Intra-prediction using a cross-component linear model in video coding
A video coding mechanism is disclosed. The mechanism includes receiving a video signal partitioned into a chroma block and a first neighboring luma block. The mechanism also includes encoding prediction information for chroma samples of the chroma block into a bitstream. The prediction information for the chroma samples is determined via intra-prediction based on down-sampled neighboring luma samples. The down-sampled neighboring luma samples are generated from a single row of luma samples from the first neighboring luma block. A bitstream including the prediction information for the chroma samples is transmitted to support decoding and display of the chroma samples.
1 . A method for cross-component prediction of a block of video data, the method comprising:
down-sampling a reconstructed luma block to obtain a down-sampled luma block, the reconstructed luma block corresponding to a chroma block;
obtaining first down-sampled luma samples of reconstructed top neighboring luma samples that are above the reconstructed luma block and second down-sampled luma samples of reconstructed left neighboring luma samples that are left of the reconstructed luma block, wherein the reconstructed top neighboring luma samples are in a single row adjacent to the reconstructed luma block and the reconstructed left neighboring luma samples are in three columns adjacent to the reconstructed luma block, wherein video data is in YUV4:2:0 format, wherein the first down-sampled luma samples are obtained by applying a three tap downsampling filter to the reconstructed top neighboring luma samples according to A[i]=(A1[2i]*2+A1[2i−1]+A1[2i+1]+2)>>2, wherein A1[i] is an ith reconstructed top neighboring luma sample in a single row adjacent to the reconstructed luma block and A[i] is an ith down-sampled luma sample of the reconstructed top neighboring luma samples, wherein the second down-sampled luma samples are obtained by applying a six tap downsampling filter to the reconstructed left neighboring luma samples according to L[i]=(L2[2i]*2+L1[2i]+L3[2i]+L2[2i+1]*2+L1[2i+1]+L3[2i+1]+4)>>3, wherein L1[i], L2[i] and L3[i] are ith reconstructed left neighboring luma samples in three columns adjacent to the reconstructed luma block, respectively, and L[i] is an ith down-sampled luma sample of reconstructed left neighboring luma samples;
deriving parameters of a linear model (LM) from:
the first down-sampled luma samples and the second down-sampled luma samples, and
reconstructed neighboring chroma samples that are above the chroma block and reconstructed neighboring chroma samples that are left of the chroma block; and
generating predicted chroma values of the chroma block based on the parameters of the LM and a downsampled luma block.
2 . The method of claim 1 , wherein the LM is a linear model of a cross-component linear model (CCLM) mode, wherein the predicted chroma values of the chroma block are determined according to predc(i,j)=α·recL′(i,j)+β, wherein predc(i,j) are the predicted chroma values of the chroma block, recL′(i,j) are down-sampled luma samples of the down-sampled luma block, and α and β are parameters of the LM.
3 . The method of claim 1 , wherein a max luma value is determined based on values of one or more first down-sampled luma samples that are larger than values of remaining first down-sampled luma samples.
4 . The method of claim 1 , wherein a min luma value is obtained based on values of one or more first down-sampled luma samples that are smaller than values of remaining first down-sampled luma samples.
5 . The method of claim 1 , wherein a max luma value is determined based on values of one or more second down-sampled luma samples that are larger than values of remaining second down-sampled luma samples.
6 . The method of claim 1 , wherein a min luma value is obtained based on values of one or more first down-sampled luma samples that are smaller than values of a rest of the first down-sampled luma samples.
7 . The method of claim 1 , wherein the LM comprises one or more of a cross-component linear model (CCLM) mode, a multi-directional linear model (MDLM) mode, and a multiple model linear model (MMLM) mode.
8 . A non-transitory computer readable medium comprising a computer program product for use by a video coding device, the computer program product comprising computer executable instructions stored on the non-transitory computer readable medium such that when executed by a processor cause the video coding device to:
down-sample a reconstructed luma block to obtain a down-sampled luma block, the reconstructed luma block corresponding to a chroma block;
obtain first down-sampled luma samples of reconstructed top neighboring luma samples that are above the reconstructed luma block and second down-sampled luma samples of reconstructed left neighboring luma samples that are left of the reconstructed luma block, wherein the reconstructed top neighboring luma samples are in a single row adjacent to the reconstructed luma block and the reconstructed left neighboring luma samples are in three columns adjacent to the reconstructed luma block, wherein video data is in YUV4:2:0 format, wherein the first down-sampled luma samples are obtained by applying a three tap downsampling filter to the reconstructed top neighboring luma samples according to A[i]=(A1[2i]*2+A1[2i−1]+A1[2i+1]+2)>>2, wherein A1[i] is an ith reconstructed top neighboring luma sample in a single row adjacent to the reconstructed luma block and A[i] is an ith down-sampled luma sample of the reconstructed top neighboring luma samples, wherein the second down-sampled luma samples are obtained by applying a six tap downsampling filter to the reconstructed left neighboring luma samples according to L[i]=(L2[2i]*2+L1[2i]+L3[2i]+L2[2i+1]*2+L1[2i+1]+L3[2i+1]+4)>>3, wherein L1[i], L2[i] and L3[i] are ith reconstructed left neighboring luma samples in three columns adjacent to the reconstructed luma block, respectively, and L[i] is an ith down-sampled luma sample of reconstructed left neighboring luma samples;
derive parameters of a linear model (LM) from:
the first down-sampled luma samples and the second down-sampled luma samples, and
reconstructed neighboring chroma samples that are above the chroma block and reconstructed neighboring chroma samples that are left of the chroma block; and
generate predicted chroma values of the chroma block based on the parameters of the LM and a downsampled luma block.
