Adaptive Weighted Bi-Prediction for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in accurately predicting current blocks due to luminance variations between reference blocks, leading to reduced prediction accuracy and increased bandwidth usage in encoded bitstreams.
Innovation Solution
A template-based adaptive weighting method is employed to dynamically adjust the weights of reference blocks during bi-prediction, using luminance and other characteristics to enhance prediction accuracy and efficiency by blending reference blocks based on their similarity to the current block.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional bi-prediction with equal weighting is used, then the encoding process is simple, but prediction accuracy deteriorates due to luminance variations in reference blocks
Solution Approach 1:
The patent changes the weighting parameters dynamically based on luminance characteristics. Instead of using fixed equal weights, the system calculates luminance values for reference blocks and current block, then adjusts weights accordingly - giving higher weight to reference blocks with luminance closer to the current block, thereby improving prediction accuracy while accounting for illumination variations
Solution Approach 2:
The patent applies different weighting strategies to different reference blocks based on their local luminance characteristics. Each reference block is evaluated individually for luminance similarity to the current block, allowing localized adaptation of prediction weights rather than applying a uniform approach across all reference blocks
2Loss of energy
If reference blocks with luminance variations are used for prediction, then bandwidth usage is reduced by using inter prediction, but prediction accuracy deteriorates leading to larger residuals
Solution Approach 1:
The system dynamically adjusts the weighting parameters based on luminance characteristics of reference blocks. By calculating luminance values and comparing them to the current block, the system adapts the prediction weights to compensate for illumination changes, thereby maintaining higher prediction accuracy while still utilizing inter-prediction to reduce bandwidth usage
Solution Approach 2:
The patent introduces a feedback mechanism where luminance values of reference blocks are evaluated and used to adjust prediction weights. This feedback loop allows the system to identify which reference blocks are more suitable for prediction based on their luminance similarity to the current block, thereby improving residual reduction efficiency
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In one embodiment, a method selects a first reference block in a first reference frame and a second reference block in a second reference frame. The first reference block and the second reference block are used to predict a current block in a current frame. A first region is selected based on the first reference block, a second region is selected based the second reference block, and a third region is selected based on the current block. The method compares a characteristic of the first region to a characteristic of the third region and compares a characteristic of the second region to the characteristic of the third region to adaptively determine a first weight for the first reference block and a second weight for the second reference block. Then, a prediction block is calculated for the current block using the first weight and the second weight.