Abnormal detection based on gait analysis
Chao Wang, Xinyu Wu, Nannan Li, Yen-Lun Chen · 2012
Abnormal behavior detection has recently gained growing interest from computer vision researchers. In this paper, the gait-analysis-based abnormal detection is proposed for walking scenes, where gaits of people are analyzed in all kinds of situations and the gait data are utilized to construct the basic gait model. Walking people in the crowd are tracked and their activities silhouettes are abstracted and compared with the basic gait model. Some of those activities which are significantly difference with the basic gait models are defined as abnormal behavior, where the activities silhouettes and gait models are measured by chamfer distance. The experiments verify that our system could effectively detect several kinds of activities different with walking.