Models of Vehicle Speeds Measurement with a Single Camera

The rapid development of hardware technology make it possible for computers or even micro-computers to handle the problem of traffic information gathering utilizing image/video processing and pattern recognition. In order to measure the speed of a vehicle, several models are proposed in this paper. Firstly, the background subtraction method with a proper background updating model is used to detect the moving vehicle, whose position in the image is then traced with the obtained distinct features. The real position of the vehicle in the world coordination system corresponding to the traced pixel position can then be calculated according to one of the two proposed imaging models in the paper. With the real distance and the time period between successive image frames obtained, the speed of a vehicle can then be calculated easily. The proposed models are used in an embedded traffic information gathering system and results show the convenience and the high accuracy of the proposed methods.