Video Block Prediction Using Motion-Shift Candidate Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding methods struggle to construct optimal candidate lists for inter prediction, particularly in dynamic or high-resolution video content, leading to increased residual errors and bitrate requirements due to suboptimal handling of complex motion patterns.
Innovation Solution
A method for determining motion information by calculating a motion shift based on a vector sum of multiple motion vectors of reference blocks, using template matching to select neighboring blocks with the smallest cost, and reconstructing block units with predicted samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional inter prediction methods are used to construct candidate lists, then the coding process remains simple, but prediction accuracy deteriorates in dynamic or high-resolution video content
Solution Approach 1:
The patent segments the candidate list construction into multiple stages: initial candidate selection from neighboring blocks, refinement through template matching, and final selection based on rate-distortion optimization. This segmentation allows the system to handle complex motion patterns systematically without overwhelming computational complexity.
Solution Approach 2:
The patent performs preliminary actions by pre-identifying potential candidate blocks from neighboring regions before the actual prediction process. Template matching is performed in advance to evaluate candidate quality, and the best candidates are pre-selected and stored for use in the prediction step, thereby improving accuracy without increasing real-time complexity.
2Manufacturing precision
If suboptimal candidate lists are used for inter prediction, then the coding process remains fast, but residual errors increase
Solution Approach 1:
The patent implements feedback mechanisms where the quality of candidate blocks is continuously evaluated using template matching metrics and rate-distortion optimization. The system uses this feedback to iteratively refine the candidate list, selecting only those blocks that provide the best prediction accuracy while maintaining coding efficiency. This feedback loop ensures that residual errors are minimized without sacrificing productivity.
3Loss of information
If conventional motion vector selection is used, then the processing speed remains high, but bitrate requirements increase due to higher residual errors
Solution Approach 1:
The patent changes key parameters in the candidate selection process, including using template matching costs as a selection criterion, adjusting the number and type of neighboring blocks considered, and optimizing the motion vector precision. These parameter changes enable the system to achieve lower residual errors while maintaining acceptable processing times by focusing computational effort on the most promising candidates.
Data Source
AI summary
An electronic device and a corresponding method for decoding/encoding video data is provided. The electronic device includes at least one processor and at least one non-transitory computer-readable medium coupled to the at least one processor and storing one or more computer-executable instructions that, when executed by the at least one processor, cause the electronic device to: receive the video data; determine a block unit from an image frame based on the video data; determine, based on a vector sum of multiple motion vectors of multiple reference blocks, a motion shift that indicates a collocated block for the block unit; determine multiple predicted samples of the block unit based on motion information of the collocated block; and reconstruct the block unit based on the predicted samples of the block unit. In addition, a non-transitory machine-readable medium for decoding/encoding video data is also provided.