Background modeling based on region segmentation

Background modeling is an important problem in automated video surveillance systems. Nonparametric models have promising results. But these models have high computational load and large memory requirement because a large set of background samples is usually needed to model the background. In this paper, a background model based on region segmentation is proposed. An adaptive single Gaussian background model is used in the stable region with gradual changes and nonparametric model is used in the variable region with jumping changes. A generalized agglomerative scheme is used to merge the pixels in the variable region and fill the small interspaces. A two-threshold sequential algorithmic scheme is used to group the background samples of the variable region into distinct Gaussian distributions. The kernel density computation complexity is largely reduced by arranging the computation order of these groups according to their proximity in mean value to the current pixel sample being estimated. Experimental results show that the proposed method is computationally more efficient than existing nonparametric model, but achieves a comparable result.