9 . The non-transitory computer readable medium of claim 8 , wherein the LM is a linear model of a cross-component linear model (CCLM) mode, wherein the predicted chroma values of the chroma block are determined according to predc(i,j)=α·recL′(i,j)+β, wherein predc(i,j) are the predicted chroma values of the chroma block, recL′(i,j) are down-sampled luma samples of the down-sampled luma block, and α and β are parameters of the LM.
10 . The non-transitory computer readable medium of claim 8 , wherein a max luma value is determined based on values of one or more first down-sampled luma samples that are larger than values of remaining first down-sampled luma samples.
11 . The non-transitory computer readable medium of claim 8 , wherein a min luma value is obtained based on values of one or more first down-sampled luma samples that are smaller than values of remaining first down-sampled luma samples.
12 . The non-transitory computer readable medium of claim 8 , wherein a max luma value is determined based on values of one or more second down-sampled luma samples that are larger than values of remaining second down-sampled luma samples.
13 . The non-transitory computer readable medium of claim 8 , wherein a min luma value is obtained based on values of one or more first down-sampled luma samples that are smaller than values of a rest of the first down-sampled luma samples.
14 . The non-transitory computer readable medium of claim 8 , wherein the LM comprises one or more of a cross-component linear model (CCLM) mode, a multi-directional linear model (MDLM) mode, and a multiple model linear model (MMLM) mode.
15 . A video coding device comprising a processor configured to:
down-sample a reconstructed luma block to obtain a down-sampled luma block, the reconstructed luma block corresponding to a chroma block;
obtain first down-sampled luma samples of reconstructed top neighboring luma samples that are above the reconstructed luma block and second down-sampled luma samples of reconstructed left neighboring luma samples that are left of the reconstructed luma block, wherein the reconstructed top neighboring luma samples are in a single row adjacent to the reconstructed luma block and the reconstructed left neighboring luma samples are in three columns adjacent to the reconstructed luma block, wherein video data is in YUV4:2:0 format, wherein the first down-sampled luma samples are obtained by applying a three tap downsampling filter to the reconstructed top neighboring luma samples according to A[i]=(A1[2i]*2+A1[2i−1]+A1[2i+1]+2)>>2, wherein A1[i] is an ith reconstructed top neighboring luma sample in a single row adjacent to the reconstructed luma block and A[i] is an ith down-sampled luma sample of the reconstructed top neighboring luma samples, wherein the second down-sampled luma samples are obtained by applying a six tap downsampling filter to the reconstructed left neighboring luma samples according to L[i]=(L2[2i]*2+L1[2i]+L3[2i]+L2[2i+1]*2+L1[2i+1]+L3[2i+1]+4)>>3, wherein L1[i], L2[i] and L3[i] are ith reconstructed left neighboring luma samples in three columns adjacent to the reconstructed luma block, respectively, and L[i] is an ith down-sampled luma sample of reconstructed left neighboring luma samples;
derive parameters of a linear model (LM) from:
the first down-sampled luma samples and the second down-sampled luma samples, and
reconstructed neighboring chroma samples that are above the chroma block and reconstructed neighboring chroma samples that are left of the chroma block; and
generate predicted chroma values of the chroma block based on the parameters of the LM and a downsampled luma block.
16 . The video coding device of claim 15 , wherein the LM is a linear model of a cross-component linear model (CCLM) mode, wherein the predicted chroma values of the chroma block are determined according to predc(i,j)=α·recL′(i,j)+β, wherein predc(i,j) are the predicted chroma values of the chroma block, recL′(i,j) are down-sampled luma samples of the down-sampled luma block, and α and β are parameters of the LM.
17 . The video coding device of claim 15 , wherein a max luma value is determined based on values of one or more first down-sampled luma samples that are larger than values of remaining first down-sampled luma samples.
18 . The video coding device of claim 15 , wherein a min luma value is obtained based on values of one or more first down-sampled luma samples that are smaller than values of remaining first down-sampled luma samples.
19 . The video coding device of claim 15 , wherein a max luma value is determined based on values of one or more second down-sampled luma samples that are larger than values of remaining second down-sampled luma samples.
20 . The video coding device of claim 15 , wherein the LM comprises one or more of a cross-component linear model (CCLM) mode, a multi-directional linear model (MDLM) mode, and a multiple model linear model (MMLM) mode.