CCLM-based intra-prediction method and apparatus therefor
A picture decoding method, performed by a decoding apparatus according to the present disclosure, comprises the steps of: when an intra-prediction mode of a chroma block is a cross-component linear model (CCLM) mode, deriving neighboring chroma reference samples with respect to the chroma block; deriving neighboring luma reference samples of a luma block corresponding to the chroma block and luma samples in the luma block; deriving down-sampled neighboring luma reference samples and downsampled luma samples by down sampling the neighboring luma reference samples and the luma samples, respectively; deriving a linear model parameter on the basis of the down-sampled neighboring luma reference samples and the neighboring chroma reference samples; generating prediction samples with respect to the chroma block on the basis of the linear model parameter and the down-sampled luma samples of the luma block; and reconstructing the chroma block on the basis of the prediction samples with respect to the chroma block.
1. A picture decoding method performed by a decoding apparatus, the method comprising:
deriving neighboring chroma reference samples for a chroma block based on an intra prediction mode of the chroma block being a cross-component linear model (CCLM) mode;
deriving neighboring luma reference samples of a luma block corresponding to the chroma block and luma samples in the luma block;
deriving downsampled neighboring luma reference samples and downsampled luma samples by downsampling the neighboring luma reference samples and the luma samples;
deriving a linear model parameter based on the downsampled neighboring luma reference samples and the neighboring chroma reference samples;
generating prediction samples for the chroma block based on the linear model parameter and the downsampled luma samples of the luma block; and
reconstructing the chroma block based on the prediction samples for the chroma block,
wherein the neighboring luma reference samples include top neighboring luma reference samples located above a top boundary of the luma block and left neighboring luma reference samples located to a left of a left boundary of the luma block,
wherein the downsampled neighboring luma reference samples include a downsampled left neighboring luma reference sample and a downsampled top neighboring luma reference sample,
wherein based on the top boundary of the luma block being overlapped with a boundary of a coding tree unit (CTU), the downsampled top neighboring luma reference sample is derived based on an equation below,
pTopDsY[ x ]=( pY [2* x− 1][−1]+2* pY [2* x ][−1]+ pY [2* x+ 1][−1]+2)>>2,
where pTopDsY[x] represents the downsampled top neighboring luma reference sample, x represents an x-axis direction position of the downsampled top neighboring luma reference sample, pY[2*x−1][−1], pY[2*x][−1] and pY[2*x+1][−1] represent sample values of the top neighboring luma reference samples, respectively, and a y-axis direction position of the top neighboring luma reference samples is −1.
2. The method of claim 1 , wherein whether the top boundary of the luma block is overlapped with the boundary of the CTU is determined based on whether an equation below is TRUE or FALSE,
bCTUboundary= y CbC &((1<<( Ctb Log 2SizeY−1)−1)==0,
where yCbC represents a y-axis direction position of a top-left sample of a current block for a top-left sample of a current picture, and CtbLog2SizeY represents a log value of a luma CTB size.
3. The method of claim 1 , wherein based on the top boundary of the luma block being overlapped with the boundary of the CTU, the x-axis direction position of the downsampled top neighboring luma reference sample is 0, and based on a existence of a sample value of a luma reference sample corresponding to coordinates (−1, 1), a sample value of the downsampled top neighboring luma reference sample is derived based on an equation below,
pTopDsY[0]=( pY [−1][−1]+2* pY [0][−1]+ pY [1][−1]+2)>>2,
where pTopDsY[0] represents the sample value of the downsampled top neighboring luma reference sample based on the x-axis direction position of the downsampled top neighboring luma reference sample being 0, pY[−1][−1], pY[0][−1] and pY[1][−1] represent sample values of the top neighboring luma reference samples, respectively, and the y-axis direction position of the top neighboring luma reference samples is −1.
4. The method of claim 1 , wherein based on bCTUboundary=yCbC & ((1<<(CtbLog2SizeY−1)−1)==0 being TRUE, the downsampled top neighboring luma reference sample is derived based on the equation of claim 1 ,
where yCbC represents a y-axis direction position of a top-left sample of a current block for a top-left sample of a current picture, and CtbLog2SizeY represents a log value of a luma CTB size.
5. The method of claim 3 , wherein based on bCTUboundary=yCbC & ((1<<(CtbLog2SizeY−1)−1)==0 being TRUE, and based on the x-axis direction position of the downsampled top neighboring luma reference sample being 0, the sample value of the downsampled top neighboring luma reference sample is derived based on the equation of claim 3 ,
where yCbC represents a y-axis direction position of a top-left sample of a current block for a top-left sample of a current picture, and CtbLog2SizeY represents a log value of a luma CTB size.
6. The method of claim 1 , wherein the linear model parameter comprises a first linear model parameter representing a scaling factor and a second linear model parameter representing an offset.
