| Now rapidly increasing amount of vehicles makes the problem of traffic jams and traffic accident serious.Road resources can provide to vehicles for convenience is also limited.In ITS,cars need to be identified,tracked.A kind of effective vehicle tracking algorithm can provide fast information for ITS strategy.Automatic event detection and traffic flow monitoring system can collect rich and accurate traffic event information,realize traffic early warning for drivers under conditions,avoid accidents,and reduce traffic congestion.Intelligent Transportation System is a kind of solution.More Than This,vehicle-tracking intel can also be accessable to traffic and also Public security system.The upgrading computers and detection equipment,the coming 5G system make vehicle tracking more reliabile.In this case,a kind of new algorithm is needed.The vehicle tracking algorithm is to identify the vehicles in the video stream,and calculate the recognition results to get the running coordinates and speed of the vehicles.At present,the efficiency of vehicle recognition and tracking can be improved.Aiming at the performance of vehicle tracking system,an improved scheme is proposed in this paper.The main work of this paper is as follows:1.A vehicle recognition algorithm based on hog feature and support vector machine(SVM)algorithm is proposed.By extracting the hog features of the image,the multi-scale and multiple scanning of the image is carried out to identify the vehicle.For the initial recognition results,multi-point fusion method is used to determine the vehicle position.2.A vehicle tracking algorithm based on Kalman filter is proposed.Based on the prediction results of the previous frame and the observation results of this frame,the gain is calculated,and then more accurate positioning is obtained.3.Design and implement a real-time vehicle information monitoring platform based on road video.With the technology presented,collect,statistic vehicle and lane information and show them. |
PDF Full Text Request