Estimating Extremal Planes to Extract Crest Lines from Volumetric Data

摘要

The aim of this research is to study and investigate crest points extraction technique from 3D volumetric data. Our goal is to propose an enhancement to extremal planes estimation technique for crest lines extraction and to test the method for medical purpose. Differential geometry approach is applied to the 3D surfaces based on 3rd order derivatives computation of principal curvatures and principal directions to estimate extremal planes for crest lines extraction. Gaussian smoothing is applied into the proposed method with various scales to filter the noise from volumetric data. These techniques are incorporated into marching tetrahedra algorithm to trace and extract the crest lines. We present and evaluate the results of the extracted crest lines using 3D visualization with several sample volumetric data.