|
|
Simon Wilson
Ancien Chercheur invité, Trinity College Dublin, Ireland
 Contact :
|
| Dernières publications dans le projet Ariana :
Unsupervised segmentation of textured satellite and aerial images with Bayesian methods.
S. Wilson et J. Zerubia. Dans Proc. European Signal Processing Conference (EUSIPCO), Toulouse, France, septembre 2002.
@INPROCEEDINGS{wilsonjz,
|
| author |
= |
{Wilson, S. and Zerubia, J.}, |
| title |
= |
{Unsupervised segmentation of textured satellite and aerial images with Bayesian methods}, |
| year |
= |
{2002}, |
| month |
= |
{septembre}, |
| booktitle |
= |
{Proc. European Signal Processing Conference (EUSIPCO)}, |
| address |
= |
{Toulouse, France}, |
| url |
= |
{http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=7071914}, |
| keyword |
= |
{} |
| } |
Segmentation of textured satellite and aerial images by Bayesian inference and Markov Random Fields.
S. Wilson et J. Zerubia. Rapport de Recherche 4336, INRIA, France, décembre 2001.
@TECHREPORT{wilsonJZ01,
|
| author |
= |
{Wilson, S. and Zerubia, J.}, |
| title |
= |
{Segmentation of textured satellite and aerial images by Bayesian inference and Markov Random Fields}, |
| year |
= |
{2001}, |
| month |
= |
{décembre}, |
| institution |
= |
{INRIA}, |
| type |
= |
{Research Report}, |
| number |
= |
{4336}, |
| address |
= |
{France}, |
| url |
= |
{https://hal.inria.fr/inria-00072251}, |
| pdf |
= |
{https://hal.inria.fr/file/index/docid/72251/filename/RR-4336.pdf}, |
| ps |
= |
{https://hal.inria.fr/docs/00/07/22/51/PS/RR-4336.ps}, |
| keyword |
= |
{} |
| } |
Résumé :
Nous étudions un modèle markovien double, initialement proposé par Melas et Wilson, pour la segmentation d'image. Le nombre de classes de l'image est obtenu par inférence bayésienne via un algorithme de Metropolis à saut réversible. Les mouvements habituellement utilisés dans une telle dynamique consistent en la fission ou la fusion de classes. Mais cela peut nécessiter beaucoup de temps de calcul, en particulier sur des images de grande taille. Ici, nous étudions des mouvements plus simples qui sont rapides à mettre en oeuvre, mais dont la mélangeance peut être longue. Nous proposons alors un schéma de fission/fusion plus complexe et comparons les performances obtenues. Nous effectuons des tests sur des images satellitai- res et aériennes. |
Abstract :
We investigate Bayesian solutions to image segmentation based on the double Markov random field model, originally proposed by Melas and Wilson. Inference on the number of classes in the image is done via reversible jump Metropolis moves. These moves, usually implemented by splitting and merging classes, can be very slow, making them impractical for large images. We investigate simpler reversible jump moves that are quick to implement but show that they may mix very slowly. We propose a more complex split and merge scheme and compare its performance. Tests are conducted on satellite and aerial images. |
|
Liste complète des publications dans le projet Ariana
|
|
Nouvelle page de démos
