Stereo Vision System Selective Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional stereo vision systems require high power consumption and cost due to continuous operation for distance information acquisition, and they struggle with noise in disparity calculations and calibration checks, especially when used in mobile systems like robots.
Innovation Solution
A stereo vision system and method that utilize image compression to acquire low-resolution distance information and selectively power the stereo matching module only when high-resolution information is needed, using a motion vector to detect objects and verify camera calibration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If stereo matching is continuously performed to acquire distance information, then distance measurement capability is improved, but power consumption increases
Solution Approach 1:
The patent implements periodic action by alternately switching between image compression processing and stereo matching processing based on whether an object is detected. The system performs image compression continuously, then selectively activates stereo matching only when objects are detected in specific distance ranges, rather than running stereo matching continuously. This periodic activation pattern reduces overall power consumption while maintaining necessary measurement capabilities.
Solution Approach 2:
The system dynamically adjusts its processing mode based on real-time detection results. When objects are detected within critical distance ranges, the system transitions to high-precision stereo matching mode. When no objects are detected or objects are outside critical ranges, it switches to lower-power image compression mode. This dynamic adaptation allows the system to optimize power consumption according to actual operational needs.
2Measurement precision
If stereo matching is performed for all pixels to acquire high-resolution distance information, then measurement precision is improved, but processing time increases
Solution Approach 1:
The patent applies partial action by performing stereo matching only on specific regions of interest rather than processing all pixels in the entire image. When objects are detected through image compression, the system activates stereo matching only for those detected objects within specific distance ranges, leaving other regions to be processed only through the faster image compression algorithm. This selective approach maintains high measurement precision for critical objects while significantly reducing overall processing time.
Solution Approach 2:
The system segments the image processing into two distinct stages: first, rapid image compression processing to detect objects and their approximate locations; second, selective stereo matching processing applied only to detected objects within critical distance ranges. This segmentation allows the system to use the fast compression algorithm for screening and reserve the computationally intensive stereo matching for only those regions where high-precision distance measurement is actually needed.
3Speed
If dedicated hardware is used for stereo matching, then processing speed is improved, but device complexity increases
Solution Approach 1:
The patent implements multi-functionality by designing the stereo matching unit to serve dual purposes: it performs both the computationally intensive stereo matching operations and the image compression operations. By making the same hardware unit versatile enough to handle multiple processing tasks, the system avoids duplicating hardware resources for each function, thereby reducing overall device complexity while maintaining high processing speeds when stereo matching is required.
Data Source
AI summary
A stereo vision system includes an image pre-processing unit for pre-processing the right and left images, and a stereo matching unit for carrying out stereo matching of the right and left images to acquire low-resolution distance information of the right and left images and high-resolution distance information of the right and left images upon detection of an object within a distance range through the low-resolution distance information.