Vehicle Vision System False Positive Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Vision systems for motor vehicles often experience false positive detections, leading to unnecessary safety actions that can distract drivers or cause more severe risks, due to limited decision time and processing resources, which complicates separating true objects from false ones.
Innovation Solution
A method that performs a detailed analysis of detected objects after the decision horizon to identify false positives, storing data in a database for future comparisons, allowing for reduced false positive detections by using additional computational resources and image data collected over a longer time period, and suppressing unnecessary safety actions in similar situations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If the detection and estimation means operates within limited decision time and processing resources, then real-time safety decisions can be made, but false positive detections increase
Solution Approach 1:
The system performs preliminary analysis of detected objects after the decision horizon has passed. By conducting detailed analysis in advance (retrospectively) when time constraints are relaxed, the system prepares information that can prevent future false positives without delaying real-time safety decisions. The analysis means examines objects that were previously detected and acted upon to determine if they were false positives.
Solution Approach 2:
The vision system is divided into separate functional modules: detection and estimation means for real-time operation, and an analysis means for detailed retrospective analysis. This segmentation allows each module to operate with appropriate time constraints - the detection module maintains real-time response while the analysis module performs comprehensive evaluation without time pressure.
2Reliability
If false positive detections are suppressed using stored data from previous analyses, then future false positives are reduced, but system complexity increases
Solution Approach 1:
The system implements feedback by storing the results of false positive analyses in a database and using this information to influence future detection decisions. When the detection means identifies an object, it can query the database to check if similar objects were previously identified as false positives, thereby suppressing unnecessary safety actions without requiring complex real-time analysis.
3Measurement precision
If detailed analysis is performed on detected objects, then false positive identification accuracy improves, but processing time increases
Solution Approach 1:
The detailed analysis is performed as a preliminary action after the decision horizon, when time constraints are no longer applicable. This retrospective analysis prepares information for future detections without interfering with real-time decision-making. The system accepts that processing time increases for analysis but not for critical safety decisions.
Data Source
Figure 1
Figure 2~3
AI summary
A vision system (10) for a motor vehicle comprises an imaging means (11) adapted to gather images from a region surrounding the motor vehicle and an electronic processing means (14) adapted to perform image processing of images gathered by said imaging means (11) and initiate a safety action based on the result of said image processing, wherein said electronic processing means (14) comprises a detection and estimation means (30) adapted to detect an object in the surrounding of the motor vehicle through said image processing and to estimate whether a detected object is a positive object requiring a safety action. The electronic processing means (14) comprises a false positive analysis means (32) adapted to perform an analysis to determine whether an object has been falsely estimated as a positive object by the detection and estimation means (30), to store data depending on the result of the false positive analysis in a data storage means (25), and to use said stored data in said detection and estimation means (30) for future object detection and estimation.