Introduction

The 3D-CSC is an inherently parallel hierarchical region growing method for segmentation of three-dimensional color images. The method based on the two-dimensional Color Structure Code (CSC), that was developed by Volker Rehrmann.

The CSC combines the advantages of local region growing algorithms (simplicity and quickness) and global methods (robustness and accuracy). Due to its inherent parallel abilities its running time may be curtailed by using a multi processor computer. Therefore the method is particularly suitable for segmentation of three-dimensional images (e.g. Magnet Resonance Tomography and Computer Tomography).

The two-dimensional CSC operates on a hierarchy of hexagonal islands, that overlap each other. To make the CSC work with 3D-images, a comparable three-dimensional island hierarchy had to be developed first (see next section).