Synthetic Aperture Sonar Image Segmentation using the Fuzzy C-Means Clustering Algorithm

Synthetic aperture side-scan sonar (SAS) provides an imaging modality for detecting objects on the sea floor. It is also an excellent tool for shallow water characterization where immobile, submerged threats would not be detected by conventional forward-looking sonar range-doppler techniques. SAS images provide an image of an object and its shadow, both of which can be used in the classification and localization of potential threats. This document discusses the development of an image segmentation algorithm that was capable of segmenting (detecting) the image of an object and its acoustic shadow in the presence of reverberation noise. As a component of an autonomous deployable active sonar system, no human input was required. An unsupervised form of cluster analysis, the Fuzzy C-Means Algorithm (FCM) was used to implement the segmentation procedure. FCM is a generalization of the classical K-Means or Hard C-Means (HCM) clustering algorithm and the FCM outperformed the HCM in the segmentation of SAS images. Operating in an