Crowd behaviour monitoring on the escalator exits

This paper presents an approach to detect an abnormal situation in a crowd scene. The proposed approach estimates sudden changes and abnormal motion variations in a set of interest points. The number of tracked points of interest is reduced by using a mask that corresponds to the hot areas of the built motion heat map. Optical flow technique tracks the points of interest. There are sufficient variations in the optical flow patterns in a crowd scene when there are cases those showing abnormal situations. Statistical treatment of optical flow information has been thresholded. To demonstrate the interest of this approach, we present the results based on the detection of collapsing events in real videos of airport escalator exits.