Abnormal Event Detection Based on Saliency Information

Abstract

Abnormal event detection is a challenging task in video analysis. In this paper, we propose a new abnormal event detection algorithm for surveillance videos. It is well accepted that human eyes are extremely sensitive to abnormal events and they can quickly pay attention to the locations of these abnormal events in visual scenes. Thus, the characteristics of the Human Visual System (HVS) can be used for abnormal event detection. By exploiting the characteristics of the HVS, we propose an abnormal event detection algorithm based on saliency information. Firstly, the saliency information is extracted from video frames based on the feature contrast. The motion information of video frames is calculated by the multiscale histogram optical flow (MHOF). Based on the features of saliency information and MHOF, the Support Vector Machine (SVM) is used to train and predict the abnormal events in visual scenes. Experimental results show that the proposed abnormal event detection method can obtain much better performance than the existing ones over the public video database.