2948 articles  [english version]
 HAL : hal-00079148, version 1
 arXiv : math.ST/0606219
 Versions disponibles : v1 (09-06-2006) v2 (07-12-2006)
 Significant edges in the case of a non-stationary Gaussian noise
 (09/06/2006)
 In this paper, we propose an edge detection technique based on some local smoothing of the image followed by a statistical hypothesis testing on the gradient. An edge point being defined as a zero-crossing of the Laplacian, it is said to be a significant edge point if the gradient at this point is larger than a threshold $s(\eps)$ defined by: if the image $I$ is pure noise, then $\P(\norm{\nabla I}\geq s(\eps) \bigm| \Delta I = 0) \leq\eps$. In other words, a significant edge is an edge which has a very low probability to be there because of noise. We will show that the threshold $s(\eps)$ can be explicitly computed in the case of a stationary Gaussian noise. In images we are interested in, which are obtained by tomographic reconstruction from a radiograph, this method fails since the Gaussian noise is not stationary anymore. But in this case again, we will be able to give the law of the gradient conditionally on the zero-crossing of the Laplacian, and thus compute the threshold $s(\eps)$. We will end this paper with some experiments and compare the results with the ones obtained with some other methods of edge detection.
 1 : Département des Sciences de la Simulation et de l'Information (DSSI) CEA : DAM/DIF 2 : Mathématiques - Analyse, Probabilités, Modélisation - Orléans (MAPMO) Université d'Orléans – CNRS : UMR7349 3 : Mathématiques appliquées Paris 5 (MAP5) CNRS : UMR8145 – Université Paris V - Paris Descartes 4 : Centre de Mathématiques et de Leurs Applications (CMLA) CNRS : UMR8536 – École normale supérieure de Cachan - ENS Cachan
 Domaine : Mathématiques/StatistiquesMathématiques/Probabilités
 Mots Clés : Edge detection – Significant edges – Inverse problem – Statistical hypothesis testing
Liste des fichiers attachés à ce document :
 PS
 contours3.ps(9.6 MB)
 PDF
 contours3.pdf(1.4 MB)
 hal-00079148, version 1 http://hal.archives-ouvertes.fr/hal-00079148 oai:hal.archives-ouvertes.fr:hal-00079148 Contributeur : Romain Abraham <> Soumis le : Vendredi 9 Juin 2006, 11:44:09 Dernière modification le : Vendredi 9 Juin 2006, 16:36:58