Transform-based image coding method and apparatus therefor
An image decoding method according to the present document comprises the steps of: deriving a transform coefficient for a current block on the basis of residual information; deriving a residual sample by applying, to the transform coefficient, at least one of LFNST and MTS; and generating a reconstructed picture on the basis of the residual samples, wherein the LFNST is performed on the basis of an LFNST index indicating an LFNST kernel, the MTS is performed on the basis of an MTS index indicating an MTS kernel, and the LFNST index may be parsed before the MTS index.
1. An image decoding method performed by a decoding apparatus, the method comprising:
obtaining prediction related information for a current block from a bitstream;
deriving prediction samples for a current block based on the prediction related information;
obtaining residual information from the bitstream;
deriving transform coefficients for the current block based on the residual information;
deriving residual samples by applying at least one of a low frequency non-separable transform (LFNST) kernel derived from an LFNST index or a primary transform kernel derived from a multiple transform selection (MTS) index to the transform coefficients; and
generating a reconstructed picture based on the prediction samples and the residual samples,
wherein the LFNST index is parsed before the MTS index, and
wherein the MTS index is signaled based on a value of the LFNST index.
2. The image decoding method of claim 1 , wherein when the current block is a luma block and the LFNST index is 0, the MTS index is parsed.
3. The image decoding method of claim 1 , wherein when a tree type of the current block is a dual tree, the LFNST index for each of a luma block and a chroma block is parsed.
4. The image decoding method of claim 1 , wherein the deriving the transform coefficient comprises deriving a width and a height for a top-left region in which a last significant coefficient may exist in the current block by zero-out, and
wherein the width and height for the top-left region are derived before parsing the MTS index.
5. The image decoding method of claim 1 , wherein the last significant coefficient position is derived based on the width and the height for the top-left region,
wherein last significant coefficient position information is binarized based on the width and the height for the top-left region.
6. An image encoding method performed by an encoding apparatus, the method comprising:
deriving prediction samples for a current block;
deriving residual samples for the current block based on the prediction sample;
deriving transform coefficients by applying at least one of a low frequency non-separable transform (LFNST) kernel or a primary transform kernel to the residual samples; and
encoding i) prediction related information, ii) residual information related to the transform coefficients and iii) at least one of an LFNST index related to the LFNST kernel and an MTS index related to the primary transform kernel,
wherein the LFNST index is encoded before the MTS index, and
wherein the MTS index is encoded based on a value of the LFNST index.
7. The image encoding method of claim 6 , wherein when the current block is a luma block and the LFNST index is 0, the MTS index is encoded.
8. The image encoding method of claim 6 , wherein when a tree type of the current block is a dual tree, the LFNST index for each of a luma block and a chroma block is encoded.
9. The image encoding method of claim 6 , wherein the deriving the transform coefficient comprises deriving a width and a height for a top-left region in which a last significant coefficient may exist in the current block by zero-out, and
wherein the width and the height for the top-left region are derived before signaling the MTS index.
10. The image encoding method of claim 9 , wherein the last significant coefficient position is derived based on the width and the height for the top-left region, and
wherein the last significant coefficient position information is binarized based on the width and the height of the top-left region.
11. A transmitting method comprising obtaining a bitstream, wherein the bitstream is generated by deriving prediction samples for a current block; deriving residual samples for the current block based on the prediction sample; deriving transform coefficients by applying at least one of a low frequency non-separable transform (LFNST) kernel or a primary transform kernel to the residual samples; and encoding i) prediction related information, ii) residual information related to the transform coefficients and iii) at least one of an LFNST index related to the LFNST kernel and an MTS index related to the primary transform kernel to generate the bitstream,
wherein the LFNST index is encoded before the MTS index, and
wherein the MTS index is encoded based on a value of the LFNST index.