Image Prediction Using Adaptive Attenuation for Texture Mismatch
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image prediction methods in video coding struggle with achieving good prediction accuracy when the horizontal and vertical texture features of a block are significantly different, as they often use the same attenuation rate factors for horizontal and vertical weighting coefficients, leading to suboptimal performance.
Innovation Solution
The proposed method involves performing intra-frame prediction on a current block using a reference block and applying weighted filtering with different attenuation rate factors for horizontal and vertical weighting coefficients, which are adjusted based on the difference between the horizontal and vertical texture features of the block.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the same attenuation rate factor is used for horizontal and vertical weighting coefficients, then the prediction method is simple and efficient, but the prediction accuracy deteriorates when horizontal and vertical texture features are significantly different
Solution Approach 1:
The patent applies dynamics by making the attenuation rate factors adaptive rather than fixed. The system dynamically selects different attenuation rate factors for horizontal and vertical weighting coefficients based on the texture feature differences detected in the current block, allowing the prediction method to adapt to varying image characteristics while maintaining good prediction accuracy across different scenarios
Solution Approach 2:
The patent changes the parameter of attenuation rate factors from being identical for both directions to being different based on texture feature analysis. By modifying this key parameter according to the detected differences between horizontal and vertical texture features, the system optimizes prediction accuracy without excessive complexity
2Measurement precision
If different attenuation rate factors are used for horizontal and vertical weighting coefficients, then the prediction accuracy improves for blocks with different texture features, but the complexity of determining and controlling attenuation rates increases
Solution Approach 1:
The patent applies preliminary action by detecting and determining the texture feature differences before finalizing the prediction process. The system analyzes the current block's texture characteristics in advance and uses this information to select appropriate attenuation rate factors, ensuring optimal prediction accuracy is achieved through preparatory analysis
Solution Approach 2:
The patent implements feedback by using the detected texture feature differences to guide the selection of attenuation rate factors. The system continuously monitors the texture characteristics and adjusts the attenuation rates accordingly, creating a closed-loop process that optimizes prediction based on actual block properties
Data Source
AI summary
An image prediction method and a related product, where the image prediction method includes performing intra-frame prediction on a current block using a reference block to obtain an initial predicted pixel value of a pixel in the current block, and performing weighted filtering on the initial predicted pixel value of the pixel in the current block to obtain a predicted pixel value of the pixel in the current block. Weighting coefficients used for the weighted filtering include a horizontal weighting coefficient and a vertical weighting coefficient, and a first attenuation rate factor acting on the horizontal weighting coefficient is different from a second attenuation rate factor acting on the vertical weighting coefficient.