Extraction of Connected Region Boundary in Multidimensional Images

David Coeurjolly (1) , Bertrand Kerautret (2) , Jacques-Olivier Lachaud (3)

1 M2DisCo - Geometry Processing and Constrained Optimization
2 ADAGIO - Applying Discrete Algorithms to Genomics and Imagery
3 LAMA - Laboratoire de Mathématiques

This paper presents an algorithm to extract the boundary of a connected region(s) using classical topology definitions. From a given adjacency definition, the proposed method is able to extract the boundary of an object in a generic way, independently of the dimension of the digital space. Source Code The implementation of the algorithm is available through the DGtal library 1 . The source code and the demonstration are based on a special version of DGTal containing no external dependencies. They are both available on the IPOL web page of this article 2 . Supplementary Material The DGtalTools 3 project gives several additional tools exploiting the proposed algorithms. These tools are defined to process both 2D and 3D images.

Mots clés: Nd connected components; topology; discrete geometry

Domaines: Synthèse d'image et réalité virtuelle [cs.GR]

David Coeurjolly, Bertrand Kerautret, Jacques-Olivier Lachaud. Extraction of Connected Region Boundary in Multidimensional Images. Image Processing On Line, 2014, 4, pp. 30-43. ⟨10.5201/ipol.2014.74⟩. ⟨hal-01112943⟩