Hierarchical Motion Estimation Using Original Frames
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion estimation algorithms in video compression often rely on reconstructed frames, which can lead to loss of structural information, especially at low bit rates, affecting the accuracy of motion vector calculation and overall video quality.
Innovation Solution
Calculating motion vectors using sub-sampled data and employing the original frame as the reference frame instead of the reconstructed frame in the hierarchical search block matching algorithm.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If reconstructed frames are used as reference frames in motion estimation, then the encoding process is simplified, but the accuracy of motion vector calculation deteriorates due to loss of structural information
Solution Approach 1:
The patent segments the reference frame into two distinct types: reconstructed frames for computational efficiency and original frames for accuracy-critical operations. Specifically, sub-sampled regions use reconstructed frames while full-resolution regions use original frames, allowing the system to optimize different parts of the motion estimation process differently
Solution Approach 2:
The patent applies different quality levels of reference frames to different regions and stages of the motion estimation process. Original frames are used where high precision is needed (full-search regions, final refinement), while reconstructed frames are used where lower precision is acceptable (coarse initial search, sub-sampled regions), optimizing the trade-off between quality and computational cost
2Measurement precision
If original frames are used as reference frames instead of reconstructed frames, then motion estimation accuracy is improved, but the computational complexity and memory requirements increase
Solution Approach 1:
The patent applies original frames partially rather than universally. Instead of using original frames for the entire motion estimation process, it uses them selectively for specific regions (full-search vs sub-sampled) and stages (refinement vs initial search), achieving improved accuracy where needed while limiting the computational overhead to only the necessary portions
Solution Approach 2:
The patent implements a nested structure where multiple levels of motion estimation are performed with different reference frame qualities. The coarse search using reconstructed frames is nested within the finer search using original frames, with each level building upon and refining the results of the previous level, allowing progressive refinement without fully committing to the higher computational cost throughout
3Productivity
If sub-sampled data is used in hierarchical search, then the processing speed is improved, but the video quality deteriorates due to information loss
Solution Approach 1:
The patent segments the search space into sub-sampled regions and full-resolution regions. Sub-sampling is applied only to the initial coarse search to reduce computational load, while the final motion vector refinement is performed on full-resolution data, ensuring that the critical final measurement uses complete information without subsampling losses
Solution Approach 2:
The patent uses sub-sampled data for preliminary coarse search to quickly eliminate obviously poor candidates and establish an initial search region. This preliminary action with reduced data prepares the ground for the subsequent full-resolution search, allowing the system to benefit from fast preliminary filtering while ensuring final accuracy with complete information
Data Source
AI summary
This invention calculates motion vectors during video encoding for sub-sampled data using a hierarchical search. The motion vector calculation employs a prior original frame as the reference frame rather than a reconstructed frame according to the prior art.