In this paper, we propose a smart security camera system for the automated detection of snatching incidents in crowds. Although half of all snatching incidents happened in the presence of other people and witnesses exist, this has not been considered in previous studies. Thus, SAMC (snatching action model in crowds) is presented in this paper for the automated detection of snatching incidents in crowded areas. Certain characteristics of motion are determined from video streams, and using a nonlinear support vector machine the system automatically classifies the situation of the video streams into criminal or non-criminal scenes. We consider many types of scenarios to perform experiments regarding snatching incidents in crowds. The experimental results show that the system can effectively detect snatching incidents with an accuracy of 96.87%.
J-STAGEがリニューアルされました！https://www.jstage.jst.go.jp/browse/-char/ja/
