Granted Patent B2

US 12,526,424 · App. 18/501,526 · Granted Jan 13, 2026

Inventors: Zhipin Deng (Beijiing, CN); Kai Zhang (San Diego, CA); Li Zhang (San Diego, CA); Hongbin Liu (Beijing, CN); Jizheng Xu (San Diego, CA)

Assignees: BEIJING BYTEDANCE NETWORK TECHNOLOGY CO., LTD.; BYTEDANCE INC.

H04N19/132H04N19/105H04N19/117H04N19/12H04N19/159H04N19/176H04N19/186H04N19/60H04N19/70H04N19/82

Abstract

Claims (142)

1 . A method for video processing, comprising:

determining, for a first conversion between a current video block of a video and a bitstream of the video, a spatial neighboring block of the current video block, wherein the current video block is an intra block, and wherein the spatial neighboring block is applied with a first prediction mode in which prediction samples of the spatial neighboring block are derived by performing a boundary down-sampling operation on reference samples of the spatial neighboring block based on a size of the spatial neighboring block, followed by a matrix vector multiplication operation, and selectively followed by an up-sampling operation;

converting the first prediction mode to a second prediction mode for the spatial neighboring block, wherein the second prediction mode is different from the first prediction mode;

constructing, based on the second prediction mode, a mode candidate list for the current video block;

determining a third prediction mode for the current video block, wherein the third prediction mode is different from the first prediction mode; and

performing the first conversion based on the third prediction mode;

wherein whether to apply the first prediction mode is specified by multiple syntax elements,

wherein a first syntax element is included in a sequence parameter set, and a second syntax element is a coding unit level syntax element, and

wherein when the second syntax element is not included in the bitstream, the first prediction mode is not applied.

2 . The method of claim 1 , wherein at least one bin of the second syntax element is context-based coded,

wherein no more than K contexts are used for the second syntax element, wherein K is equal to 4,

wherein when a width-height ratio of the spatial neighboring block is greater than 2, a context with an index of 3 is used for a first bin of the second syntax element,

wherein when a width-height ratio of the spatial neighboring block is smaller than or equal to 2, a single context selected from contexts with indices of 0, 1, or 2 is used for a first bin of the second syntax element,

wherein the first prediction mode includes multiple types, and a type index for the spatial neighboring block is derived excluding referring to type indices of previous video blocks,

wherein the type index for the spatial neighboring block is explicitly included in the bitstream,

wherein a down-sampling factor derived in the boundary down-sampling operation based on the size of the spatial neighboring block is larger than or equal to 1, and wherein a one-dimensional vector array is further derived based on concatenating down-sampled reference samples derived from the boundary down-sampling operation and the one-dimensional vector array is used as input of the matrix vector multiplication operation,

wherein in the second prediction mode, distance-based weighted calculations are applied on reference values in a vertical direction and a horizontal direction to derive prediction values, and

wherein the second prediction mode includes a planar mode.

3 . The method of claim 1 , further comprising:

determining, for a second conversion between a chroma block of the video and a bitstream of the video, whether to apply the first prediction mode on the chroma block;

generating prediction samples for the chroma block based on the determining; and

performing the second conversion between the chroma block and the bitstream;

wherein in responding to determining to apply the first prediction mode on the chroma block, the prediction samples are generated by performing a boundary down-sampling operation based on a size of the chroma block, followed by the matrix vector multiplication operation, and selectively followed by the up-sampling operation.

4 . The method of claim 3 , wherein when determining not to apply the first prediction mode on the chroma block, the prediction samples are generated by performing a luma intra prediction mode conversion operation, followed by a chroma intra prediction mode derivation operation, and followed by a general intra samples derivation operation,

wherein determining to apply the first prediction mode on the chroma block is based on a first luma prediction mode of a first corresponding luma block being same as the first prediction mode,

wherein determining to apply the first prediction mode on the chroma block is further based on a value of a syntax element intra_chroma_pred_mode, and wherein the syntax element intra_chroma_pred_mode is used to specify a mapping relation between a chroma component intra mode and a luma component intra mode,

wherein the syntax element intra_chroma_pred_mode specifies that a Derived Mode (DM) mode is applied on the chroma block, and wherein in the DM mode, the chroma component intra mode is identical to the luma component intra mode,

wherein the first corresponding luma block is a luma block covering a luma sample corresponding to a top-left sample of the chroma block,

wherein in the luma intra prediction mode conversion operation, a converted luma prediction mode is derived based on a second luma prediction mode of a second corresponding luma block,

wherein the second corresponding luma block is a luma block covering a luma sample corresponding to a center sample of the chroma block,

wherein in the chroma intra prediction mode derivation operation, a second chroma prediction mode is derived for the chroma block based on the converted luma prediction mode, and wherein the second chroma prediction mode is different from the first prediction mode, and

wherein when a size of the chroma block is same as a size of the first corresponding luma block, same matrix coefficients are used in the first prediction mode and the first luma prediction mode.

