Noise-Aware Sub-Pixel Motion Vector Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video processing technologies face challenges in processing high-definition video data at full resolution with real-time optical flow estimation due to increased processing time and computational complexity, particularly when using resource-constrained mobile devices.
Innovation Solution
A method and device for sub-pixel refinement of motion vectors that utilize a predefined noise model to generate a noise prediction map, perform block-based motion estimation, and determine the need for sub-pixel refinement based on this map, minimizing direct image analysis and reducing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If in-depth image analysis is performed to improve motion estimation accuracy, then measurement precision is improved, but device complexity and processing time increase
Solution Approach 1:
The patent applies preliminary action by generating a noise prediction map before performing motion estimation. The noise model is pre-established based on image quality and noise characteristics, allowing the system to predict noise levels in advance and adjust motion estimation parameters accordingly, rather than performing complex real-time analysis during motion estimation
Solution Approach 2:
The patent changes parameters by adjusting motion estimation precision dynamically based on the noise prediction map. In high-noise regions, the system reduces motion estimation precision to save computational resources, while in low-noise regions, it maintains high precision. This parameter adaptation resolves the contradiction between accuracy and complexity
2Measurement precision
If full resolution processing is performed to improve measurement precision, then measurement precision is improved, but productivity decreases
Solution Approach 1:
The patent applies local quality by differentiating processing precision across different spatial regions. The noise prediction map identifies high-noise and low-noise regions, and motion estimation is performed with appropriate precision levels for each region. This allows full precision only where necessary, improving overall processing speed while maintaining accuracy where it matters
Solution Approach 2:
The patent uses partial action by performing motion estimation at reduced precision in regions where high precision is not necessary (high-noise areas). Instead of applying full precision processing uniformly across the entire image, the system applies precision selectively, achieving sufficient accuracy with less computational effort
3Manufacturing precision
If sub-pixel refinement is performed on all blocks to improve manufacturing precision, then manufacturing precision is improved, but loss of time increases
Solution Approach 1:
The patent applies local quality by determining sub-pixel refinement needs on a block-by-block basis using the noise prediction map. Only blocks located in low-noise regions undergo sub-pixel refinement, while blocks in high-noise regions skip this computationally intensive step. This selective approach maintains precision where needed while significantly reducing overall processing time
Data Source
AI summary
A method performed by an electronic device for sub-pixel refinement of motion vectors is provided. The method includes obtaining, by the electronic device, a pair of adjacent video frames, generating, by the electronic device, a noise prediction map on a frame from the pair of adjacent frames based on a predefined noise model, obtaining, by the electronic device, the motion vectors by performing block-based motion estimation between the adjacent video frames, and determining, by the electronic device, whether to perform the sub-pixel refinement of the motion vectors based on the noise prediction map.