Hi Gaetan,<br><br>I found that in the itkSaltAndPepperNoiseImageFilter, it would be more useful if the values m and M be computed from the minima and minima in the image. <br><br>Currently, you are taking the pixel type maxima and minima which precludes the use of int or double or float pixel types in the image model with a narrow dynamic range of data.<br>
<br>Does that sound reasonable ?<br><br>Kishore<br> <br><br><div class="gmail_quote">2010/4/1 Gaëtan Lehmann <span dir="ltr"><<a href="mailto:gaetan.lehmann@jouy.inra.fr">gaetan.lehmann@jouy.inra.fr</a>></span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Hi,<br>
<br>
Le 30 mars 10 à 17:40, Oleksandr Dzyubak a écrit :<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Hi Gaëtan,<br>
<br><div class="im">
I have done my homework.<br>
<br>
1) I studied the ITK header.<br>
<br>
InsightToolkit/Common/itkMersenneTwisterRandomVariateGenerator.h<br>
<br>
In that header the link to the inventors' web page<br>
<a href="http://www.math.keio.ac.jp/%7Ematumoto/emt.html" target="_blank">www.math.keio.ac.jp/~matumoto/emt.html</a> is dead.<br>
The correct and still alive one is below.<br>
<br>
<a href="http://www.math.sci.hiroshima-u.ac.jp/%7Em-mat/MT/emt.html" target="_blank">http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html</a><br>
<br>
2) I have also read the article by the inventors<br>
M. Matsumoto and T. Nishimura, "Mersenne Twister: A 623-dimensionally equidistributed uniform pseudorandom number generator", ACM Trans. on Modeling and Computer Simulation Vol. 8, No. 1, January pp.3-30 (1998).<br>
<br>
This classical article (along with others) can be downloaded from<br>
the Makoto Matsumoto's home page.<br>
<br>
<a href="http://www.math.sci.hiroshima-u.ac.jp/%7Em-mat/MT/ARTICLES/earticles.html" target="_blank">http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/ARTICLES/earticles.html</a><br>
<br>
3) From that article: "A new algorithm called Mersenne Twister<br>
is proposed for generating uniform pseudorandom numbers."<br>
Period is (2^ 19937) - 1, 632-dimensional equidistribution up to 32-bit accuracy.<br>
So the period is a such huge number<br>
that this generator could be considered as a "true random".<br>
</div></blockquote>
<br>
Sounds good :-)<div class="im"><br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
4) New debugged and updated release 2009/11/17 of the generator<br>
can be downloaded using link below.<br>
<br>
<a href="http://www.math.sci.hiroshima-u.ac.jp/%7Em-mat/MT/SFMT/index.html" target="_blank">http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html</a><br>
<br>
<br>
5) I downloaded and tested the updated code version from the link you provided.<br>
<a href="http://mima2.jouy.inra.fr/darcs/contrib-itk/noise/" target="_blank">http://mima2.jouy.inra.fr/darcs/contrib-itk/noise/</a><br>
<br>
a) Compiles -- fine.<br>
b) Works -- fantastic.<br>
c) Tested images are identical to those provided in the submission.<br>
</blockquote>
<br></div>
Great!<div class="im"><br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
6) Could you upload the revised version to the Insight Journal?<br>
I would be more than happy to write a review and<br>
recommend to add it to the ITK library.<br>
<br>
</blockquote>
<br></div>
I just tried, but I'm still stuck at that stage <a href="http://insight-journal.com/submit/step10/59" target="_blank">http://insight-journal.com/submit/step10/59</a>, as for the fftconv contrib.<br>
Julien, did you have a chance to look at that?<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Thanks for nice work,<br>
</blockquote>
<br>
Thanks a lot for your very useful report and comments<br><font color="#888888">
<br>
Gaëtan</font><div><div></div><div class="h5"><br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Alex<br>
<br>
<br>
Gaëtan Lehmann wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Le 29 mars 10 à 15:54, Oleksandr Dzyubak a écrit :<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Hi Gaëtan,<br>
<br>
</blockquote>
<br>
Hi Alex,<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Whenever I try to send images using a regular email client,<br>
my emails get bounced. So I used the web page uploader<br>
from <a href="http://nabble.com" target="_blank">nabble.com</a> to send the images. I hope you also got a copy.<br>
If not, below is the link to the images.<br>
<br>
<a href="http://old.nabble.com/Noise-simulation-to27990624.html#a28069421" target="_blank">http://old.nabble.com/Noise-simulation-to27990624.html#a28069421</a><br>
</blockquote>
<br>
I'd say there is a small difference between the two images, but that's a random process :-)<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
About median.cxx. In your submission you use it<br>
just as a denoiser to demonstrate that it works nice on, in particular, images<br>
with "salt-and-pepper", right?<br>
</blockquote>
<br>
I included it to experiment a bit with PSNR calculator and denoising. That's really a not much useful program - I didn't event mentioned it in the paper!<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
About the generator. How good is it? Is it random or chaotic?<br>
What is the cycle?<br>
</blockquote>
<br>
I can't say at all. I simply modified the Mersenne Twister generator from ITK to make it thread safe.<br>
I actually don't have much skills in random generator.<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
You say that the Gauss has been fixed.<br>
Is it possible the get the updated version?<br>
<br>
</blockquote>
<br>
Sure! It is in the main development repository.<br>
<br>
