The Design Of Vehicle Detection System

Now, with the rapid development of society, traffic violations are more and more; people require intelligent detection traffic incidents higher. Because manual work detects traffic video; not only this way waste a lot of manpower and material resources, but also people’s attention and energy are limited in the higher intensity work environment; omissions occur frequently; people can’t be real-time to deal with emergency events in this way. Vehicle detection system belongs to the scope of the intelligent transportation system and has a relatively large social need. So vehicle detection system is designed in this paper.In this paper, the method of the three frames difference and the mixed Gauss model combined with each other is used in the vehicle detection system. The noise of detecting motion vehicles is less than the conventional mixed Gauss model, and the effect is better. The system uses the client/server structure. In the server and client, the main modules are as follows:On the side of the server, first of all, image morphological processes is used, because there will be much noise in the video image, and the noise component causes a lot of effect in the image processes behind the current process; then the method of the three frames difference and the mixed Gauss model combined with each other is used to detect motion vehicles; in the process of detecting motion vehicles, the pixel values of motion vehicle area in the current video frame are set to1, and the pixel values of background area in the current video frame are set to0; thereby the coordinates of the motion vehicle centroid are determined; the movement direction of the motion vehicle is obtained compared with the previous moment video frame; it is judged that whether the motion vehicle runs the red light or not combined with traffic light signals; if the motion vehicle runs the red light, the current video frame is encoded with the JPEG format to capture; the original video stream of the YUV format is encoded to the video stream of the H.264format with hardware compression encoding format, because not only the data of stream media is large, but also the real-time requirements for transmitting video stream is relatively high; the video stream of the H.264format is sent from the server side with RTP protocol and RTCP protocol.On the side of the client, first of all, the video stream of the H.264format is received with RTP protocol and RTCP protocol; then the video stream of the H.264format is decoded to the YUV format; the motion vehicle is detected in the video stream while the motion vehicle is marked with the blue color rectangle; the video frame is broadcasted.In the end, the vehicle detection system is tested and analyzed, mainly including two aspects of the server and the client. In the server, it is successfully detected that whether the motion vehicle runs the red light or not combined with traffic light signals; the video stream of the H.264format could be sent from the server side with RTP protocol and RTCP protocol. In the client, the video frame could be broadcasted while broadcasted video is relatively clear, and the motion vehicle is successfully detected while the motion vehicle is marked with the blue color rectangle.