Automatic maritime surveillance with visual target detection

In this paper an automatic maritime surveillance system is presented. Boat detection is performed by means of an Haar-like classifier in order to obtain robustness with respect to targets having very different size, reflections and wakes on the water surface, and apparently motionless boats anchored off the coast. Detection results are filtered over the time in order to reduce the false alarm rate. Experimental results show the effectiveness of the approach with different light conditions and camera positions. The system is able to provide the user a global view adding a visual dimension to AIS data.