The Detection And Tracking Of Moving Vehicle

The traffic information collection technology plays an important part in the intelligent transportation system. The technology based on the video has a much brighter future for the traffic information processing system due to its wide detection scope, more flexible setup, easier adjustment, convenient install and maintenance and richer traffic flow parameters.The parameters of the moving vehicles are originally detected in the natural scene. Since the background is not only complex but also some regions of it are similar to the objects of interest, it is very difficult to update the background in detecting moving objects. Meanwhile, methods such as background subtraction, template matching and feature extraction can hardly extract the moving objects accurately in a long period of time due to the scale and the orientation of the object are likely changed. According to continuity of gray-value of targets in inter-frame, the frame-difference method is proposed to detect the moving objects in this paper. In order to realize the reliability of the vehicle identification and tracking, a novel algorithm is proposed for tracking the boundary of a moving vehicle by using the fast level set method. The moving object information and pixel reliability are combined to improve the veracity of fast level set method. The efficiency and accuracy of the algorithm is verified by the experiment results.Finally, the moving objects are tracked by using the position of the object contour, the area information and the external moment. Motion estimation is useful to forecast the motion at next time. By integrating motion estimation into motion detection, the latter will be able to be more accurate in the future. In this thesis the motion estimation method based on kalman filter is implemented. Experiment has shown that this method can effectively track the vehicle.