is, the weighted sum of the first reference sample and the second reference sample. Alternatively, the prediction sample of the target sample may be output by a weighted sum of the first reference sample and the third reference sample, by a weighted sum of the second reference sample and the third reference sample, or by a weighted sum of the first reference sample, the second reference sample and the third reference sample. Alternatively, the prediction sample of the target sample may be derived through a weighted sum of the first reference sample, the second reference sample, and the third reference sample. The target sample prediction sample may be generated based on the above-described examples and combinations of the first reference sample, the second reference sample and/or the third reference sample different from the above-described examples.

[143] Specifically, for example, a target sample prediction sample may be generated as follows. If a directed intra-prediction mode, in which intra-prediction is performed based on a fractional sample position reference sample, is performed on the current block, adjacent integer sample position samples can be interpolated based on a cubic filter, so that the first reference sample, adjacent integer position samples can be output the samples may be interpolated based on the Gaussian filter so that a second reference sample of the target sample may be output, and a prediction sample of the target sample may be generated based on the first reference sample and the second reference sample. Here, the directed intra-prediction mode in which intra-prediction is performed based on the position reference sample of the fractional sample may represent one of the directed intra-prediction modes except intra-prediction modes #2, #18, #34, #50 and # 66. In addition, neighboring samples at the whole sample position may represent neighboring samples near the fractional sample position located in the directional intra-prediction mode prediction direction of the current block based on the target sample among the neighboring samples of the current block.

[144] In another example, the prediction accuracy of intra-prediction increases as the distance between the target sample and the reference sample decreases, and thus, a method for deriving a first weight for the first reference sample generated based on the first interpolation filter, which is a complex interpolation filter, and a second weight for a second reference sample generated based on the second interpolation filter, which is an interpolation filter having a low-pass filter effect based on the distance between the target sample and the reference sample, and a weighted summation of the first reference sample and the second reference samples based on the first weight and the second weight to generate a target sample prediction sample. For example, the first weight may be output as being inversely proportional to the distance between the target sample and the reference sample, and the second weight may be output as the value obtained by subtracting the first weight from 1. Alternatively, the first weight and second weight used herein may be scaled with increasing in whole units to avoid calculation with decimal point. Accordingly, the first weight may be output as a larger value as the distance between the target sample and the reference sample increases, and as the distance between the target sample and the reference sample increases, the first weight may be output as a smaller value. The distance between the target sample and the reference sample may be calculated based on the prediction angle of the intra-prediction mode of the current block and the position of the target sample. Alternatively, a table for block size and intra-prediction mode may be stored, and the distance between the target sample and the reference sample may be output with reference to the table. Here, the method for deriving a reference sample of a target sample based on a plurality of interpolation filters disclosed in the above-described embodiments may be selectively applied. For example, whether to output a reference sample of a target sample based on a plurality of interpolation filters may be inferred based on the size of the current block, the intra-prediction mode of the current block, the variance of the values of neighboring samples of the current block, and the like. Alternatively, a mark indicating whether to output a reference sample of the target sample based on the plurality of interpolation filters from the encoder may be transmitted, and whether or not to output the reference sample of the target sample based on the plurality of interpolation filters may be determined based on the label.

[145] FIG. 12 illustrates an example of deriving a reference sample for a target sample of a current block based on a plurality of interpolation filters, and deriving a prediction sample of a target sample based on the reference sample. The encoding device/decoding device may output an intra-prediction mode for the current block and determine whether the intra-prediction mode is a directed intra-prediction mode (S1200). The directional intra-prediction mode may be referred to as angular prediction.

[146] If the intra-prediction mode is a non-directional intra-prediction mode, the encoding/decoding apparatus may perform intra-prediction of the current block based on the non-directional intra-prediction mode (S1210).

[147] When the intra-prediction mode is a directed intra-prediction mode, the encoding device/decoding device may interpolate adjacent samples at the whole sample positions based on the cubic filter to output the first reference sample of the target sample (S1220). Here, neighboring samples of whole sample positions may represent neighboring samples about a fractional sample position located in the directional intra-prediction mode prediction direction of the current block relative to the target sample among the neighboring samples of the current block. The encoding device/decoding device may perform intra-prediction of the target sample based on the first reference sample (S1230). The encoder/decoder may generate a first prediction time sample by copying the first reference sample.

[148] When the intra-prediction mode is a directed intra-prediction mode, the encoding device/decoding device interpolates neighboring position samples of the whole sample based on the Gaussian filter to output a second reference sample of the target sample (S1240). The encoding device/decoding device may perform intra-prediction of the target sample based on the second reference sample (S1240). The encoder/decoder may generate a second prediction time sample by copying the second reference sample.

[149] The encoding device/decoding device may weightly add the first prediction time sample and the second prediction time sample to output a prediction sample of the target sample (S1250). The prediction sample may be output as the sum of a value obtained by multiplying the first weight α of the first prediction time sample by the first prediction time sample and a value obtained by multiplying 1-α of the second prediction time sample by the second prediction time sample. The first weight may be output as being inversely proportional to the distance between the target sample and the reference samples as described above, and the second weight may be output as the value obtained by subtracting the first weight from 1. Alternatively, the first weight and the second weight as used herein , can be scaled up in integer units and output to avoid decimal point calculations. The distance between the target sample and the reference samples can be calculated based on the intra-prediction mode prediction angle of the current block and the position of the target sample. Alternatively, a table regarding block size and intra-prediction mode may be stored in advance, and the distance between the target sample and the reference samples may be output with reference to the table.

[150] FIG. 13 schematically illustrates a method for encoding video by an encoding apparatus according to the present disclosure. The method disclosed in FIG. 13 may be performed by the encoding device disclosed in FIG. 1. Specifically, for example, steps S1300 to S1340 according to FIG. 13 may be performed by the prediction unit of the encoding device, and S1350 may be performed by the entropy encoding unit of the encoding device.

[151] The encoder determines the intra-prediction mode for the current block (S1300). The encoder may perform various intra-prediction modes to output the intra-prediction mode having the optimal RD cost as the intra-prediction mode for the current block. The intra-prediction mode may be one of two non-directional prediction modes and 33 directed prediction modes. As described above, the two non-directional prediction modes may include an intra-DC mode and an intra-planar mode. Alternatively, the intra-prediction mode may be one of two non-directional intra-prediction modes and 65 directed intra-prediction modes. As described above, the two non-directional prediction modes may include an intra-DC mode and an intra-planar mode. In addition, the 65 directional intra-prediction modes may include vertical directional intra-prediction modes and horizontal directional intra-prediction modes. Vertical directional intra-prediction modes may include intra-prediction mode #34 to intra-prediction mode #66, and horizontal directional intra-prediction modes may include intra-prediction mode #2 to intra-prediction mode #33 .

[152] The encoder outputs neighbor samples including left neighbor samples and top neighbor samples of the current block (S1310). The encoder can output adjacent samples of the current block. Neighbor samples may include left neighbor samples and top neighbor samples. In addition, neighboring samples may include top-left samples. Left neighbor samples, left top neighbor samples, and top neighbor samples may be derived from neighboring blocks already reconstructed during decoding of the current block. The 2N top neighbor samples, top left neighbor samples, and 2N left neighbor samples of the current block can be output. If the size of the current block is N×N and the x component of the top left sample of the current block is 0 and its y component is 0, the left neighboring samples can be p[-1][0] to p[-1][2N- 1], the upper left neighboring sample may be p[-1][-1], and the upper neighboring samples may be p[0][-1] to p[2N-1][-1].

[153] Alternatively, when the size of the current block is M×N, and the x component of the top left sample of the current block is 0, and its y component is 0, M+N top neighbor samples, top left neighbor samples, and M-N left neighbor samples may be output current block. If the size of the current block is non-square M×N, and the x component of the top left sample of the current block is 0, and its y component is 0, the left neighboring samples can be p[-1][0] to p[-1] [M+N-1], the upper left neighboring sample may be p[-1][-1], and the upper neighboring samples may be p[0][-1] to p[M+N-1] [-1].

