Optical Flow Prediction Refinement for Video Sub-Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding methods face challenges in efficiently compressing high-resolution videos due to increased bandwidth demands and complexity in motion vector prediction and refinement, particularly in handling sub-blocks within video coding units.
Innovation Solution
The implementation of prediction refinement with optical flow (PROF) techniques, which calculates motion vector differences based on sub-block dimensions and applies phase-variant affine subblock motion compensation to enhance prediction accuracy and efficiency in video processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If prediction refinement with optical flow is applied to sub-blocks, then prediction accuracy and video compression performance are improved, but computational complexity increases
Solution Approach 1:
The patent divides the video block into multiple sub-blocks (e.g., 4x4, 8x8, or 16x16 pixel sub-blocks) and applies prediction refinement independently to each sub-block. This segmentation allows the complex optical flow calculation to be distributed across smaller units, improving prediction accuracy for regions with non-uniform motion while reducing the computational burden compared to refining the entire block as a single unit.
Solution Approach 2:
The patent applies different motion vector refinement strategies to different sub-blocks based on their specific characteristics. By calculating optical flow and motion vector differences locally for each sub-block, the method adapts the prediction refinement to the local motion patterns, achieving higher prediction accuracy where needed while maintaining computational efficiency through selective processing.
2Measurement precision
If motion vector differences are calculated with high precision, then prediction accuracy improves, but bit width requirements and data processing load increase
Solution Approach 1:
The patent employs bit-shift operations to scale and adjust the precision of motion vector differences. By using shift amounts that can be configured or fixed, the method achieves high precision motion vector representation when needed while controlling the bit width requirements. The parameter change through bit-shifting allows flexible precision management without proportionally increasing data processing load.
Data Source
AI summary
A video processing method is disclosed to include: generating one or more initial predictions for a sub-block of a current video block of the video; refining the one or more initial predictions to generate a final prediction block using a prediction refinement with optical flow (PROF) procedure by calculating motion vector differences based on variables having a bit width not exceeding a predetermined maximum bit width; and performing a conversion between the current video block and a coded representation of the video using the variables having the bit width not exceeding the predetermined maximum bit width.