AUTOMATIC TARGET CLASSIFICATION USING UNDERWATER ACOUSTIC SIGNALS

In this study, automatic target classification was performed using underwater acoustic signals. The targets were classified using the Low Frequency Analysis and Recording algorithm, which is used for detecting underwater and surface ships from long distances and for target recognition by analyzing their noise. While the proposed method uses 6 different frequency ranges separately for target classification, instead of a single frequency range, comparing to existing methods, it presents an innovative approach by presenting a LOFAR image to the SONAR operator in 7 different frequency ranges. Additionally, LOFAR data is visualized by the t-SNE algorithm and it is observed that grouping the ships based on noise characteristics forms a more compact clusters compared to groupings based on the sizes of the ships. We show that the proposed approach improves the performance of target classification.