5 . The method of claim 1 , further comprising:

performing a third conversion between the current video block of the video and the bitstream of the current video block using a matrix based intra prediction (MIP) mode in which prediction samples of the current video block are determined by performing, on reference samples of the current video block, a boundary down-sampling operation based on a size of the current video block, followed by a matrix vector multiplication operation, and selectively followed by an up-sampling operation;

wherein the prediction samples are determined according to following formulas in the up-sampling operation:

predSamples[ x Hor+ dX][y Hor]=((upHor− dX )*predSamples[ x Hor][ y Hor]+ dX *predSamples[ x Hor+upHor][ y Hor]+offsetHor)/upHor, and

predSamples[ x Ver][ y Ver+ dY ]=((up Ver− dY )*predSamples[ x Ver][ y Ver]+ dY *predSamples[ x Ver][ y Ver+up Ver]+offset Ver)/up Ver,

wherein offsetHor and offset Ver are integers,

wherein upHor is a function of a current video block width and a pre-defined value based on a current video block size,

wherein up Ver is a function of a current video block height and the pre-defined value based on the current video block size;

wherein dX is 1 . . . upHor−1,

wherein dY is 1 . . . upVer−1, and

wherein xHor are positions based on the upHor and yHor are positions based on the up Ver.

6 . The method of claim 5 , wherein offsetHor=upHor/2, and offset Ver=up Ver/2,

wherein the boundary down-sampling operation includes deriving, according to a rule, boundary samples by applying a left bit shift operation or a right bit shift operation on a sum of at least one reference boundary sample, and wherein the rule determines whether to apply the left bit shift operation or the right bit shift operation,

wherein the rule defines that the right bit shift operation is applied when a number of shifted bits is greater than zero,

wherein the boundary samples redBdryS[x] are calculated using one of following equations:

redBdryS[ x ]=(Σ i=0 bDwn-1 refS[ x *bDwn+ i ]+(1<<(Log 2(bDwn)−1)))>>Log 2(bDwn), if bDwn>1, or

redBdryS[ x ]=refS[ x ]if bDwn=1,

wherein bDwn is equal to a function of the current video block size and a boundary size, wherein refS [x] indicates a number of reference samples x, wherein >> indicates the right bit shift operation, and wherein << indicates the left bit shift operation,

wherein the boundary size is predefined based on the current video block size,

wherein bDwn is calculated as bDwn=nTbs/boundarySize, wherein nTbs and boundarySize represent the current video block size and the boundary size, respectively,

wherein the MIP mode includes multiple types, and a type index for the current video block is derived excluding referring to type indices of previous video blocks, and

wherein the type index for the current video block is explicitly included in the bitstream.

7 . The method of claim 1 , further comprising:

determining, for a fourth conversion between a first video block of the video and the bitstream of the video, that a first intra mode is applied on the first video block;

invoking a reference sample filtering operation to derive reference samples of the first video block;

generating prediction samples for the first video block based on the reference samples;

performing the fourth conversion based on the prediction samples of the first video block;

determining, for a fifth conversion between a second video block of the video and the bitstream of the video, that a second intra mode is applied on the second video block of the video, wherein the second intra mode is different from the first intra mode;

deriving reference samples of the second video block without invoking the reference sample filtering operation;

performing a boundary down-sampling operation on the reference samples of the second video block based on a size of the second video block, followed by a matrix vector multiplication operation, and selectively followed by an up-sampling operation to generate prediction samples for the second video block; and

performing the fifth conversion based on the prediction samples of the second video block.

8 . The method of claim 7 , wherein when the second intra mode is applied on the second video block, a third mode is not applied on the second video block, wherein in the third mode, a difference between a quantized residual and a prediction of the quantized residual is represented in the bitstream using a pulse coding modulation representation, and prediction samples in the third mode are derived by invoking the reference sample filtering operation,

wherein a second syntax element indicating whether to apply the second intra mode is included in the bitstream,

wherein a third syntax element indicating whether to apply the third mode is not included in the bitstream,

wherein at least one bin of the second syntax element is context-based coded,

wherein no more than K contexts is used for the second syntax element, wherein K is equal to 4,

wherein when a width-height ratio of the second video block is greater than 2, a context with an index of 3 is used for a first bin of the second syntax element, and

wherein when a width-height ratio of the second video block is smaller than or equal to 2, a single context selected from contexts with indices of 0, 1, or 2 is used for a first bin of the second syntax element.

