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

Visual surveillance especially for humans and vehiclesis currently one of the most active research topics in computervision. In this paper, a new method is introduced for thedetection of moving objects in surveillance applications. Theproposed method relies on a model assigning a vector of graylevels to every pixel location of the current image. The vectorrepresents information on the neighborhood region of that pixel.Using norm of the vectors in two consecutive frames and theBayesian change detection algorithm, we introduce a novelmethod for moving object detection which is robust to noise andillumination changes. Also, it is insensitive to repeated motions inthe background.