Bernard UEI/AVA, Thierry et Manzanera UEI/AVA, Antoine et Prêteux, Françoise (2001) Markovian-based modeling on programmable retina In: Proceedings Conference on Controle Quality by Artificial Vision (QCAV'01), Mai 2001, Le Creusot, France.
Fichier(s) associé(s) à ce document :
![[img]](/style/images/fileicons/application_pdf.png) | PDF Restricted to Administrateur de l'archive uniquement 428Kb |
Résumé
This paper addresses the implementation of a Markovian model on a programmable digital retina. The original contribution is to adapt the coding of the local energy in order to get compact implementations of deterministic (Iterated Conditional Mode), and stochastic (Metropolis Dynamics) relaxation algorithrns for energy minimization. The logical conciseness of the procedures, together with the fine-grainsed paralielism of the retinal architecture, makes possible the real-time computation of the simulated annealing on a low-power vision system. the proposed model is applied to the segmentation of moving objects over a static background.
| Type de document: | Document issu d'une conférence ou d'un atelier (Conférence) |
|---|---|
| Sujets: | Sciences et technologies de l'information et de la communication |
| Unité d'appartenance: | |
| Code ID : | 3364 |
| Déposé par : | Julien Karachehayas |
| Déposé le : | 07 mars 2008 01:20 |
| Dernière modification: | 05 juin 2013 09:13 |
