File Exchange

image thumbnail

Image Edge Enhancing Coherence Filter Toolbox

version 1.7 (10.1 MB) by

Advanced 2D/3D noise removal and edge enhancing with anisotropic diffusion filtering ( Weickert )

4.86667
15 Ratings

69 Downloads

Updated

View License

Introduction:
This toolbox will perform Anisotropic Non-Linear Diffusion filtering on a 2D gray/color or 3D image. This filtering will reduce the image noise while preserving the region edges, and also enhancing the edges by smoothing along them.

This is one of the more advanced image enhancement methods available, and also contains HDCS from october 2009. The result looks like an artist painted the image, with clear brush strokes along the image edges and ridges, see screen-shot.

My papers about the code are included:
- "Coherence Filtering to Enhance the Mandibular Canal in Cone-Beam CT Data", IEEE-EMBS Benelux Chapter Symposium, 2009.
- "Optimized Anisotropic Rotational Invariant Diffusion Scheme on Cone-Beam CT", MICCAI, 2010

Method:
The basis of the method used is the one introduced by Weickert.
1, Calculate Hessian from every pixel of the Gaussian smoothed input image
2, Gaussian Smooth the Hessian, and calculate its eigenvectors and values (Image edges give large eigenvalues, and the eigenvectors corresponding to those eigenvalues describe the direction of the edge)
3, The eigenvectors are used as diffusion tensor directions. The amplitude of the diffusion in those 3 directions is based on the eigen values and determined by Weickerts equation.
4, An Finite Difference scheme is used to do the diffusion
5, Back to step 1, till a certain diffusion time is reached.

Diffusion schemes:
There are several diffusion schemes available: standard, implicit, nonegative discretization, and also a rotation invariant scheme, and a novel diffusion scheme with new optimized derivatives.

Mex Files:
All 3D files are not only available as Matlab but also as C-code /MEX files, to increase speed and reduce the amount of memory used. Compile the c-code by executing compile_c_files.m.

Literature (Full list in the included paper):
- Weickert : "A Scheme for Coherence-Enhancing Diffusion Filtering with Optimized Rotation Invariance"
- Mendrik et al, "Noise Reduction in Computed Tomography Scans Using 3-D Anisotropic Hybrid Diffusion With Continuous Switch", October 2009
- Weickert : "Anisotropic Diffusion in Image Processing", Thesis 1996
- Laura Fritz : "Diffusion-Based Applications for Interactive Medical Image Segmentation"
- Siham Tabik, et al. : "Multiprocessing of Anisotropic Nonlinear Diffusion for filtering 3D image"

Usage:
Read the help of CoherenceFilter, compile the c-code and try the examples in the help.

Please report bugs, successes and questions.

Comments and Ratings (32)

Abdullah

now the compiler can run how can I apply it on random images ???

Aasd

Aasd (view profile)

good job

Ilya Belevich

David

David (view profile)

Nice toolbox but I have two issues:
First your default T value is 2 not 5 as described in the help text.

Second in the main while loop for 3D you compare the Scheme strings and look for R:
if(strcmpi(Options.Scheme,'R'))
u=CoherenceFilterStep3D(u,Options);
else
u=Anisotropic_step3D(u,Options);
end

in Anisotropic_step3D you then look again for R:
switch lower(Options.Scheme)
case 'r'
u=diffusion_scheme_3D_rotation_invariant(u,Dxx,Dxy,Dxz,Dyy,Dyz,Dzz,Options.dt);

So diffusion_scheme_3D_rotation_invariant will actually never be called.
I don't know if this intentional, since for 2D your first comparison looks very different:

if(strcmpi(Options.Scheme,'R')&&(Options.eigenmode==0)&&(exist('CoherenceFilterStep2D')==3))
u=CoherenceFilterStep2D(u,Options);
else
u=Anisotropic_step2D(u,Options);
end

Christopher

I'd like to agree with / confirm Michelle's comment. compile_c_files gives an error " mex: no file name given." unless you change "diffusion_scheme_3D_novel_getUpdate" to "diffusion_scheme_3D_novel_getUpdate.c"
Thanks!

Michele

Hi,
there's a little typo in compile_c_files.m.
Last file to be compiled has not the .c extension.
This: "diffusion_scheme_3D_novel_getUpdate.c"
Thanks.

Sourin

Sourin (view profile)

solved! it was some mac os problem with gcc!

Sourin

Sourin (view profile)

Hi

I am getting the allowing error…. any help is greatly appreciated. I am using matlab 2012b on Mac os. thanks

Sourin

derivatives.c: In function 'gradient2Dx':
derivatives.c:24: warning: incompatible implicit declaration of built-in function 'malloc'
derivatives.c: In function 'gradient2Dy':
derivatives.c:57: warning: incompatible implicit declaration of built-in function 'malloc'
derivatives.c: In function 'gradient3Dx_double':
derivatives.c:110: warning: incompatible implicit declaration of built-in function 'malloc'
derivatives.c: In function 'gradient3Dx_float':
derivatives.c:195: warning: incompatible implicit declaration of built-in function 'malloc'
derivatives.c: In function 'gradient3Dy_float':
derivatives.c:280: warning: incompatible implicit declaration of built-in function 'malloc'
derivatives.c: In function 'gradient3Dz_float':
derivatives.c:364: warning: incompatible implicit declaration of built-in function 'malloc'
derivatives.c: In function 'gradient3Dy_double':
derivatives.c:449: warning: incompatible implicit declaration of built-in function 'malloc'
derivatives.c: In function 'gradient3Dz_double':
derivatives.c:533: warning: incompatible implicit declaration of built-in function 'malloc'

mex: compile of ' "derivatives.c"' failed.

