Video Image Compression via Foreground Background Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-speed, high-resolution microscopy systems generate vast amounts of video data quickly, exceeding affordable storage capacity, while funding agencies require all data from published experiments to be retained for re-analysis, necessitating improved video image compression methods that preserve analysis results.
Innovation Solution
Classifying image pixels in video images as foreground or background, replacing background pixels with time-averaged values, and using a compression algorithm to achieve significant compression ratios without impacting analysis results, such as up to 100× compression with scientifically lossless results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If video data is compressed using traditional compression algorithms, then storage capacity is improved, but information loss occurs that impacts analysis results
Solution Approach 1:
The patent segments the video data into foreground pixels (containing relevant information) and background pixels (containing mostly noise). By applying different processing strategies to each segment - preserving foreground pixels and compressing/averaging background pixels - the system achieves high compression ratios while maintaining analysis integrity for the important foreground content.
Solution Approach 2:
The patent applies different quality levels to different regions of the video data. Foreground regions maintain high quality with minimal compression to preserve analysis information, while background regions accept higher compression since they contain mostly noise that doesn't impact analysis results. This local differentiation resolves the contradiction between overall compression and information preservation.
2Reliability
If all video data is retained for re-analysis, then reliability is improved, but storage requirements exceed affordable capacity
Solution Approach 1:
The patent extracts and retains only the essential foreground information that is necessary for re-analysis, while removing or heavily compressing the background noise. This extraction approach maintains reliability for scientific re-analysis by preserving the critical data elements while eliminating redundant information that consumes storage space.
Solution Approach 2:
The patent changes the parameter of data representation by replacing background pixel values with time-averaged values. This parameter transformation significantly reduces the storage requirements for background data while maintaining the statistical properties needed for analysis, thereby resolving the contradiction between retention and storage capacity.
3Productivity
If high-speed microscopy is used to increase productivity, then data acquisition rate is improved, but data generation exceeds storage capacity
Solution Approach 1:
The patent applies preliminary processing during the data acquisition phase by classifying pixels into foreground and background categories in real-time. This preliminary segmentation enables immediate compression of background data as it is acquired, preventing the accumulation of excessive data volumes even at high acquisition rates.
Solution Approach 2:
The patent uses time-averaging of background pixels, which inherently involves periodic sampling and averaging over time intervals. This periodic processing reduces the data volume generated by high-speed acquisition while preserving the temporal characteristics needed for analysis, thereby managing the data volume burden of high productivity systems.
Data Source
AI summary
According to one exemplary method, the method includes classifying image pixels in each video image of a series of video images as either foreground or background. The method also includes replacing each pixel classified as background with a time-averaged value of pixel intensity for that pixel. The method further includes compressing the video images using a compression algorithm.