The Use of Multi-Way Analisys in the Classification Task of Passive SONAR Contacts

Abstract

When a ship sails, it irradiates a particular noise that categorizes it in a specific class. The identification of the ship classes is a non trivial task and it is performed employing the passive SONAR systems. The information processed by the SONAR sensors is usually evaluated by an operator that identifies the contact. Recently, intelligent systems capable to extract important features from the acoustic signal irradiated by ships have been proposed to increase the reliability and speed up the process of decision making. In this work, it is employed a new methodology applied in the process of analyzing and mining databases based in tensor analysis. This methodology showed to be an interesting treatment for this type of problem overcoming other methods.