Matrix-Weighted Intra Prediction for Video Block Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in improving the accuracy and efficiency of intra prediction, particularly in determining optimal intra prediction modes and merge candidates, which affects the overall compression efficiency and quality of video encoding/decoding processes.
Innovation Solution
The proposed methods involve fusing at least two predictors through matrix-multiplication of weight coefficients and neighboring reconstructed samples, refining merge candidates via intra template matching, and employing a spectrum domain-based intra prediction mode to enhance the accuracy and efficiency of intra prediction in video coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple predictors are fused using matrix-multiplication of weight coefficients and neighboring reconstructed samples, then the accuracy of intra prediction is improved, but the computational complexity increases
Solution Approach 1:
The patent applies parameter changes by using matrix-multiplication of weight coefficients to transform the prediction process. Instead of simple arithmetic operations, the system uses matrix operations with learned weight parameters to combine multiple predictors, thereby improving prediction accuracy while managing computational complexity through efficient parameter utilization
Solution Approach 2:
The patent combines multiple predictors (angular mode predictor, planar mode predictor, and DC mode predictor) into a composite prediction system. These predictors are fused using matrix-multiplication operations, creating a composite prediction mechanism that leverages the strengths of each individual predictor while achieving superior overall performance
2Manufacturing precision
If pre-defined merge candidates are refined through intra template matching, then the selection of merge candidates is improved, but the processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-defining merge candidates before the actual prediction process. These pre-defined candidates serve as initial selections that are then refined through template matching. This preliminary preparation reduces the search space and guides the refinement process, improving selection accuracy while controlling processing time
Solution Approach 2:
The patent implements feedback through intra template matching, where the predicted block is compared with reference blocks to evaluate and refine merge candidates. This feedback mechanism allows the system to iteratively improve the selection of merge candidates based on actual prediction quality, ensuring optimal candidates are chosen
3Productivity
If a spectrum domain-based intra prediction mode is employed, then the compression efficiency is improved, but the algorithm complexity increases
Solution Approach 1:
The patent applies dimensionality change by transforming the prediction process into the spectrum domain. Instead of operating only in the spatial domain, the system uses spectral transforms to analyze and predict block data in the frequency domain. This dimensional transformation enables more effective compression by capturing patterns that are not visible in the spatial domain, thereby improving compression efficiency
Data Source
AI summary
A video bitstream including coded information of a current block in a current picture is received. The coded information indicates a plurality of candidate intra prediction modes for the current block. Two or more predictors are determined based on the plurality of candidate intra prediction modes for the current block according to a pre-defined condition. At least one of the two or more predictors is generated based on a matrix-multiplication mode of the plurality of candidate intra prediction modes such that the at least one of the two or more predictors is obtained by a matrix multiplication of a matrix of weight coefficients and neighboring reconstructed samples in a template of the current block. The current block is reconstructed based on a weighted combination of the two or more predictors.