Selective transform set in video coding
A video bitstream including coded information of a current block in a current picture is received. The coded information includes prediction samples of the current block. The prediction samples are obtained by a matrix multiplication of a matrix of coefficients and reference samples of the current block. A transform set is determined for the current block based on which of a subset of intra prediction modes is applied to the current block when one of the subset of the intra prediction modes is applied to the current block. The subset of intra prediction modes is derived based on the prediction samples of the current block. The current block is reconstructed based on the prediction samples and the transform set.
1 . A method of video decoding, comprising:
receiving a video bitstream including coded information of a current block in a current picture, the coded information including prediction samples of the current block, the prediction samples being obtained by a matrix multiplication of a matrix of coefficients and reference samples of the current block;
determining a transform set for the current block based on which of a subset of intra prediction modes is applied to the current block when one of the subset of the intra prediction modes is applied to the current block, the subset of intra prediction modes being derived based on the prediction samples of the current block; and
reconstructing the current block based on the prediction samples and the transform set.
2 . The method of claim 1 , wherein the determining the transform set further comprises:
determining the transform set from a look-up table that indicates a mapping between the subset of the intra prediction modes and a plurality of transform sets.
3 . The method of claim 1 , wherein:
the determining the transform set includes determining the transform set for the current block when the one of the subset of the intra prediction modes is applied to the current block and one of a size of the current block and a size of a template that includes the reference samples meets a size condition; and
the size condition includes one of:
the size of the current block is smaller than a pre-defined block size,
a ratio between a width of the current block and a height of the current block is smaller than a pre-defined ratio,
the width of the current block and the height of the current block is smaller than a pre-defined dimension, and
the size of the template is equal to a pre-defined template size.
4 . The method of claim 1 , wherein the subset of intra prediction modes includes one of a planar mode, a DC mode, a mode equal to (2+4×K) in which K is an integer from 0 to 16, and a mode equal to (2+2×L) in which L is an integer from 0 to 32.
5 . The method of claim 1 , wherein the transform set is one of a primary transform set and a secondary transform set.
6 . The method of claim 1 , wherein the transform set includes a set of transform kernels that is trained for the subset of the intra prediction modes.
7 . The method of claim 1 , wherein the determining the transform set for the current block further comprises:
determining an offset from a candidate list to an initial index to obtain an updated index, the initial index indicating an initial transform set, the candidate list including a plurality of candidate offsets obtained from offsets in an offset transition table; and
determining the transform set based on the updated index.
8 . The method of claim 7 , wherein the candidate list includes:
a first candidate offset that is applied to a previously reconstructed block of the current block,
a second candidate offset being linked to the first candidate offset according to a first edge in the offset transition table, and
a third candidate offset being linked to the first candidate offset according to a second edge in the offset transition table.
9 . The method of claim 7 , further comprising:
collecting a plurality of offsets from the offset transition table that is applied to previously reconstructed blocks of the current block; and
updating one of more offsets in the offset transition table based on one or more most frequent ones of the plurality of offsets.
10 . The method of claim 9 , wherein the previously reconstructed blocks are adjacent to the current block.
11 . The method of claim 9 , wherein the updating is performed when one CTU is coded.
12 . The method of claim 7 , wherein a total number of edges associated with each of the offsets in the offset transition table is same.
13 . The method of claim 7 , wherein a total number of edges associated with each of the offsets in the offset transition table are different.
14 . The method of claim 7 , wherein the offset transition table is defined based on an intra prediction mode associated with the current block.
15 . A method of video encoding, comprising:
deriving a subset of intra prediction modes based on prediction samples of a current block in a current picture, the prediction samples being obtained by a matrix multiplication of a matrix of coefficients and reference samples of the current block;
determining a transform set for the current block based on which of a subset of intra prediction modes is applied to the current block when one of the subset of the intra prediction modes is applied to the current block, the subset of intra prediction modes being derived based on the prediction samples of the current block; and
encoding the prediction samples of the current block into a bitstream based on the transform set.
16 . The method of claim 15 , wherein the determining the transform set further comprises:
determining the transform set from a look-up table that indicates a mapping between the subset of the intra prediction modes and a plurality of transform sets.
17 . The method of claim 15 , wherein:
the determining the transform set includes determining the transform set for the current block when the one of the subset of the intra prediction modes is applied to the current block and one of a size of the current block and a size of a template that includes the reference samples meets a size condition; and
the size condition includes one of:
the size of the current block is smaller than a pre-defined block size,
a ratio between a width of the current block and a height of the current block is smaller than a pre-defined ratio,
the width of the current block and the height of the current block is smaller than a pre-defined dimension, and
the size of the template is equal to a pre-defined template size.
18 . The method of claim 15 , wherein the subset of intra prediction modes includes one of a planar mode, a DC mode, a mode equal to (2+4×K) in which K is an integer from 0 to 16, and a mode equal to (2+2×L) in which L is an integer from 0 to 32.
19 . The method of claim 15 , wherein the transform set is one of a primary transform set and a secondary transform set.
20 . A method of processing visual media data, the method comprising:
processing a bitstream of the visual media data according to a format rule, wherein:
the bitstream includes coded information of a current block in a current picture, the coded information including prediction samples of the current block, the prediction samples being obtained by a matrix multiplication of a matrix of coefficients and reference samples of the current block; and
the format rule specifies that:
a transform set is determined for the current block based on which of a subset of intra prediction modes is applied to the current block when one of the subset of the intra prediction modes is applied to the current block, the subset of intra prediction modes being derived based on the prediction samples of the current block; and
the current block is processed based on the prediction samples and the transform set.