Research On Vehicle Detection And Matching Tracking Based On The Video Image

Vehicle detection and tracking is an important component of intelligent transportationsystem, which provides important data basis for subsequent traffic control and management.Compared with the traditional system, vehicle detection and tracking system based on videohas significant advantages, such as intuitive, anti-interference, wide range of detection andlower cost etc. With the rapid development of information technology, the research on vehicledetection and tracking method based on video has been focused on by many researchers.In this thesis, the current development of video vehicle detection and trackingtechnology has been introduced, and the structure of the whole test system framework hasbeen given. Combined with the related knowledge of image processing, some key technologyof the video sequence are studied.In the aspect of vehicle detection, the background extraction methods (such as, statisticalmethod, average method and the median method) and vehicle extraction method (such as,optical flow method, the frame differential method and background differential method) arestudied. Because the vehicle information contains a lot of shadow information by abovemethod, a algorithm based on shadow feature is used to remove the shadow in this paper.First of all, the direction of shadows is determined. For the special case of dark vehicle, thedirection of shadows is put forward by using the variance of the image boundary grey value.Then the grey value distribution range of vehicle shadow is calculated so as to remove theshadow. After that, the outline of the moving vehicle is obtained by using the knowledge suchas connected component and image morphology.In the aspect of vehicle tracking, through research on several common methods oftracking, a kind of matching tracking method based on the regional gray is adopted. By usingthe area of the vehicle, the scope of the match is reduced so as to identify the matching areaeffectively. It is shown that the vehicle matching tracking based on regional grey has goodperformance.