Skeletonization is a transformation of an object in a digital image into a simplified representation of the original object. The skeleton of an image object is an abstraction of the original object which largely preserves the extent and connectivity of the original region while throwing away most of the boundary and interior pixels. In this paper, we propose a new method to calculate skeleton from 3D space instead of image space which has only two dimensions. Our method start with a contour of an object in an image, then inflate this two dimensional shape to a three dimensional mesh, and then apply a 3D mesh curve skeleton extraction algorithm to this intermediate three dimension mesh model. Finally, we project the resulting 3D curve skeleton back to image space and get the skeleton of the original shape or object in the image. Our method is noise insensitive. A little perturbation on shape would not change the structure of the resulting skeleton. Our method is relatively fast because it only generates a geometry mesh in contrast to compute a Voronoi graph. Our method preserves the topology as well as the shape. © 2011 Springer-Verlag.
CITATION STYLE
Hu, X., Sun, B., Zhao, H., Xie, B., & Wu, H. (2011). Image skeletonization based on curve skeleton extraction. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6761 LNCS, pp. 580–587). https://doi.org/10.1007/978-3-642-21602-2_63
Mendeley helps you to discover research relevant for your work.