Abstract : The segmentation of networks is important in several imaging domains, and models incorporating prior shape knowledge are often essential for the automatic performance of this task. We incorporate such knowledge via phase fields and higher-order active contours (HOACs). In this paper: we introduce an improved prior model, the phase field HOAC ‘inflection point' model of a network; we present an improved data term for the segmentation of road networks; we confirm the robustness of the resulting model to choice of gradient descent initialization; and we illustrate these points via road network extraction results on VHR satellite images.
https://hal.inria.fr/inria-00390446 Contributeur : Aymen El GhoulConnectez-vous pour contacter le contributeur Soumis le : mardi 2 juin 2009 - 11:07:21 Dernière modification le : samedi 25 juin 2022 - 23:01:03 Archivage à long terme le : : vendredi 11 juin 2010 - 00:09:43
Aymen El Ghoul, Ian Jermyn, Josiane Zerubia. INFLECTION POINT PHASE FIELD HIGHER-ORDER ACTIVE CONTOUR MODEL FOR NETWORK EXTRACTION FROM VHR SATELLITE IMAGES. EUSIPCO, Aug 2009, GLASGOW, United Kingdom. ⟨inria-00390446⟩