A model of an intelligent video-based security surveillance system for general indoor/outdoor environments
Wayne Chelliah Naidoo, Jules‐Raymond Tapamo · 2008
Over a decade ago, simply recording a few minutes of CCTV footage required special hardware. Today, with the emergence of new sensors and improved processing hardware, a relatively inexpensive personal computer can process and store video in real-time, which fundamentally enables our research. Automated visual surveillance is poised to be a key technology in the fight against crime, particularly in monitoring security sensitive areas. A significant advantage of this technology lies in its non-intrusive nature in multi-target tracking. In this paper, we present an automated attention mechanism that allows for the operation of vision-based surveillance systems in a wide variety of environments typical of general indoor/outdoor settings. Different applications of our system are demonstrated including real-time abandoned luggage detection and general outdoor person/vehicle classification.