Error using mex (line 206)
Unable to complete successfully.

Error in compile_c_files (line 8)
mex(filename,'-v');

Anton Semechko

Anton Semechko (view profile)

When filtering 3D images I noticed that there is not option to specify the voxel resolution. To me this implies that in creating this function the author assumed that the voxels of the input image have isotropic resolution. Unfortunately this is usually not the case and must be accounted for when discretizing the diffusion equation.

Aside from this, I think this is a great submission. Many thanks!.

Ron

Ron (view profile)

Hi,

I have found an error in the code.
The 2D "standard" diffusion algorithm may be incorrect. The stencil shown in Weickertt (2002) is not the same as that referenced (from the handbook volume 2). That stencil has both terms in the top-left and bottom left corners negative. I beleive it is a typo in the Weickertt paper unless I am missing something.

Ron

Ron (view profile)

Hi,

I have found an error in the code.
The 2D "standard" diffusion algorithm may be incorrect. The stencil shown in Weickertt (2002) is not the same as that referenced (from the handbook volume 2). That stencil has both terms in the top-left and bottom left corners negative. I beleive it is a typo in the Weickertt paper unless I am missing something.

leila

leila (view profile)

I think it is useful to explain how to tune the parameters and constants. The default configuration over-smooth my image. I have a very thin object in a noisy image that disappeared by the filter.
thank you.

leila

leila (view profile)

hi, can I request everyone to report run-time of their experiment and size of image volume.
thanks.

where can imresize2 (that's called in hessian) be found?

xiuhong

it is a useful tool. Thanks the author very much!

ITTelkom

Sun SHENG

it is a elaborate tool. it is very usefull to my research. Thanks the author a lot!

Matthias Schabel

To solve the mex compilation error :

on OSX, add the following as line 3 of diffusion_scheme_3D_novel_getUpdate.c :

#include "string.h"

and re-run the mex script

Wouter

Wouter (view profile)

Compliments for this excellent code

I experience the same problem as Anthony Kilburg:

diffusion_scheme_3D_novel_getUpdate.c:126: warning: incompatible implicit declaration of built-in function 'memcpy'

Probably due to OS X or my compiler? (gcc-4.2)

Google 'says' I have to include <String.h>, but this does not seems to solve it.

Ding Yuan

Dirk,

Thanks for sharing, your profession in matlab is very impressive. the package is really nonlinear anisotropic diffusion, not only coherence diffusion, so I strongly suggest changing the name. I found a problem in functions2D/EigenVectors2D.m the eigenvectors v1 v2 corresponds to mu2 and mu1, please check it, it can be a serious bug.

Hello,

I reached an error while attempting to compile the C-code files. Error message is as follows:

??? Error using ==> mex at 221
Unable to complete successfully.

Error in ==> compile_c_files at 32
mex diffusion_scheme_3D_novel_getUpdate -v;

Upon review of the code I downloaded, I found that all mex funtions call for the function name .c -v for all the except the function I am having trouble with. Is the call in the function meant to be:
mex diffusion_scheme_3D_novel_getUpdate.c -v;

Thank you

wang ning

in diffusion_scheme_3D_standard.c , /* Compute tensor-driven diffusion (as in [1] pp. 80-82) */ which Literature is quoted ? What's [1] ?

nurul

nurul (view profile)

hi sir,
i just downloaded ur work. i find an error on line 218 . it says
Error in ==> CoherenceFilter at 218
if(size(u,3)<4), u=double(u); else u=single(u);

bin yao

A great code for studying diffusion filtering implementation. Thanks a lot for sharing.

I have two questions about the scheme
1) implicit scheme : does it means using semi-implicit AOS scheme to implementation?

2) diffusion_scheme_3D_implicit.m , you said "!! Scheme is unstable, and not ready to use yet." what is the problem?

Dirk-Jan Kroon

Dirk-Jan Kroon (view profile)

I uploaded a new version with derivative boundary check bug fixed, probably tomorrow online. thnx Elmar.

Dirk-Jan Kroon

Dirk-Jan Kroon (view profile)

*Elmar
Thank you for sharing the solution for your NaN problem.
Now, I know it's probably caused by a boundary check problem.

Dirk-Jan

Elmar

Elmar (view profile)

I permuted the directions of the volume so that it is 147x227x227, now it works perfectly

Dirk-Jan Kroon

Dirk-Jan Kroon (view profile)

* Elmar,

Thank you for your report, I never experienced NaN's my self.
Maybe you can add a small value to rule out division by zero.

If you still have problems, maybe you can share a part of the volume data with me?

Thanx, Dirk-Jan

Elmar

Elmar (view profile)

Thanks for sharing this tool! works very well and fast, but i'm experiencing some problems:
I'm trying to filter a 3D MRI volume of size 227x227x147. Depending on the value of 'rho' I use I get a variable number of slices in the third dimension that are all 'NaN'. The bigger rho, the more NaN slices are in the output image volume. For every value of T used the number of NaN-slices accumulates, so that after several timesteps the whole image volume consists of NaNs.
What could be the reason for the NaNs in the output?

My original image is single and normalized to a range of 0 to 1 and does not contain any NaNs.
thx, Elmar

Updates

1.7

Added paper, some minor changes in values

1.6

Added new Diffusion scheme which uses optimized 5x5 second order derivatives...

1.5

LCC compiler fix

1.4

Added Adriënne M. Mendrik et al. "Hybrid Diffusion with Continuous Switch (HDCS)", october 2009 IEEE transactions on medical imaging.

1.3

Literature

1.2

Fixed boundary check bug in backward derivatives.

1.1

Tested Linux Ubuntu

MATLAB Release
MATLAB 7.9 (R2009b)

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

» Watch video