Event detection based on "common fate" principle: application to vehicles detection from aerial sequences of road traffic

This paper introduces a vision system for road traffic surveillance from sequences acquired from an unmanned aerial vehicle (UAV). During the navigation of the UAV, the vision system acquires sequences which are treated in order to detect vehicles. The dynamic behavior of the UAV-camera system makes a fixed background impossible : we present a new approach based on the "common fate" principle : image primitives which have the same type of movement (or displacement) are grouped. The detection of vehicles is then based on the spatiotemporal grouping of primitives formulated as a normalized cuts problem. A verification step based on the Dempster-Shafer theory is also proposed in order to recognize vehicles.