Dynamic Background Modeling for Foreground Segmentation
Shaoqiu Xu · 2009
This paper presents a dynamic background modeling approach for foreground segmentation. The classification between foreground and background is based on Bayes decision rule. The posterior probability of a pixel being observed as a background or a foreground is directly estimated based on the occurrence frequency of its quantized version. Experimental results show that the presented method can be performed in real time and has good performance in complex and dynamic environments.