Dynamic model behavior analysis of small groups based on particle video

Video surveillance is becoming more and more significant in the detection of abnormal events for public security. As a usual kind of crowd activity mode, the research on the motion analysis of small groups is under increasing attention. This paper presents a particle video-based abnormal behavior detection method of small groups. First, use a particle tracking algorithm to obtain the video stream of particles. Since the particle stream contains too much redundant information in the same group, therefore propose to use the longest common subsequence algorithm to make trajectory clustering in order to obtain main information. Then use location information to build effective particle dynamic model, which means changes in the state space. Finally, Hidden Markov Model is used for small groups of abnormal behavior detection.