Options
1997
Conference Paper
Titel
Texture-based segmentation for automatic surface reconstruction
Abstract
This paper reports a segmentation pipeline for automatic analysis of multi-modal tomographic images. It aims as a computer based support at the localization of pathological tissue such as brain tumors. The segmentation pipeline includes texture analysis, classification with Kohonen Feature Map and knowledge based morphological postprocessing. Furthermore the paper sketches a statistical investigation of different texture discrimination methods such as the wavelet transform. Segmentation results that have been reached with this pipeline are used for surface reconstruction and 3D-visualisation of pathological tissue.