Image Prediction Signal Smoothing via Correlation Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image predictive encoding technologies suffer from reduced encoding efficiency due to prediction signals being polluted by encoding strain, leading to increased code length and deteriorated image quality, and existing methods to mitigate this require additional information for smoothing candidate prediction signals.
Innovation Solution
An image predictive encoding device that searches for highly correlated prediction adjacent regions in a previously-reproduced image, derives combinations of these regions, processes their pixel signals, and selects candidate prediction signals based on correlation, generating a prediction signal without increasing the information amount, thereby smoothing the prediction signal effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional prediction signal generation methods are used, then encoding process is simple, but prediction signal quality deteriorates due to encoding strain pollution
Solution Approach 1:
The patent segments the prediction signal generation process into multiple candidate signal generations and evaluates their correlation with the target block separately. By dividing the prediction process into multiple correlated prediction signals and evaluating their individual correlations, the system can select or combine the most appropriate prediction signal, thereby improving prediction quality without significantly increasing overall system complexity.
2Reliability
If smoothing processing is applied to candidate prediction signals, then prediction signal quality improves, but information amount increases
Solution Approach 1:
The patent employs a correlation evaluation mechanism that automatically assesses each candidate prediction signal's correlation with the target block and selects or combines signals based on this evaluation. This self-evaluating system improves prediction quality through intelligent selection rather than blanket smoothing processing, avoiding the need to transmit additional smoothing parameters and thus preventing information increase.
3Measurement precision
If multiple candidate prediction signals are generated and evaluated, then prediction accuracy improves, but encoding complexity increases
Solution Approach 1:
The patent generates multiple candidate prediction signals but does not process all of them equally. Instead, it evaluates their correlations with the target block and selectively uses or combines only the most correlated signals. This partial processing approach achieves high prediction accuracy by focusing computational resources on the most relevant candidates rather than exhaustively processing all possible candidates.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An objective is to provide an image predictive encoding device, image predictive encoding method, image predictive encoding program, image predictive decoding device, image predictive decoding method, and image predictive decoding program for selecting a plurality of candidate prediction signals, without increase in information amount. A weighting unit 234 and an adder 235 process pixel signals extracted by a prediction adjacent region extractor 232 by a predetermined synthesis method, for example, by averaging to generate a comparison signal to an adjacent pixel signal for each combination. A comparison-selection unit 236 selects a combination with a high correlation between the comparison signal generated by the weighting unit 234 and others and the adjacent pixel signal acquired by a target adjacent region extractor 233. A prediction region extractor 204, weighting unit 205, and adder 206 generate candidate prediction signals and process them by a predetermined synthesis method to generate a prediction signal.