Context-Based Affine Mode Signaling for Video Block Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing video coding standards, such as HEVC, face challenges in efficiently handling complex motions like zoom, rotation, and perspective transformations due to the complexity and increased signal overhead of affine motion mode, which complicates the encoding and decoding processes.
Innovation Solution
The method involves determining an inter prediction mode for a portion of a bitstream and signaling affine motion mode based on the status of a skip flag, enabling affine mode if certain conditions are met, and optimizing the use of affine motion mode through context encoding and candidate list management to improve coding efficiency and reduce complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If affine motion mode is applied to handle complex motions (zoom, rotation, perspective), then motion prediction accuracy is improved, but encoding complexity and signal overhead increase
Solution Approach 1:
The patent applies affine motion mode selectively only to specific blocks that exhibit complex motion patterns, rather than uniformly to all blocks. The encoder determines whether to apply affine mode on a block-by-block basis, allowing complex motion compensation where needed while maintaining simple translation-based compensation elsewhere, thus resolving the contradiction between accuracy and complexity
Solution Approach 2:
The patent uses a limited set of affine parameters (typically 2-3 control points) rather than full affine transformation parameters for all blocks. By applying partial affine transformation only to necessary blocks and using simpler modes for others, the patent achieves improved motion prediction where required without incurring the full complexity cost universally
2Measurement precision
If affine motion mode is applied to handle complex motions, then motion prediction accuracy is improved, but signal overhead increases
Solution Approach 1:
The patent signals affine motion parameters only for blocks that require complex motion compensation, rather than for all blocks. This selective signaling approach reduces the overall bitstream overhead while maintaining accurate motion prediction for blocks where affine transformation is beneficial
Solution Approach 2:
The patent extracts and transmits only the essential affine parameters (control point positions and motion vectors) rather than complete transformation matrices. By transmitting a minimal set of parameters that capture the dominant motion characteristics, the patent reduces signal overhead while preserving prediction accuracy
3Productivity
If affine motion mode is enabled for all blocks, then compression efficiency improves, but processing complexity increases
Solution Approach 1:
The patent evaluates and applies affine motion mode independently for each block based on local motion characteristics, allowing compression efficiency to be maximized locally without requiring complex global optimization. This block-level independence simplifies processing while maintaining overall compression efficiency
Solution Approach 2:
The patent applies a simplified decision-making process for each block (e.g., comparing affine vs. non-affine rates) rather than exhaustive optimization. This partial evaluation approach achieves good compression efficiency with reduced processing complexity by not exploring all possible motion modes for every block
Data Source
AI summary
The invention relates to signalling affine mode in an encoded video stream; in particular determining a list of merge candidates corresponding to blocks neighbouring a current block; and signalling affine mode for said current block; wherein signalling said affine mode comprises decoding a context encoded flag from the data stream, and wherein the context variable for said flag is determined based on whether or not said neighbouring blocks use affine mode. Related encoding and decoding methods and devices are also disclosed.