[154] The encoding device outputs reference samples for predicting the target sample among neighboring samples based on the position of the target sample of the current block and the prediction angle of the intra-prediction mode (S1320). The encoding apparatus may output the position of the reference sample relative to the target sample based on the position of the target sample of the current block and the prediction angle of the intra-prediction mode, and when the position of the reference sample is the position of a fractional sample, the encoding apparatus may output adjacent samples located around the position derived based on the position of the target sample of the current block and the prediction angle of the intra-prediction mode, as reference samples of the target sample. That is, the encoding device can output a plurality of neighboring samples as reference samples of the target sample based on the position of the target sample of the current block and the prediction angle of the intra-prediction mode. For example, four adjacent samples may be output as reference samples of the target sample. Here, the target sample may represent a sample in the current block on which intra-prediction is performed. The prediction angle of the intra-prediction mode may be derived based on Table 1 described above, and intraPredAngle may be a variable indicating the prediction angle derived from the intra-prediction mode.

[155] The encoder determines an interpolation filter for the target sample (S1330). The encoder may determine an interpolation filter for the target sample based on the size of the current block and/or the intra-prediction mode of the current block. Also, for example, the interpolation filter can be determined when the position of the reference sample is the position of a fractional sample, that is, when a plurality of reference samples are output.

[156] For example, an interpolation filter for the target sample may be determined based on the size of the current block. For example, if the current block size is 4x4, a complex interpolation filter may be defined as the interpolation filter for the target sample. Specifically, when the size of the current block is 4x4, a cubic filter can be defined as an interpolation filter for the target sample. A cubic filter is one of the complex interpolation filters, and a cubic filter may be called a spline filter.

[157] If the current block is a square block, the width and height are equal to each other. That is, since the current block is a square block having size N×N, the reference size (ie, reference value) when selecting the interpolation filter may be N for the intra-prediction directional mode of any prediction direction. Moreover, if the current block is a non-square block, that is, if the current block is a non-square block having a size of M×N, when the intra-prediction mode of the current block is an intra-prediction mode having a vertical direction, the size of the reference block (i.e. ., reference value) for selecting the interpolation filter may be M. Similarly, if the current block is a non-square block having size M×N, and the intra-prediction mode of the current block is an intra-prediction mode having a horizontal orientation, the size of the current block (m .e., reference value) as the reference for selecting the interpolation filter may be N. Alternatively, on the contrary, if the current block is a non-square block having size M×N, and the intra-prediction mode of the current block is an intra-prediction mode having a vertical directionality, the interpolation filter of the current block can be selected based on N, and similarly, if the current block is a non-square block having size M×N, and the intra-prediction mode of the current block is an intra-prediction mode having horizontal directionality, the interpolation filter of the current block can be selected based on M. However, in a specific example to be described later, when an intra-prediction mode having a vertical directionality is applied to a current block having a size of M×N, the size of the current block as a reference for selecting an interpolation filter can be represented by M, and similarly, if an intra-prediction mode having a horizontal directionality is applied to the current block, the size of the current block can be represented by N. Here, when the intra-prediction mode includes 65 directed intra-prediction modes and two non-directional modes intra-prediction, an intra-prediction mode having a vertical focus may represent intra-prediction modes #34-#66, and an intra-prediction mode having a horizontal focus may represent intra-prediction modes #2-#33.

[158] For example, it may be determined whether the size of the current block, indicated by the shape of the current block and the directionality of the intra-prediction mode, is smaller than a specific value. If the current block size is smaller than a specific value, a complex interpolation filter can be defined as an interpolation filter for the target sample, and if the current block size is not smaller than a specific value, an interpolation filter having a low-pass filter effect can be defined as an interpolation filter for the target sample. Specifically, when the current block size is smaller than a specific value, a cubic filter may be defined as an interpolation filter for the target sample. If the current block size is not smaller than a specific value, a Gaussian filter can be specified as an interpolation filter for the target sample. Alternatively, when the current block size is not smaller than a particular value, a linear filter may be defined as an interpolation filter for the target sample. The interpolation filter having a low-pass filter effect may include a Gaussian filter and a linear filter. Here, the specific value can be set to 4, 8, 16, 32 or the like.

[159] Specifically, when the width and height of the current block are equal (i.e., when the current block is a square block), it can be determined whether the width of the current block is smaller than a specific value, and when the width of the current block is smaller than a specific value. value, the interpolation filter for the target sample can be output as a cubic filter for the target sample, and when the width of the current block is not smaller than the specific value, the interpolation filter for the target sample can be output as a Gaussian filter.

[160] In addition, when the width and height of the current block are different (that is, when the current block is a non-square block) and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, it can be determined whether the width of the current block is block less than a specific value. When the width of the current block is less than a specific value, the interpolation filter for the target sample may be output as a cubic filter, and when the width of the current block is not less than the specific value, the interpolation filter for the target sample may be output as a Gaussian filter.

[161] In addition, when the width and height of the current block are different (that is, when the current block is a non-square block) and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, it can be determined whether the height of the current block is block less than a specific value. When the height of the current block is less than a specific value, the interpolation filter for the target sample may be output as a cubic filter, and when the height of the current block is not less than a specific value, the interpolation filter for the target sample may be output as a Gaussian filter.

[162] In addition, when the width and height of the current block are different (that is, when the current block is a non-square block) and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, it can be determined whether the height of the current block is block less than a specific value. When the height of the current block is less than a specific value, the interpolation filter for the target sample may be output as a cubic filter, and when the height of the current block is not less than a specific value, the interpolation filter for the target sample may be output as a Gaussian filter.

[163] In addition, when the width and height of the current block are different (that is, when the current block is a non-square block) and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, it can be determined whether the width of the current block is block less than a specific value. When the width of the current block is less than a specific value, the interpolation filter for the target sample may be output as a cubic filter, and when the width of the current block is not less than the specific value, the interpolation filter for the target sample may be output as a Gaussian filter.

[164] In another example, an interpolation filter for the target sample may be determined based on the intra-prediction mode of the current block.

[165] For example, it may be determined whether the intra-prediction mode prediction angle of the current block is smaller than a specific value. If the prediction angle of the intra-prediction mode is smaller than the specific value, a complex interpolation filter can be determined as an interpolation filter for the target sample, and if the prediction angle of the intra-prediction mode is not smaller than the specific value, an interpolation filter having a low-pass filter effect is can be defined as an interpolation filter for the target sample. Specifically, when the prediction angle of the intra-prediction mode is smaller than a specific value, a cubic filter may be defined as an interpolation filter for the target sample, and when the prediction angle of an intra-prediction mode is not smaller than a specific value, a Gaussian filter may be defined as an interpolation filter for the target sample. Alternatively, when the prediction angle of the intra-prediction mode is not less than a specific value, a linear filter may be defined as an interpolation filter for the target sample. The interpolation filter having a low-pass filter effect may include a Gaussian filter and a linear filter. Here, the specific value can be set to 4, 8, 16, 32 or the like. The prediction angle of the intra-prediction mode may be derived based on Table 1 described above, and intraPredAngle may indicate the prediction angle of the intra-prediction mode. Also, for example, a specific value can be set to 11.

