The vehicle speed detection is an important research of the Intelligent Transportation System, i.e. ITS. And the vehicle speed detection based on video is very popular because of its low costN,easy to use. So, we let the vehicle speed detection based on video be the important content to research.The camera calibration algorithm and vehicle detection and tracking algorithm are researched in this paper. While in the part of vehicle detection and tracking, firstly the traditional methods are introduced, then according to the defect such as be easily affected by environment changing and processing delay, we propose a background difference method fusing the background updating in rear-time and virtual test line, which can offset the defect mentioned above to some extent. To improve the speed detection accuracy further, a new vehicle tracking method based on SIFT corner detection and matching are proposed and implemented after the characters of some corner detection methods are analysed, which can avoid the process of ground detection and enhance immunity. |