A Pedestrian Abnormal Behavior Detection Algorithm Based on the Angle Change of Flow Points

Intelligent video surveillance is of great importance on cracking down on illegal crime, ensuring social stability, and it is the leading edge topic. In order to detect the pedestrians' abnormal behavior, such as fighting, violence and running, a pedestrians' abnormal behavior detection algorithm based on the angle change of flow points in fixed area is presented. Firstly, we get the points by the same step size in the area, calculate the optical flow of the points through the former later computing optical flow point of consistency, the method was new compared with other detecting method; secondly, we calculate the angle change of the same points, and then calculate the histogram of the angle, according to the size of change of angle histogram to determine whether there is any abnormal behavior. Experimental results show that, the method can detect the pedestrians' abnormal behavior effectively.