Methods and devices for intra sub-partition coding mode
The present disclosure relates to an intra sub-partition (ISP) method of encoding a video signal. The method includes partitioning a block of video data into a plurality of sub-partitions, obtaining a number of most probable intra predictions based on intra prediction modes of neighboring blocks of a current block and determining an intra prediction mode to predict the samples of the block sub-partitions, selecting an optimal transform combination from DCT-II, DST-VII and DCT-VIII transform kernels for each sub-partition, obtaining residual coefficients of a first sub-partition and deriving dequantized residual coefficients of the first sub-partition, deriving residual samples of the first sub-partition by performing inverse transform on dequantized residual coefficients of the first sub-partition in horizontal and vertical directions, deriving prediction samples of the first sub-partition, and deriving the reconstructed samples of the first sub-partition, and deriving prediction samples of a second sub-partition.
1. An intra sub-partition (ISP) method of encoding a video signal, the method comprising:
partitioning a block of video data into a plurality of sub-partitions, wherein each sub-partition comprises NM samples, where N and M are number of samples in horizontal or vertical direction;
obtaining a number of most probable intra predictions based on intra prediction modes of neighboring blocks of a current block and determining an intra prediction mode to predict the samples of the block sub-partitions;
selecting an optimal transform combination from discrete cosine transform II (DCT-II), discrete sine transform VII (DST-VII), and discrete cosine transform VIII (DCT-VIII) transform kernels for each sub-partition, wherein selecting the optimal transform combination comprises applying a multiple transform selection (MTS) mode to the ISP;
obtaining residual coefficients of a first sub-partition and deriving dequantized residual coefficients of the first sub-partition by performing inverse-quantization of the residual coefficients of the first sub-partition;
deriving residual samples of the first sub-partition by performing inverse transforms on dequantized residual coefficients of the first sub-partition in horizontal and vertical directions, wherein the inverse transforms are based on the optimal transform combination of the first sub-partition;
deriving prediction samples of the first sub-partition by performing intra prediction of the first sub-partition based on reconstructed samples adjacent to the first sub-partition;
deriving the reconstructed samples of the first sub-partition by adding the residual samples of the first sub-partition to the prediction samples of the first sub-partition; and
deriving prediction samples of a second sub-partition by performing intra prediction of the second sub-partition based on the reconstructed samples of the first sub-partition, wherein the optimal transform combination is indicated by one MTS flag and one MTS index, wherein the MTS flag corresponds to the block and is shared by sub-partitions and the MTS index corresponds to each sub-partition.
2. The method of claim 1 , further comprising:
selecting transform kernels from DCT-II and DST-VII based on a sub-partition size in response to determining that the MTS flag is zero; and
selecting transform kernels from DST-VII and DCT-VIII that are applied in horizontal and vertical directions for one sub-partition and determining the value of the corresponding MTS index in response to determining that the MTS flag is not zero.
3. The method of claim 2 , wherein selecting transform kernels from DCT-II and DST-VII based on the sub-partition size further comprises:
in response to determining that the number of samples in one dimension of the sub-partition is equal or smaller than 16, applying the DST-VII transform in the dimension;
in response to determining that the number of samples in one dimension is larger than 16, applying the DCT-II transform in the dimension.
4. A computing device comprising:
one or more processors;
a non-transitory computer-readable memory storing instructions executable by the one or more processors, wherein the one or more processors are configured to:
partition a block of video data into a plurality of sub-partitions, wherein each sub-partition comprises NM samples where N and M are number of samples in horizontal or vertical direction;
obtain a number of most probable intra predictions based on intra prediction modes of neighboring blocks of a current block and determining an intra prediction mode to predict the samples of the block sub-partitions;
select an optimal transform combination from discrete cosine transform II (DCT-II), discrete sine transform VII (DST-VII), and discrete cosine transform VIII (DCT-VIII) transform kernels for each sub-partition, wherein selecting the optimal transform combination comprises applying a multiple transform selection (MTS) mode to an intra sub-partition (ISP) mode;
obtain residual coefficients of a first sub-partition and deriving dequantized residual coefficients of the first sub-partition by performing inverse-quantization of the residual coefficients of the first sub-partition;
derive residual samples of the first sub-partition by performing inverse transforms on dequantized residual coefficients of the first sub-partition in horizontal and vertical directions, wherein the inverse transforms are based on the optimal transform combination of the first sub-partition;
derive prediction samples of the first sub-partition by performing intra prediction of the first sub-partition based on reconstructed samples adjacent to the first sub-partition;
derive the reconstructed samples of the first sub-partition by adding the residual samples of the first sub-partition to the prediction samples of the first sub-partition; and
derive prediction samples of a second sub-partition by performing intra prediction of the second sub-partition based on the reconstructed samples of the first sub-partition,
wherein the optimal transform combination is indicated by one MTS flag and one MTS index, wherein the MTS flag corresponds to the block and is shared by sub-partitions and the MTS index corresponds to each sub-partition.
