estimation methods for video vehicle tracking systems

One of the major advantages of the video cameras’ usage for tracking of vehicles is to reduce the costs of Intelligent Transport Systems. However, this requires the development of software techniques allowing an automatic extraction of the vehicle or group of vehicles from the current video frame, which is possible by using the background estimation methods, assuming a fixed camera installed over or at the side of the road. Background estimation based on the linear image filtering algorithms can be performed by averaging a certain number of video frames. However, this technique is relatively slow, which complicates its use, especially in variable lighting conditions. The paper presents an alternative background estimation technique, utilised for its further replacement, based on the nonlinear image filtering algorithms.