Research On The Technology Of Moving Vehicle Detection And Tracking In Traffic Route Guidance

With the development of science and technology, the development of transportation has also increased rapidly. However, the increase of vehicles is much faster than the growth of roads and other traffic facilities, which causes serious traffic problems, such as traffic jam, environmental pollution and traffic accidents and so on. Moving vehicle detection and tracking technology based on the computer vision is an effective way to solve the intelligent transportation problem. This paper focuses on studying the methods and problems existing in the moving vehicle detection and tracking in the process of traffic guidance. It covers the following aspects:First, the three common principles of the moving vehicle detection methods are elaborated, and the advantages and disadvantages of each method are analyzed. The video image background, under the traffic guidance environment, is static relatively. According to this characteristic, the author mainly researches the detection algorithms of moving vehicles, including the background reconstruction algorithm based on average time, the background reconstruction algorithm based on pixel intensity classification and the background reconstruction algorithm based on gaussian mixture models. After comparing and analyzing the results of the three detection algorithms, the background reconstruction algorithm based on gaussian mixture models is chosen as the method of the detection moving vehicles.Second, to the problem of shadow in the video image captured in the process of detecting moving vehicles, the removing shadow algorithms are studied and a method of removing shadow based on the HSV color space is given. Then the algorithm is tested, and the result obtained is good. Furthermore, a mathematical morphological filtering method based on the expansion-corrosion is introduced in order to renovate effectively incomplete moving vehicle area.Third, the four basic principles of the moving vehicle tracking methods are introduced, and the advantages and disadvantages of each method are analyzed. It lays emphasis on the research of the method of moving vehicle tracking based on SIFT characteristics. The large-dimensional eigenvector based on SIFT algorithm will affect real-time tracking. According to the problem, an algorithm, named SPA, which improves real-time tracking through PCA reducing dimensions on SIFT eigenvectors is presented.Fourth, the experiment principles and steps of the moving vehicle detection and tracking in the traffic flow guidance are described; a flow chart and the experiment result are provided. Meanwhile, the experiment results are discussed and analyzed. Besides, this paper describes the condition that vehicles are occluded in the tracking, and offers some usual solutions.