Detecting Pedestrian Abnormal Behavior Based on Fuzzy Associative Memory

Visual analysis of human motion in video sequences has attracted more and more attention to computer visions in recent years. In order to indicate the pedestrian movement in Intelligent Security Monitoring System, an articulated model of human is presented. According to the contour angle movement of bodypsilas major organs, a fuzzy function is designed. Fuzzy Associative Memory (FAM) is proposed to infer abnormal behavior of the walker. The overall degree of the anomaly is resulted from the fuzzy membership of the pedestrian´s organ using a three layer FAM. In the realization of the system a combined method of centroid and fuzzy discriminant is presented. Fuzzy discriminant can detect irregularities and implements initiative analysis of body´s behaviors in visual surveillance. Therefore, we can recognize some abnormal behaviors and then alarm. The results show that the new algorithm has better performance.