Suspicious Loitering detection using a contour-based Object Tracking and Image Moment for Intelligent Video Surveillance System

Video surveillance provides for the safety of the people in the public environment by monitoring unusual events. This system only monitors the scenario but don't detect the suspicious events occur and not to prevents unusual activities. Hence, this system is essential to upgrade and adapt the intelligent techniques that automatically track and detect the suspicious loitering person in the surveillance. The aim of this paper is to propose a technique for loitering detection based on the contour features and contour-based tracking method. First, foreground objects are segmented using the frame difference method. Identify the static objects from detected objects and thereby compute the centroid using image moments. The frame threshold detects the loitering person by tracking the trajectory of the centroid coordinates through a certain period of time. The benchmark dataset and the real-time own dataset videos are utilized for testing to evaluate the efficiency of the system. The experimental result shows that the proposed method archives high detection rate.