Affine Merge Candidate Derivation for Lower-Load Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing picture coding technologies that apply affine transforms for inter prediction in moving-picture coding/decoding face high processing loads.
Innovation Solution
Implementing a moving-picture coding and decoding process that derives affine inheritance, construction, and fixation merging candidates by fixing motion vectors of affine control points to (0, 0) for coding and decoding blocks, reducing processing load while maintaining efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If affine transform is applied for inter prediction in moving-picture coding/decoding, then coding efficiency is improved, but processing load increases
Solution Approach 1:
The patent segments the affine transform process into three distinct merging candidate types: affine inheritance merging candidates (inheriting affine model from neighboring blocks), affine construction merging candidates (constructed from motion information elements of neighboring blocks), and affine fixation merging candidates (with fixed motion information). This segmentation allows the system to select appropriate candidate types based on scene requirements, improving coding efficiency while managing processing load through selective application.
Solution Approach 2:
The patent introduces parameter changes by fixing the motion vector of affine control points to (0, 0) in affine fixation merging candidates. This parameter fixation reduces the search space and computational complexity for certain block types while maintaining the flexibility to use full affine transforms when needed, thereby balancing coding efficiency improvement with processing load reduction.
2Measurement precision
If affine transform is applied for inter prediction, then prediction accuracy for deformed objects is improved, but computational complexity increases
Solution Approach 1:
The patent implements a dynamic candidate selection mechanism where the encoder/decoder can choose among three types of affine merging candidates based on the characteristics of the current block and neighboring blocks. This dynamic approach allows the system to apply full affine transforms only when necessary for accurate prediction of deformed objects, while using simpler candidates for other cases, thus maintaining prediction accuracy while reducing overall computational complexity.
Solution Approach 2:
The patent introduces affine fixation merging candidates with fixed motion information (0, 0) as a computationally inexpensive option. These simplified candidates serve as a lightweight alternative to full affine transforms for blocks where high prediction accuracy is less critical, effectively providing a low-cost solution that reduces computational complexity while maintaining acceptable prediction quality for standard cases.
Data Source
AI summary
A moving-picture coding device, for performing an affine transform in units of coding blocks and forming a bitstream, includes a subblock-based temporal merging candidate derivation unit configured to derive a subblock-based temporal merging candidate which comprises a motion vector and a flag indicating whether or not a prediction subblock is valid for each reference list in units of subblocks, an affine inheritance merging candidate derivation unit configured to derive an affine inheritance merging candidate for inheriting an affine model of blocks neighboring a coding target block in a space domain, and an affine construction merging candidate derivation unit configured to derive an affine construction merging candidate from a plurality of motion information elements of blocks neighboring the coding target block in a space or time domain.