Shadow Detection Method in Video Surveillance System
Li Zhang, Changjun Wang · 2010
A shadow detection approach based on the proposed horizontal and vertical histograms was established. The method firstly constructs the horizontal and vertical histograms and gets the clustering centres of the pixel position in the upper human body, and then two groups of thresholds are chosen. Two initial shadow detection images are gotten based on the two groups of thresholds. Final shadow deletion image is attained by fusing the two initial shadow detection images. The method doesn't depend on the image texture characteristics and has reliable results. The approach lays the foundation for the human tracking. These merits prove that the method would be a potentially significant contribution to the active area of shadow deletion in the video surveillance system.