Abandoned object detection using thermal and visible band image fusion

Packages that are left unattended in public spaces are a security concern and timely detection of these packages is important for prevention of potential threats. However continuous surveillance of such places is difficult. Therefore, it is important to aid operators with automatic detection of abandoned items. Most of the methods define abandoned items as items newly added to the scene and stayed stationary for a predefined time. Hence other stationary objects, such as people sitting on a bench, are also detected as suspicious objects resulting in a high number of false alarms. These false alarms could be prevented by discriminating suspicious items as living/nonliving objects. In this paper, a system using a thermal camera in addition to a visible band camera is proposed. By analysis of information from different cameras, people and the objects left behind can be detected separately to reduce the false alarm rate.