Template matching prediction for versatile video coding
Template matching prediction is combined with additional coding tools to provide a robust set of coding/decoding tools by advantageous use of syntax elements. Some of the additional coding tools comprise matrix intra prediction, intra sub-partitioning, low frequency non-separable transform, multiple transform selection.
1 . A method, comprising:
signaling that at least a first and a second coding tool are used on a video block; and
encoding the video block using said at least first and second coding tool on said video block wherein template matching prediction is used on a luma component of said video block and one or more LFNST transform kernels corresponding to planar modes are used on a chroma component of said video block, and transform used is DCT-II, disallowing multiple transform selection with template matching prediction.
2 . The method of claim 1 , wherein a general constraint flag is used for TMP to indicate deactivation.
3 . The method of claim 1 , wherein TMP is used where LFNST transform kernels corresponding to planar modes are used.
4 . An apparatus, comprising:
a processor, configured to:
signal that at least a first and a second coding tool are used on a video block; and
encode the video block using said at least first and second coding tool on said video block wherein template matching prediction is used on a luma component of said video block and one or more LFNST transform kernels corresponding to planar modes are used on a chroma component of said video block, and transform used is DCT-II, disallowing multiple transform selection with template matching prediction.
5 . The apparatus of claim 4 , wherein TMP is used for luma and chroma components, and syntax is used to signal use of TMP for both luma and chroma components.
6 . The apparatus of claim 4 , wherein template matching is used on luma components with a same position of matching block used for chroma components.
7 . The apparatus of claim 4 , wherein
prediction candidates from inside and outside a coding tree unit are allowed.
8 . The apparatus of claim 4 , wherein an SIMD scheme is used for computation of template differences.
9 . A method, comprising:
parsing a video bitstream to determine at least a first and a second decoding tool for a video block; and
decoding the video block using said at least first and second coding tool on said video block wherein template matching prediction is used on a luma component of said video block and one or more LFNST transform kernels corresponding to planar modes are used on a chroma component of said video block, and transform used is DCT-II, disallowing multiple transform selection with template matching prediction.
10 . The method of claim 9 , wherein a general constraint flag is used for TMP to indicate deactivation.
11 . The method of claim 9 , wherein TMP is used where LFNST transform kernels corresponding to planar modes are used.
12 . An apparatus, comprising:
a processor, configured to:
parse a video bitstream to determine at least a first and a second decoding tool for a video block; and
decode the video block using said at least first and second coding tool on said video block wherein template matching prediction is used on a luma component of said video block and one or more LFNST transform kernels corresponding to planar modes are used on a chroma component of said video block, and transform used is DCT-II, disallowing multiple transform selection with template matching prediction.
13 . The apparatus of claim 12 , wherein TMP is used for luma and chroma components, and syntax is used to signal use of TMP for both luma and chroma components.
14 . The apparatus of claim 12 , wherein template matching is used on luma components with a same position of matching block used for chroma components.
15 . The apparatus of claim 12 , wherein
prediction candidates from inside and outside a coding tree unit are allowed.
16 . The apparatus of claim 12 wherein an SIMD scheme is used for computation of template differences.