Overlapping Camera Diagnosis for Rapid Sensor Abnormal Region Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods fail to accurately and promptly identify abnormal regions in sensors due to erroneous detections, such as dirt or lens scratches, which can lead to incorrect environmental recognition.
Innovation Solution
An abnormality diagnosis device utilizing multiple sensors or cameras arranged to overlap in visual or observation fields, determining target reliability and abnormal regions through shared imaging or sensing regions, and updating abnormality scores based on reliability for rapid identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional target detection is used to identify abnormal regions, then the system can detect potential issues, but erroneous detections occur when dirt or lens scratches are mistaken for target objects
Solution Approach 1:
The patent introduces an abnormality diagnosis device as an intermediary system that receives sensor information and performs separate analysis to verify target detections. This intermediary system includes a target detection unit, a target reliability determination unit, and an abnormal region determination unit that work together to filter out false detections caused by dirt or lens scratches before they affect the main target tracking system.
Solution Approach 2:
The system implements feedback by using tracking results from the monocular region to update the abnormality score map, which then feeds back into the target reliability determination. When a target is consistently detected across multiple frames and confirmed by tracking, the abnormality score decreases, reinforcing the validity of the detection. Conversely, if tracking fails or detection is inconsistent, the abnormality score increases, triggering further verification.
2Reliability
If the system determines abnormal regions by checking detection results at multiple time points, then erroneous detections can be reduced, but the time required to confirm abnormal regions increases
Solution Approach 1:
The system performs preliminary action by continuously maintaining an abnormality score map that is updated in real-time with each detection. Instead of waiting to accumulate multiple detection results before starting analysis, the system proactively updates the abnormality scores with each new detection and tracking result, preparing the foundation for rapid abnormal region determination without requiring post-hoc analysis of multiple time points.
Solution Approach 2:
The abnormality score map is dynamically updated based on current detection results and tracking outcomes. The system adapts the abnormality scores in real-time, increasing scores when detections are suspicious and decreasing them when tracking confirms valid targets. This dynamic adjustment allows the system to rapidly respond to changing conditions and determine abnormal regions quickly without fixed time delays.
Data Source
AI summary
Provided is an abnormality diagnosis device capable of determining an abnormal region generated in a sensor in a short time. The abnormality diagnosis device includes: a target reliability determination unit 400 which determines reliability of a target based on a detection result in a common imaging region; and an individual abnormal image region determination unit 500 which determines an abnormal image region in a monocular region using a tracking result of the target in the monocular region, wherein the individual abnormal image region determination unit 500 changes an abnormality score of an abnormality score map assigned to an abnormal image region according to the reliability of the target.