On extracting skeletons from binary 3D images

Abstract

This paper explores the subject of extracting skeletons from 3D binary images. It draws attention to the flaws found in the available skeletonization implementations. The most popular algorithm devised by Lee, Kashyap and Chu is analyzed. It is shown that despite good results, some aspects of said algorithm can be further improved upon. An own implementation, with some modifications is made and compared against the existing ones.