Automated acoustic classification of sidescan images

Dividing sidescan images into regions that have similar sediments is often done by expert interpretation. Automated classification systems are not widely used at present. This paper describes techniques, based on amplitudes and texture, that lead to useful and practical automated classifications of sidescan and multibeam images. Sediment type affects amplitudes and texture, but so do system operating details and survey geometry. Effects of the last two must be compensated to isolate the effects of sediment type. Images from multibeam surveys are accompanied by bathymetric data from which grazing angles across rasters can be calculated. By compiling tables of amplitude against range and grazing angle, systematic changes in amplitude with these two variables can be removed consistently. Sidescan images have no accompanying depth profile, forcing a flat-bottom assumption that degrades the ability to compensate fully for angular effects. This degradation was studied, using Reson 8101 data from Portsmouth, NH (Fig. 1)