Research On Techniques Of Video-Based Moving Vehicle Detection And Tracking

The moving vehicle detection and tracking are important part of modern Intelligent Transportation System. This paper pays great attention to investigate the techniques for moving vehicle detection and tracking under the condition of a fixed camera based on analyzing and summarizing existing techniques. The following research have been accomplished:(1) Research on moving vehicle detectionThis paper investigates and presents a vehicle detection method based on combining inter-frame difference method and background subtraction method. First, we pick up and update the background by using self-adaptive background updating model based on frame difference method, and then obtain the foreground moving area. A shadow detection method based on HSV color feature is adopted to detect and eliminate vehicle shadow, and then the vehicle's real moving area is obtained. In the end, the external rectangle of a moving vehicle is exactly obtained by using morphological filtering and the method of obtaining external rectangle.(2) Research on moving vehicle trackingThis paper investigates and presents a vehicle tracking method based on combining GM(1,1) model and multi-feature matching. GM(1,1) model is used as motion estimation model to predict the position of moving vehicle, and the position is used for estimation and match of the tracking object in the consecutive frames, and then the trajectory of moving vehicles and the important traffic parameters such as the vehicle speed, the quantity of the traffic flow can be obtained. This paper uses multi-feature matching such as Grey-histogram, size and shape in the matching process in order that the tracking object can be exactly matched.(3) Design and implementation of video-based vehicle detection and tracking systemA video-based vehicle detection and tracking system is designed and implemented based on above vehicle detection and tracking algorithm. The system can detects exactly and tracks efficiently moving vehicles. And the experimental results of above vehicle detection and tracking algorithm can be obtained intuitively.