Inter-Color Component Prediction for Image Encoding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image encoding/decoding methods do not effectively utilize the correlation between color components, leading to inefficiencies in encoding and decoding high-resolution images.
Innovation Solution
An image encoding apparatus and method that perform inter-color prediction and compensation by generating residual signals based on differences between luminance and chrominance components, and selectively applying these predictions based on predetermined conditions, such as transform unit size, to improve encoding efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional image encoding/decoding methods are used without inter-color prediction, then the encoding process is simpler, but image encoding/decoding efficiency deteriorates
Solution Approach 1:
The patent performs inter-color prediction in advance by generating residual signals between color components (e.g., luminance and chrominance) before the main encoding process. This preliminary action captures correlated data between color components, allowing the main encoding process to work with already-optimized data, thereby improving overall encoding efficiency without proportionally increasing complexity.
Solution Approach 2:
The patent selectively applies inter-color prediction based on predetermined conditions such as transform unit size. By changing the parameter of when and where to apply the prediction (based on TU size thresholds), the system optimizes the balance between encoding efficiency improvement and computational complexity, applying the complex operation only where it provides the most benefit.
2Productivity
If inter-color prediction is applied to all transform units, then image encoding efficiency improves, but computational complexity increases
Solution Approach 1:
The patent applies inter-color prediction selectively to specific transform units based on their size relative to a predetermined threshold. Rather than uniformly applying the prediction across all transform units, the system identifies local regions (transform units larger than the threshold) where inter-color correlation is most beneficial, applying the prediction operation only there. This local quality approach improves encoding efficiency in critical areas while conserving computational energy in less critical areas.
3Measurement precision
If residual signals are generated for all color components, then prediction accuracy improves, but data processing load increases
Solution Approach 1:
The patent extracts and processes only the essential residual signals needed for inter-color prediction. Specifically, it generates residual signals between color components (such as luminance and chrominance) and uses these extracted residual signals for prediction, rather than processing all possible color component combinations. This extraction approach maintains prediction accuracy by focusing on the most significant correlations while reducing the overall data processing load.
Data Source
AI summary
The present invention provides an image encoding apparatus carrying out inter-color prediction, comprising a residual block acquisition module obtaining a residual block with respect to a first color component and a residual block with respect to a second color component from a difference between an input block and a prediction block; an inter-color component prediction module carrying out inter-color component prediction by generating a residual signal reflecting a difference between a residual block with respect to the first color component and a residual block with respect to the second color component; a transform module generating a transformat coefficient by carrying out transformation with respect to the residual signal; a quantization module generating quantized data by carrying out quantization with respect to the transform coefficient; and an entropy encoding module carrying out entropy encoding by removing statistical redundancy of the quantized data.