Abstract
Deployment of image processing in traffic surveillance is problematic, because inductive loop detectors outperform video based solutions in respect of accuracy and reliability. Here, we discuss the use of wide area video surveillance for tasks like measuring wide area traffic flow parameters, where inductive loop detectors cannot be deployed. Among these, traffic density is essential for determining traffic state. Wide area traffic surveillance by means of image processing is in principle feasible, but complicate because of occlusion problems. An approach for occlusion robust measurement of traffic density is described in this paper, which is based on motion measurements and the evaluation of the continuity equation.