[166] In another example, the interpolation filter for the target sample may be determined based on the size of the current block and the intra-prediction mode. If the current block is a square block, the width and height are the same. If the current block is a square block, the width and height are equal to each other. That is, since the current block is a square block having size N×N, the reference size (ie, reference value) when selecting the interpolation filter may be N for the intra-prediction directional mode of any prediction direction. Moreover, if the current block is a non-square block, that is, if the current block is a non-square block having a size of M×N, when the intra-prediction mode of the current block is an intra-prediction mode having a vertical direction, the size of the reference block (i.e. ., reference value) for selecting the interpolation filter may be M. Similarly, if the current block is a non-square block having size M×N, and the intra-prediction mode of the current block is an intra-prediction mode having a horizontal orientation, the size of the current block ( i.e., the reference value) as the reference for selecting the interpolation filter may be N. Alternatively, in contrast, if the current block is a non-square block having size M×N, and the intra-prediction mode of the current block is the intra-prediction mode, having a vertical directionality, the interpolation filter of the current block can be selected based on N, and similarly, if the current block is a non-square block having size M×N, and the intra-prediction mode of the current block is an intra-prediction mode having a horizontal directionality, the interpolation filter of the current block may be selected based on M. However, in a specific example to be described later, when an intra-prediction mode having a vertical direction is applied to a current block having a size of M×N, the size of the current block as a reference for selecting an interpolation filter may be represented by M, and similarly, if an intra-prediction mode having a horizontal directionality is applied to the current block, the size of the current block can be represented by N. Here, when the intra-prediction mode includes 65 directional intra-prediction modes and two non-directional intra-prediction modes, an intra-prediction mode having a vertical focus may represent intra-prediction modes #34-#66, and an intra-prediction mode having a horizontal focus may represent intra-prediction modes #2-#33.

[167] For example, the encoding device may determine whether the size of the current block is smaller than the first specific value, and when the size of the current block is not smaller than the first specific value, the encoding device may determine an interpolation filter having a low-pass filter effect as interpolation filter for the target sample. Specifically, when the size of the current block is not smaller than the first specific value, the encoding device may determine a Gaussian filter as an interpolation filter. Alternatively, when the size of the current block is not smaller than the first specific value, the encoding device may define a linear filter as an interpolation filter.

[168] Specifically, when the width and height of the current block are equal to each other, the encoding device can determine whether the width of the current block is smaller than the first specific value. When the width of the current block is not smaller than the first specific value, the encoder may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample.

[169] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the encoding device can determine whether the width of the current block is smaller than the first specific value. When the width of the current block is not smaller than the first specific value, the encoder may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample.

[170] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the encoding device can determine whether the height of the current block is smaller than the first specific value. When the height of the current block is not less than the first specific value, the encoder may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample.

[171] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the encoding device can determine whether the height of the current block is smaller than the first specific value. When the height of the current block is not less than the first specific value, the encoder may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample.

[172] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the encoding device can determine whether the width of the current block is smaller than the first specific value. When the width of the current block is not smaller than the first specific value, the encoder may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample.

[173] If the size of the current block is smaller than the first specific value, the encoding device may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the encoding device may determine a complex interpolation filter as an interpolation filter for the target sample. Specifically, when the size of the current block is smaller than the first specific value, the encoding device may define a cubic filter as an interpolation filter.

[174] Specifically, when the width and height of the current block are equal and the width of the current block is smaller than the first specific value, the encoding device can determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value, and when the mode prediction angle intra-prediction is less than the second specific value, the encoder may define a cubic filter as an interpolation filter.

[175] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the encoding device can determine whether the width of the current block is smaller than the first specific value. If the width of the current block is smaller than the first specific value, the encoding device may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the encoding device may define a cubic filter as an interpolation filter.1

[176] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the encoding device can determine whether the height of the current block is smaller than the first specific value. If the height of the current block is smaller than the first specific value, the encoding device may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the encoding device may determine the cubic filter as an interpolation filter.

[177] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the encoding device can determine whether the height of the current block is smaller than the first specific value. If the height of the current block is smaller than the first specific value, the encoding device may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the encoding device may determine the cubic filter as an interpolation filter.

[178] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the encoding device can determine whether the width of the current block is smaller than the first specific value. If the width of the current block is smaller than the first specific value, the encoding device may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the encoding device may determine the cubic filter as an interpolation filter.

[179] If the prediction angle of the intra-prediction mode is not smaller than the second specific value, the encoding device may determine an interpolation filter having a low-pass filter effect as an interpolation filter for the target sample. Specifically, if the prediction angle of the intra-prediction mode is not smaller than the second specific value, the encoding device may determine a Gaussian filter as an interpolation filter. Alternatively, if the prediction angle of the intra-prediction mode is not smaller than the second specific value, the encoding device may determine the linear filter as the interpolation filter.

[180] Specifically, if the width and height of the current block are equal and the width of the current block is smaller than the first specific value, the encoding device may determine whether the intra-prediction mode prediction angle of the current block is smaller than the specific value, and if the intra-prediction mode prediction angle -prediction is not less than the second specific value, the encoder may define a Gaussian filter or a linear filter as the interpolation filter.

[181] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the encoding device can determine whether the width of the current block is smaller than the first specific value. If the width of the current block is smaller than the first specific value, the encoding device may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is not smaller than the second specific value, the encoding device may determine a Gaussian filter or a linear filter as the interpolation filter.

[182] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the encoding device can determine whether the height of the current block is smaller than the first specific value. If the height of the current block is smaller than the first specific value, the encoding device may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is not smaller than the second specific value, the encoding device may determine a Gaussian filter or a linear filter as the interpolation filter.

[183] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the encoding device can determine whether the height of the current block is smaller than the first specific value. If the height of the current block is smaller than the first specific value, the encoding device may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is not smaller than the second specific value, the encoding device may determine a Gaussian filter or a linear filter as the interpolation filter.

[184] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the encoding device can determine whether the width of the current block is smaller than the first specific value. If the width of the current block is smaller than the first specific value, the encoding device may determine whether the intra-prediction mode prediction angle of the current block is not smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the encoding device may determine a Gaussian filter or a linear filter as the interpolation filter.

[185] In another example, the current block may be divided into multiple regions, and the interpolation filter of each region may be determined based on the distance between each region and adjacent samples of the current block. In this case, the interpolation filter for the target sample may be output as an interpolation filter of the region including the target sample. In addition, if the size of the current block is greater than or equal to a specific size, the current block may be divided into multiple regions. Specifically, interpolation filters for regions closer to neighboring samples of the current block than a particular value among regions may be defined as complex interpolation filters, and interpolation filters for regions further away from neighboring samples of the current block than a particular value among regions. can be defined as interpolation filters having a low-pass filter effect.

[186] Here, the size of a plurality of areas divided from the current block can be preset. Alternatively, the size of the plurality of regions may be inferred based on the size of the current block, the intra-prediction mode, or the like. For example, in the case where the intra-prediction mode of the current block is one of the intra-prediction modes #35 to #66, the size of the divided areas of the current block may be output as a size of 4×4. Here, if the current block is a block having a size of 16×16, the current block may be divided into regions having a size of 4×4, a complex interpolation filter may be defined as an interpolation filter for regions #0 to #7 in raster scanning order, and an interpolation filter having a low-pass filter effect can be defined as interpolation filters for other areas. For example, interpolation filters for regions #0-#7 in raster scan order, among the regions, may be defined as cubic filters, and interpolation filters for other regions may be defined as Gaussian filters or linear filters. Here, the 16 area numbers having a size of 4x4 according to the raster scanning order may be sequentially output in order from the top line to the bottom line, and sequentially output in the order from left to right in each line. That is, the areas included in the first row among the 16 areas having a size of 4x4 of the current block can be represented as area #0, area #1, area #2 and area #3 in order from left to right, areas included in the second line can be represented as area #4, area #5, area #6 and area #7 in order from left to right, areas included in the third line can be represented as area #8, area #9, area #10 and area #11 is in order from left to right, and the areas included in the fourth row can be represented as area #12, area #13, area #14 and area #15 in order from left to right. This may generate the dimensions of the divided regions of the current block and information indicating the interpolation filter for each of the regions.

[187] In another example, an interpolation filter for the target sample may be determined based on the distance between the target sample and the reference samples. The distance between the target sample and the reference samples can be derived based on the position of the target sample and the prediction angle of the intra-prediction mode of the current block. For example, the distance can be calculated based on the position of the target sample and the value of a trigonometric function (eg, tanθ) according to the prediction angle of the intra-prediction mode. Alternatively, the distance may be inferred based on a predefined table for block size and intra-prediction mode. Alternatively, the distance between the target sample and the reference samples may indicate a vertical distance or a horizontal distance. That is, the distance between the target sample and the reference samples may be inferred based on the vertical distance, or the distance between the target sample and the reference samples may be inferred based on the horizontal distance. For example, if the distance between the target sample and the reference samples indicates a vertical distance, the distance may be inferred based on the y component of the target sample. Also, if the distance between the target sample and the reference samples indicates a horizontal distance, the distance may be inferred based on the x component of the target sample.

