Segmentation of Radiographic Images of Weld Applying Traditional and GVF Snakes.

Type : Article de conférence
Auteur(s) :  M. BENREBAYA, M. Halimi
Année :  2007
Domaine : Génie électrique
Conférence: International Conference on Electrical Engineering Design and Technologies (ICEEDT) HAMMAMET, Tunisia
Résumé en PDF :  (résumé en pdf)
Fulltext en PDF :  (.pdf)
Mots clés :  Active contour models, Edge detection, gradient vector flow, weld radiographic defects

Résumé : 

In this paper, we employ the active contour models (Snakes) for edge detection and segmentation of welds radiographic defects. These models are widely used in many applications, including edge, shape modeling, segmentation, and motion tracking. The first model defined and implemented is the classical snake formulated by Kass & al. Snake is a method of deformation a closed contour to the boundary of an object in an image. The snake model is a controlled continuity closed contour that deforms under the influence of internal forces, image forces and external constraint forces. The snake model algorithms suffer from the inability to converge a contour to severe object concavities. Another problem is the generation of false contours due to the creation of unwanted contour loops. In order to remedy to these drawbacks, we apply a the method, called Gradient Vector Flow (GVF). This method proposes gradient vector flow as the external force. GVF snake gives good results on radiographic images of weld.