nicolas-dufour / markov-field-segmentation

We use markov fields to do unsupervised image segmentation

Markov Field Segmentation

This project was done at Télécom SudParis as part of the MSA major. The goal is to compare hidden markov models for image segmentation. Project done in french

We do Hidden Markov chains segmentation (supervised and unsupervised) trained with the backward-forward algorithm. In the unsupervised setting, hyper-parameters are estimated using the EM Algorithms

We also do Hidden Markov Field Segmentation (supervised and unsupervised) trained using MCMC and Gibbs Sampling. In the unsupervised setting, hyper-parameters are estimated using the EM Algorithms

Unsupervised segmentation of the European continent