Otolith image analysis by computer vision

Otoliths are small stone located in fish inner ears and characterised by an accretionnary growth. They act as a biological archive and are of much use in marine biology and ecology. In this article a computer vision framework is presented which recover the successive shapes of the otolith and the significant ridges and valleys from a 2D grayscale image. Seeing vision processes as complex systems, an iterated process is presented using two perceptual information, contrast and good continuity. The successive concentric shapes of the otoliths are recovered as the level-sets of a dome shaped potential function, computed in a variational framework. Potential applications includes in particular fish age estimation, otoliths morphogenesis modelling, otolith proxy fusion.