5. The computing device of claim 4 , wherein the one or more processors are further configured to:
select transform kernels from DCT-II and DST-VII based on a sub-partition size in response to determining that the MTS flag is zero; and
select transform kernels from DST-VII and DCT-VIII that are applied in horizontal and vertical directions for one sub-partition and determine the value of the corresponding MTS index in response to determining that the MTS flag is not zero.
6. The computing device of claim 5 , wherein the one or more processors configured to select transform kernels from DCT-II and DST-VII based on the sub-partition size are further configured to:
in response to determining that the number of samples in one dimension of the sub-partition is equal or smaller than 16, apply the DST-VII transform in the dimension;
in response to determining that the number of samples in one dimension is larger than 16, apply the DCT-II transform in the dimension.
7. A non-transitory computer-readable storage medium storing a plurality of programs for execution by a computing device having one or more processors, wherein the plurality of programs, when executed by the one or more processors, cause the computing device to perform:
partitioning a block of video data into a plurality of sub-partitions, wherein each sub-partition comprises N×M samples where N and M are number of samples in horizontal or vertical direction;
obtaining a number of most probable intra predictions based on intra prediction modes of neighboring blocks of a current block and determining an intra prediction mode to predict the samples of the block sub-partitions;
selecting an optimal transform combination from discrete cosine transform II (DCT-II), discrete sine transform VII (DST-VII), and discrete cosine transform VIII (DCT-VIII) transform kernels for each sub-partition, wherein selecting the optimal transform combination comprises applying a multiple transform selection (MTS) mode to an intra sub-partition (ISP) mode;
obtaining residual coefficients of a first sub-partition and deriving dequantized residual coefficients of the first sub-partition by performing inverse-quantization of the residual coefficients of the first sub-partition;
deriving residual samples of the first sub-partition by performing inverse transforms on dequantized residual coefficients of the first sub-partition in horizontal and vertical directions, wherein the inverse transforms are based on the optimal transform combination of the first sub-partition;
deriving prediction samples of the first sub-partition by performing intra prediction of the first sub-partition based on reconstructed samples adjacent to the first sub-partition;
deriving the reconstructed samples of the first sub-partition by adding the residual samples of the first sub-partition to the prediction samples of the first sub-partition; and
deriving prediction samples of a second sub-partition by performing intra prediction of the second sub-partition based on the reconstructed samples of the first sub-partition,
wherein the optimal transform combination is indicated by one MTS flag and one MTS index, wherein the MTS flag corresponds to the block and is shared by sub-partitions and the MTS index corresponds to each sub-partition.
8. The non-transitory computer readable storage medium of claim 7 , wherein the plurality of programs further cause the computing device to perform:
selecting transform kernels from DCT-II and DST-VII based on a sub-partition size in response to determining that the MTS flag is zero; and
selecting transform kernels from DST-VII and DCT-VIII that are applied in horizontal and vertical directions for one sub-partition and determining the value of the corresponding MTS index in response to determining that the MTS flag is not zero.
9. The non-transitory computer readable storage medium of claim 8 , wherein the plurality of programs caused the computing device to perform selecting transform kernels from DCT-II and DST-VII based on the sub-partition size further cause the computing device to perform:
in response to determining that the number of samples in one dimension of the sub-partition is equal or smaller than 16, applying the DST-VII transform in the dimension;
in response to determining that the number of samples in one dimension is larger than 16, applying the DCT-II transform in the dimension.