Geometric Motion Prediction for Arbitrary-Shape Video Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques struggle with efficient prediction of blocks with arbitrary shapes due to limitations in square or rectangular block partitioning, leading to suboptimal coding efficiency and video quality, particularly in advanced standards like VVC.
Innovation Solution
The method employs geometric motion prediction by bisecting a current block into arbitrary shapes, performing different motion predictions on each partition, and blending the predictors based on predefined pixel units to generate a final inter predictor.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If square or rectangular block partitioning is used for inter prediction, then the encoding and decoding process is simple, but prediction accuracy for objects with arbitrary shapes deteriorates
Solution Approach 1:
The current block is divided into two triangular partitions using a diagonal line instead of traditional square or rectangular partitions. This segmentation allows each partition to independently represent different motion characteristics, improving prediction accuracy for arbitrary shaped objects while maintaining manageable complexity through systematic processing of the two partitions
Solution Approach 2:
The patent introduces asymmetric triangular partitioning where a diagonal line divides the block into two unequal triangular regions. This asymmetric division better conforms to the natural shapes of objects in video content, allowing more accurate motion representation compared to symmetric square partitions, thereby improving prediction accuracy for arbitrary shaped objects
2Measurement precision
If triangle partition mode (TPM) is used to partition blocks, then prediction accuracy for arbitrary shapes improves, but coding efficiency deteriorates due to limited merge mode usage
Solution Approach 1:
The block is segmented into two triangular partitions that can be independently processed. Each partition can select its own motion information from the merge candidate list, allowing flexible adaptation to different motion characteristics in different regions while maintaining coding efficiency through selective use of merge mode
Solution Approach 2:
The patent introduces dynamic selection of motion information for each triangular partition. Instead of using a fixed partitioning approach, the system dynamically selects motion vectors and merge candidates based on the specific content and motion characteristics of each partition, improving both prediction accuracy and coding efficiency
3Adaptability or versatility
If geometric partition mode (GPM) is expanded with diverse partitioning forms, then versatility in handling arbitrary shapes improves, but device complexity increases
Solution Approach 1:
The complex GPM is simplified by dividing the block into two triangular partitions. This segmentation reduces the complexity of handling diverse partitioning forms while maintaining versatility, as each triangular partition can be processed independently with standardized operations
Solution Approach 2:
Different motion information and blending operations are applied to different triangular partitions based on their local characteristics. Each partition can have its own motion vector and blending weight, allowing localized optimization without increasing overall system complexity, thereby maintaining versatility while managing device complexity
Data Source
AI summary
A method and an apparatus are disclosed for video coding using geometric motion prediction. In the disclosed embodiments, a video decoding device decodes motion information for a first block partition of the current block and generates a first prediction block of the current block by using the motion information of the first block partition. The video decoding device decodes a merge candidate index for a second block partition of the current block and generates a second prediction block of the current block by using the merge candidate index. The video decoding device derives a geometric block partition shape and blends the first prediction block and the second prediction to generate a final prediction block of the current block.