The Gradient of the Maximal Curvature Estimation for Crest Lines Extraction

Crest lines are one of the many types of feature lines that illustrate the prominent characteristics of an object’s surface. In this study, we investigate one vital component to extract crest lines, the gradient of the maximal curvature. Most of geometry properties required to calculate crest lines can be obtained from the volume data during the process of the surface construction using implicit surface polygonizer. Nevertheless the gradient of the maximal curvature cannot be obtained due to the nature of the surface construction algorithm. Hence we proposed three weight function based methods in accordance with the knowledge of the surface mesh. We implemented our methods and conducted both qualitative and quantitative analysis on our methods to find the most appropriate method. We also put forward a simple filtering mechanism as a post-processing procedure to enhance the accuracy of the crest lines, which is addressed at the end of this study.