Angular Table Optimization for Diagonal Texture Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing angular table in high efficiency video coding (HEVC) is generated in an ad hoc manner and is not optimized for video coding, particularly for diagonal textures, leading to decreased coding performance in video sequences with varied diagonal textures.
Innovation Solution
A new angular table is derived using a mathematical model, with entries determined via uniform probability distribution to optimize coding performance, allowing for improved coding priorities for diagonal modes and other directions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the existing angular table is used in HEVC intra prediction, then the implementation is simple, but the coding performance deteriorates for video sequences with diagonal textures
Solution Approach 1:
The patent changes the parameters of the angular table by deriving entries through mathematical optimization rather than using fixed ad hoc values. The angular table entries are optimized based on probability distribution models to maximize coding performance for diagonal textures while maintaining the same intra prediction mode structure, thus resolving the contradiction between implementation simplicity and coding accuracy.
2Productivity
If the angular table is optimized for diagonal textures, then coding gain improves, but the complexity of deriving the angular table increases
Solution Approach 1:
The patent applies preliminary action by pre-optimizing the angular table entries using mathematical models and probability distributions before actual video coding. The optimization is performed offline to derive the optimal angular table, which is then used directly in coding without adding runtime complexity, thus achieving coding gain without increasing operational complexity.
Solution Approach 2:
The patent substitutes the mechanical/ad hoc approach of angular table generation with a mathematical model-based system. Instead of manually or empirically setting angular table values, the patent uses probability distribution models and mathematical optimization to automatically derive the entries, replacing the mechanical process with a systematic mathematical approach that reduces derivation complexity.
3Measurement precision
If more directional intra prediction modes are used, then prediction accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent changes the parameters of the angular table to optimize prediction accuracy without adding more directional modes. By adjusting the angular table entries based on probability distributions, the existing set of directional modes achieves higher prediction accuracy, avoiding the need to increase the number of modes and thus preventing computational complexity from increasing.
Data Source
AI summary
A method of intra prediction comprising computing a prediction block for a luma block, wherein computing the prediction block includes using a plurality of directional intra prediction modes and an angular table (angTable) comprising a plurality of entries, and wherein a value of each entry is determined via a mathematical derivation. An apparatus comprising a memory configured to store an angTable comprising a plurality of entries, wherein a value of each entry is determined via a mathematical derivation, and a processor configured to compute a prediction block for a luma block using a plurality of directional intra prediction modes and the angTable.