Vehicle Detection under Day and Night Illumination

Effective detection of vehicles in urban traffic scenes can be achieved by exploiting image analysis techniques. Nevertheless, vehicle detection in daytime and at night can’t be approached with the same image analysis algorithms, due to the strongly different illumination conditions. This paper describes the two different sets of image analysis algorithms that have been used in the VTTS system (Vehicular Traffic Tracking System) for extracting vehicles from image sequences acquired in daytime and at night. In the system, a supervising level selects the set of algorithms to apply and performs vehicle tracking under control of a rule-based decision module. The paper describes the tracking module, and reports experimental results for both vehicle detection and tracking.