Affine Motion Model Derivation Using Spatial Neighbor Gradients
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards face challenges in efficiently encoding and decoding high-resolution video data while maintaining image quality, as the amount of video data grows exponentially with resolution.
Innovation Solution
The implementation of a method to improve coding efficiency by deriving an affine motion model based on motion information from spatial neighboring blocks, using gradients of motion vectors to estimate parameters and construct motion vectors for the current coding block.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional motion estimation is used for high-resolution video encoding, then compression can be achieved, but coding efficiency deteriorates as resolution increases to 4K×2K or 8K×4K
Solution Approach 1:
The patent changes the parameters used for motion modeling from simple translational motion vectors to affine motion models with multiple control points. This allows the system to capture complex motion patterns in high-resolution video more accurately, improving coding efficiency without requiring proportionally more data to be transmitted.
Solution Approach 2:
The patent divides the current video block into multiple sub-blocks and uses separate motion vectors for each sub-block. By segmenting the motion representation, the system can more accurately model local motion variations in high-resolution video, achieving better compression ratios while maintaining image quality.
2Measurement precision
If affine motion model with multiple control points is used, then motion modeling accuracy improves, but computational complexity increases
Solution Approach 1:
The patent performs preliminary actions by deriving motion information from spatial neighboring blocks before encoding the current block. This pre-computation of motion candidates reduces the computational burden during actual encoding, as the affine parameters can be inferred from already-decoded neighboring regions without requiring complex real-time calculations.
Solution Approach 2:
The patent copies motion information from spatial neighboring blocks to construct affine motion models for the current block. By reusing motion parameters from adjacent blocks that have already been processed, the system achieves accurate motion modeling while avoiding redundant computations, thus reducing overall computational complexity.
3Productivity
If spatial prediction with neighboring blocks is used, then compression performance improves, but the amount of reference data required increases
Solution Approach 1:
The patent extracts only the essential motion information from spatial neighboring blocks that is needed for constructing affine motion models. Instead of storing or processing all reference data from neighboring blocks, the system selectively extracts motion vectors and parameters that are most useful for predicting the current block, reducing the effective reference data requirement while maintaining compression performance.
Data Source
AI summary
An electronic apparatus performs a method of coding video data. The method includes receiving, from a bitstream of the video data, a first syntax that indicates an affine motion model enabled for a current coding block, estimating parameters of the affine motion model using gradients of motion vectors of multiple spatial neighboring blocks of the current coding block, and constructing motion vectors of the affine motion model for the current coding block by using the estimated parameters. In some embodiments, constructing motion vectors further includes converting the estimated parameters into control point motion vectors (CPMVs), and adding the CPMVs into a current affine merge candidate list. In some embodiments, constructing motion vectors further includes deriving a motion vector predictor for an affine mode.