Active Random Fields demo

Author:  Adrian Barbu, Department of Statistics, Florida State University
Contact: abarbu@stat.fsu.edu
WWW:     http://www.stat.fsu.edu/~abarbu/

The code, filters present in this package demonstrate the Active Random Field computational model 
described in:

 Adrian Barbu. Learning Real-Time MRF Inference for Image Denoising. CVPR 2009

Usage:
From the Matlab command line call
testarf
and it will evaluate the ARF performance on the images specified in the file test.txt,
currently 68 test images from the Berkeley dataset. The images should be present in
the same directory with the matlab file.

The default parameters are sigma=25 and niter=1.
Other values can be changed in testarf.m

Possible values of sigma: 10,15,20,25,50
Possible values of niter: 1,2,3,4
