A new method of shadow detection based on edge information and HSV color information

In the process of shadow detection, the method with HSV color information has a problem: it often mistakes the moving object with dark color for shadow. To this problem, the paper presented an algorithm based on edge information and HSV color information. At first, we made a edge detection at the Region-of-Interest (ROI), then in vertical direction we cumulated the number of the pixel points. Because the edge of both sides of the vehicle was evident, the cumulation usually was the largest. We got the axis of symmetry and the edge of the vehicle by the character. After determining the scope of vehicles, we used the method with HSV color information to detective shadow out of the region of the vehicle. The comparison experiment proves that the algorithm can effectively solve the problem of mistaking moving dark object for shadow. It promotes the accuracy of identifying shadow.