Automatic detection of sand ripple features in sidescan sonar imagery

A novel image processing technique, based on fingerprint analysis, is applied to sidescan sonar data in order to automatically extract seabed ripple orientation, wavelength, and defect density parameters. The technique is applied to seabed imagery collected on repeat passes over the same ripple field at headings distributed over 360° in order to evaluate ripple parameter extraction across a range of relative angles between the sonar and dominant ripple orientation. The presence of ripples increases the difficulty of identifying objects of interest in sidescan sonar seabed imagery. The potential for utilizing the density of ripple defects as an objective parameter for the quantification of this difficulty in the context of object detection operations is discussed. Results suggest a number of sidescan sonar data analysis applications that are highly compatible with unsupervised detectors and autonomous mission planning processes.