This file contains the implementation of the triplet Markov random field presented in the paper:

Ouali S, Courbot J-B, Pierron R and Haeberle O (2024), "Bayesian image segmentation under varying blur with triplet Markov random field", Inverse Problems. Vol. 40, pp. 095010. IOP Publishing.

The preprint of this paper is available here.

The main entry point of this package is the notebook 'Segmentation example.ipynb', that can be run to test the implemented algorithm. It requires having a 3D PSF ('PSF GL.tif') and an image ('90.png') that are both provided.

The package also contains the following files:

• TMRF_functions: it contains the necessary functions for the implementation of the TMRF model using the chromatic Gibbs sampler. The file also contains the implementation of the MAP and MPM estimators
• GMRF_functions: it contains the necessary functions for the simulation of a Gaussian random field
• MH_functions: it contains the necessary functions to implement the Metroplis-Hastings algorithm
• parameter_estimation: this file contations the different function necessary to estimates the parameters of the proposed TMRF model
• Segmentation example: this notebook can be run to test the implemented algorithm. It requires having a 3D PSF. An example image named 90.png is also added.