Traffic Incident Detection Method Based On Vehicle Detection And Tracking

Establishing Intelligent Transportation System(ITS)is the main development trend of most cities in the future.Traffic incident detection in the crossroad which has become the research direction of many scholars is the key and difficult point in it.Therefore,these article focuses on the three aspects about vehicle detection,vehicle tracking and traffic incident detection at crossroad,and puts forward the corresponding technical methods.The main work and achievements are as follows:First,vehicle detection is the basis of subsequent vehicle tracking and detection.To address the problem of vehicle occlusion,light changing and other factors on the detection,proposes a new motion detection algorithm,which combined the sensitivity of frame difference method and the accuracy of the ViBe algorithm,can accurately extract the foreground from suvillance video.On the basis foreground extraction,apply a fast license plate detection algorithm,and locating vehicle accords to algorithm result,finally complete the vehicle detection.Second,vehicle tracking is the core of the whole process.According the feature of complex environment and serious vehicle occlusion of crossroad,proposes a vehicle tracking algorithm based on compressed sensing,which can effectively deal with the change of vehicle size effectively and updating model adaptively.On the other hand,according result of fast vehicle occlusion decision algorithm,the ORB feature points matching algorithm is used to assist tracking to prevent losing problem when the vehicle is seriously covered.Thrid,traffic incident detection includes vehicle violation detection and traffic parameters calculating.First,establish mathematical model for calibration data,for example,lane intersection,as well as paramerizing vehicle tracing results.And then puts the vehicle violation situation according the illegal judgement formula.Finally calculating the traffic parameters.