Cast Vehicle Shadow Segmentation Based on Contour Analysis

This paper presents a novel method for cast vehicle shadow segmentation in a sequence of traffic images taken from a stationary camera mounted on a tall building or pedestrian crossings bridge. Based on a simplified vehicle/shadow joint contour model, which is represented by a polygon, we propose a two-step method: first, estimate the environmental parameters of current traffic scene through analysis of illumination distributions of abundant passing vehicles and, second, extract geometrical information of the contour model to detect the boundary between vehicles and their associated shadows. Experiments on practical real-world traffic-image sequences demonstrate that our method is simple, robust and efficient under traffic scenes with different lighting conditions.