Sediment Classification Based on Least-Squares Support Vector Machine and Phase-Plane Analysis

A new acoustic sediment classification technique based on least-squares support vector machine (LSSVM) and phase-plan analysis method is presented in this paper. The main task of our proposed technique is to identify the seafloor sediment type in terms of the phase-point distributions characteristics of acoustic echo signals. Our proposed method includes two important issues: Firstly, the phase point distributions (PPDs) of echo data, which are acquired using an echo sounding system from two similar types of seafloor sediment in Yellow Sea Experiment, are analyzed and compared. Then four characteristic parameters are extracted from these PPDs. Secondly, these parameters are directly used as the input vector of the LSSVM. Finally, a LSSVM classifier is trained and applied to an experimental data set. And the classification results demonstrate its feasibility in classifying the sediment types.