Matrix-Based Video Prediction for High-Resolution Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression technologies face inefficiencies in predicting video signals, particularly in high-resolution images, as they lack effective methods for intra and inter prediction that can adapt to varying block sizes and shapes.
Innovation Solution
The method involves matrix-based prediction, where a prediction block is generated through matrix-based prediction by determining a reference region, applying a downsampling filter, and interpolating samples based on neighboring samples and matrix values, allowing for variable block properties and interpolation directions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional intra prediction methods are used, then encoding complexity is kept simple, but prediction accuracy deteriorates for high-resolution videos with varying block sizes and shapes
Solution Approach 1:
The patent applies parameter changes by using matrix-based prediction with variable matrix values and downsampling filter configurations adapted to different block properties (size, shape, position, component type). This allows the prediction method to dynamically adjust parameters to match the specific characteristics of each block, thereby improving prediction accuracy for high-resolution videos without requiring a complete overhaul of the encoding framework
Solution Approach 2:
The patent segments the prediction process into distinct stages: determining reference regions, applying downsampling filters to generate input vectors, performing matrix-based prediction, and conducting interpolation. This segmentation allows each stage to be optimized independently, improving overall prediction accuracy while managing encoding complexity through modular processing
2Adaptability or versatility
If fixed-size blocks are used for prediction, then processing is simpler, but adaptability to different block properties deteriorates
Solution Approach 1:
The patent implements dynamics by making the prediction process adaptive to varying block properties. The matrix values, downsampling filter configurations, and reference region selections are dynamically determined based on block size, shape, position, and component type. This dynamic adaptation enables the system to handle diverse block configurations effectively while maintaining structured processing through defined determination rules
3Measurement precision
If more reference samples are used for prediction, then prediction accuracy improves, but computational complexity increases
Solution Approach 1:
The patent extracts only the essential reference information needed for accurate prediction by applying downsampling filters to generate compact input vectors. Instead of using all available reference samples directly, the method extracts key features and represents them in a compressed vector form, thereby maintaining prediction accuracy while significantly reducing the number of computations required for matrix operations
Data Source
AI summary
An image encoding/decoding method and device according to the present invention may comprise: acquiring information for matrix-based prediction of a current block; and generating a prediction block of the current block on the basis of the information for the matrix-based prediction.