[188] For example, the encoder may output the distance between the target sample and the reference samples and determine whether the distance is less than a particular value. If the distance is less than a specific value, the encoding device may determine a complex interpolation filter as an interpolation filter for the target sample, and if the distance is not less than a specific value, the encoding device may determine an interpolation filter having a low-pass filter effect as an interpolation filter for the target sample. Specifically, if the distance is less than a specific value, the encoding device may determine a cubic filter as an interpolation filter for the target sample, and if the distance is not less than a specific value, the encoding device may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample. . A specific value can be inferred based on the size of the current block. Alternatively, the specific value may be output based on the intra-prediction mode of the current block, determining whether the current block is a square/non-square block, or the like. For example, when the current block size is N×N, the specific value may be output as N/2. Additionally, information about a particular value can be generated, entropy encoded, and transmitted.

[189] In another example, a plurality of interpolation filters may be defined as an interpolation filter for the target sample. For example, the interpolation filters for the target sample may include one of complex interpolation filters and one of interpolation filters having a low-pass filter effect. Alternatively, the interpolation filters for the target sample may include one of complex interpolation filters and two of interpolation filters having a low-pass filter effect. Alternatively, the interpolation filters for the target sample may include two of complex interpolation filters and one of interpolation filters having a low-pass filter effect. Specifically, the interpolation filters for the target sample may include a cubic filter and a Gaussian filter.

[190] The encoding device outputs a prediction sample of the target sample based on the interpolation filter and the reference samples (S1340). The encoding device can output filter coefficients of the interpolation filter based on the position of the target sample and the prediction angle of the intra-prediction mode, and output the prediction sample of the target sample based on the filter coefficients and reference samples. For example, four adjacent samples among adjacent samples of the current block may be output as reference samples, and four filter coefficients of the interpolation filter may be output. The encoder may interpolate the reference samples based on the filter coefficients to output a prediction sample. The prediction sample can be derived based on Equation 1 described above.

[191] Additionally, when a plurality of interpolation filters are defined as interpolation filters for the target sample, the encoding device may output prediction (temporal) samples based on each interpolation filter and output a prediction sample of the target sample based on the output (temporal) prediction sample. For example, the prediction sample of the target sample may be derived by averaging the (temporal) prediction samples, or may be derived by weighting the summation of the (temporal) prediction samples. Moreover, whether a plurality of interpolation filters is defined as an interpolation filter for the target sample can be inferred based on the size of the current block, the intra-prediction mode of the current block, the deviation (variance) of the values of neighboring samples of the current block, and the like. In addition, a mark may be generated indicating whether a plurality of interpolation filters are defined as interpolation filters for the target sample.

[192] For example, interpolation filters for the target sample may include a cubic filter and a Gaussian filter. In this case, the encoding device can output the filter coefficients of a cubic filter based on the position of the target sample and the prediction angle of the intra-prediction mode, and can output the filter coefficients of the Gaussian filter based on the position of the target sample and the prediction angle of the intra-prediction mode. The encoding device may output a first prediction sample for the target sample based on the filter coefficients of the cubic filter and the reference samples, output a second prediction sample for the target sample based on the filter coefficients of the Gaussian filter and the reference samples, and output a prediction sample of the target sample based on the first prediction sample and the second sample. predictions. The prediction sample of the target sample may be derived by averaging the first prediction sample and the second prediction sample. Alternatively, the prediction sample of the target sample may be derived by weighting the sum of the first prediction sample and the second prediction sample. In this case, the weight for the first prediction sample may be inversely proportional to the distance between the target sample and the reference samples, and the weight for the second prediction sample may be derived as the value obtained by subtracting the weight for the first prediction sample from 1. Alternatively, the first weight and the second weights used here can be scaled up in integer units and output to avoid decimal point calculations.

[193] Also, for example, interpolation filters for the target sample may include a cubic filter and a linear filter. In this case, the encoding device can output cubic filter filter coefficients based on the position of the reference sample and can output linear filter filter coefficients based on the position of the reference sample. The encoding device may output a first prediction sample for the target sample based on the filter coefficients of the cubic filter and the reference samples, output a second prediction sample for the target sample based on the filter coefficients of the linear filter and the reference samples, and output a prediction sample of the target sample based on the first prediction sample and the second sample. predictions. The prediction sample of the target sample may be derived by averaging the first prediction sample and the second prediction sample. Alternatively, the prediction sample of the target sample may be derived by weighting the sum of the first prediction sample and the second prediction sample. In this case, the weight for the first prediction sample may be inversely proportional to the distance between the target sample and the reference samples, and the weight for the second prediction sample may be derived as the value obtained by subtracting the weight for the first prediction sample from 1. Alternatively, the first weight and the second weights used here can be scaled up in integer units and output to avoid decimal point calculations.

[194] In another example, if the MPM (most probable mode) mode is applied to the current block to infer the intra-prediction mode of the current block based on the intra-prediction mode of the adjacent block of the current block, and the intra-prediction mode of the current block is a directed mode intra-prediction rather than planar or DC mode, the interpolation filter for the target sample can be determined based on neighboring blocks sampled through the neighboring MPM mode. That is, the interpolation filter used in the adjacent block can be output as an interpolation filter for the target sample. If the MPM mode is applied to the current block, the encoder may determine the MPM list based on the intra-prediction mode for the left or upper adjacent block of the current block, and determine the intra-prediction mode based on the MPM list.

[195] The encoding device generates prediction information for the current block, encodes the prediction information, and outputs the encoded prediction information (S1350). The encoding apparatus may encode prediction information for the current block and output the encoded information in the form of a bit stream. The prediction information may include information about the intra-prediction mode of the current block. The encoding device may generate intra-prediction mode information indicating the intra-prediction mode, encode the intra-prediction mode information, and output the information as a bit stream. The intra-prediction mode information may include information directly indicating the intra-prediction mode for the current block, or may include information indicating any one candidate in the intra-prediction mode candidate list obtained based on the left or top intra-prediction mode block of the current block. The intra-prediction mode candidate list may indicate a list of MPMs.

[196] In addition, when a current block is divided into a plurality of regions, the prediction information may include information indicating the size of the divided regions of the current block and an interpolation filter of each region. In addition, when the interpolation filter for the target sample is selected based on the size of the current block, the intra-prediction mode of the current block, or the distance between the target sample and the reference samples, the prediction information may include information regarding the specific value used to select the interpolation filter for the target sample . When the interpolation filter is selected based on the size of the current block and the intra-prediction mode of the current block, the prediction information may include information regarding the first specific value and information regarding the second specific value. Also, the prediction information may include a mark indicating whether a plurality of interpolation filters are defined as interpolation filters for the target sample. When the mark indicates that a plurality of interpolation filters are defined as interpolation filters for the target sample, the prediction sample for the target sample may be derived based on the plurality of interpolation filters, and when the mark indicates that the plurality of interpolation filters is not defined as interpolation filters for the target sample, the sample predictions for the target sample may not be derived from multiple interpolation filters. For example, when the label value is 1, the label may indicate that a plurality of interpolation filters are defined as interpolation filters for the target sample, and when the label value is 0, the label may indicate that a plurality of interpolation filters are not defined as interpolation filters for the target sample. samples. The prediction information may be signaled via a video parameter set (VPS), a sequence parameter set (SPS), a picture parameter set (PPS) or a slice segment header, or may be signaled on a block basis.

[197] FIG. 14 schematically illustrates a method for decoding video by a decoding apparatus according to the present disclosure. The method disclosed in FIG. 14 may be performed by the decoding apparatus disclosed in FIG. 4. Specifically, for example, steps S1400 to S1440 according to FIG. 14 may be performed by the prediction module of the decoding device.

