Comparisons of crest line extraction algorithms applied to marine dunes

Marine dunes are bedforms that, because of their locations and dimensions, can infer with human activities at sea. New MultiBeam Echo-sounder Systems (MBES) enable to collect always more accurate and denser data revealing the diversity of shape, size and dynamics of these features with finer details. Until now, research on dunes have mostly consisted in describing or modeling their morphology and migratory behavior. The development of new automatic tools allowing to quantitatively describe their morphology and dynamics seems necessary. The dune crest extraction is the first step towards the automation of marine dune analysis. Crest detection algorithms relying on differential geometry concepts were designed for discrete surfaces (meshes). In this paper, we endeavor to quantitatively compare the results of four of these algorithms applied on marine dunes and conclude on their applicability for dune crest detection.