Chroma from luma prediction using neighbor luma samples
This disclosure relates to video processing that includes a video processing device that: determines that a Chroma from Luma (CfL) prediction mode is to be applied to a luma block in a received coded bitstream; generates a neighbor luma average for the luma block by averaging a set of reconstructed luma samples, wherein the set of reconstructed luma samples comprises a plurality of reconstructed neighbor luma samples in at least one neighbor luma block that neighbors the luma block; generates an alternating current (AC) contribution of a plurality of prediction samples of a chroma block co-located with the luma block based on a plurality of luma samples in the luma block and the neighbor luma average; and reconstructs the chroma block at least by applying the CfL prediction mode based on the AC contribution.
1. A method for video decoding, the method comprising:
determining that a Chroma from Luma (CfL) prediction mode is to be applied to a chroma block in a received coded bitstream;
generating a neighbor luma average for the chroma block by averaging a set of reconstructed luma samples, wherein the set of reconstructed luma samples comprises a plurality of reconstructed neighbor luma samples in at least one neighbor luma block of the chroma block;
generating an alternating current (AC) contribution of a plurality of cross-color prediction samples for the chroma block based on a plurality of luma samples of a luma block co-located with the chroma block and the neighbor luma average; and
reconstructing the chroma block by applying the CfL prediction mode based on at least the plurality of cross-color prediction samples having the AC contribution.
2. The method of claim 1 , further comprising:
sub-sampling the plurality of reconstructed neighbor luma samples before averaging the set of reconstructed luma samples to generate a plurality of sub-sampled reconstructed neighbor luma samples,
wherein the set of reconstructed luma samples comprises the plurality of sub-sampled reconstructed neighbor luma samples.
3. The method of claim 1 , wherein generating the AC contribution comprises subtracting the neighbor luma average from the plurality of luma samples of the luma block co-located with the chroma block.
4. The method of claim 1 , wherein reconstructing the chroma block comprises:
generating a scaled AC contribution by multiplying the AC contribution with a scaling parameter; and
adding the scaled AC contribution to a DC contribution of the plurality of cross-color prediction samples of the chroma block.
5. The method of claim 1 , wherein the plurality of reconstructed neighbor luma samples comprises at least one of: one or more left reconstructed neighbor luma samples, one or more above reconstructed neighbor luma samples, or one or more above-left reconstructed neighbor luma samples of the chroma block.
6. The method of claim 1 , wherein the plurality of reconstructed neighbor luma samples comprises only one or more above reconstructed neighbor luma samples and one or more left reconstructed neighbor luma samples of the chroma block.
7. The method of claim 1 , wherein the plurality of reconstructed neighbor luma samples comprises at least one of: one or more left reconstructed neighbor luma samples, one or more above reconstructed neighbor luma samples, one or more right reconstructed neighbor luma samples, one or more bottom reconstructed neighbor luma samples, or one or more corner reconstructed neighbor luma samples of the chroma block.
8. The method of claim 1 , wherein the plurality of reconstructed neighbor luma samples comprises at least one of: one or more above-right reconstructed neighbor luma samples, or one or more bottom-left reconstructed neighbor luma samples of the chroma block.
9. The method of claim 1 , wherein one or more types of reconstructed neighbor luma samples to comprise the plurality of reconstructed neighbor luma samples used to generate the neighbor luma average is one of:
explicitly indicated via signaling; or
implicitly indicated based on coded information.
10. The method of claim 9 , wherein the one or more types of reconstructed neighbor luma samples are implicitly indicated based on the coded information, and the coded information comprises at least one of: an intra prediction mode of the luma block, a block shape, a block size, or a block aspect ratio.
11. The method of claim 1 , wherein the plurality of reconstructed neighbor luma samples comprises only one or more above reconstructed neighbor luma samples when only the above reconstructed neighbor luma samples being available.
12. The method of claim 1 , wherein the plurality of reconstructed neighbor luma samples comprises only one or more left reconstructed neighbor luma samples when only the left reconstructed neighbor luma samples being available.
13. The method of claim 1 , further comprising: when none of a predetermined type of reconstructed neighbor luma samples are available for the chroma block, padding one or more reconstructed neighbor luma samples of the predetermined type, wherein the padding is performed according to a padding method used in intra angular prediction mode, and wherein the neighbor luma average is determined based on the one or more padded reconstructed neighbor luma samples.
14. The method of claim 1 , wherein the plurality of reconstructed neighbor luma samples comprises one or more above reconstructed neighbor luma sample of the chroma block, wherein the one or more above reconstructed neighbor luma samples are only in a nearest above reference line when is located at a super block boundary.
15. The method of claim 1 , wherein the plurality of reconstructed neighbor luma samples are in a nearest adjacent reference line.
16. The method of claim 1 , wherein the plurality of reconstructed neighbor luma samples are not sub-sampled prior to being averaged to generate the neighbor luma average.
17. The method of claim 1 , wherein the set of reconstructed luma samples further comprises one or more boundary luma samples in the luma block co-located with the chroma block.
18. The method of claim 17 , wherein the one or more boundary luma samples comprises at least one of: one or more top boundary luma samples or one or more left boundary luma samples.
19. A video encoding device, comprising
a memory storing a set of instructions; and
a processor configured to execute the set of instructions, and upon execution of the set of instructions, configured to:
determine that a Chroma from Luma (CfL) prediction mode is to be applied to a chroma block in a video bitstream;
average a set of reconstructed luma samples to generate a neighbor luma average for the luma block, wherein the set of reconstructed luma samples comprise a plurality of reconstructed neighbor luma samples in at least one neighbor luma block of the chroma block;
generate an alternating current (AC) contribution of a plurality of cross-color prediction samples for the chroma block based on a plurality of luma samples of a luma block co-located with the chroma block and the neighbor luma average;
generate a scaling factor for scaling the AC contribution of the plurality of cross-color prediction samples for predicting AC components of the chroma block; and
encode the scaling factor in the video bitstream.
20. A method for processing a video block, comprising converting the video block to a bitstream, wherein the bitstream comprises:
encoded data of a luma block of the video block;
an encoded Chroma-from-Luma mode indicator for indicating that a chroma block of the video block is cross-color predicted from at least an alternating current (AC) component of the luma block relative to a DC level determined from an average of at least a plurality of reconstructed neighboring luma samples of the luma block; and
an encoded scaling factor for scaling the AC component of the luma block for predicting the chroma block.