Intra Prediction Weighting for Complex Texture Video Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The accuracy of prediction in template-based intra mode derivation (TIMD) solutions for intra prediction in video coding standards like H.266/VVC needs improvement, particularly in handling blocks with complex textures.
Innovation Solution
Determine different weights for different units of a current block based on sub-templates, using first, second, and fourth weights to accurately calculate the intra prediction value, enhancing the compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If template-based intra mode derivation (TIMD) is used to determine intra prediction value, then the prediction effect is improved, but the accuracy of prediction effect estimation needs further improvement
Solution Approach 1:
The patent applies local quality by dividing the current block into multiple units (e.g., 4×4 sub-blocks) and determining different weights for prediction modes on different units. This allows the prediction accuracy to be optimized locally for each unit based on its specific characteristics, rather than using a uniform approach for the entire block. The weights are determined based on the correlation between sub-templates and corresponding units, enabling adaptive local optimization of prediction accuracy.
Solution Approach 2:
The patent segments the current block into multiple smaller units and processes each unit separately with its own weight determination. This segmentation allows for more granular control over prediction accuracy and enables the system to handle complex textures within different regions of the block differently, improving overall prediction reliability.
2Measurement precision
If different weights are set for different units of current block, then the intra prediction value accuracy is improved, but the computational complexity increases
Solution Approach 1:
The patent changes the parameter of weights from a uniform value to position-dependent values. The weight for each unit is determined by parameters such as the correlation between sub-templates and units, allowing the system to adaptively adjust weights based on local characteristics. This parameter change enables improved prediction accuracy while the weight determination process is integrated into the existing TIMD framework to manage complexity.
Solution Approach 2:
The patent performs preliminary determination of weights for each unit before finalizing the intra prediction value. By pre-calculating weights based on sub-template correlations and storing them for use in the prediction process, the system avoids repeated complex calculations during prediction, thus managing computational complexity while maintaining high accuracy.
3Measurement precision
If multiple prediction modes are evaluated with different weights, then the prediction accuracy is improved, but the processing time increases
Solution Approach 1:
The patent applies partial action by evaluating multiple prediction modes with different weights only for units where it provides significant benefit, rather than uniformly applying the full complexity to all blocks. The weight determination and multi-mode evaluation are selectively applied based on the specific characteristics of each unit, reducing unnecessary processing time while maintaining accuracy where it matters most.
Data Source
AI summary
An intra prediction method, an encoder, and a storage medium are provided. In the method, after determining the weights of at least two prediction modes on at least two units in a current block, the rate of change of a weight in a certain direction can be determined further according to the weights of the at least two units, and according to the rate of change, the weights on other units on the current block can then be determined by means of a smooth transition to determine an intra prediction value of the current block.