Image Processing for Moving Camera Change Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional change detection algorithms are not applicable to video data taken with a moving camera, as the changing background and perspective of the camera render them ineffective.
Innovation Solution
A method using a Self Adaptive Discriminant filter to track a region of interest in a sequence of images, comparing histograms of red, green, and blue intensity values to detect significant changes, and storing or displaying images only when the difference exceeds a threshold, thereby addressing the challenges of a moving camera.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional change detection algorithms are used with fixed cameras, then change detection is effective for detecting people and vehicles, but the algorithms become ineffective when applied to moving cameras due to changing background and perspective
Solution Approach 1:
The patent divides the image into a region of interest and a background region. By segmenting the image and processing only the region of interest, the algorithm becomes insensitive to background changes caused by camera movement, thereby maintaining reliability for moving camera applications while reducing computational complexity.
Solution Approach 2:
The patent applies different processing strategies to different parts of the image: the region of interest undergoes histogram comparison for change detection, while the background is either excluded or processed differently. This local differentiation allows the system to maintain high detection accuracy in the region of interest while tolerating background changes from camera motion.
2Reliability
If the entire image is processed for change detection, then comprehensive monitoring is achieved, but computational resources and processing time increase significantly
Solution Approach 1:
The patent extracts and isolates the region of interest from the rest of the image for specialized processing. By taking out only the relevant portion of the image that contains potential changes, the system achieves comprehensive monitoring of important areas while significantly reducing the computational burden compared to processing the entire image.
Solution Approach 2:
The patent applies partial action by processing only the region of interest rather than the entire image. This selective processing provides sufficient monitoring coverage for critical areas while avoiding the excessive computational resources required for full-image processing, achieving an optimal balance between monitoring coverage and processing efficiency.
3Measurement precision
If histogram comparison is performed on the entire image, then all changes are detected, but false alarms increase due to background variations
Solution Approach 1:
The patent segments the image into region of interest and background, applying histogram comparison only to the region of interest. This segmentation prevents background variations from generating false alarms while maintaining precise change detection in the relevant area, thereby improving measurement precision without increasing false positives.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Apparatus and method for processing a sequence of images of a scene, the method comprising: tracking a region of interest (14) in the sequence of images (e.g. using a Self Adaptive Discriminant filter), selecting a particular image (12) in the sequence, selecting a set of images from the sequence,the set of images comprises one or more images that precede the particular image (12) in the sequence of images; and determining a value indicative of the level of change between the region of interest (14) in the particular image (12) and the regions of interest (14) in the images in the set of images (e.g. using a Change Detection Process).