Non-separable transform for inter-coded blocks
Systems and techniques are described for processing video data. For example, an apparatus can obtain a block of video data encoded using inter-prediction. In some examples, the apparatus can determine a direction associated with the block of video data, for instance based on pixels of the block of video data, based on pixels of at least one neighboring block of the block of video data, or based on information associated with a geometric partitioning mode (GPM) associated with the block of video data. The apparatus can apply a non-separable transform to the block of video data, for instance to decode or encode the video data. In some examples, the apparatus can apply the non-separable transform to the block according to the direction associated with the block of video data.
1 . An apparatus for processing video data, the apparatus comprising:
at least one memory; and
at least one processor coupled to the at least one memory, the at least one processor being configured to:
obtain a prediction block corresponding to a block of video data encoded using inter-prediction;
analyze one or more pixels of the prediction block to identify a prediction block direction associated with the prediction block;
approximate a block direction associated with the block of video data based on the prediction block direction associated with the prediction block; and
apply an inverse low frequency non-separable transform (LFNST) transform to residual coefficients of the block of video data according to the block direction associated with the block of video data.
2 . The apparatus of claim 1 , wherein a kernel associated with the inverse LFNST transform is dependent on the block direction associated with the block of video data.
3 . The apparatus of claim 1 , wherein the block direction associated with the block of video data is equivalent to the prediction block direction.
4 . The apparatus of claim 1 , wherein, to analyze the one or more pixels of the prediction block to identify the prediction block direction, the at least one processor is configured to:
determine at least one gradient of the one or more pixels of the prediction block; and
determine a dominant direction associated with the at least one gradient.
5 . The apparatus of claim 1 , wherein the prediction block direction is a direction of an edge in the prediction block.
6 . The apparatus of claim 1 , wherein the at least one processor is configured to:
analyze neighboring block pixels of at least one neighboring block adjacent to the block of video data, wherein at least one of the prediction block direction or the block direction is based on the neighboring block pixels.
7 . The apparatus of claim 6 , wherein the at least one neighboring block includes a first neighboring block above the block and a second neighboring block on a left side of the block.
8 . The apparatus of claim 6 , wherein, to analyze the neighboring block pixels of the at least one neighboring block, the at least one processor is configured to:
determine a gradient of the neighboring block pixels included in the at least one neighboring block; and
determine a direction associated with the gradient.
9 . The apparatus of claim 6 , wherein the at least one neighboring block includes an intra-coded block.
10 . The apparatus of claim 1 , wherein at least one of the prediction block direction or the block direction is based on information associated with a geometric partitioning mode (GPM) associated with the block.
11 . The apparatus of claim 1 , wherein the at least one processor is configured to:
determine a direction of a partition edge, wherein the block of video data is partitioned along the partition edge according to a geometric partitioning mode (GPM) associated with the block, and wherein at least one of the prediction block direction or the block direction is based on the direction of the partition edge.
12 . The apparatus of claim 1 , wherein the at least one processor is configured to:
apply the inverse LFNST transform directly to one or more block pixels of the block of video data.
13 . The apparatus of claim 1 , wherein the at least one processor is configured to:
apply a separable transform to one or more block pixels of the block of video data as a primary transform, wherein the inverse LFNST transform is a secondary transform.
14 . The apparatus of claim 1 , wherein the at least one processor is configured to:
apply a separable transform to one or more block pixels of the block of video data as a secondary transform, wherein the inverse LFNST transform is a primary transform.
15 . The apparatus of claim 1 , wherein the at least one processor is configured to:
determine that a size of the block of video data is greater than a size threshold, wherein application of the inverse LFNST transform to the residual coefficients of the block of video data is based on determination that the size of the block of video data is greater than the size threshold.
16 . The apparatus of claim 1 , wherein the at least one processor is configured to:
transmit one or more non-separable transform candidates for the block of video data.
17 . The apparatus of claim 1 , wherein the block of video data includes a residual block of video data.
18 . The apparatus of claim 1 , wherein the at least one processor is configured to:
apply a separable transform to the block of video data to generate the residual coefficients before applying the inverse LFNST transform to the residual coefficients of the block of video data.
19 . The apparatus of claim 18 , wherein the separable transform is a discrete cosine transform (DCT).
20 . The apparatus of claim 1 , wherein the at least one processor is configured to:
apply a separable transform to the block of video data after applying the inverse LFNST transform to the residual coefficients of the block of video data.
21 . The apparatus of claim 20 , wherein the separable transform is an inverse discrete cosine transform (DCT).
22 . The apparatus of claim 1 , wherein the at least one processor is configured to:
apply a LFNST transform to residual coefficients of the block of video data according to the block direction associated with the block of video data.
23 . The apparatus of claim 1 , wherein the at least one processor is configured to:
decode the video data using reconstructed residuals generated from the inverse LFNST transform applied to the residual coefficients of the block of video data.
24 . The apparatus of claim 1 , wherein the at least one processor is configured to:
encode the video data using residuals generated from the inverse LFNST transform applied to the residual coefficients of the block of video data.
25 . The apparatus of claim 1 , wherein applying the inverse LFNST transform to the residual coefficients of the block of video data is configured to generate reconstructed residuals associated with the block of video data.
26 . A method of processing video data, the method comprising:
obtaining a prediction block corresponding to a block of video data encoded using inter-prediction;
analyzing one or more pixels of the prediction block to identify a prediction block direction associated with the prediction block;
approximating a block direction associated with the block of video data based on the prediction block direction associated with the prediction block; and
applying an inverse low frequency non-separable transform (LFNST) transform to residual coefficients of the block of video data according to the block direction associated with the block of video data.
27 . The method of claim 26 , wherein the block direction associated with the block of video data is equivalent to the prediction block direction.
28 . The method of claim 26 , wherein at least one of the prediction block direction or the block direction is based on neighboring block pixels of at least one neighboring block adjacent to the block of video data.
29 . The method of claim 26 , wherein at least one of the prediction block direction or the block direction is based on information associated with a geometric partitioning mode (GPM) associated with the block.
30 . The method of claim 26 , wherein analyzing the one or more pixels of the prediction block to identify the prediction block direction includes:
determining at least one gradient of the one or more pixels of the prediction block; and
determining a dominant direction associated with the at least one gradient.