Abstract
Motion Object Detection is a fundamental and crucial step in the applications of intelligent video surveillance. Various techniques including background subtraction are proposed to achieve good results of foreground extraction. This paper presents a novel background model that relies on neighboring information of pixel. The main innovation concern is that we bring life value concept to each background sample and adopt different update strategies based on different life values of background samples. Neighboring information is used to construct and update the background model. Experiment results on Change Detection 2014 dataset demonstrate that the proposed method works more robustly and outperforms some state-of-the-art methods.