Clutter map detector for active diver detection sonar

Abstract

This study presents a temporal approach based on the clutter map for the automatic detection of divers in the active sonar system. The noisy and non-stationary surrounding environment of shallow water makes the statistical parameters estimation of classical constant false alarm rate (CFAR) detectors more difficult. In the authors' system, they take advantage of the fixed positions of the transmitting and receiving arrays in order to improve the statistical parameters estimation over time. The resulting detection algorithm is analysed on simulated data for slow-moving targets and compared with classical spatial CFAR techniques on real data. The authors show that the proposed technique based on the clutter map provides better results than the classical CFAR techniques, particularly due to a reduction in false alerts. They finally analyse the spatial distribution of false alarms for which randomness is important for the performance of the tracking algorithm. They show that this distribution is more random with the clutter map detector than with classical CFAR techniques.