Techniques for Automatic Mapping Coral Reef Habitats by Using Side-Scan Sonar

Abstract

This manuscript presents an image processing procedure for underwater habitat classification at various geological levels, based on the use of extremaly high-resolution sonograph from side scan sonar. We classify successfully the seafloor such as rocky, sandy and rippled sand structure as well as texture in discriminating coral from rocky habitat. This procedure illustrated the ability in mapping complexity benthos assemblage types, and interactions at appropriate scales of geological and biological issues. Multi environmental factors and evidence were found continuously affecting in the study area, which explains the diversification in specific topography. The seafloor classification maps, based on image processing procedure, assessment by optical still images, showing an overall accuracy of 93% in the geological domain (sandy, rocky, rippled sand), and overall accuracy of 59% in rugosity domain (biological coverage, coral types, and rate).