Weighted Image Block Prediction for Temporal-Distance Motion Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional moving picture coding systems using B pictures face inefficiencies due to approximated motion vectors and averaging of motion-compensated blocks without considering temporal distances between reference pictures, leading to reduced accuracy and coding efficiency, especially in sequences with fading scenes.
Innovation Solution
A block prediction method that calculates forward and backward motion vectors based on co-located blocks in reference pictures, applying variable weight factors to these vectors to account for temporal distances and display order information, enabling interpolative prediction to improve accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the conventional block prediction method uses approximated motion vectors from co-located blocks in backward reference pictures, then the coding complexity is reduced, but the prediction accuracy deteriorates
Solution Approach 1:
The patent applies dynamics by making the prediction method adaptive rather than static. It dynamically selects between direct mode and overlay mode based on temporal distance thresholds, and within overlay mode, dynamically adjusts the blending ratio parameter alpha based on the temporal distances to forward and backward reference pictures. This dynamic adaptation resolves the contradiction by optimizing prediction accuracy for different scene conditions while maintaining computational efficiency through threshold-based mode selection.
Solution Approach 2:
The patent changes parameters to resolve the contradiction by introducing a blending ratio parameter alpha that varies based on temporal distances. When overlay mode is selected, alpha is calculated as a function of the temporal distances to forward and backward reference pictures, allowing the system to adjust the weight of each reference picture's contribution. This parameter change enables accurate motion compensation without requiring complex motion vector calculations, thus improving prediction accuracy while maintaining coding efficiency.
2Device complexity
If the conventional method averages two motion-compensated blocks without considering temporal distance, then the processing is simplified, but the prediction accuracy for fading scenes deteriorates
Solution Approach 1:
The patent resolves this contradiction by making the prediction process dynamic through threshold-based mode selection and adaptive parameter adjustment. Instead of always using simple averaging, the system dynamically determines whether to use direct mode or overlay mode based on temporal distance thresholds. Within overlay mode, it dynamically adjusts the blending ratio to account for fading scenes, where one reference picture may be significantly brighter or darker than the other. This dynamic approach maintains processing simplicity while dramatically improving prediction accuracy for fading scenes.
Solution Approach 2:
The patent changes the blending ratio parameter alpha based on temporal distances to resolve the contradiction. Instead of using a fixed 50-50 average, the system calculates alpha as a function of the temporal distances to forward and backward reference pictures. This parameter change allows the system to give more weight to the temporally closer reference picture, accurately handling fading scenes where brightness changes occur, while maintaining processing simplicity through a closed-form calculation rather than complex optimization.
3Measurement precision
If the method uses overlay mode with temporal distance consideration, then the prediction accuracy is improved, but the computational complexity increases
Solution Approach 1:
The patent resolves this contradiction by implementing a dynamic threshold-based mode selection mechanism. The system calculates temporal distances and compares them against predefined thresholds to determine whether to use computationally simple direct mode or more accurate overlay mode. This dynamic selection ensures that the computationally intensive overlay mode is only used when necessary (when temporal distances indicate potential prediction errors), thereby improving prediction accuracy while controlling computational complexity through intelligent mode switching.
Solution Approach 2:
The patent changes the blending ratio parameter alpha based on temporal distances to resolve the contradiction between accuracy and complexity. Instead of using complex motion vector optimization, the system uses a closed-form calculation of alpha as a function of temporal distances. This parameter change provides high prediction accuracy with minimal computational overhead, as the blending ratio is directly calculated from temporal distance measurements rather than requiring iterative optimization or complex processing.
4Speed
If the conventional method does not consider temporal distance between reference pictures, then the processing is faster, but the coding efficiency deteriorates
Solution Approach 1:
The patent resolves this contradiction by implementing dynamic threshold-based mode selection that adapts to temporal distance conditions. When temporal distances are small and conditions are favorable, the system uses fast direct mode processing. When temporal distances indicate potential prediction errors, it switches to overlay mode with temporal distance consideration. This dynamic adaptation maintains high processing speed for simple cases while improving coding efficiency for challenging cases, achieving an optimal balance between speed and efficiency.
Solution Approach 2:
The patent changes the blending ratio parameter alpha based on temporal distances to resolve the contradiction between processing speed and coding efficiency. The parameter alpha is calculated as a simple function of temporal distances, requiring minimal computation while significantly improving prediction accuracy and coding efficiency. This parameter change enables the system to achieve high coding efficiency without substantial increases in processing time, as the calculation involves only basic arithmetic operations on temporal distance measurements.
Data Source
AI summary
In the method for predicting an image, a reference image block for a current image block of a moving picture is obtained using motion vector information, and a variable weight factor is applied to the reference image block to form a product value of the weight factor and the reference image block. The current image block of the moving picture is predicted using the product value.