Matlab code by Antonis Matakos for his image deblurring paper

Antonios Matakos, Sathish Ramani, J A Fessler.
Accelerated edge-preserving image restoration without boundary artifacts. 
IEEE Trans. Im. Proc., 22(5):2019-29, May 2013.
http://doi.org/10.1109/TIP.2013.2244218

The main code files (functions) are in toolbox/im_restore/

figures/ : input images for restoration work

data/ : xinf_* (converged results)


toolbox/im_restore/ :

image_restore: function used in paper with ADMM method for analysis formulation

image_restore_synth: Function for ADMM formulation for synthesis formulation.
Not used in paper and not certain if fully complete and/or debugged.
@BlurMatrix: Toolbox used for DCT restoration.
Used in the DCT options of image_restore.
Not essential for the proposed ADMM algorithm.


toolbox/systems/ :

block_fatrix: modified to create fatrix with only one block and allow blocks
in any diagonal apart from the main.

Gdiag2: Extension of Gdiag to allow non-square diagonal matrices.
Used as mask for image restoration code
and also for some parts of ghost correction testing.

sparse_bccb: creates a sparse bccb matrix from psf.
Unsure if used for image restoration code.

zero_fatrix: matrix of zeros.
More straightforward implementation than forcing a diagonal with all zeros.


toolbox/utilities/ :

Some utility functions that may already be included in the toolbox.
Not sure where they are used.
I think nufft_sinc_nopi is from Michael and was used for the QS algorithm.


script/ :
contains some test scripts for image restoration
Some of the scripts are self-contained but others depend on data files
that are included in the data folder for completeness.