Computer-Aided Security Surveillance : design of the Quo Vadis Object Tracker

In recent years, use of video surveillance in public places has skyrocketed due to increasing security concerns. To help in coping with the task of monitoring this huge amount of data, a software system to track people and objects seen in order to be able to detect possible threats such as left luggage would be useful. We examine the issues involved in the design of com- puter vision software systems for people and object tracking, and present the design of a system that can track people and objects in an indoor environment, and detect their inaction. The tracking algorithm is based on a combination of matching overlapping regions of motion and matching colour profiles. It is able to follow multiple people or objects simultaneously and still tell them apart after occlusion. Experimental results will be presented to demonstrate the system’s performance.