Video Block Prediction With Affine Motion Candidates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in improving coding efficiency, particularly in handling complex motion patterns beyond simple translations, such as zooms, rotations, and perspective motions, which are common in real-world scenarios.
Innovation Solution
Implementing affine motion prediction methods that utilize block-based affine transform motion compensation, incorporating non-adjacent spatial candidates, history-parameter-based affine models, and regression-based affine candidates to enhance motion vector prediction in video coding systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If simple translation motion models are used for video coding, then the device complexity is low, but the coding efficiency is insufficient for complex motion patterns
Solution Approach 1:
The patent divides the motion compensation process into multiple stages: first performing translation motion compensation, then separately handling rotation and perspective transformations. This segmentation allows the system to manage complex motion patterns through a series of simpler operations rather than a single complex model, thereby improving coding efficiency while controlling implementation complexity.
Solution Approach 2:
The patent implements adaptive motion compensation where the system dynamically selects between different motion models (translation only, translation plus rotation, translation plus perspective) based on the actual motion characteristics of the video content. This dynamic adaptation enables the system to achieve high coding efficiency for complex motions while maintaining low complexity for simple cases.
2Measurement precision
If multiple prediction methods (BV, MV, intra prediction) are combined to improve prediction accuracy, then the coding efficiency improves, but the processing complexity increases
Solution Approach 1:
The patent merges multiple prediction methods by combining block vector (BV) based prediction, motion vector (MV) based prediction, and intra prediction into a unified prediction framework. The system generates multiple prediction candidates from different methods and selects or combines them to produce the final prediction, thereby improving prediction accuracy while managing processing complexity through structured integration.
Solution Approach 2:
The patent employs parameter changes by adjusting prediction parameters adaptively based on the specific motion characteristics and coding conditions. The system modifies prediction parameters such as motion vectors, block vectors, and prediction modes dynamically to optimize the balance between prediction accuracy and processing complexity for different video content types.
Data Source
AI summary
Embodiments of the present disclosure provide a solution for video processing. A method for video processing is proposed. In the method, for a conversion between a target video block of a video and a bitstream of the video, a first prediction of the target video block is determined based on a first block vector (BV). A final prediction of the target video block is determined based on the first prediction and a second prediction of the target video block. The second prediction is determined based on at least one of: a second BV, a motion vector (MV), or an intra prediction. The conversion is performed based on the final prediction.