7. A picture encoding method performed by an encoding apparatus, the method comprising:
deriving neighboring chroma reference samples for a chroma block based on an intra prediction mode of the chroma block being a cross-component linear model (CCLM) mode;
deriving neighboring luma reference samples of a luma block corresponding to the chroma block and luma samples in the luma block;
deriving downsampled neighboring luma reference samples and downsampled luma samples by downsampling the neighboring luma reference samples and the luma samples;
deriving a linear model parameter based on the downsampled neighboring luma reference samples and the neighboring chroma reference samples;
generating prediction samples for the chroma block based on the linear model parameter and the downsampled luma samples of the luma block;
deriving residual samples for the chroma block based on the prediction samples for the chroma block; and
encoding picture information including information about the residual samples,
wherein the neighboring luma reference samples include top neighboring luma reference samples located above a top boundary of the luma block and left neighboring luma reference samples located to a left of a left boundary of the luma block,
wherein the downsampled neighboring luma reference samples include a downsampled left neighboring luma reference sample and a downsampled top neighboring luma reference sample,
wherein based on the top boundary of the luma block being overlapped with a boundary of a coding tree unit (CTU), the downsampled top neighboring luma reference sample is derived based on an equation below,
pTopDsY[ x ]=( pY [2* x −1][−1]+2* pY [2* x ][−1]+ pY [2* x+ 1][−1]+2)>>2,
where pTopDsY[x] represents the downsampled top neighboring luma reference sample, x represents an x-axis direction position of the downsampled top neighboring luma reference sample, pY[2*x−1][−1], pY[2*x][−1] and pY[2*x+1][−1] represent sample values of the top neighboring luma reference samples, respectively, and an y-axis direction position of the top neighboring luma reference samples is −1.
8. The method of claim 7 , wherein whether the top boundary of the luma block is overlapped with the boundary of the CTU is determined based on whether an equation below is TRUE or FALSE,
bCTUboundary= y CbC &((1<<( Ctb Log 2SizeY−1)−1)==0,
where yCbC represents a y-axis direction position of a top-left sample of a current block for a top-left sample of a current picture, and CtbLog2SizeY represents a log value of a luma CTB size.
9. The method of claim 7 , wherein based on the top boundary of the luma block being overlapped with the boundary of the CTU, the x-axis direction position of the downsampled top neighboring luma reference sample is 0, and based on a existence of a sample value of a luma reference sample corresponding to coordinates (−1, 1), a sample value of the downsampled top neighboring luma reference sample is derived based on an equation below,
pTopDsY[0]=( pY [−1][−1]+2* pY [0][−1]+ pY [1][−1]+2)>>2,
where pTopDsY[0] represents the sample value of the downsampled top neighboring luma reference sample based on the x-axis direction position of the downsampled top neighboring luma reference sample being 0, pY[−1][−1], pY[0][−1] and pY[1][−1] represent sample values of the top neighboring luma reference samples, respectively, and the y-axis direction position of the top neighboring luma reference samples is −1.
10. The method of claim 7 , wherein based on bCTUboundary=yCbC & ((1<<(CtbLog2SizeY−1)−1)==0 being TRUE, the downsampled top neighboring luma reference sample is derived based on the equation of claim 7 ,
where yCbC represents a y-axis direction position of a top-left sample of a current block for a top-left sample of a current picture, and CtbLog2SizeY represents a log value of a luma CTB size.
11. The method of claim 9 , wherein based on bCTUboundary=yCbC & ((1<<(CtbLog2SizeY−1)−1)==0 being TRUE, and based on the x-axis direction position of the downsampled top neighboring luma reference sample being 0, the sample value of the downsampled top neighboring luma reference sample is derived based on the equation of claim 9 ,
where yCbC represents a y-axis direction position of a top-left sample of a current block for a top-left sample of a current picture, and CtbLog2SizeY represents a log value of a luma CTB size.
12. The method of claim 7 , wherein the linear model parameter comprises a first linear model parameter representing a scaling factor and a second linear model parameter representing an offset.
13. A non-transitory decoder-readable storage medium storing picture information generated by a picture encoding method, the method comprising:
deriving neighboring chroma reference samples for a chroma block based on an intra prediction mode of the chroma block being a cross-component linear model (CCLM) mode;
deriving neighboring luma reference samples of a luma block corresponding to the chroma block and luma samples in the luma block;
deriving downsampled neighboring luma reference samples and downsampled luma samples by downsampling the neighboring luma reference samples and the luma samples;
deriving a linear model parameter based on the downsampled neighboring luma reference samples and the neighboring chroma reference samples;
generating prediction samples for the chroma block based on the linear model parameter and the downsampled luma samples of the luma block;
deriving residual samples for the chroma block based on the prediction samples for the chroma block; and
encoding picture information including information about the residual samples,
wherein the neighboring luma reference samples include top neighboring luma reference samples located above a top boundary of the luma block and left neighboring luma reference samples located to a left of a left boundary of the luma block,
wherein the downsampled neighboring luma reference samples include a downsampled left neighboring luma reference sample and a downsampled top neighboring luma reference sample,
wherein based on the top boundary of the luma block being overlapped with a boundary of a coding tree unit (CTU), the downsampled top neighboring luma reference sample is derived based on an equation below,
pTopDsY[ x ]=( pY [2* x −1][−1]+2* pY [2* x ][−1]+ pY [2* x+ 1][−1]+2)>>2,
where pTopDsY[x] represents the downsampled top neighboring luma reference sample, x represents an x-axis direction position of the downsampled top neighboring luma reference sample, pY[2*x−1][−1], pY[2*x][−1] and pY[2*x+1][−1] represent sample values of the top neighboring luma reference samples, respectively, and an y-axis direction position of the top neighboring luma reference samples is −1.