Pedestrian Detection based on Adaptive Background Modeling
Xin Ni · Jisuanji fangzhen · 2009
Pedestrian detection has been widely applied in many fields, which is one of the reserch focus of computer vision. And background subtraction is one of the most popular methods in pedestrian detection. A new adaptive background modeling initialization method is proposed, which computing non-motion pixels and its persistence time in consecutive frames to initialize background model. Moreover, a method of background updating and detection of pedestrian is proposed. Experimental results show that through this method of background modeling, it can get a accurate background image, and update backgroud image quickly, which can adapt for change of illuminate and move-in or move-off object.