Reconstruction of Surfaces from Cross Sections Using Skeleton Information

Abstract

Surface reconstruction from parallel cross sections is an important problem in medical imaging and other object-modeling applications. Shape and topological differences between object contours in adjacent sections cause severe difficulties in the reconstruction process. A way to approach this problem is using the skeleton to create intermediate sections that represent the place where the ramifications occur. Several authors have proposed previously the use of some type of skeleton to face the problem, but in an intuitive way and without giving a basis that guarantees a complete and correct use. In this paper, the foundations of the use of the skeleton to reconstruct a surface from cross sections are expounded. Some results of an algorithm that is based on these foundations and has been recently proposed by the authors are shown that illustrate the excellent performance of the method in especially difficult cases not solved previously.