Coral shape estimation from corallite textures

Abstract

Coral rugosity is known to be correlated to fish abundance. In this paper we estimate coral shape from single digital images alone using the perspective transformation of corralite patterns as cues for surface reconstruction. Recovering shape information from image can replace current rugosity indexing methods that are tedious to implement. The technique is demonstrated on corals belonging to the genera Favia, Favites, and Porites.