Scale space methods for recognition of vehicles

Abstract

The demand for transportation, especially the demand for cars, is still increasing worldwide. Correspondingly, the number of traffic jams is increasing, too. However, jams often occur although the maximum road capacity has not been exceeded. Therfore, it is important to obtain more detailed data about the state of the traffic to bei able to, e. g., adjust electronic velocity signs and traffic lights in an appropriate way in order to provide a higher traffic flow, and to reduce the number of jams from nowhere. An appropriate way to get detailed traffic data, especialy to receive data about the routes of drivers, is to use machine vision. In co-operation with universities and a number of smaller and larger companies from Berlin and Brandenburg (OptecnetBB and FAV, we are interested to monitor the entire city of Berlin with its about 40 thousand cars usually travelling at the same time. Three possibilities are of interest: The first one is to use a large number of video cameras fixed at high buildings, the second one is to receive data from above, e. g., using SAR on an aeroplane, and third, in future, to use satellites. In the following text we will concentrate on the first possibility.