From 8a8f0d10f86faeccbdf91ecebce5c2c35d40253d Mon Sep 17 00:00:00 2001
From: Paul Irofti <bulibuta@users.noreply.github.com>
Date: Tue, 14 Feb 2017 18:21:51 +0200
Subject: [PATCH] README.md documentation

---
 README.md | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/README.md b/README.md
index f4831e3..ba4daf5 100644
--- a/README.md
+++ b/README.md
@@ -1 +1,31 @@
 # Regularized K-SVD Algorithm
+
+Implementation of the [Regularized K-SVD](http://ieeexplore.ieee.org/abstract/document/7831399/) Dictionary Learning Algorithm described in
+B. Dumitrescu and P. Irofti, "Regularized K-SVD," in IEEE Signal Processing Letters, vol. 24, no. 3, pp. 309-313, March 2017.
+
+## Prerequisite
+[OMP](http://www.cs.technion.ac.il/~ronrubin/Software/ompbox10.zip) implementation by Ron Rubinstein
+
+## Usage
+### INPUTS:
+- Y -- training signals set
+- D -- current dictionary
+- X -- sparse representations
+- iter -- current DL iteration
+
+### PARAMETERS:
+- reg -- regularization factor (default: 0.01)
+- vanish -- regularization vanishing factor (default: 0.95)
+- regstop -- stop regularization from this iteration on (default: Inf)
+
+### OUTPUTS:
+- D -- updated dictionary
+- X -- updated representations
+
+Sample call
+
+    [D,X] = ksvd_reg(Y,D,X,iter)
+    
+to be used within a dictionary learning loop (see [DL](DL.m)).
+
+Have a look at the [test script](test_ksvd_reg.m) for a full example.