Seafloor Classification based on Sub-bottom Profiler Data using Random Forest

Abstract

Research on Seafloor classification and recognition is of great significance. Sub-bottom profiler is a kind of equipment used to prospect seabed shallow profiler. Its echo signal has rich sediment characteristics and can be used for seabed sediment classification. Based on the data characteristics of dual frequency sub-bottom profiler, this paper accurately extracts the echo intensity sequence containing rich characteristics of seabed sediments by using high-frequency data and threshold detection method. Dynamic and static methods are used to collect sediment data of gravel, silt and cement in the lab testing pond, The dynamic data set, static data set are classified by random forest algorithm. The classification accuracy of the optimized classification model is 98.15%, 85.67%. It proved that this method can be effectively used in the classification of seabed sediment.