Underwater Target Feature Extraction and Classification Based on Gammatone Filter and Machine Learning

Abstract

Underwater target radiated noise feature extraction and classification are important issues in underwater acoustic applications. In this paper., feature extraction is processed based on Gammatone filter and the target classification is processed using machine learning (ML). From the processed result of the real underwater target data, it showed that Gammatone filter is an efficient way to do feature extraction and it also has better classification accuracy compared with some other feature extracting methods. It also showed that machine learning is an efficient tool when applied in underwater target radiated noise classification where the assignment is a label to given input value.