The Bregman Cookbook
The Bregman Cookbook is a collection of Matlab functions which use the
different Bregman Iteration methods to solve some particular algorithms
based on sparsity via L1-norm minimization. Except for some generalities,
the Bregman Cookbook is not devoted to the theoretical aspects of
Bregman iteration methods but is more focused on their numerical
implementations.
Feel free to contact me if you need some help, report a bug, or any
other suggestions.
If you want to contribute to the Bregman Cookbook, you can send me your
code + a LateX file containing explanations on the numerical aspects of
your algorithm (eventually with some references to specific
publications).
Optional toolboxes
Some functions use the Framelet and Curvelet expansions. These
expansions are available in separate Matlab Toolboxes:
Some examples obtained with the Bregman Cookbook code:
- Anisotropic ROF denoising:
- Framelet Nonblind deconvolution:
The Bregman Cookbook archive
The current version is
- v3.2 (June 23, 2013) - Convergence criteria updated for all functions (bugs fixed) + Test script for 1D case.
- v3.0 (March 4, 2012) - Complete reorganization into several
subfolders. New 3D functions and example scripts are added.
- v2.0 (October 30, 2011) - Isotropic ROF is added and both
Isotropic and Anisotropic ROF are now solved in the Fourier domain.
- v1.2 (July 18, 2011) - Now deconvolution functions are accepting
kernels defined both in the spatial or Fourier domains.
- v1.0 (May 8, 2011)
Toolbox now available on
Matlab
Central.