Triangle Merge Motion Prediction for Low-Load Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing picture coding techniques involving affine transforms lead to heavy processing loads, particularly in handling object deformations such as enlargement, reduction, or rotation in moving pictures.
Innovation Solution
A technique that includes a merging candidate list constructor and a triangle merging candidate selector to derive uni-prediction motion information candidates, reducing processing load while maintaining efficient coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If affine transform is applied for inter prediction to handle object deformations, then coding efficiency is improved, but processing load increases heavily
Solution Approach 1:
The patent divides the prediction process into two distinct modes: uni-prediction mode for blocks with simple motion characteristics and bi-prediction mode for blocks with complex motion characteristics. This segmentation allows the system to apply the computationally intensive bi-prediction only where necessary, while using the simpler uni-prediction for the majority of blocks, thereby reducing overall processing load while maintaining coding efficiency for deforming objects.
Solution Approach 2:
The patent applies the complex bi-prediction technique partially - only to specific blocks that exhibit deformation characteristics. By using a flag-based mechanism to identify and process only those blocks requiring affine transform, the system avoids applying excessive computation to all blocks, thus reducing processing load while still achieving good coding efficiency for the deforming regions.
2Measurement precision
If bi-prediction is used for triangle merging candidates, then prediction accuracy improves, but processing load increases
Solution Approach 1:
The patent introduces a dynamic switching mechanism that adapts the prediction mode based on block characteristics. A flag indicates whether uni-prediction or bi-prediction should be used for each block, allowing the system to dynamically adjust the processing complexity according to the actual motion characteristics of each block, thereby optimizing the balance between prediction accuracy and processing load.
Solution Approach 2:
The patent applies different prediction qualities to different regions: uni-prediction is used for blocks with simple motion characteristics while bi-prediction is used for blocks with complex deformation characteristics. This local differentiation ensures high prediction accuracy where needed while avoiding unnecessary computational overhead in regions where simple prediction suffices.
Data Source
AI summary
There is provided a technique that includes a merging candidate list constructor that constructs a merging candidate list including spatial merging candidates, and a triangle merging candidate selector that selects, from the merging candidate list, a first triangle merging candidate that is uni-prediction and a second triangle merging candidate that is uni-prediction, in which the triangle merging candidate selector derives a uni-prediction motion information candidate having a same priority in the first triangle merging candidate and the second triangle merging candidate.