Seabed Sediment Classification based on Multi-features Fusion and Feature Selection Framework

An improved framework used to classify sediment using the signal reflected by the seafloor is put forward in this paper. The proposed framework is verified by the data collected by 400-kHz Reson 7125 multibeam echo sounder(MBES) in the experiment carried out by California Seabed Mapping Project. After fusing the 21 features extracted firstly from the seabed sediments signals acquired from MBES image, the four most important features which still retain clear physical meanings are selected by the random forest feature algorithm to calculate features importance. Finally, the classification results are obtained by support vector machine(SVM), k-nearest neighbor(KNN) and decision tree(DT), respectively. The results show that the presented method could achieve excellent classification accuracy on different classifiers by combining feature fusion and feature selection.