Development of an underwater target classifier using target specific features

Abstract

In Sonar, the detection and estimation functions are performed by signal processors, which involve the computation of various statistics, for enhancing the overall performance of the system. This also takes into account all the undesirable propagation effects caused by the underwater channel. Underwater targets can be classified by using certain target specific features such as target strength, target dynamics, and the signatures of the noise generated by the targets. Rough identification of the targets is carried out with target strength values at known aspects while for precise identification, classification clues from target dynamics and target signatures are generated. Databases for the engine noise spectra of various underwater targets, propeller noises, machinery noises and cavitation noises, speed-noise characteristics, etc., have been developed. The signal energy estimated within a finite-time interval is compared with the earlier detection/estimation decisions, which are stored in the target data record and the relevant target data are updated. The algorithm for identification of target from the most matching signature patterns in the database will generate the classification clues, which will help in target identification. Salient highlights of an underwater target classifier using the above-discussed target specific features are presented in this paper.