The invention discloses a computer-vision-based traffic light control method which comprises the following steps: processing a video acquired by a camera mounted at a traffic light control intersection, extracting a series of frames from the video, processing the frames by virtue of an image processing technology, extracting the position and number information of vehicles, calculating real-time distances between the vehicles and the intersection, the real-time speed of the vehicles and the average speed of all the vehicles, predicting the time when the vehicles reach the intersections, analyzing the positions of the vehicles to determine the time when the vehicles certainly pass through the intersection, and flexibly selecting a K-medoids clustering algorithm or a Clara clustering algorithm to perform clustering analysis on the predicted time when the vehicles reach the intersection to obtain time for which a green light is required to be kept on to determine the switching time of traffic lights. According to the method, the traffic lights can be dynamically and adaptively controlled, the probability of traffic congestion can be reduced, and the aim of optimizing traffic light control can be fulfilled.
Computer-vision-based traffic light control method
2015-07-15
Patent
Elektronische Ressource
Englisch
Machine vision-based intelligent traffic light control system
Europäisches Patentamt | 2015
|Computer vision application: Real time smart traffic light
Tema Archiv | 2005
|Intelligent traffic light control system based on machine vision
Europäisches Patentamt | 2024
|Intelligent traffic light control system and method based on machine vision
Europäisches Patentamt | 2024
|