Traffic Incident Detection And Its Application Based On Video Technology

Intelligent transportation systems, it is that computer vision technology and digital imageprocessing technology are applied to the transport system. In this article: At first,theknowledge of computer vision and digital image processing have been described; Then, onthe basis of computer vision and digital image processing, the research of imagepre-processing, video-based motion-vehicle detection and motion-vehicle tracking isdescribed in detail; Finally, on the basis of the video vehicle detection and the video vehicletracking,it research and analysis the four kinds of traffic incident detection algorithms.In front of the image processing, some preparation work is done at first. It is a necessarypart, because the images are affected all aspects in the before of the acquisition and the way oftransfer. It is an important part of the vehicle detection systems and vehicle tracking system,by making these preparations, it provid a good basis for subsequent testing work.Vehicle detection, through the research and analysis of existing moving target detectionalgorithm, the paper presents a mixed background frame difference method. It is proposed atthe set of the advantages of background subtraction and frame difference method. andultimately it can detect motion-vehicle well. The simulation shows that the system has goodreal-time performance and accuracy.Vehicle tracking, through the describe of the existing moving target tracking algorithm indetail, this paper presents a moving vehicle tracking system based on improved Kalman filter.The crux of it is that, through the establishment of inter-relation matrix, raise the idea oftracking of "prediction-match-fix" to track the five different movement states. The greatestadvantage of this method is that, for the occlusion problem of multi-target tracking, it is agood solution. Simulation results show that the system has good real-time performance andaccuracy.For the detection of events, based on the video vehicle detection and the video vehicletracking, this paper analyzes the four kinds of video-based traffic incident detection algorithms, these four algorithms are as follows:①detection algorithm of vehicle direction,②detection algorithm of vehicle retrograde,③detection algorithm of vehicle parking,④detection algorithm of speeding. These results show that the algorithm can detect these eventsout.