Real-time video surveillance by a hybrid static/active camera mechatronic system

In this paper we present an effective real-time video surveillance system for real-life outdoor surveillance scenarios. The system is an integration of two subsystems: the static camera and moving one. The approaches employed address properly the challenges that might arise in a typical outdoor scene, such as local and global lighting changes, variations in objects' appearance and occlusions. Our aim is to detect and follow abnormal behaviors, which may occur when vehicles and pedestrians interact typical urban environment. Static camera subsystem operates multiple object tracking and classification. In case that an object of special interest is identified, the operation of the moving camera subsystem initiates, in order to track this object and record its activity.