An object detection technique for very high resolution remote sensing images

A problem of major interest for local governmental authorities is the monitoring of the number of cars invading the streets and the open parking lots. Different technologies have been used for such purpose. An emerging one is based on the use of remotely sensed images of very high spatial resolution, in particular those acquired from unmanned aerial vehicles. In this paper, we present a method to count objects in urban environments, with a focus on cars. It relies on an opportune sequence of processing and analysis steps so that to handle images of very few centimeters of resolution. Experimental results on real images are reported and discussed.