VESUV - a Computer Vision System for Integrated Traffic Analysis

This paper describes the Computer Vision System VESUV as a video-based system for traffic analysis and control. Using standard video hardware the presented approach enables the rplacement of traditional traffic sensors such as inductive loops or radar sensors which deliver traffic data in terms of numbers, classification and velocity of passing vehicles or environmental conditions. VESUV focuses on the functional integration of all the formaer types of sensors into one: the camera. Furthermore, VESUV provides additional features for route tracing which is realized by connecting locally acquired data at central traffic control stations. VESUV was successfully tested in a field trial on a German highway. Existing conventional traffic analysis systems provided the reference data. The promising results demonstrate that online object tracing can be realized with a minimal amount of error. As the first of its kind, the introduced system provides route-tracing information for traffic control systems.