Skeletonization of Grayscale Volumes for Shape Description

Skeletons as important shape features of an object are useful for shape description. Unfortunately, methods obtaining skeletons of a grayscale volume are lacking due to no clear boundary between object and background. In this paper, we present a new segmentation-free skeletonization method on grayscale volumes based on marching cubes, topological thinning and a novel pruning routine. Our method is capable of generating a family of skeletal curves and surfaces that lie centered at rod-like and plate-like parts in the grayscale volume for shape description. It is demonstrated on biomedical datasets.