A trend in observation systems for surveillance is to go from large single platform static systems to smaller, possibly mobile, systems with more adaptable complementary sensors.

Approaches usually focus either on intelligent networks of relatively simple sensor capabilities and accompanying simple processing methods or on sophisticated ‘smart’ sensors with accompanying sophisticated detection, tracking and recognition algorithms connected through simple networks that are used to combine the information.

Promising is the research being done on fusing measurements from sophisticated dissimilar and distributed sensors while tracking. This puts higher demands, however, on the bandwidth and latency of the network connecting them.

Besides intelligence, like detecting unwanted behaviour, tracking people in crowds or recognising vehicles or faces, also adaptability to changing circumstances or changing information needs is desired. Sometimes tracking of a number of people is much more important than detecting unwanted behaviour or vice versa.

The requirement of optimal use of bandwidth and adaptability of the system calls for algorithms that are capable of selecting the most valuable measurements, originating from the local sensors, and /or adapting the processing of these measurements based on the current information needs of the system.

The project investigates the applicability of various techniques and develop algorithms for a specific or limited set of specific applications.

Results
stories
publications

PhD Student:Eelke van Foeken
Promotor:Prof. dr. ir. Frans C. A. Groen
Internal Supervisor: Dr. Leon Kester