Traffic Signal Recognition Using Adaptive Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing traffic signal recognition methods experience delays in processing transitions from a 'green signal' to other signals due to reliance on time-series image analysis, leading to delayed vehicle control responses.
Innovation Solution
A traffic signal recognition device and method that processes a sequence of images to determine the display state of traffic signals, using a threshold-based system to quickly identify changes in signal states, reducing delay times by adjusting thresholds based on signal types and vehicle proximity to stop lines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a display state of the traffic signal is determined by referring to a plurality of images arranged in time-series order, then recognition accuracy is improved, but delay time from actual transition to processing start increases
Solution Approach 1:
The system performs preliminary detection by analyzing a plurality of images arranged in time-series order to identify traffic signal transitions before they occur in real-time. By pre-processing and pre-identifying signal changes in the image sequence, the system prepares the recognition result in advance, thereby reducing the delay time when the actual transition occurs while maintaining high recognition accuracy through multi-frame analysis
2Measurement precision
If the number of captured images for determination is increased, then recognition accuracy is improved, but processing time increases
Solution Approach 1:
The system applies partial action by determining the display state using a predetermined number of captured images rather than analyzing all available images. This selective approach processes only the necessary portion of the image sequence required for accurate recognition, thereby maintaining high recognition accuracy while avoiding excessive processing time that would result from analyzing every captured frame
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A traffic signal recognition method and a traffic signal recognition device acquire a result sequence consisting of a plurality of determination results in time-series order obtained by determining a display state of a traffic signal based on a plurality of images of a traveling direction of a vehicle, set a first threshold as a number threshold if a specific determination result which is a latest determination result among the determination results is a determination result of a permission display that allows a passage of the stop line corresponding to the traffic signal, set a second threshold smaller than the first threshold as the number threshold if the specific determination result is other than a determination result of the permission display, and output the specific determination result if a number of determination results identical to the specific determination result is greater than the number threshold.