|
|
EcoNet
Approches de la Physique Statistique pour l'Analyse d'Image : Application à la Gestion Forestière.
 Résumé (en anglais) :
The goal of this project is to develop and compare two approaches in the context of tree crown extraction from remotely sensed images. As a result, we obtain a description of each individual tree by its location and the size of its crown. This description allows the counting of tree and to forescast the development of the population.
The two approaches concern the Markov Random Field and the Marked Point Process modeling. For each approach, the computation time issue will be considered. For this, two optimization techniques, the Monte Carlo approach (MCMC and RJMCMC) and the diffusion process approach (Langevin dynamic and Birth and Death process) will be studied. |
Participants :
Informations complémentaires :
|
|
Nouvelle page de démos
