Cross-Component Intra Prediction for Higher-Accuracy Image Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face challenges in achieving high precision intra-frame prediction due to the limitations of using manually designed filters with a small quantity of reconstructed neighboring pixels, leading to low image frame prediction accuracy.
Innovation Solution
Implement cross-component prediction using a refined model based on the mapping relationship between different components of reconstructed pixels, constructing a cross-component prediction model to enhance prediction accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a manually designed filter with a small quantity of reconstructed neighboring pixels is used for intra-frame prediction, then the device complexity is reduced, but the prediction accuracy deteriorates
Solution Approach 1:
The patent changes the parameters of the prediction model from simple manual filters to data-driven cross-component prediction models with multiple parameters (offsets, gradients, curvature terms) that adapt to local image characteristics, thereby improving prediction accuracy while maintaining manageable complexity through systematic parameter estimation
Solution Approach 2:
The patent introduces an intermediate cross-component prediction model that maps relationships between different color components (e.g., YCbCr) to enhance prediction accuracy. This intermediary model uses reconstructed pixels from one component to predict another component, adding predictive power without directly increasing filter complexity
2Productivity
If a manually designed simple filter is used for intra-frame prediction, then the coding efficiency is improved, but the prediction quality deteriorates
Solution Approach 1:
The patent employs parameter changes by transitioning from fixed simple filters to adaptive cross-component prediction models with multiple tunable parameters (offsets, gradients, curvature coefficients) that are estimated from local image data, significantly improving prediction quality while maintaining coding efficiency through compact parameter representation
Solution Approach 2:
The patent segments the prediction process into multiple stages: basic prediction using reconstructed pixels, cross-component mapping, gradient calculation, and curvature adjustment. This segmentation allows each stage to contribute specifically to prediction quality while keeping individual operations computationally efficient
3Speed
If only a small quantity of reconstructed neighboring pixels are used for prediction, then the processing speed is improved, but the prediction accuracy deteriorates
Solution Approach 1:
The patent changes the approach from using more pixels to using pixels more effectively by extracting multiple parameters (offsets, gradients, curvature terms) from a small set of reconstructed neighboring pixels. This parameter extraction methodology maximizes the information content from limited pixels, improving prediction accuracy without sacrificing processing speed
Solution Approach 2:
The patent replaces the mechanical approach of simply averaging or filtering neighboring pixels with a data-driven cross-component prediction model that uses mathematical relationships between color components. This substitution transforms the prediction process from a spatial averaging operation to a component-mapping operation, achieving higher accuracy with the same pixel count
Data Source
AI summary
This application provide an image processing method performed by a computer device. The method includes: determining a current coding block in an image bitstream, the current coding block comprising a first component and a second component; obtaining a cross-component prediction model, the cross-component prediction model indicating a mapping relationship between the first component of the current coding block and the second component of the current coding block; performing cross-component prediction on the current coding block based on the mapping relationship by inputting a reconstructed value of the first component of the current coding block to the cross-component prediction model to obtain a predicted value of the second component of the current coding block; and reconstructing the current coding block using the reconstructed value of the first component and the predicted value of the second component of the current coding block.