Extraction of traffic information from images at DEIS

This paper reports about the research work currently going on at DEIS, University of Bologna, on the use of computer vision for extraction of information from urban traffic images. Two algorithms are described: the first, named “virtual inductive loop” (VIL), computes image differences on small windows of the traffic scene between an updated background and the actual window and is specific for vehicle counting and detection. The second algorithm, based on the block matching algorithm (BMA), is particularly suitable for vehicle motion detection and tracking. It consists of a BMA implementation (improved with respect to the standard version) followed by a regularisation step based on a custom median filter, of a labelling algorithm and of a tracking step. Finally, we present our future plan for the research developments concerning these topics