[198] The decoder outputs the intra-prediction mode for the current block (S1400). The decoding apparatus may obtain prediction information about the current block through the bit stream. The prediction information may include information directly indicating the intra-prediction mode for the current block, or include information indicating any of the candidates in the intra-prediction mode candidate list derived based on the intra-prediction mode of the left or top block of the current block. The intra-prediction mode candidate list may be referred to as an MPM candidate list. The decoding apparatus may output an intra-prediction mode for the current block based on the received prediction information. The intra-prediction mode may be one of two non-directional prediction modes and 33 directed prediction modes. As described above, the two non-directional prediction modes may include an intra-DC mode and an intra-planar mode. Alternatively, the intra-prediction mode may be one of two non-directional intra-prediction modes and 65 directed intra-prediction modes. As described above, the two non-directional prediction modes may include an intra-DC mode and an intra-planar mode. In addition, the 65 directional intra-prediction modes may include vertically directional intra-prediction modes and horizontally directional intra-prediction modes. Vertically directed intra-prediction modes may include intra-prediction mode #34 to intra-prediction mode #66, and horizontally directed intra-prediction modes may include intra-prediction mode #2 to intra-prediction mode #33 .

[199] The decoding apparatus outputs neighboring samples including left neighboring samples and upper neighboring samples of the current block (S1410). The decoder can output adjacent samples of the current block. Neighbor samples may include left neighbor samples and top neighbor samples. In addition, neighboring samples may include top-left samples. Left neighbor samples, left top neighbor sample, and top neighbor samples may be derived from neighboring blocks already reconstructed during decoding of the current block. The 2N top neighbor samples, the top left neighbor samples, and the 2N left neighbor samples of the current block can be output. If the size of the current block is N×N, and the x component of the top left sample of the current block is 0, and its y component is 0, the left neighboring samples can be p[-1][0] to p[-1] [2N-1], the top left neighbor sample may be p[-1][-1], and the top neighbor samples may be p[0][-1] to p[2N-1][-1] .

[200] Alternatively, when the size of the current block is M×N, and the x component of the top left sample of the current block is 0, and its y component is 0, M+N top neighbor samples, top left neighbor samples, and M-N left neighboring samples of the current block. If the size of the current block is non-square M×N, and the x component of the top left sample of the current block is 0, and its y component is 0, the left neighboring samples can be p[-1][0] to p[- 1][M+N-1], the upper left neighboring sample may be p[-1][-1], and the upper neighboring samples may be p[0][-1] to p[M+N- 1][-1].

[201] The decoding apparatus outputs reference samples for predicting the target sample among neighboring samples based on the position of the target sample of the current block and the prediction angle of the intra-prediction mode (S1420). The decoding apparatus may output the position of the reference sample relative to the target sample based on the position of the target sample of the current block and the prediction angle of the intra-prediction mode, and when the position of the reference sample is the position of a fractional sample, the decoding apparatus may output adjacent samples located around the position derived based on the position of the target sample of the current block and the prediction angle of the intra-prediction mode, as reference samples of the target sample. That is, the decoding device can output a plurality of neighboring samples as reference samples of the target sample based on the position of the target sample of the current block and the prediction angle of the intra-prediction mode. For example, four adjacent samples may be output as reference samples of the target sample. Here, the target sample may represent a sample in the current block on which intra-prediction is performed. The prediction angle of the intra-prediction mode may be derived based on Table 1 described above, and intraPredAngle may be a variable indicating the prediction angle derived from the intra-prediction mode.

[202] The decoding apparatus determines an interpolation filter for the target sample (S1430). The decoding apparatus may determine an interpolation filter for the target sample based on the size of the current block and/or the intra-prediction mode of the current block. Also, for example, an interpolation filter may be determined when the position of the reference sample is the position of a fractional sample.

[203] For example, an interpolation filter for the target sample may be determined based on the size of the current block. For example, if the current block size is 4x4, a complex interpolation filter may be defined as the interpolation filter for the target sample. Specifically, when the current block size is 4x4, a cubic filter can be defined as an interpolation filter for the target sample. A cubic filter is one of the complex interpolation filters, and a cubic filter may be called a spline filter.

[204] Also, if the current block is a square block, the width and height are equal to each other. That is, since the current block is a square block having size N×N, the reference size (ie, reference value) when selecting the interpolation filter may be N for the intra-prediction directional mode of any prediction direction. Moreover, if the current block is a non-square block, that is, if the current block is a non-square block having a size of M×N, when the intra-prediction mode of the current block is an intra-prediction mode having a vertical orientation, the size of the reference block (i.e. e., reference value) for selecting an interpolation filter may be M. Similarly, if the current block is a non-square block having size M×N, and the intra-prediction mode of the current block is an intra-prediction mode having a horizontal orientation, the size of the current block (i.e., the reference value) as the reference for selecting the interpolation filter may be N. Alternatively, on the contrary, if the current block is a non-square block having size M×N, and the intra-prediction mode of the current block is an intra-prediction mode -prediction having a vertical directionality, the interpolation filter of the current block can be selected based on N, and similarly, if the current block is a non-square block having size M×N, and the intra-prediction mode of the current block is an intra-prediction mode having a horizontal directionality , the interpolation filter of the current block can be selected based on M. However, in the specific example that will be described later, when the intra-prediction mode having a vertical direction is applied to the current block having size M×N, the size of the current block as a reference for interpolation filter selection can be represented by M, and similarly, if an intra-prediction mode having a horizontal directionality is applied to the current block, the size of the current block can be represented by N. Here, when the intra-prediction mode includes 65 directional intra-mode -prediction and two non-directional intra-prediction modes, an intra-prediction mode having a vertical focus may represent intra-prediction modes #34-#66, and an intra-prediction mode having a horizontal focus may represent intra-prediction modes #2- #33.

[205] Also, for example, it can be determined whether the size of the current block, indicated by the shape of the current block and the directionality of the intra-prediction mode, is smaller than a specific value. If the size of the current block is smaller than a specific value, a complex interpolation filter can be defined as an interpolation filter for the target sample, and if the size of the current block is not smaller than a specific value, an interpolation filter having a low-pass filter effect can be defined as an interpolation filter for the target sample. Specifically, when the size of the current block is smaller than a specific value, a cubic filter may be defined as an interpolation filter for the target sample. If the current block size is not smaller than a specific value, a Gaussian filter can be specified as an interpolation filter for the target sample. Alternatively, when the current block size is not smaller than a particular value, a linear filter may be defined as an interpolation filter for the target sample. The interpolation filter having a low-pass filter effect may include a Gaussian filter and a linear filter. Here, the specific value can be set to 4, 8, 16, 32 or the like. Also, the prediction information for the current block may include information about a specific value. In this case, a specific value may be inferred based on information about the specific value.

[206] Specifically, when the width and height of the current block are equal in size (i.e., when the current block is a square block), it can be determined whether the width of the current block is smaller than a specific value, and when the width of the current block is smaller, than a specific value, the interpolation filter for the target sample can be output as a cubic filter for the target sample, and when the width of the current block is not smaller than the specific value, the interpolation filter for the target sample can be output as a Gaussian filter.

[207] In addition, when the width and height of the current block are different (that is, when the current block is a non-square block), and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, it can be determined whether the width is of the current block is less than a specific value. When the width of the current block is less than a specific value, the interpolation filter for the target sample may be output as a cubic filter, and when the width of the current block is not less than the specific value, the interpolation filter for the target sample may be output as a Gaussian filter.

[208] In addition, when the width and height of the current block are different (that is, when the current block is a non-square block) and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, it can be determined whether the height of the current block is block less than a specific value. When the height of the current block is less than a specific value, the interpolation filter for the target sample may be output as a cubic filter, and when the height of the current block is not less than the specific value, the interpolation filter for the target sample may be output as a Gaussian filter.

