edwin 0.1.0

Module related to Bayesian approach to inverse problems

The Bayesian inversion (edwin) package provides algorithm developed during my scientific research work in numerical computation for inverse problems (signal, image processing). Feel free to use them as you want. Any comments and contributions are welcome.

The name edwin is in reference to Edwin T. Jaynes, a great Bayesian Analysis scientific.

Acknowledgements

The use of edwin software package should be explicitly acknowledged in publications in the following form:

1. an acknowledgment statement: “Some of the results in this paper have been derived using some of the edwin package algorithms From F. Orieux et al. published in citations.

2. at the first reference, a footnote placed in the main body of the paper referring to the edwin web site, currently http://bitbucket.org/forieux/edwin

The citations are mentioned in documentation, References section of this file and are available in bibtex file.

Info

• Author: François Orieux

• Contact: orieux at iap dot fr

• Project homepage: http://bitbucket.org/forieux/edwin

• Downloads page: https://bitbucket.org/forieux/edwin/downloads

Contents

improcessing.py

A module that implement the algorithm described in [2] for unsupervised myopic image deconvolution. However the myopic part is not actually available.

inversion.py

A module that implement the algorithm described in [1] and use in [3-4] and other papers. It’s implement an unsupervised general inverse problem algorithm estimation, based on MCMC algorithm.

sampling.py

Implementation of stochastic sampling algorithm, specially [1].

optim.py

A module that implement classical optimisation algorithm for use of other module. They are design for very large system resolution (dim > 1e6).

Requirements

This package depends on my free otb package (utility functions).

• Numpy version >= 1.4.1

• otb version >= 0.2.1

Installation

The pip version:

pip install edwin

If you have not pip, download the archive, decompress it and to install in your user path, run in a command line:

python setup.py install --user

or for the system path, run as root:

python setup.py install

Development

This package follow the Semantic Versionning convention http://semver.org/. To get the development version you can clone the mercurial repository available here http://bitbucket.org/forieux/edwin

The ongoing development depends on my research activity but is open. I try to fix bugs.

License

edwin is free software distributed under the MIT license, see LICENSE.txt

References

A bibtex file is provided in the archive.

[1]

F. Orieux, O. Féron and J.-F. Giovannelli, “Sampling high-dimensional Gaussian distributions for general linear inverse problems”, IEEE Signal Processing Letters, 2012

[2]

François Orieux, Jean-François Giovannelli, and Thomas Rodet, “Bayesian estimation of regularization and point spread function parameters for Wiener-Hunt deconvolution”, J. Opt. Soc. Am. A 27, 1593-1607 (2010)

[3]

F. Orieux, E. Sepulveda, V. Loriette, B. Dubertret and J.-C. Olivo-Marin, “Bayesian Estimation for Optimized Structured Illumination Microscopy”, IEEE trans. on Image Processing. 2012

[4]

F. Orieux, J.-F. Giovannelli, T. Rodet, and A. Abergel, “Estimating hyperparameters and instrument parameters in regularized inversion Illustration for Herschel/SPIRE map making”, Astronomy & Astrohpysics, 2013