Video Coding Prediction Bandwidth Reduction via Chroma Motion Vector Modification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video coding technologies face challenges in reducing memory bandwidth, particularly for non-4:2:0 chroma formats, which increase hardware costs due to higher memory bandwidth requirements for interpolating chroma pixels, leading to increased costs for larger buses, memory, and faster memory to meet these specifications.
Innovation Solution
The techniques involve restricting bi-prediction mode in video coding for non-4:2:0 formats, converting bi-directional motion vectors to uni-directional motion vectors, and differently partitioning video block components to reduce memory bandwidth, allowing for reduced hardware specifications and costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If bi-prediction mode is used for non-4:2:0 chroma formats, then coding efficiency is improved, but memory bandwidth requirements increase
Solution Approach 1:
The patent applies different prediction modes to different color components locally. Specifically, bi-prediction is applied to luma components while uni-prediction is applied to chroma components in non-4:2:0 formats. This local differentiation maintains coding efficiency for luminance information while reducing memory bandwidth requirements for chrominance information, resolving the contradiction between coding efficiency and memory bandwidth usage.
2Manufacturing precision
If higher memory bandwidth is provided to meet non-4:2:0 format requirements, then chroma pixel interpolation quality is improved, but hardware costs increase
Solution Approach 1:
The patent segments the video data processing into separate handling for luma and chroma components. By partitioning the prediction process into distinct pathways (bi-prediction for luma, uni-prediction for chroma), the system achieves adequate chroma interpolation quality without requiring the full memory bandwidth that would be needed if bi-prediction were applied uniformly to all components, thereby reducing hardware costs.
Solution Approach 2:
The patent changes the prediction mode parameter from bi-prediction to uni-prediction specifically for chroma components in non-4:2:0 formats. This parameter change reduces the memory bandwidth requirement while maintaining acceptable chroma interpolation quality, thus lowering hardware costs without significantly compromising manufacturing precision for chroma components.
3Quantity of substance
If bi-directional motion vectors are converted to uni-directional motion vectors, then memory bandwidth is reduced, but prediction accuracy may be affected
Solution Approach 1:
The patent applies different motion vector prediction approaches to different color components. Bi-directional motion vectors are maintained and converted only for chroma components where prediction accuracy is less critical, while luma components retain bi-prediction for higher accuracy. This local differentiation resolves the contradiction by preserving prediction accuracy where needed while reducing memory bandwidth where acceptable.
Data Source
AI summary
In one example, an apparatus for coding video data comprises a video coder configured to obtain a motion vector for predicting a video block with a non-4:2:0 chroma format, determine a video block size for the video block, modify the motion vector to generate a modified motion vector for obtaining samples of at least one reference picture with which to predict the video block if the video block size meets a size criterion, and generate a prediction block for the video block using the samples of the at least one reference picture and the modified motion vector.