Research On Vehicle Detection And Tracking Methods For Highway

In recent years, socio-economic development promote the rapid development of road transport facilities. With the increasing number of motor vehicles, city's population and urban areas,urban development and economic growth are in some trouble. Intelligent transportation systems have been paid more attention because of its easeing traffic congestion, improving traffic efficiency to great extent,while video-based vehicle detection and tracking plays a key role in the intelligent transportation system. In this paper, vehicle movement detection, tracking in highway for the research background, we have a in-depth research in the key algorithms and techniques which are on the basis of research results at home and abroad , seeking to explore a method . based on simplified video data, can detecte and trackof highway vehicles fast, effective and Robust .The main work of the paper is as follows:Based on the comparison of advantages and disadvantages among several classic detection algorithms of moving cars , then an improved mixture Gaussian Background model is proposed to motor vehicle detection, and then.Bayer color filter array is chosen for background modeling to reduce the computational time; the characteristic of shadow in HSV space is analysed to remove the shadow from foreground image.,which can extract moving vehicles area from the video of highway effectively.Using morphological filter, reduce the noise of obtained binary images, connecting the splitting vehicle area and filling the empty area, so that,to get a more accurate vehicle movement areas,On the basis of which, label the connected components of the obtained vehicle regions, to extract the minimum bounding rectangle of vehicle.By improving the traditional Camshift algorithm, tracking window is automatically initialized by detected vehicle region to track vehicle; and a Camshift vehicle tracking method based on Kalman prediction is presented, improving the efficiency and precision of tracking and solving the problem of large area of color interference and occlusion. So,this method can track the vehicles on highway robustBased on the above work, highway vehicle detection and tracking prototype system was designed .Not only can easily test for this algorithm, but also get some traffic data by preliminary processing the traffic video on highway.