Variational Approaches in Shape Partitioning

Wednesday 3 May 2017 h.14:30, Room 2BC30
Martin Huska (Padova, Dip. Mat.)
“Variational Approaches in Shape Partitioning"

Abstract
The rapid development of 3D scanning technology has incredibly increased the availability of digital models exploited for a wide range of applications varying from computer graphics and medical imaging up to industrial production. One fundamental procedure that processes the raw acquired data for further manipulation, e.g. in product design, animation, deformation and reverse engineering, is the shape partitioning. This process consists in the decomposition of an object into non-overlapping salient sub-parts determined by a shape attribute.
In this seminar, we will introduce the concept of Shape Partitioning together with the wide range of partitioning methods. Next, we will observe a few partitioning/segmentation models in the field providing some results. At last, if the time allows, we will introduce the concept of Convex-Nonconvex segmentation over surfaces.
The seminar will be held at introductory level, thus, general audience is welcome to participate.