A New Illumination-Invariant Method of Moving Object Detection for Video Surveillance Systems

Visual surveillance especially for humans and vehicles is currently one of the most active research topics in computer vision. In this paper, a new method is introduced for the detection of moving objects in surveillance applications. The proposed method relies on a model assigning a vector of gray levels to every pixel location of the current image. The vector represents information on the neighborhood region of that pixel. Using norm of the vectors in two consecutive frames and the Bayesian change detection algorithm, we introduce a novel method for moving object detection which is robust to noise and illumination changes. Also, it is insensitive to repeated motions in the background.