Shape skeletonization by identifying discrete local symmetries

This paper presents a new skeletonization method based on a novel concept - discrete local symmetry. A skeleton obtained from discrete local symmetrie s approaches the skeleton of the underlying continuous shape if the samplin g is dense enough. Discrete local symmetries can be obtained by computing t he constrained Delaunay triangulation of the underlying image. Internal tri angles of a triangulation are divided into isolated triangles, end triangle s, normal triangles and junction triangles. A discrete local symmetry corre sponds to an isolated triangle, or an end triangle, or a normal triangle. S everal measures are taken to remove skeletonization artifacts and suppress image noise. The proposed method can produce correct centre lines and junct ions. It is efficient and robust against noise. The method is suitable for skeletonizing high-resolution images. (C) 2001 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved.