Underwater acoustic target classification based on modified GFCC features

A major challenge for underwater acoustic target classification relates to significant performance decrease in complex underwater environment. Recent researches have shown that the auditory feature extracted from Gammatone filter has remarkable ability on robust speaker identification. If this remarkable ability can be simulated, the accuracy of underwater acoustic target classification will be improved significantly in noisy underwater environment. Aiming at this purpose, a novel implementation of the Gammatone filter-based feature is proposed and applied to underwater acoustic target classification in this paper. Support Vector Machine (SVM) is used as the classifier in our experiments. Classification results indicates that the proposed feature, namely Modified Gammatone Frequency Cepstrum Coefficients (MGFCC) features are more robust than conventional acoustic features in underwater acoustic target classification.