Seabed classification from multibeam echosounder data using statistical methods

Abstract

The development of reliable methods for automatic seabed classification enjoys widespread interest at the present time. In this article, statistical methods for seabed classification from backscatter sonar data are investigated. The classification rule is derived from the Bayes decision rule and involves a probability model of the features extracted from multibeam echosounder data. The features are based on the backscatter distribution, the spectral distribution, and the backscatter-level co-occurence. The authors also present procedures for detection of seabed of unknown type and classification of pixels as a mixture of two different classes. Raw backscatter data from the Simrad EM 1000 Multibeam Echo Sounder are used. The results show that it is possible to differentiate between seabeds of various sediment types.<>