Bayesian network based multi stream fusion for automated online video surveillance

Video Surveillance is an omnipresent topic when it comes to enhancing security in public places and transportation systems. Fully automated behavior detection systems are desirable when it comes to cutting costs for analysing video and audio streams online. These will initiate an alarm signal autonomously if a possibly dangerous situation is detected. The particular investigated scenario is monitoring passengers' behaviors in aircrafts. In order to work robustly in unconstrained environments many subsystems have to be developed. Though in the last years reliable approaches for required systems have been brought up, there exists a gap between reliability and computational effort. Hence a Low Level Activity representation of behaviors will be presented, which can be detected with so called weak classifiers in real time. These outputs will be interpreted by a highly sophisticated probabilistic Bayesian Network.