Fast Image Deconvolution using Hyper-Laplacian Priors
Presented at Neural Information Processing Systems 2009, Vancouver.
Download Matlab Code that implements the deconvolution algorithm presented in the paper. The file is a .tgz archive and has been tested on Linux and Windows under Matlab 7.9 (R2009a and R2009b). Please send any bug reports to my email address.
Update January 1, 2010: A bug was found in the previous version uploaded on December 1, 2009. Please re-download the new code from the link above.
Update June 12, 2011: A very fast GPU implementation of our algorithm has been recently developed.