A hybrid architecture for intelligent video surveillance
Vincenzo Di Massa, Marco Gori, Igor L.S. Russo · 2005
This paper presents a hybrid architecture for intelligent video surveillance which is able to detect complex events on the basis of a strongly-based learning approach. We describe briefly the main components used for motion detection, segmentation, tracking, and clustering, along with the solution adopted for their hybrid combination. Finally, we emphasize the approach adopted for classifying video sequences which is based on hidden Markov models