9 . The method of claim 1 , further comprising:

determining that a first intra mode is applied on a video block of the video,

performing a boundary down-sampling operation on reference samples of the video block based on a size of the video block, followed by the matrix vector multiplication operation, and selectively followed by an up-sampling operation to generate prediction samples for the video block; and

performing the first conversion based on the prediction samples and residual samples of the video block;

wherein a secondary transform tool is applied on the video block, and a secondary transform matrix is selected to generate the residual samples for the video block, and

wherein the secondary transform tool includes applying, during encoding, a forward secondary transform to an output of a forward primary transform before applying a quantization process, or applying, during decoding, an inverse secondary transform to an output of a dequantization process before applying an inverse primary transform.

10 . The method of claim 9 , wherein the secondary transform tool includes a non-separable secondary transform.

11 . The method of claim 9 , wherein the secondary transform tool includes a Reduced Secondary Transform (RST) or a rotation transform.

12 . The method of claim 9 , wherein based on the first intra mode being applied on the video block, a second intra mode is converted from the first intra mode, and wherein a secondary transform classification is derived based on the second intra mode,

wherein the selected secondary transform matrix is determined based on the secondary transform classification,

wherein a down-sampling factor derived in the boundary down-sampling operation is larger than or equal to 1, and wherein a one-dimensional vector array is further derived based on concatenating down-sampled reference samples derived from the boundary down-sampling operation, and the one-dimensional vector array is used as an input to the matrix vector multiplication operation,

wherein in the second intra mode, distance-based weighted calculations are applied on reference values in vertical direction and horizontal direction to derive prediction values,

wherein the second intra mode includes a planar mode,

wherein whether to apply the secondary transform tool is based on a height (H) or a width (W) of the video block.

13 . The method of claim 1 , wherein the first conversion includes encoding the current video block into the bitstream.

14 . The method of claim 1 , wherein the first conversion includes decoding the current video block from the bitstream.

15 . An apparatus for processing video data comprising a processor and a non-transitory memory with instructions thereon, wherein the instructions upon execution by the processor, cause the processor to:

determine, for a first conversion between a current video block of a video and a bitstream of the video, a spatial neighboring block of the current video block, wherein the current video block is an intra block, and wherein the spatial neighboring block is applied with a first prediction mode in which prediction samples of the spatial neighboring block are derived by performing a boundary down-sampling operation on reference samples of the spatial neighboring block based on a size of the spatial neighboring block, followed by a matrix vector multiplication operation, and selectively followed by an up-sampling operation;

convert the first prediction mode to a second prediction mode for the spatial neighboring block, wherein the second prediction mode is different from the first prediction mode;

construct, based on the second prediction mode, a mode candidate list for the current video block;

determine a third prediction mode for the current video block, wherein the third prediction mode is different from the first prediction mode; and

perform the first conversion based on the third prediction mode;

wherein whether to apply the first prediction mode is specified by multiple syntax elements,

wherein a first syntax element is included in a sequence parameter set, and a second syntax element is a coding unit level syntax element, and

wherein when the second syntax element is not included in the bitstream, the first prediction mode is not applied.

16 . The apparatus of claim 15 , wherein at least one bin of the second syntax element is context-based coded,

wherein no more than K contexts are used for the second syntax element, wherein K is equal to 4,

wherein when a width-height ratio of the spatial neighboring block is greater than 2, a context with an index of 3 is used for a first bin of the second syntax element,

wherein when a width-height ratio of the spatial neighboring block is smaller than or equal to 2, a single context selected from contexts with indices of 0, 1, or 2 is used for a first bin of the second syntax element,

wherein the first prediction mode includes multiple types, and a type index for the spatial neighboring block is derived excluding referring to type indices of previous video blocks,

wherein the type index for the spatial neighboring block is explicitly included in the bitstream,

wherein a down-sampling factor derived in the boundary down-sampling operation based on the size of the spatial neighboring block is larger than or equal to 1, and wherein a one-dimensional vector array is further derived based on concatenating down-sampled reference samples derived from the boundary down-sampling operation and the one-dimensional vector array is used as input of the matrix vector multiplication operation,

wherein in the second prediction mode, distance-based weighted calculations are applied on reference values in a vertical direction and a horizontal direction to derive prediction values, and

wherein the second prediction mode includes a planar mode.