[209] In addition, when the width and height of the current block are different (that is, when the current block is a non-square block) and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, it can be determined whether the height of the current block is block less than a specific value. When the height of the current block is less than a specific value, the interpolation filter for the target sample may be output as a cubic filter, and when the height of the current block is not less than the specific value, the interpolation filter for the target sample may be output as a Gaussian filter.

[210] In addition, when the width and height of the current block are different (that is, when the current block is a non-square block) and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, it can be determined whether the width is of the current block is less than a specific value. When the width of the current block is less than a specific value, the interpolation filter for the target sample may be output as a cubic filter, and when the width of the current block is not less than the specific value, the interpolation filter for the target sample may be output as a Gaussian filter.

[211] In another example, an interpolation filter for the target sample may be determined based on the intra-prediction mode of the current block.

[212] For example, it may be determined whether the intra-prediction mode prediction angle of the current block is smaller than a specific value. If the prediction angle of the intra-prediction mode is smaller than the specific value, a complex interpolation filter can be determined as an interpolation filter for the target sample, and if the prediction angle of the intra-prediction mode is not smaller than the specific value, an interpolation filter having a low-pass filter effect is can be defined as an interpolation filter for the target sample. Specifically, when the prediction angle of the intra-prediction mode is smaller than a specific value, a cubic filter may be defined as an interpolation filter for the target sample, and when the prediction angle of an intra-prediction mode is not smaller than a specific value, a Gaussian filter may be defined as an interpolation filter for the target sample. Alternatively, when the prediction angle of the intra-prediction mode is not less than a specific value, a linear filter may be defined as an interpolation filter for the target sample. The interpolation filter having a low-pass filter effect may include a Gaussian filter and a linear filter. Here, the specific value can be set to 4, 8, 16, 32 or the like. The prediction angle of the intra-prediction mode may be derived based on Table 1 described above, and intraPredAngle may indicate the prediction angle of the intra-prediction mode. Also, for example, the specific value may be set to 11. Also, the prediction information for the current block may include information about the specific value. In this case, a specific value may be inferred based on information about the specific value.

[213] In another example, the interpolation filter for the target sample may be determined based on the current block size and the intra-prediction mode.

[214] For example, the decoding apparatus may determine whether the size of the current block is smaller than the first specific value, and when the size of the current block is not smaller than the first specific value, the decoding apparatus may determine an interpolation filter having a low-pass filter effect as interpolation filter for the target sample. Specifically, when the size of the current block is not smaller than the first specific value, the decoding apparatus may determine a Gaussian filter as an interpolation filter. Alternatively, when the current block size is not smaller than the first specific value, the decoding apparatus may determine the linear filter as an interpolation filter.

[215] Specifically, when the width and height of the current block are equal to each other, the decoding apparatus may determine whether the width of the current block is smaller than the first specific value. When the width of the current block is not smaller than the first specific value, the decoding apparatus may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample.

[216] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the decoding apparatus can determine whether the width of the current block is smaller than the first specific value. When the width of the current block is not smaller than the first specific value, the decoding apparatus may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample.

[217] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the decoding apparatus can determine whether the height of the current block is smaller than the first specific value. When the height of the current block is not less than the first specific value, the decoding apparatus may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample.

[218] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the decoding device can determine whether the height of the current block is smaller than the first specific value. When the height of the current block is not less than the first specific value, the decoding apparatus may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample.

[219] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the decoding apparatus can determine whether the width of the current block is smaller than the first specific value. When the width of the current block is not smaller than the first specific value, the decoding apparatus may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample.

[220] If the size of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the decoding apparatus may determine a complex interpolation filter as an interpolation filter for the target sample. Specifically, when the prediction angle of the intra-prediction mode is smaller than the second specific value, the decoding apparatus may determine the cubic filter as the interpolation filter.

[221] Specifically, when the width and height of the current block are equal and the width of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value, and when the mode prediction angle intra-prediction is less than the second specific value, the decoding device may determine the cubic filter as an interpolation filter.

[222] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the decoding apparatus can determine whether the width of the current block is smaller than the first specific value. If the width of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the decoding apparatus may determine the cubic filter as an interpolation filter.

[223] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the decoding device can determine whether the height of the current block is smaller than the first specific value. If the height of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the decoding apparatus may determine the cubic filter as an interpolation filter.

[224] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the decoding device can determine whether the height of the current block is smaller than the first specific value. If the height of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the decoding apparatus may determine the cubic filter as an interpolation filter.

[225] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the decoding apparatus can determine whether the width of the current block is smaller than the first specific value. If the width of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the decoding apparatus may determine the cubic filter as an interpolation filter.

[226] If the prediction angle of the intra-prediction mode is not smaller than the second specific value, the decoding apparatus may determine an interpolation filter having a low-pass filter effect as an interpolation filter for the target sample. Specifically, if the prediction angle of the intra-prediction mode is not smaller than the second specific value, the decoding apparatus may determine a Gaussian filter as an interpolation filter. Alternatively, if the prediction angle of the intra-prediction mode is not smaller than the second specific value, the decoding apparatus may determine the linear filter as the interpolation filter.

[227] Specifically, if the width and height of the current block are equal and the width of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the specific value, and if the intra-mode prediction angle -prediction is not less than the second specific value, the decoding device may determine a Gaussian filter or a linear filter as an interpolation filter.

[228] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the decoding apparatus can determine whether the width of the current block is smaller than the first specific value. If the width of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is not smaller than the second specific value, the decoding apparatus may determine a Gaussian filter or a linear filter as the interpolation filter.

[229] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the decoding apparatus can determine whether the height of the current block is smaller than the first specific value. If the height of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is not smaller than the second specific value, the decoding apparatus may determine a Gaussian filter or a linear filter as the interpolation filter.

[230] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a vertical directionality, the decoding device can determine whether the height of the current block is smaller than the first specific value. If the height of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is not smaller than the second specific value, the decoding apparatus may determine a Gaussian filter or a linear filter as the interpolation filter.

[231] If the width and height of the current block are different from each other and the intra-prediction mode of the current block is a directional intra-prediction mode having a horizontal directionality, the decoding apparatus can determine whether the width of the current block is smaller than the first specific value. If the width of the current block is smaller than the first specific value, the decoding apparatus may determine whether the intra-prediction mode prediction angle of the current block is smaller than the second specific value. If the prediction angle of the intra-prediction mode is smaller than the second specific value, the decoding apparatus may determine a Gaussian filter or a linear filter as the interpolation filter.

[232] Here, the prediction information for the current block may include information regarding the first specific value and information regarding the second specific value. In this case, the first specific value may be output based on information regarding the first specific value, and the second specific value may be output based on information regarding the second specific value. Alternatively, the first specific value and the second specific value may be output based on the preset values.

[233] In another example, the current block may be divided into multiple regions, and the interpolation filter of each region may be determined based on the distance between each region and adjacent samples of the current block. In this case, the interpolation filter for the target sample may be output as an interpolation filter of the region including the target sample. In addition, if the size of the current block is greater than or equal to a specific size, the current block may be divided into multiple regions. Specifically, interpolation filters for regions closer to neighboring samples of the current block than a particular value among regions may be defined as complex interpolation filters, and interpolation filters for regions further away from neighboring samples of the current block than a particular value among regions. can be defined as interpolation filters having a low-pass filter effect.

