Automatic target recognition for naval traffic control using neural networks

Safety requirements in traffic control stress the importance of techniques devoted to automatic tracking of little crafts in harbor areas. Changeable sceneries and noise variability make hard the recognition task of an automatic target recognition system. In this paper, a modular system based on neural networks for the quasi real time detection of moving targets in seaport radar images is proposed. The neural modules resolve both noise removal and target identification tasks by processing steps including a segmentation, a filtering and a classification phase. The system performances have been evaluated on temporal sequences acquired in a real maritime environment.