Variational and Shape Prior-based Level Set Model for Image Segmentation - Archive ouverte HAL
Communication Dans Un Congrès Année : 2010

Variational and Shape Prior-based Level Set Model for Image Segmentation

Résumé

A new image segmentation model based on level sets approach is presented herein. We deal with radiographic medical images where boundaries are not salient, and objects of interest have the same gray level as other structures in the image. Thus, an a priori information about the shape we look for is integrated in the level set evolution for good segmentation results. The proposed model also accounts a penalization term that forces the level set to be close to a signed distance function (SDF), which then avoids the re-initialization procedure. In addition, a variant and complete Mumford-Shah model is used in our functional; the added Hausdorff measure helps to better handle zones where boundaries are occluded or not salient. Finally, a weighted area term is added to the functional to make the level set drive rapidly to object's boundaries. The segmentation model is formulated in a variational framework, which, thanks to calculus of variations, yields to partial differential equations (PDEs) to guide the level set evolution. Results obtained on both synthetic and digital radiographs reconstruction (DRR) show that the proposed model improves on existing prior and non-prior shape based image segmentation.
Fichier principal
Vignette du fichier
ESDiop_ICNAAM2010.pdf (486.65 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01196246 , version 1 (09-09-2015)

Identifiants

  • HAL Id : hal-01196246 , version 1

Citer

El Hadji Samba Diop, Sileye Ba, Taha Jerbi, Valérie Burdin. Variational and Shape Prior-based Level Set Model for Image Segmentation. ICNAAM 2010 : 8th International Conference of Numerical Analysis and Applied Mathematics, Sep 2010, Rhodes, Greece. ⟨hal-01196246⟩
244 Consultations
270 Téléchargements

Partager

More