Detecting the Degree of Anomal in Long Duration Videos for Surveillance

Abstract

The method to discriminate anomalous image sequences for efficiently watching monitoringvideos is proposed. Considering of applying systems composed of many monitoring cameras, the methodis required which is independent of the camera setting environment and the contents of the videos. Wepropose a method that can discriminate anomalous image sequences for more efficiently utilizing securityvideos. Considering the wide popularity of security cameras, the method is independent of the camerasetting environment and the contents of the videos. We use the spatio-temporal feature obtained byextracting the areas of change from the video. To create the input for the discrimination process, wereduce the dimensionality of the data by PCA. Discrimination is based on a 1-class SVM, which is anon-supervised learning method, and its output is the degree of anomaly of the sequence. The methodis applied to videos that simulate real environments and the results show the feasibility of determininganomalous sequences from security videos.