[234] Here, the size of a plurality of areas output from the current block can be preset. Alternatively, the size of the plurality of regions may be inferred based on the size of the current block, the intra-prediction mode, or the like. For example, in the case where the intra prediction mode of the current block is one of the intra prediction modes #35 to #66, the size of the divided areas of the current block may be output as a size of 4×4. Here, if the current block is a block having a size of 16x16, the current block may be divided into regions having a size of 4x4, a complex interpolation filter may be defined as an interpolation filter for regions #0 to #7 in raster scanning order, and an interpolation filter having a low-pass filter effect can be defined as interpolation filters for other areas. For example, interpolation filters for regions #0-#7 in raster scan order, among the regions, may be defined as cubic filters, and interpolation filters for other regions may be defined as Gaussian filters or linear filters. Here, the 16 area numbers having a size of 4x4 according to the raster scanning order may be sequentially output in order from the top line to the bottom line, and sequentially output in the order from left to right in each line. That is, the areas included in the first row among the 16 areas having a size of 4x4 of the current block can be represented as area #0, area #1, area #2 and area #3 in order from left to right, areas included in the second line can be represented as area #4, area #5, area #6 and area #7 in order from left to right, areas included in the third line can be represented as area #8, area #9, area #10 and area #11 is in order from left to right, and the areas included in the fourth row can be represented as area #12, area #13, area #14 and area #15 in order from left to right. Here, prediction information for the current block may be received, and the prediction information may include information indicating the size of each of the divided regions of the current block and information indicating an interpolation filter of each region. In this case, the sizes of the divided regions of the current block and the interpolation filters for the corresponding regions can be derived based on the information indicating the sizes of the divided regions of the current block and the information indicating the interpolation filters of the corresponding regions.

[235] In another example, an interpolation filter for the target sample may be determined based on the distance between the target sample and the reference samples. The distance between the target sample and the reference samples can be derived based on the position of the target sample and the prediction angle of the intra-prediction mode of the current block. For example, the distance can be calculated based on the position of the target sample and the value of a trigonometric function (eg, tanθ) according to the prediction angle of the intra-prediction mode. Alternatively, the distance may be inferred based on a predefined table for block size and intra-prediction mode. Alternatively, the distance between the target sample and the reference samples may indicate a vertical distance or a horizontal distance. If the distance between the target sample and the reference samples indicates a vertical distance, the distance may be inferred based on the y component of the target sample. Also, if the distance between the target sample and the reference samples indicates a horizontal distance, the distance may be inferred based on the x component of the target sample.

[236] For example, the decoding apparatus may output the distance between the target sample and the reference samples and determine whether the distance is less than a particular value. If the distance is less than a specific value, the decoding device may determine a complex interpolation filter as an interpolation filter for the target sample, and if the distance is not less than a specific value, the decoding device may determine an interpolation filter having a low-pass filter effect as an interpolation filter for the target sample. Specifically, if the distance is less than a specific value, the decoding device may determine a cubic filter as an interpolation filter for the target sample, and if the distance is not less than the specific value, the decoding device may determine a Gaussian filter or a linear filter as an interpolation filter for the target sample. . A specific value can be inferred based on the size of the current block. Alternatively, the specific value may be output based on the intra-prediction mode of the current block, whether the current block is a square block/non-square block, or the like. For example, when the current block size is N×N, the specific value may be output as N/2. Also, prediction information for the current block may be received, and the prediction information may include information regarding a specific value. In this case, a specific value may be inferred based on information regarding the specific value.

[237] In another example, a plurality of interpolation filters may be defined as interpolation filters for the target sample. For example, the interpolation filters for the target sample may include one of complex interpolation filters and one of interpolation filters having a low-pass filter effect. Alternatively, the interpolation filters for the target sample may include one of complex interpolation filters and two of interpolation filters having a low-pass filter effect. Alternatively, the interpolation filters for the target sample may include two of complex interpolation filters and one of interpolation filters having a low-pass filter effect. Specifically, the interpolation filters for the target sample may include a cubic filter and a Gaussian filter. Here, prediction information for the current block may be received, and the prediction information may include a mark indicating whether a plurality of interpolation filters are defined as interpolation filters for the target sample. Whether a plurality of interpolation filters are defined may be determined based on the label. For example, when the mark indicates that a plurality of interpolation filters are defined as interpolation filters for the target sample, the prediction sample for the target sample may be derived based on the plurality of interpolation filters, and when the mark indicates that the plurality of interpolation filters is not defined as interpolation filters for the target sample ,the prediction sample for the target sample may not be ,derived based on multiple interpolation filters. For example, when the label value is 1, the label may indicate that a plurality of interpolation filters are defined as interpolation filters for the target sample, and when the label value is 0, the label may indicate that a plurality of interpolation filters are not defined as interpolation filters for the target sample. samples.

[238] The decoding apparatus outputs a prediction sample of the target sample based on the interpolation filter and the reference samples (S1440). The decoding apparatus may output filter coefficients of the interpolation filter based on the position of the target sample and the prediction angle of the intra-prediction mode, and output the prediction sample of the target sample based on the filter coefficients of the reference samples. For example, four adjacent samples among adjacent samples of the current block may be output as reference samples, and four filter coefficients of the interpolation filter may be output. The decoder may interpolate the reference samples based on the filter coefficients to output a prediction sample. The prediction sample can be derived based on Equation 1 described above.

[239] Additionally, when a plurality of interpolation filters are defined as interpolation filters for the target sample, the decoding apparatus may output prediction (temporal) samples based on each interpolation filter and output a prediction sample of the target sample based on the output (temporal) prediction sample. For example, the prediction sample of a target sample may be derived by averaging the (temporal) prediction samples, or may be derived by a weighted summation of the (temporal) prediction samples. Whether a plurality of interpolation filters is determined as an interpolation filter for the target sample can be inferred based on the size of the current block, the intra-prediction mode of the current block, the deviation of the values of neighboring samples of the current block, and the like. In addition, a label indicating whether a plurality of interpolation filters are defined as interpolation filters for the target sample may be received, and whether a plurality of interpolation filters are defined as interpolation filters for the target sample may be determined based on the label.

[240] For example, interpolation filters for the target sample may include a cubic filter and a Gaussian filter. In this case, the decoding device can output filter coefficients of a cubic filter based on the position of the target sample and the prediction angle of the intra-prediction mode, and can output filter coefficients of the Gaussian filter based on the position of the target sample and the prediction angle of the intra-prediction mode. The decoding apparatus may output a first prediction sample for the target sample based on the filter coefficients of the cubic filter and the reference samples, output a second prediction sample for the target sample based on the filter coefficients of the Gaussian filter and the reference samples, and output a prediction sample of the target sample based on the first prediction sample and the second sample. predictions. The prediction sample of the target sample may be derived by averaging the first prediction sample and the second prediction sample. Alternatively, the prediction sample of the target sample may be derived by weighting the sum of the first prediction sample and the second prediction sample. In this case, the weight for the first prediction sample may be inversely proportional to the distance between the target sample and the reference samples, and the weight for the second prediction sample may be derived as the value obtained by subtracting the weight for the first prediction sample from 1. Alternatively, the first weight and the second weights used here can be scaled up in integer units and output to avoid decimal point calculations.

[241] Also, for example, interpolation filters for the target sample may include a cubic filter and a linear filter. In this case, the decoding device can output filter coefficients of a cubic filter based on the position of the reference sample, and can output filter coefficients of a linear filter based on the position of the reference sample. The decoding apparatus may output a first prediction sample for the target sample based on the filter coefficients of the cubic filter and the reference samples, output a second prediction sample for the target sample based on the filter coefficients of the linear filter and the reference samples, and output a prediction sample of the target sample based on the first prediction sample and the second sample. predictions. The prediction sample of the target sample may be derived by averaging the first prediction sample and the second prediction sample. Alternatively, the prediction sample of the target sample may be derived by weighting the sum of the first prediction sample and the second prediction sample. In this case, the weight for the first prediction sample may be inversely proportional to the distance between the target sample and the reference samples, and the weight for the second prediction sample may be derived as the value obtained by subtracting the weight for the first prediction sample from 1. Alternatively, the first weight and the second weights used here can be scaled up in integer units and output to avoid decimal point calculations.

[242] In another example, if the MPM (most probable mode) mode is applied to the current block to infer the intra-prediction mode of the current block based on the intra-prediction mode of the adjacent block of the current block, and the intra-prediction mode of the current block is a directed mode intra-prediction rather than planar or DC mode, the interpolation filter for the target sample can be determined based on neighboring blocks sampled via the neighboring MPM mode. That is, the interpolation filter used in the adjacent block can be output as an interpolation filter for the target sample. If the MPM mode is applied to the current block, the decoding apparatus may determine the MPM list based on the intra-prediction mode for the left or upper adjacent block of the current block, and determine the intra-prediction mode based on the MPM list.

