Research On Multi-target Detection And Intelligent Control System In Traffic Video Stream

With the rapid growth of the number of vehicles in the city,the traffic pressure at intersections gradually rises,and the improvement of the performance of the traffic light control system has a great effect on improving the efficiency of vehicle traffic at the intersection.This paper takes the intelligent control of traffic lights at single intersections in urban roads as the research direction,and studies a traffic light intelligent control system based on traffic video streams.In the research,the vehicle multi-target detection algorithm based on video stream is realized,and the traffic light fuzzy control algorithm model is established.The main work of this paper includes:1.Solve the problem of detecting multiple vehicle targets at intersections.This paper implements an improved vehicle target detection algorithm based on YOLOv3.Through the analysis of the characteristics of the vehicle detection task,the original structure of the prediction network is optimized,and the detection channel of the 8 times down-sampling feature map is deleted,and the algorithm operation speed is improved.Based on the idea of spatial pyramid pooling network,the SPP module is added to the feature extraction network structure of YOLOv3,so that the local and global features of the original feature map are retained,and the feature map information is more abundant.The experimental results show that the improved YOLOv3 algorithm has a better effect on vehicle target detection when the calculation speed is about the same.2.A method of fuzzy control of traffic lights is studied.Based on the principle of fuzzy control of traffic lights at a single intersection,this paper designs a fuzzy controller that takes the number of vehicles waiting to pass in the lanes in different phases as the input and the green light extension time as the output.The fuzzy language and domain for evaluating the number of vehicles to pass and the length of green light extension,as well as the variable membership function and control rule query table are established,and the establishment of the traffic light control model based on the fuzzy control algorithm is completed.3.Use VISSIM and MATLAB software to simulate the timing control strategy and fuzzy control strategy of traffic lights.VISSIM software carries out modeling and data output of intersection scenes,and MATLAB software carries out traffic light dynamic timing algorithm implementation..The simulation results show that the traffic light control method using fuzzy control strategy can greatly alleviate the traffic pressure at the intersection,which is beneficial to solve the congestion phenomenon in urban traffic.