Matrix-Based Intra Prediction for Asymmetric Binary Tree Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding methods struggle with inefficient intra prediction for blocks with widths and heights that are not powers of 2, limiting the effectiveness of matrix-based intra prediction tools like MIP in VVC.
Innovation Solution
Adapt the MIP process to handle blocks with non-power-of-2 dimensions by downsampling and upsampling reference samples using weight matrices, allowing for intra prediction even for asymmetric binary tree (ABT) block sizes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing video encoding methods are used for blocks with non-power-of-2 dimensions, then the encoding process is simple, but the intra prediction efficiency is poor
Solution Approach 1:
The patent segments the prediction process into distinct stages: downsampling reference samples to a reduced size, performing matrix-based prediction on the reduced block, and then upsampling the predicted block to the original size. This segmentation allows efficient handling of non-power-of-2 blocks by breaking down the complex operation into manageable steps.
Solution Approach 2:
The patent changes the parameters of the reference samples by downsampling them to a reduced size before prediction, and then changes the parameters of the predicted block by upsampling to the original size. This parameter transformation enables the use of standard matrix-based prediction tools on blocks of any size, not just power-of-2 dimensions.
2Productivity
If matrix-based intra prediction is applied to non-power-of-2 blocks, then coding efficiency is improved, but additional processing steps are required
Solution Approach 1:
The patent performs preliminary downsampling of reference samples before applying matrix-based intra prediction. This preliminary action prepares the data in a format suitable for standard prediction tools, enabling efficient compression while maintaining compatibility with existing prediction algorithms.
Solution Approach 2:
The patent introduces an intermediate reduced-size predicted block as a mediator between the reference samples and the final output block. This intermediate representation allows the use of fixed-size matrix operations on variable-size blocks, bridging the gap between different block dimensions.
3Measurement precision
If standard intra prediction is used for ABT block sizes, then the process is straightforward, but prediction accuracy is limited
Solution Approach 1:
The patent applies parameter changes by downsampling reference samples to match the aspect ratio of ABT blocks, performing matrix-based prediction with transformed parameters, and then upsampling to restore the original dimensions. This parameter transformation enables accurate prediction for asymmetric block sizes that would be difficult to handle with standard prediction methods.
Data Source
AI summary
Several methods are described to jointly use the ABT (Asymmetric Binary Tree) partitioning mode and Matrix-based Intra Prediction (MIP). In a first embodiment, we propose to forbid the use of the MIP intra prediction mode, for block sizes that are resulted from ABT partitioning. In a second embodiment, we propose to allow the MIP intra prediction for block sizes not equal to a power of two in width or height, by extending the block before MIP and crop the predicted block to the original size after MIP. In a third embodiment, we propose to adapt the down-sampling of the boundary reference samples and the up-sampling of the reduced predicted blocks, to the block sizes introduced by ABT partitioning. In a further embodiment, we set the reduced predicted block to size 8×8 in any case the initial block size is 8 and larger than 8 in a direction.