Multiple Sensor Indoor Surveillance: Problems and Solutions

The paper presents the Multiple Sensor Indoor Surveillance (MSIS) project which is a research project at the Accenture Technology Labs. It describes the objectives of the project, the problems it was designed to solve and solutions that have been currently obtained. The project environment includes 32 Web cameras, an infrared badge system, a PTZ camera, and a fingerprint reader. The solutions for the following two problems are described in details. The first problem is how to visualize events detected by 32 cameras during 24 hours. The solution is obtained using self-organizing maps. The second problem is how to localize people using fusion of multiple streams of noisy sensory data with the contextual and domain knowledge that is provided by both the physical constraints imposed by the local environment and by the people that are involved in the surveillance tasks. A Bayesian framework is suggested to solve this problem. The experimental data are provided and discussed