Moving Object Super-Resolution via Frequency Domain Co-Registration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing super-resolution techniques distort moving objects in surveillance images due to inadequate registration models, which fail to accurately align multiple frames of moving scenes, leading to poor image quality.
Innovation Solution
The method involves image co-registration using frequency domain techniques to track and align small regions around moving objects across frames, allowing for fractional pixel shifts and enhancing the resolution of moving objects through a sequence of images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If standard image registration models (translations or homographies) are used for super-resolution, then static scenes can be properly aligned, but moving objects become distorted in the super-resolved image
Solution Approach 1:
The patent segments the image into static background regions and moving object regions. Different registration models are applied to each segment: standard models for static regions and motion-compensated models for moving objects. This segmentation allows the system to maintain registration accuracy for static scenes while preventing distortion of moving objects through region-specific processing
Solution Approach 2:
The patent introduces dynamic motion compensation by tracking object motion between frames and adjusting the registration model accordingly. Instead of using a fixed registration model for all regions, the system dynamically adapts the registration parameters for moving object regions based on detected motion, thereby maintaining image quality while preserving registration accuracy
2Manufacturing precision
If multiple frames are combined to achieve super-resolution, then resolution enhancement is possible, but moving objects suffer from misalignment and distortion
Solution Approach 1:
The patent applies segmentation to separate moving objects from the static background before combining multiple frames. By identifying moving object regions in each frame and applying motion compensation specifically to these regions, the system enables super-resolution enhancement while preserving the shape accuracy of moving objects that would otherwise be distorted by standard multi-frame combining
3Productivity
If a fixed registration model is applied to all regions, then processing is simple and fast, but moving objects cannot be properly aligned
Solution Approach 1:
The patent segments the image into static and moving regions, allowing simple fixed registration for the majority static background while applying more complex motion-compensated registration only to moving object regions. This selective approach maintains high processing speed for most of the image while achieving accurate alignment for moving objects
Solution Approach 2:
The patent applies different registration qualities to different regions: simple fast registration for static regions and more accurate motion-compensated registration for moving object regions. This local quality approach ensures that processing speed is maintained for the bulk of the image while achieving the necessary alignment accuracy for moving objects where it matters most
Data Source
AI summary
In some approaches, super-resolution of static and moving objects can be performed. Results of moving object super-resolution may be improved by means of performing image co-registration. The quality of images of moving objects in an automated form may be improved. A sequence of images may be processed wherein objects can be detected and tracked in succeeding frames. A small region around a tracked object may be extracted in each frame. These regions may be co-registered to each other using frequency domain techniques. A set of co-registered images may be used to perform super-resolution of the tracked object. Also described are image processing systems and articles of manufacture having a machine readable storage medium and executable program instructions.