The Research Of Urban Traffic Signal Timing Optimization Method Based On Machine Vision

Duing to the development of the global economy and the improvement of per capita income level,the behavior of people to buy a car has become a commonplace things,at the same time,the resulting urban traffic congestion has become increasingly serious.Under the existing urban traffic resources,the problem that how to make vehicles run more reasonable and how to improve the traffic capacity effectively is becoming problem the people is increasingly thinking.Now most of the traffic lights use the mode of timing control,such as: the signals of timing control and the signals of manual control,the mode of timing control is easy to cause no vehicles running under the green light direction and so many vehicles that can’t running under the red light direction,it’s greatly affected the efficiency of urban traffic.So serious traffic congestion in the city today,traditional traffic lights have been increasingly unable to meet people’s desire to travel unimpeded.Video detection technology is an important branch of machine vision research field,and the technology that moving target detection based on video is used for the life,economy,military,and many other fields,so this paper use the video detection technology complete the part of vehicle counting in the intersection.This paper chose the interframe difference method as the main method of vehicle detection,in the process of detection,the traditional interframe difference method was improved,increased in the digital image processing technology,such as: image enhancement,morphology processing,region extraction,edge detection,and so on,and setted up many threshold,greatly increase the accuracy of the vehicle counting,solving the problem that the traditional interframe difference method cannot detect the vehicle that run the too slow or stop on halfway.In the part of the light control,this paper studies of single intersection,proposes a dynamic timing algorithm based on fuzzy control.The algorithm can get the real-time vehicle number as the input variable of fuzzy control,determining dynamically thelength of the additional green time,thus increasing the rate of traffic,reducing the average delay time of the vehicles in the intersection.Then this paper studies of some intersections based on the result of single intersection,make the result of single intersection and the vehicle number in adjacent intersection as the input variable of secondary fuzzy control determining the delay time,thus realizing real-time dynamic of multiple intersection signal timing,reducing the average vehicle delay,improving the efficiency of the road.On the software platform,this paper use Visual c + + 6.0 under the Windows operating system to complete the construction of the interface and signal timing simulation,and combining the OpenCV complete vehicle counting based on real-time video.