The department I am working in is called "images/CReATIVe". It is directed by Pr. Michel BARLAUD who is my Ph. D thesis director too.
My research interests deal with Feature Detection and 3D Reconstruction.

I am currently working on 2D/3D objects detection and segmentation using geodesic active curves and level set method.
My first work was to find some time-optimisation methods to reduce processing time of active contours algorithms.
I realised this work in collaboration with Alcatel Research Center (Marcoussis, France) during a six months training course in 1998.

After this first step in segmentation world, I am now working on finding new criterions used to drive active curves in different ways :
Beginning with a criterion and using Level Set Method, I obtain a Partial Differential Equation which allows the active curve to move
towards the objects of the image.

Working with Eric Debreuve, Michel Barlaud and Gilles Aubert, I have presented a new iterative curve evolution.
 

Publications

• "Détection et Reconstruction Automatique de Bâtiments à partir de Contours Actifs Géodésiques"

Rapport de Recherche I3S 99-05, Sophia Antipolis, France
O. Amadieu

Download(please let me know what are your researches about when downloading these papers)

Résumé
Rapport de fin d'études (Ecole d'Ingénieur ESSI et DEA Imagerie Vision "ARAVIS")
Mise en application des contours actifs géodésiques sur des images synthétiques et des Modèles Numériques d'Elévation.
Schémas numériques des contours géodésiques. Implémentation par la méthode des "Level Set" (ou courbes de niveaux), Schéma numérique de la réinitialisation. Optimisation temporelle de l'évolution par méthode de Bande Etroite (NarrowBand) et Multirésolution.
 

• "Inward and Outward Curve Evolution Using Level Set Method" 

ICIP'99, Kobe, Japan
O. Amadieu, E. Debreuve, M. Barlaud, G. Aubert

Download(please let me know what are your researches about when downloading these papers)

Abstract
Iterative curve evolution techniques are powerful methods for image segmentation. Classical methods proposed curve evolutions which guarantee close contours at convergence and, combined with the level set method, they easily handled curve topology changes. However, these methods allow only one-way curve evolutions: shrinking or growing of the curve. Thus, the initial curve must encircle all the objects to be segmented or several curves must be used, each one totally inside one object.
In this paper, we present a new approach of iterative curve evolution using the level set method based on the variational criterion of an inverse problem. Besides the closing of the final contours and the curve topology change management, our method allows a two-way curve evolution: parts of the curve evolve in the outward direction while others evolve in the inward direction. It offers much more freedom in the initial curve position than with a classical geodesic search method. Our algorithm performs accurate and precise segmentations, with length penalty. Results are shown on damaged images with complex objects (including sharp angles, deep concavities or holes).

• "Segmentation par Contours Actifs Déformables - Approche Bidirectionnelle" 

RFIA'2000, Paris, France
O. Amadieu, E. Debreuve, M. Barlaud, G. Aubert

Download(please let me know what are your researches about when downloading these papers)

Résumé
Les techniques itératives d'évolution de courbes sont des méthodes performantes pour la segmentation d'images. Jusqu'à présent ces méthodes garantissaient des contours fermés et, implémentées avec la méthode des courbes de niveaux, géraient les changements de topologie de la courbe. Malgré cela, ces méthodes permettent uniquement une évolution des contours dans une seule direction à la fois : vers l'intérieur ou vers l'extérieur.
Dans cet article, nous présentons une nouvelle approche d'évolution itérative de courbes, utilisant la méthode des courbes de niveaux, basée sur le critère variationnel d'un problème inverse. En plus de la fermeture des contours et de la gestion de la topologie, notre méthode permet une évolution dans les deux directions simultanément : des parties de la courbe peuvent évoluer vers l'intérieur tandis que d'autres peuvent évoluer vers l'extérieur, ce qui autorise ainsi une plus grande liberté dans le choix de la position initiale du contour qu'auparavant. Notre algorithme fournit une segmentation d'images précise avec pénalisation pondérée de longueur. Les résultats présentés sont ceux obtenus sur des images réelles dégradées contenant des objets complexes (comportant des angles aigus, des concavités prononcées ou des trous).

Mots Clef
Segmentation, multi-direction, courbes de niveaux, évolution, EDP, contour actif

Abstract
Iterative curve evolution techniques are powerful methods for image segmentation. From non-intrinsic to intrinsic methods, from parametric to geometric models, methods proposed curve evolutions which guarantee close contours at convergence and which allow us to apply curvature penalties on those contours. Moreover, combined with the level set method, they easily handle curve topology changes. The classical iterative geodesic search takes advantage of all these benefits. However, this method allows only one-way curve evolutions: shrinking or growing of the curve. Thus, the initial curve must encircle all the objects to be segmented or several curves must be used, each one totally inside one object.
In this paper, we present a new approach of iterative curve evolution using the level set method based on the variational criterion of an inverse problem. Besides the closing of the final contours and the curve topology change management, our method allows a two-way curve evolution: parts of the curve can evolve in the outward direction while other parts can evolve in the inward direction. It offers much more freedom in the initial curve position than with a classical geodesic search method. Our algorithm performs accurate and precise segmentations, with length penalty. Results are shown on damaged images with complex objects (including sharp angles, deep concavities or holes).

Keywords
Segmentation, inward, outward, level set, PDE, active contour
 
 

• "Binary Objects Tomographic Reconstruction From Few Noisy X-ray Radiographs Using a Region Based Curve Evolution Method" 
  MIC'2001, Californy, USA

J.P. Bruandet, F. Peyrin,J.M. Dinten, O. Amadieu and M.Barlaud

Download(please let me know what are your researches about when downloading these papers)

Abstract
This paper addresses tomographic reconstruction of binary images from a small number of noisy projections. Therefore a region based reconstruction method using curve evolution is proposed. To manage the lack of data, the tomographic reconstruction problem is defined as a domain optimization problem. A geometrical criterion directly embedded in a dynamical scheme is defined. The minimization of this criterion drives the region determination. This scheme is implemented using a level set method including regularization via local curvature. The quality of images reconstructed from noiseless projections is satisfying even without regularization. However when the number of projections or the signal to noise ratio is decreased, reconstructed images suffer from geometrical degradations. We show the improvement of the results brought by regularization in drastic conditions such as reconstruction from three noisy projections.

Presentations
 

ICIP'1999 Simultaneous Inward and Outward Curve Evolution	RFIA'2000 Segmentation par Contours Actifs Déformables - Approche Bidirectionnelle