Three-Dimensional Medical Image Segmentation Using a Graph-Theoretic Energy-Minimisation Approach

Michael Mainik, Leibniz Universität Hannover, Seminar
06/2008

Segmentation is a problem in many areas of research and application. Therefore many different approaches exist to find regions of interest in images. An interesting class of segmentation algorithms are those that follow the approach of Mumford and Shah and describe the Segmentation by a smooth function, that minimizes a so-called Energy-Functional of the original function describing the image.

This work gives an overview of the method and explains a divide-and-conquer strategy to find a minimum using a neighborhood graph and minimal merging costs for speed-up.

Kontakt: Philipp Blanke