Browsing by Subject "3D image analysis"

Now showing 1 - 1 of 1

Access status: Open Access ,
Detekcja mostów w obrazach 3D prezentujących propagację pęknięć korozyjnych naprężeniowych w stali nierdzewnej
(Wydawnictwa AGH, 2007) Janaszewski, Marcin Sławomir; Babout, Laurent
This paper presents an algorithm to identify bridges in 3D objects represented with a set of connected voxels. The notion of a bridge can be defined as a part of a material with higher resistance to stress corrosion. Therefore if a crack meets a bridge it splits itself into two branches, which surround the bridge and merge when it finishes. Thus from the 3D image analysis point of view a bridge can be treated as a hole in a volumetric object which represents a crack. The literature analysis shows that usually detection of holes is carried out based on the object skeleton and localization of loops in the skeleton as each loop implies occurrence of a hole. Unfortunately, due to unprecedented complexity of tested images, commonly used skeletonization algorithms fail. Therefore authors decided to construct a new algorithm of whole detection which consists of two stages. The first one lies in the application of a voxel coding technique to make a transformation of an object into a set of clusters, where a cluster is a set of connected voxels. The set of clusters gives important information about the object topology. In the second step the algorithm extracts and analyses clusters to detect holes. The algorithm was satisfyingly tested on several 3D crack images from a stress corrosion cracking experiment on a stainless steel sample. The discussion of obtained results and propositions for future works are also included in the paper.