Real Time Analysis of Situation Events for Intelligent Surveillance

Real time recognition of situation events is important and difficult in intelligent surveillance. A hierarchical framework based on hierarchical events fusion is proposed to analyze situation events. Situation events are decomposed into a sequence of sub-events at different levels based on the hierarchical features of the event and the relations among events at different levels. The hierarchical framework is modeled using a hierarchical dynamic Bayesian network. The corresponding RBPF method is constructed for the inference of the posterior probability of each node in the hierarchical dynamic Bayesian network in order to analyze situation events in real time. Experiments results show that this method can analyze situation events in real time, and achieve better recognition precision and less computation time than the PF method.