File Exchange

image thumbnail

Non-Local Euclidean Medians

version 1.1 (7.77 KB) by

Non-local patch-based image denoising using Non-Local Euclidean Medians (NLEM).



View License

MATLAB implementation of the Non-Local Euclidean Medians (NLEM) algorithm described in the following paper:

[1] K.N. Chaudhury, and M. Unser, "Non-Local Euclidean Medians", IEEE Signal
Processing Letters, vol. 19, no. 11, 2012.

The current implementation is meant for experimental usage, and is far from optimal in terms of efficiency. For example, it can be optimized using *parfor* loops, or using mex code for computing the Euclidean median.

Comments and Ratings (7)

Kunal Chaudhury

Ben Kang: This does something very similar, but is faster and generally performs better denoising:


Hua (view profile)

Kunal Chaudhury

Ben Kang: We have come up with something called "pruned non-local means" that is much faster than NLEM and gives better results:

We plan to share the code for PNLM sometime in the future.

Ben Kang

Worked better than non-local mean on my dataset but took much longer time. It would be great if providing a faster version. Tried parfor, but got alerts due to variables are sliced within the for loops. Any idea to dramatically improve the speed? Thanks!

Kunal Chaudhury

Royi Avital: NLEM does better than the classic NLM on images that have lot of sharp intensity jumps. Taking the Euclidean median (instead of the mean) gives better robustness for such images.

Royi Avital

Is it any better than the classic NLM?


nice.but what about NLEM version for color image ?



code update.

MATLAB Release
MATLAB 7.11 (R2010b)

Download apps, toolboxes, and other File Exchange content using Add-On Explorer in MATLAB.

» Watch video