Passive Sonar Automated Target Classification using Deep Hierarchical Feature Learning Approaches

Abstract

Oceans cover a significant expanse of our planet than it is covered by the landmass. In all five newlinedominions where human endeavours take place, land, sea, underwater, atmosphere and space, newlinethe underwater activities are the most hidden and perhaps the most difficult. Obviously newlinenaval forces take advantage of the covertness offered by the sea to carry out operations that newlineare otherwise difficult to execute in open territories. The element of surprise and stealth newlinemake underwater warfare of paramount interest among the world s leading navies. Since the newlineearly efforts in probing the oceans, acoustics has remained as the predominant mode. The newlinesingle most ubiquitous equipment referred to as SOund NAvigation and Ranging (SONAR), newlinethe underwater equivalent of RAdio Detection And Ranging (RADAR), in its many forms newlinehelped and is continuing to help in exploring the depths of the oceans. newlinePassive acoustic target recognition stood at the vanguard of underwater acoustic research newlinefor several decades in the past while considering naval defence scenario and might continue newlineso for the coming decades. Passive acoustics plays a crucial role in Naval Non Co-operative newlineTarget Recognition (NCTR) systems, especially in Anti-Submarine Warfare (ASW) by virtue newlineof its tactical advantages. The target classification processes were historically performed by newlinetrained sonar operators all the way from the passive listening tubes to the modern digital newlinesonar console. In a modern strategic scenario, the human factors are the major limiting aspect newlinethat compromises the endurance and performance of any system. Unmanned systems are newlineincreasingly being preferred in all defence verticals as well, due to their low operational cost newlineand reduced risks of collateral loss.