Underwater target classifier capable of reducing self-convolution distortion

Underwater target classification has got numerous applications in ocean engineering and technology. The suitable selection of target specific features is of prime importance, as it determines the efficiency and general performance of the classifier. Equally significant, is the selection of the classifier architecture, as it has a profound effect on the implementation complexity and system behavior. The spectral features, when suitably modified, are capable of providing certain essential clues that can be utilized in the design of underwater target classifiers. A non-stochastic underwater target classifier, which makes use of an improved feature set, under heavily distorted signal environment, is proposed in this paper. The classifier is based on modified Kaiser-Bessel window and makes use of the Matching Parameter (M_P) metric which is a functional measure of the Mahalanobis and Euclidean distances. The system also utilizes an algorithmic vector quantization approach for the formation of clusters. The proposed system reduces the ambiguity in the classification process under heavily distorted and randomly fluctuating signal variations introduced by signal self-convolution processes occurring in nonlinear underwater channels.