[243] Although not shown in the drawing, the decoding apparatus may use the prediction sample as a reconstructed sample according to a prediction mode, or add the residual sample with the prediction sample to generate a reconstructed sample. When there is a residual sample for a target block, the decoding apparatus may receive information regarding the residual for the target block, and information regarding the residual may be included in information regarding the reconstructed sample. The residual information may include a conversion factor related to the residual sample. The decoding apparatus may output a residual sample (or arrays of residual samples) for a target block based on the residual information. The decoding apparatus may generate a reconstructed sample based on the prediction sample and the residual sample, and may output a reconstructed block or a reconstructed picture based on the reconstructed sample. Thereafter, to enhance the subjective/objective image quality as desired, the decoding apparatus may apply an edge filtering procedure such as deblocking filtering and/or an SAO procedure to the reconstructed picture as described above.

[244] In addition, the decoding apparatus may receive prediction information for the current block via a bitstream and may entropy encode it. The prediction information may include information about the intra-prediction mode of the current block. The decoding apparatus may obtain information regarding the intra-prediction mode indicating the intra-prediction mode. The intra-prediction mode information may include information directly indicating the intra-prediction mode for the current block, or may include information indicating any one candidate in the intra-prediction mode candidate list obtained based on the left or top intra-prediction mode block of the current block. The intra-prediction mode candidate list may indicate a list of MPMs.

[245] In addition, when a current block is divided into a plurality of regions, the prediction information may include information indicating the size of the divided regions of the current block and an interpolation filter of each region. In addition, when the interpolation filter for the target sample is selected based on the size of the current block, the intra-prediction mode of the current block, or the distance between the target sample and the reference samples, the prediction information may include information regarding the specific value used to select the interpolation filter for the target sample . When the interpolation filter is selected based on the size of the current block and the intra-prediction mode of the current block, the prediction information may include information regarding the first specific value and information regarding the second specific value. Also, the prediction information may include a mark indicating whether a plurality of interpolation filters are defined as interpolation filters for the target sample. When the mark indicates that a plurality of interpolation filters are defined as interpolation filters for the target sample, a prediction sample of the target sample can be derived based on the plurality of interpolation filters, and when the mark indicates that a plurality of interpolation filters is not defined as interpolation filters for the target sample, the prediction sample the target sample may not be inferred based on multiple interpolation filters. For example, when the label value is 1, the label may indicate that a plurality of interpolation filters are defined as interpolation filters for the target sample, and when the label value is 0, the label may indicate that a plurality of interpolation filters are not defined as interpolation filters for the target sample. samples. The prediction information may be signaled via a video parameter set (VPS), a sequence parameter set (SPS), a picture parameter set (PPS) or a slice segment header, or may be signaled on a block basis.

[246] According to the present disclosure described above, prediction may be performed on a target sample based on an interpolation filter derived according to the current block size, the distance between the target sample and the reference sample, and/or prediction mode (prediction angle), wherein the reference the fractional sample position sample for the target sample can be accurately generated to improve the prediction accuracy for the current block, and the residual for the current block can be reduced to improve coding efficiency.

[247] Also, in accordance with the present disclosure, since the interpolation filter for the target sample can be selected based on various conditions described above, the numbers of interpolation filter selection information bits can be reduced, and the prediction accuracy for the current block can be improved, and the efficiency The encoding of the current block can be improved.

[248] In the above-described embodiment, the methods are described on the basis of a flow chart having a sequence of steps or blocks. The present disclosure is not limited to the order of the steps or blocks described above. Some steps or blocks may occur simultaneously or in a different order from other steps or blocks, as described above. Additionally, those skilled in the art will appreciate that the steps shown in the flowchart above are not exclusive, that additional steps may be included or that one or more steps in the flowchart may be deleted without affecting the volume of this disclosure.

[249] The method in accordance with the present disclosure described above may be implemented in software. An encoding device and/or a decoding device according to the present disclosure may be included in a device that performs image processing, for example, for a TV, computer, smartphone, set-top box, or display device.

[250] When embodiments of the present disclosure are implemented in software, the above-described method may be implemented through modules (processes, functions, and so on) that perform the functions described above. Such modules can be stored in memory and executed by the processor. The memory may be internal or external to the processor, and the memory may be coupled to the processor using various well known means. The processor may include an application specific integrated circuit (ASIC), other chipsets, logic circuitry, and/or data processing apparatus. The memory may include ROM (read-only memory), RAM (random access memory), flash memory, memory card, storage medium, and/or other storage device.

Claims (30)

1. A video encoding method performed by an encoding device, the method comprising: constructing a list of most probable modes (MPM) for the current block based on the intra-prediction mode of the block adjacent to the current block; determining the intra-prediction mode of the current block based on the MPM list; outputting neighboring samples including left neighboring samples and upper neighboring samples of the current block; deriving reference samples for predicting the target sample of the current block among neighboring samples based on the position of the target sample and the prediction angle of the intra-prediction mode; determining an interpolation filter for the target sample based on the intra-prediction mode and the current block size, wherein the interpolation filter is determined as one of a Gaussian filter and a cubic filter; deriving a target sample prediction sample based on the interpolation filter and reference samples; And encoding a bit stream including prediction mode information for the current block representing the intra-prediction mode of the current block among the MPM candidates in the MPM list, in this case, the definition of an interpolation filter for the target sample based on the intra-prediction mode and the size of the current block contains: determining whether the size of the current block is smaller than the first particular value; And based on the current block size being smaller than the first specific value, determining whether the intra-prediction mode prediction angle is smaller than the second specific value, wherein the intra-prediction mode is one of the directional intra-prediction modes with a prediction angle greater than 0, wherein, based on the current block size being greater than or equal to the first specific value, the interpolation filter for the target sample is output as a Gaussian filter, wherein, based on the current block size being smaller than the first specific value and the intra-prediction mode prediction angle being smaller than the second specific value, the interpolation filter for the target sample is output as a cubic filter, and wherein, based on the current block size being smaller than the first specific value and the intra-prediction mode prediction angle being not smaller than the second specific value, the interpolation filter for the target sample is output as a Gaussian filter. 2. A non-transitory computer-readable storage medium storing a bit stream generated by the method, the method comprising: constructing a list of most probable modes (MPM) for the current block based on the intra-prediction mode of the block adjacent to the current block; determining the intra-prediction mode of the current block based on the MPM list; outputting neighboring samples including left neighboring samples and upper neighboring samples of the current block; deriving reference samples for predicting the target sample of the current block among neighboring samples based on the position of the target sample and the prediction angle of the intra-prediction mode; determining an interpolation filter for the target sample based on the intra-prediction mode and the current block size, wherein the interpolation filter is determined as one of a Gaussian filter and a cubic filter; deriving a target sample prediction sample based on the interpolation filter and reference samples; And encoding a bit stream including prediction mode information for the current block representing the intra-prediction mode of the current block among the MPM candidates in the MPM list, in this case, the definition of an interpolation filter for the target sample based on the intra-prediction mode and the size of the current block contains: determining whether the size of the current block is smaller than the first particular value; And based on the current block size being smaller than the first specific value, determining whether the intra-prediction mode prediction angle is smaller than the second specific value, wherein the intra-prediction mode is one of the directional intra-prediction modes with a prediction angle greater than 0, wherein, based on the current block size being greater than or equal to the first specific value, the interpolation filter for the target sample is output as a Gaussian filter, wherein, based on the current block size being smaller than the first specific value and the intra-prediction mode prediction angle being smaller than the second specific value, the interpolation filter for the target sample is output as a cubic filter, and wherein, based on the current block size being smaller than the first specific value and the intra-prediction mode prediction angle being not smaller than the second specific value, the interpolation filter for the target sample is output as a Gaussian filter.