17 . A non-transitory computer-readable storage medium storing instructions that cause a processor to:

determine, for a first conversion between a current video block of a video and a bitstream of the video, a spatial neighboring block of the current video block, wherein the current video block is an intra block, and wherein the spatial neighboring block is applied with a first prediction mode in which prediction samples of the spatial neighboring block are derived by performing a boundary down-sampling operation on reference samples of the spatial neighboring block based on a size of the spatial neighboring block, followed by a matrix vector multiplication operation, and selectively followed by an up-sampling operation;

convert the first prediction mode to a second prediction mode for the spatial neighboring block, wherein the second prediction mode is different from the first prediction mode;

construct, based on the second prediction mode, a mode candidate list for the current video block;

determine a third prediction mode for the current video block, wherein the third prediction mode is different from the first prediction mode; and

perform the first conversion based on the third prediction mode;

wherein whether to apply the first prediction mode is specified by multiple syntax elements,

wherein a first syntax element is included in a sequence parameter set, and a second syntax element is a coding unit level syntax element, and

wherein when the second syntax element is not included in the bitstream, the first prediction mode is not applied.

18 . The non-transitory computer-readable storage medium of claim 17 , wherein at least one bin of the second syntax element is context-based coded,

wherein no more than K contexts are used for the second syntax element, wherein K is equal to 4,

wherein when a width-height ratio of the spatial neighboring block is greater than 2, a context with an index of 3 is used for a first bin of the second syntax element,

wherein when a width-height ratio of the spatial neighboring block is smaller than or equal to 2, a single context selected from contexts with indices of 0, 1, or 2 is used for a first bin of the second syntax element,

wherein the first prediction mode includes multiple types, and a type index for the spatial neighboring block is derived excluding referring to type indices of previous video blocks,

wherein the type index for the spatial neighboring block is explicitly included in the bitstream,

wherein a down-sampling factor derived in the boundary down-sampling operation based on the size of the spatial neighboring block is larger than or equal to 1, and wherein a one-dimensional vector array is further derived based on concatenating down-sampled reference samples derived from the boundary down-sampling operation and the one-dimensional vector array is used as input of the matrix vector multiplication operation,

wherein in the second prediction mode, distance-based weighted calculations are applied on reference values in a vertical direction and a horizontal direction to derive prediction values, and

wherein the second prediction mode includes a planar mode.

19 . A method for storing a bitstream of a video, comprising:

determining, for a current video block of the video, a spatial neighboring block of the current video block, wherein the current video block is an intra block, and wherein the spatial neighboring block is applied with a first prediction mode in which prediction samples of the spatial neighboring block are derived by performing a boundary down-sampling operation on reference samples of the spatial neighboring block based on a size of the spatial neighboring block, followed by a matrix vector multiplication operation, and selectively followed by an up-sampling operation;

converting the first prediction mode to a second prediction mode for the spatial neighboring block, wherein the second prediction mode is different from the first prediction mode;

constructing, based on the second prediction mode, a mode candidate list for the current video block;

determining a third prediction mode for the current video block, wherein the third prediction mode is different from the first prediction mode;

generating the bitstream from the current video block based on the third prediction mode; and

storing the bitstream in a non-transitory computer-readable recording medium;

wherein whether to apply the first prediction mode is specified by multiple syntax elements,

wherein a first syntax element is included in a sequence parameter set, and a second syntax element is a coding unit level syntax element, and

wherein when the second syntax element is not included in the bitstream, the first prediction mode is not applied.

20 . The method of claim 19 ,

wherein at least one bin of the second syntax element is context-based coded,

wherein no more than K contexts are used for the second syntax element, wherein K is equal to 4,

wherein when a width-height ratio of the spatial neighboring block is greater than 2, a context with an index of 3 is used for a first bin of the second syntax element,

wherein when a width-height ratio of the spatial neighboring block is smaller than or equal to 2, a single context selected from contexts with indices of 0, 1, or 2 is used for a first bin of the second syntax element,

wherein the first prediction mode includes multiple types, and a type index for the spatial neighboring block is derived excluding referring to type indices of previous video blocks,

wherein the type index for the spatial neighboring block is explicitly included in the bitstream,

wherein a down-sampling factor derived in the boundary down-sampling operation based on the size of the spatial neighboring block is larger than or equal to 1, and wherein a one-dimensional vector array is further derived based on concatenating down-sampled reference samples derived from the boundary down-sampling operation and the one-dimensional vector array is used as input of the matrix vector multiplication operation,

wherein in the second prediction mode, distance-based weighted calculations are applied on reference values in a vertical direction and a horizontal direction to derive prediction values, and

wherein the second prediction mode includes a planar mode.

Assignments (2)

Priority Claims (1)

Continuity (4)

References Cited (400)