Abnormal behavior detection in crowd scenes
Zheng Xu, Songhao Zhu, Dongliang Jin · 2018
Abnormal activity detection is divided into global abnormal activity detection and local abnormal activity detection. First, the hybrid histogram of optical flow feature is extracted. Then, for global abnormal activity detection, the method of double sparse representation is proposed to tackle this problem in this paper. Finally, for local abnormal activity, by first detecting the foreground of the region of interest of the current frame, the method of online weighted clustering is utilized to detect the local abnormal activity. Experimental results demonstrate the advantages of this method.