Fine scale analyses of a coralline bank mapped using multi-beam backscatter data

In this work, we have developed a classification technique to characterize the seafloor of the Gaveshani (coralline) bank area using multi-beam backscatter data. Soft-computational techniques like the artificial neural networks (ANNs) based unsupervised self-organizing maps (SOM) architecture is used to determine the existence of six classes. Thereafter, 55 segments were identified for data segmentation, employing six profiles selected from the backscatter maps, using the fuzzy c-means (FCM) algorithms. The gridded map of the segmented seafloor backscatter data was overlaid on the co-registered backscatter map to understand the distribution of the overlying sediment material on the bank.