Chroma from luma prediction using mapping and different types
This disclosure relates to video processing that includes: determining that a chroma block is to be predicted in a Chroma from Luma (CfL) mode, wherein the chroma block corresponds to a luma block; mapping a plurality of luma samples of the luma block to a plurality of neighbor chroma samples in at least one neighbor chroma block that neighbors the chroma block; and performing a CfL prediction of the chroma block using the plurality of neighbor chroma samples as a plurality of prediction samples of the chroma block. This disclosure also relates to video processing that includes: determining a type of chroma from luma (CfL) prediction process from among a plurality of different types of CfL prediction processes; and performing a CfL prediction process according to the type of CfL prediction process.
1. A method of video decoding, the method comprising:
determining that a chroma block is to be predicted in a Chroma from Luma (CfL) mode, wherein the chroma block corresponds to a luma block;
determining a plurality of reconstructed neighbor luma samples by identifying a plurality of smallest-difference reconstructed neighbor luma samples of the luma block, including:
identifying multiple smallest-difference reconstructed neighbor luma samples;
selecting a target smallest-difference reconstructed neighbor luma sample from among the multiple smallest-different reconstructed neighbor luma samples in accordance with the target smallest-difference reconstructed neighbor luma sample having a smallest distance to the luma sample among the multiple smallest-difference reconstructed neighbor luma samples; and
including the target smallest-difference reconstructed neighbor luma sample in the plurality of smallest-difference reconstructed neighbor luma samples:
mapping the plurality of reconstructed luma samples of the luma block to a plurality of neighbor chroma samples in at least one neighbor chroma block that neighbors the chroma block; and
performing a CfL prediction of the chroma block using the plurality of neighbor chroma samples as a plurality of prediction samples of the chroma block.
2. The method of claim 1 , wherein the mapping comprises:
mapping the plurality of reconstructed luma samples to a plurality of reconstructed neighbor luma samples; and
mapping the plurality of reconstructed neighbor luma samples to the plurality of neighbor chroma samples.
3. The method of claim 1 , wherein identifying the plurality of smallest-difference reconstructed neighbor luma samples comprises: determining the plurality of smallest-difference reconstructed neighbor luma samples according to at least one of: sum of absolute difference (SAD) measurements or sum of squared error (SSE) measurements.
4. The method of claim 1 , wherein mapping the plurality of reconstructed luma samples to the plurality of neighbor chroma samples comprises: mapping a patch of reconstructed luma samples to a patch of neighbor reconstructed luma samples.
5. The method of claim 4 , wherein the patch of neighbor reconstructed luma samples comprises a smallest-difference patch, the smallest-difference patch corresponding to the patch of reconstructed luma samples.
6. The method of claim 5 , further comprising: determining the smallest-distance patch based on a center luma sample of the patch of reconstructed luma samples and a center luma sample of the smallest-difference patch.
7. The method of claim 2 , wherein the plurality of reconstructed neighbor luma samples comprises less than all types of reconstructed neighbor luma samples.
8. The method of claim 7 , wherein the less than all types of reconstructed luma samples comprises at least one of: left reconstructed neighbor luma samples, above reconstructed neighbor luma samples, or above-left reconstructed neighbor luma samples.
9. The method of claim 2 , wherein the plurality of reconstructed neighbor luma samples are in a nearest adjacent reference line.
10. A method of video encoding, the method comprising:
determining that a chroma block is to be predicted in a Chroma from Luma (CfL) mode, wherein the chroma block corresponds to a luma block;
determining a plurality of reconstructed neighbor luma samples by identifying a plurality of smallest-difference reconstructed neighbor luma samples of the luma block, including:
identifying multiple smallest-difference reconstructed neighbor luma samples;
selecting a target smallest-difference reconstructed neighbor luma sample from among the multiple smallest-different reconstructed neighbor luma samples in accordance with the target smallest-difference reconstructed neighbor luma sample having a smallest distance to the luma sample among the multiple smallest-difference reconstructed neighbor luma samples; and
including the target smallest-difference reconstructed neighbor luma sample in the plurality of smallest-difference reconstructed neighbor luma samples;
mapping the plurality of reconstructed luma samples of the luma block to a plurality of neighbor chroma samples in at least one neighbor chroma block that neighbors the chroma block; and
performing a CfL prediction of the chroma block using the plurality of neighbor chroma samples as a plurality of prediction samples of the chroma block.
11. The method of claim 10 , wherein the mapping comprises:
mapping the plurality of reconstructed luma samples to a plurality of reconstructed neighbor luma samples; and
mapping the plurality of reconstructed neighbor luma samples to the plurality of neighbor chroma samples.
12. The method of claim 11 , wherein the plurality of reconstructed neighbor luma samples are in a nearest adjacent reference line.
13. The method of claim 11 , wherein the plurality of reconstructed neighbor luma samples comprises less than all types of reconstructed neighbor luma samples.
14. The method of claim 10 , wherein mapping the plurality of reconstructed luma samples to the plurality of neighbor chroma samples comprises mapping a patch of reconstructed luma samples to a patch of neighbor reconstructed luma samples.
15. The method of claim 14 , wherein the patch of neighbor reconstructed luma samples comprises a smallest-difference patch, the smallest-difference patch corresponding to the patch of reconstructed luma samples.
16. The method of claim 15 , further comprising determining the smallest-distance patch based on a center luma sample of the patch of reconstructed luma samples and a center luma sample of the smallest-difference patch.
17. A method of visual processing, the method comprising:
obtaining a source video sequence that comprises a plurality of frames; and
performing a conversion between the source video sequence and a video bitstream of visual media data according to a format rule,
wherein the video bitstream comprises a set of encoded blocks including a chroma block and a corresponding luma block; and
wherein the format rule specifies that:
the chroma block is to be predicted in a Chroma from Luma (CfL) mode;
a plurality of reconstructed neighbor luma samples is to be determined by identifying a plurality of smallest-difference reconstructed neighbor luma samples of the luma block, including:
identifying multiple smallest-difference reconstructed neighbor luma samples;
selecting a target smallest-difference reconstructed neighbor luma sample from among the multiple smallest-different reconstructed neighbor luma samples in accordance with the target smallest-difference reconstructed neighbor luma sample having a smallest distance to the luma sample among the multiple smallest-difference reconstructed neighbor luma samples; and
including the target smallest-difference reconstructed neighbor luma sample in the plurality of smallest-difference reconstructed neighbor luma samples;
the plurality of luma samples of the luma block is to be mapped to a plurality of neighbor chroma samples in at least one neighbor chroma block that neighbors the chroma block; and
a CfL prediction for the chroma block is to be performed using the plurality of neighbor chroma samples as a plurality of prediction samples of the chroma block.
18. The method of claim 17 , wherein mapping the plurality of reconstructed luma samples to the plurality of neighbor chroma samples comprises: mapping a patch of reconstructed luma samples to a patch of neighbor reconstructed luma samples.
19. The method of claim 18 , wherein the patch of neighbor reconstructed luma samples comprises a smallest-difference patch, the smallest-difference patch corresponding to the patch of reconstructed luma samples.
20. The method of claim 17 , wherein the plurality of reconstructed neighbor luma samples comprises less than all types of reconstructed neighbor luma samples.