Dynamic Background Subtraction Based on Local Dependency Histogram

Traditional background subtraction methods perform poorly when scenes contain dynamic backgrounds such as waving tree, spouting fountain, illumination changes, camera jitters, etc. In this paper, a novel and effective dynamic background subtraction method is presented with three contributions. First, we present a novel local dependency descriptor, called local dependency histogram (LDH), to effectively model the spatial dependencies between a pixel and its neighboring pixels. The spatial dependencies contain substantial evidence for dynamic background subtraction. Second, based on the proposed LDH, an effective approach to dynamic background subtraction is proposed, in which each pixel is modeled as a group of weighted LDHs. Labeling the pixel as foreground or background is done by comparing the new LDH computed in current frame against its model LDHs. The model LDHs are adaptively updated by the new LDH. Finally, unlike traditional approaches which use a fixed threshold to define whether a pixel matches to its model, an adaptive thresholding technique is also proposed. Experimental results on a diverse set of dynamic scenes validate that the proposed method significantly outperforms traditional methods for dynamic background subtraction.