Research On Freeway Vehicle Detection And Tracking Based On Airborne Video Camera

The technology of freeway vehicle detection and tracking based on airborne video camera has wide applications and great significance in contemporary freeway management, which can implement traffic monitoring in large area on the freeway. The paper studies vehicle detection and tracking technology based on the video of the aerial photography. The main contents of this paper include the following aspects:(1) Research of freeway identification technology. Because the two lines of freeway are approximately parallel and freeway width is constant, thresholding segmentation method, boundary detection algorithm and line detection algorithm are combined to detect freeway boundary. Finally we extract the region of interest prepared for the follow-up vehicle detection and tracking.(2) Research of vehicle detection technology. In order to solve the problem of background movement, the paper firstly studies Vibe background updating algorithm. In view of the shortcomings of the fixed threshold of Vibe background updating algorithm, the paper improves new Vibe self-adaptive threshold background updating algorithm. Distance threshold can adaptively update with the change of the sample value in the background model and background learning rate can adaptively adjust the speed of the background update with the change of the noise. Noise in binary mask images processed by our paper decreases compared with Vibe background updating algorithm. Then, binary mask images are optimized. Finally we extract the contour of vehicles.(3) Research of vehicle tracking technology. The paper firstly studies Camshift algorithm based on color information. Because color information is limited to the description of the characteristics of vehicles, vehicles can be described by the contour and texture features in addition to color feature. An adaptive fusion tracking method based on color feature, contour feature and texture feature is proposed by the paper. Finally we realize vehicles of different colours and shapes tracking lastingly and stably.(4) Software design and implementation. The moving vehicle detection and tracking software is designed based on the above essential algorithms. This software mainly realizes freeway detection, vehicle detection and vehicle tracking. Finally, we complete the software test from several aspects such as functional test and performance test by several aerial videos.