A real-time event classification system for a fibre-optic perimeter intrusion detection system
Seedahmed S. Mahmoud, Jim Katsifolis · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2009
The most important challenge for distributed fibre-optic intrusion detection systems is to minimise the nuisance alarm rate without compromising the system sensitivity. Event classification and discrimination is a powerful tool which can be used to minimise nuisance alarm rates whilst maintaining optimum sensitivity and probability of detection. A novel event classification system using supervised neural networks together with a level crossing based feature extraction algorithm is presented for a fence-based fibre-optic intrusion detection system. Performance results are presented showing accurate classification of and discrimination between different intrusion and non-intrusion events such as fence-climbing, fencecutting, stone-throwing and stick-dragging.