Method for coding image on basis of secondary transform and device therefor
An image decoding method according to the present document comprises a step of deriving modified transform coefficients on the basis of an inverse reduced secondary transform (RST) regarding transform coefficients regarding a target block. The inverse RST is performed on the basis of transform sets determined on the basis of a mapping relation following an intra prediction mode applied to the target block, and a transform kernel matrix selected from two transform kernel matrices included in the transform sets, respectively, and is performed on the basis of whether or not the inverse RST is applied, and a transform index indicating one of the transform kernel matrices included in the transform sets. A syntax element bin string regarding the transform index is decoded on the basis of a different context set according to the tree type regarding the division structure of the target block.
1. An image decoding method performed by a decoding apparatus, the method comprising:
receiving residual information, a transform index, and information on an intra prediction mode from a bitstream;
deriving transform coefficients for a target block based on the residual information; and
deriving modified transform coefficients based on an inverse reduced secondary transform (RST) for the transform coefficients,
wherein the inverse RST is performed based on a transform kernel matrix,
wherein the transform kernel matrix is determined based on a transform index and a transform set, and
wherein the transform index is decoded based on a context adaptive binary arithmetic coding (CABAC) and by:
deriving, based on context sets, context information of a syntax element for the transform index, the context sets including a first context set and a second context set,
decoding, based on the context information, bins of the syntax element for the transform index, the bins including a first bin and a second bin, and
deriving, based on the decoded bins, a value of the syntax element for the transform index,
wherein (i) based on a tree type of the target block being a dual tree type, the context information is derived based on the first context set, and (ii) based on the tree type of the target block being a single tree type, the context information is derived based on the second context set,
wherein the first context set comprises (i) first context information for the first bin of the syntax element for the transform index and (ii) second context information for the second bin of the syntax element for the transform index, and
wherein the second context set comprises (i) third context information for the first bin of the syntax element for the transform index and (ii) fourth context information for the second bin of the syntax element for the transform index.
2. The image decoding method of claim 1 , wherein the first context information and the third context information are different from each other.
3. The image decoding method of claim 1 , wherein the value of the syntax element comprises any one of 0 indicating that the inverse RST is not applied to the target block, 1 indicating a first transform kernel matrix as the transform kernel matrix, and 2 indicating a second transform kernel matrix as the transform kernel matrix.
4. The image decoding method of claim 3 , wherein the value of the syntax element is binarized by a truncated unary code, and
wherein the value of the syntax element of 0 is binarized into ‘0’, the value of the syntax element of 1 is binarized into ‘10’, and the value of the syntax element of 2 is binarized into ‘11’.
5. An image encoding method performed by an image encoding apparatus, the method comprising:
deriving residual samples for a target block;
deriving transform coefficients for the target block based on a primary transform for the residual samples; and
deriving modified transform coefficients based on a reduced secondary transform (RST) of the transform coefficients,
wherein the RST is performed based on a transform kernel matrix,
wherein the transform kernel matrix is determined based on a transform index and a transform set, and
wherein the transform index is generated based on a CABAC and by:
deriving a value of a syntax element for the transform index,
deriving, based on context sets, context information of the syntax element for the transform index, the context sets including a first context set and a second context set, and
encoding, based on the context information, bins of the syntax element for the transform index, the bins including a first bin and a second bin,
wherein (i) based on a tree type of the target block being a dual tree type, the context information is derived based on the first context set, and (ii) based on the tree type of the target block being a single tree type, the context information is derived based on the second context set,
wherein the first context set comprises (i) first context information for the first bin of the syntax element for the transform index and (ii) second context information for the second bin of the syntax element for the transform index, and
wherein the second context set comprises (i) third context information for the first bin of the syntax element for the transform index and (ii) fourth context information for the second bin of the syntax element for the transform index.
6. The image encoding method of claim 5 , wherein
the first context information and the third context information are different from each other.
7. A non-transitory computer-readable storage medium storing encoded information generated by an encoding method, the method comprising:
deriving residual samples for a target block;
deriving transform coefficients for the target block based on a primary transform for the residual samples; and
deriving modified transform coefficients based on a reduced secondary transform (RST) of the transform coefficients,
wherein the RST is performed based on a transform kernel matrix,
wherein the transform kernel matrix is determined based on a transform index and a transform set, and
wherein the transform index is generated based on a CABAC and by:
deriving a value of a syntax element for the transform index,
deriving, based on context sets, context information of the syntax element for the transform index, the context sets including a first context set and a second context set, and
encoding, based on the context information, bins of the syntax element for the transform index, the bins including a first bin and a second bin,
wherein (i) based on a tree type of the target block being a dual tree type, the context information is derived based on the first context set, and (ii) based on the tree type of the target block being a single tree type, the context information is derived based on the second context set,
wherein the first context set comprises (i) first context information for the first bin of the syntax element for the transform index and (ii) second context information for the second bin of the syntax element for the transform index, and
wherein the second context set comprises (i) third context information for the first bin of the syntax element for the transform index and (ii) fourth context information for the second bin of the syntax element for the transform index.
8. The image decoding method of claim 1 , wherein the inverse RST is an inverse transform of a transform that derives a smaller number of transform coefficients than transform coefficients for a predetermined top-left area of the target block based on the transform coefficients for the top-left area.
9. The image encoding method of claim 5 , wherein the RST is a transform that derives a smaller number of transform coefficients than transform coefficients for a predetermined top-left area of the target block based on the transform coefficients for the top-left area.
10. The image decoding method of claim 2 , wherein the second context information is the same as the fourth context information.
11. The image encoding method of claim 6 , wherein the second context information is the same as the fourth context information.
12. The image encoding method of claim 5 , wherein the value of the syntax element comprises any one of 0 indicating that the inverse RST is not applied to the target block, 1 indicating a first transform kernel matrix as the transform kernel matrix, and 2 indicating a second transform kernel matrix as the transform kernel matrix.
13. The image encoding method of claim 5 , wherein the value of the syntax element is binarized by a truncated unary code, and
wherein the value of the syntax element of 0 is binarized into ‘0’, the value of the syntax element of 1 is binarized into ‘10’, and the value of the syntax element of 2 is binarized into ‘11’.
14. The non-transitory computer-readable storage medium of claim 7 , wherein the first context information and the third context information are different from each other.
15. The non-transitory computer-readable storage medium of claim 14 , wherein the second context information is the same as the fourth context information.
16. The non-transitory computer-readable storage medium of claim 7 , wherein the value of the syntax element comprises any one of 0 indicating that the inverse RST is not applied to the target block, 1 indicating a first transform kernel matrix as the transform kernel matrix, and 2 indicating a second transform kernel matrix as the transform kernel matrix.
17. The non-transitory computer-readable storage medium of claim 7 , wherein the value of the syntax element is binarized by a truncated unary code, and
wherein the value of the syntax element of 0 is binarized into ‘0’, the value of the syntax element of 1 is binarized into ‘10’, and the value of the syntax element of 2 is binarized into ‘11’.