Volumic segmentation using hierarchical representation and triangulated surface

Abstract

This article presents a new algorithm for segmenting 3D images. It is based on a dynamic triangulated surface and on a pyramidal representation. The triangulated surface, which can as well modify its geometry as its topology, segments images into their components by altering its shape according to internal and external constraints. In order to speed up the whole process, the surface performs a coarse-to-fine approach by evolving in a specifically designed pyramid of 3D images.