Effective Shadow Detection in Traffic Monitoring Applications

Alessandro Bevilacqua
University of Bologna
ARCES-DEIS (Department of Electronics, Computer Science and Systems) - Viale Risorgimento, 2
40136 Bologna


e-mail: abevilacqua@deis.unibo.it http://


Keywords: binary edge matching, gradient analysis, shadows detection, motion analysis, motion detection, visual surveillance, background difference, image division, traffic monitoring


This paper presents work we have done in detecting moving shadows in the context of an outdoor traffic scene for visual surveillance purposes. The algorithm just exploits some foreground photometric properties concerning shadows. The input of the system is constituted by the blobs previously detected and by the division image between the current frame and the background of the scene. The method proposed is essentially based on multi-gradient operations applied on the division image which aim to discover the most likely shadow regions. Further on, the subsequent “smart” binary edge matching we devised is performed on each blob’s boundary and permits to effectively discard those regions inside the blob which are either too far from the boundary or too small. We demonstrate the effectiveness of our method by using a gray level sequence taken from a sunny, daytime, traffic scene. Since no a priori knowledge is used in order to detect, and remove, shadows, this method represents one of the most general purpose systems to date for detecting outdoor shadows.