Recursive Motion Estimator Convergence via Predominant Motion Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Recursive motion estimators struggle to capture fast and sudden global motion scenes, leading to visible picture artifacts due to inadequate motion compensation, as the sampling of motion vectors often misses the points of global motion and focuses on local motion instead.
Innovation Solution
A method and apparatus that improve the convergence speed of motion estimation by generating global motion vectors through a predominant motion detection process, which includes histogram formation, filtering, and distribution of motion vectors, allowing for more accurate prediction of pixel changes across frames and improved motion compensation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If motion vectors are sampled at fixed positions to estimate global motion, then the motion estimation process can be simplified, but the convergence speed decreases and fast global motion cannot be captured accurately
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing motion vectors at multiple positions before the actual motion estimation process. These pre-computed motion vectors are then used during recursive estimation to accelerate convergence, allowing the system to quickly capture fast global motion without performing complex calculations in real-time during the estimation process itself
Solution Approach 2:
The patent uses copying by creating a predictor field that copies and distributes motion vectors from previously determined positions to multiple locations. This predictor field contains replicated motion vector information that can be directly applied during recursive estimation, avoiding the need to recompute motion vectors from scratch and thereby significantly improving convergence speed
2Ease of manufacture
If motion vectors are sampled at fixed positions, then the sampling process is simple, but the quality of global motion detection deteriorates when fast global motion occurs
Solution Approach 1:
The patent applies dynamics by transitioning from static fixed-position sampling to a dynamic adaptive sampling approach. The system determines sampling positions based on the actual motion characteristics detected in previous frames, allowing sampling points to dynamically adjust to where global motion is occurring. This enables accurate capture of fast global motion while maintaining simplicity through automated adaptive positioning rather than complex manual configuration
Solution Approach 2:
The patent implements feedback by using previously determined motion vectors to influence the sampling process in subsequent frames. The system analyzes motion characteristics from earlier frames and uses this feedback information to determine optimal sampling positions for current frame estimation. This feedback mechanism ensures that sampling points are strategically positioned to capture global motion accurately while keeping the overall process simple through automated adaptive adjustment
3Reliability
If strict quality criteria are applied to global motion vectors, then the reliability of motion compensation improves, but the frequency of global motion vector usage decreases
Solution Approach 1:
The patent applies preliminary action by pre-processing and filtering motion vectors before they are used in motion compensation. The system performs preliminary quality assessment and selection of motion vectors based on predetermined criteria, creating a ready-to-use set of reliable motion vectors that can be directly applied without re-evaluation. This preliminary action ensures high reliability while increasing usage frequency by eliminating the need for repeated quality verification during actual compensation processing
Solution Approach 2:
The patent uses copying by creating a predictor field that copies verified motion vectors to multiple positions and time points. Once motion vectors pass quality criteria and are stored in the predictor field, they can be replicated and applied throughout the image and across multiple frames, significantly increasing their usage frequency. This copying mechanism maintains reliability by using pre-verified vectors while enabling widespread application through efficient replication rather than repeated validation
Data Source
AI summary
A method and apparatus for motion estimation of at least a first and a second image frame by estimating at least one motion vector correlating a portion of pixels of the at least first and second image frame, the first and second image frame being part of an image frame sequence. The at least one motion vector is obtained by a predominant motion detection generating at least one global motion vector based on at least one previously determined motion vector, the previously determined motion vector correlating a portion of pixels of earlier image frames of the image frame sequence, and an estimation estimating the at least one motion vector based on the at least one global motion vector.