<a href="http://mima2.jouy.inra.fr/darcs/contrib-itk/noise/" target="_blank">http://mima2.jouy.inra.fr/darcs/contrib-itk/noise/</a><br>
<br>
I think I didn't say that you can use darcs to get it (<a href="http://www.darcs.net" target="_blank">http://www.darcs.net</a>)<br>
<br>
darcs get <a href="http://mima2.jouy.inra.fr/darcs/contrib-itk/noise/" target="_blank">http://mima2.jouy.inra.fr/darcs/contrib-itk/noise/</a><br>
<br>
Thanks,<br>
<br>
Gaëtan<br>
<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Thanks,<br>
<br>
Alex<br>
<br>
Gaëtan Lehmann wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Le 26 mars 10 à 16:27, Oleksandr Dzyubak a écrit :<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Hi Gaëtan,<br>
<br>
I downloaded the updated version from your repository that you provided.<br>
<a href="http://mima2.jouy.inra.fr/darcs/contrib-itk/noise/" target="_blank">http://mima2.jouy.inra.fr/darcs/contrib-itk/noise/</a><br>
<br>
Below is the result of my tests.<br>
<br>
1)<br>
The compilation was perfect.<br>
</blockquote>
<br>
great :-)<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
2)<br>
dzyubak@helium: /Original_Images$ ./median cthead1.tif cthead1_median.png | grep " Number Of Threads"<br>
<br>
After Running the median program, I do see any difference<br>
between original and processed images.<br>
As I can see, in the median.cxx you just removed the neighborhood radius.<br>
What is the purpose of being that utility in the submission?<br>
Can't be the median.cxx just removed from the package?<br>
<br>
</blockquote>
<br>
It does something here - by default, the radius is 1.<br>
Just try on the output of the salt and pepper filter :-)<br>
<br>
./sp ../images/cthead1.tif sp.png<br>
./median sp.png out.png<br>
<br>
it is very efficient for this kind of noise.<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
3)<br>
dzyubak@helium: /Original_Images$ ./gauss cthead1.tif cthead1_gauss_s22_8_m0.png 22.8 0.0 | grep " Number Of Threads"<br>
Number Of Threads: 16<br>
Global Maximum Number Of Threads: 128<br>
Global Default Number Of Threads: 16<br>
<br>
The image has the same seeding problem as in the previous version.<br>
No changes at all.<br>
<br>
</blockquote>
<br>
err, yes, my fault. I forgot that this filter was not using the mersenne twister generator.<br>
It's fixed now, by using the same random generator as in the other filters :-)<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
4)<br>
dzyubak@helium: /Original_Images$ ./sp cthead1.tif cthead1_sp_001.png 0.01 | grep " Number Of Threads"<br>
<br>
Number Of Threads: 16<br>
Global Maximum Number Of Threads: 128<br>
Global Default Number Of Threads: 16<br>
<br>
Surprisingly, now cthead1_sp_001.png looks similar to that<br>
from your submission. The incorrect seeding problem is gone.<br>
Noise really looks random. However, I noticed that your image is "noisier".<br>
Did you use the same parameter "0.01" or that is still the seeding problem?<br>
</blockquote>
<br>
noiser? can your provide the images?<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
5)<br>
The shot and spekle noise generators work fine and my images<br>
are identical to yours.<br>
</blockquote>
<br>
great :-)<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
6)<br>
Peak signal to noise ratio. In your submission you have 20 dB. Mine is below.<br>
<br>
dzyubak@helium: /Original_Images$ ./psnr cthead1.tif cthead1_median.png<br>
PSNR: 37.3425 dB<br>
<br>
dzyubak@helium: /Original_Images$ ./psnr cthead1.tif cthead1_gauss_s22_8_m0.png<br>
PSNR: 20.0251 dB<br>
<br>
dzyubak@helium: /Original_Images$ ./psnr cthead1.tif cthead1_shot_s015.png<br>
PSNR: 20.0429 dB<br>
<br>
dzyubak@helium: /Original_Images$ ./psnr cthead1.tif cthead1_spekle_s024.png<br>
PSNR: 20.0901 dB<br>
<br>
dzyubak@helium: /Original_Images$ ./psnr cthead1.tif cthead1_sp_001.png<br>
PSNR: 22.4873 dB<br>
<br>
Interesting, the utility says that my salt and pepper is noisier,<br>
but just visual test yours is as twice noisier. Subjective judgment?<br>
</blockquote>
<br>
PSNR is only a (not so bad) approximation of the image degradation as perceived by a human.<br>
Our brain very capable to ignore the noise, and for some case like the salt and pepper case, the PSNR is not very representative of our feeling.<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Summing up.<br>
1) What is the need for median.cxx ?<br>
Can't be this utility just removed from the submission?<br>
<br>
2) The seeding problem in the Gauss noise generator is still there.<br>
<br>
Regards,<br>
<br>
Alex<br>
<br>
<br>
Gaëtan Lehmann wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Le 25 mars 10 à 23:27, Gaëtan Lehmann a écrit :<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
Hi Alex,<br>
<br>
I think there is a problem with multithreading and the random generator - more specifically, the seeds are the same in all the threads.<br>
I didn't noticed this on my macbook with only 2 cores, even when increasing the number of threads to a higher number than the number of cores, but I see the same behavior on my workstation with 16 threads.<br>
<br>
How many threads are running in parallel on your system? The number is displayed while running one of the noise generator<br>
<br>
[glehmann@gbook build]$ ./shot ../images/cthead1.tif toto.png | grep " Number Of Threads:"<br>
Number Of Threads: 2<br>
Number Of Threads: 2<br>
<br>
You can limit the number of threads to a small number, or even to 1, with<br>
<br>
export ITK_GLOBAL_DEFAULT_NUMBER_OF_THREADS=1<br>
<br>
to workaround the problem.<br>
<br>
I will restore the (non thread safe) seeding behavior in ThreadSafeMersenneTwisterRandomVariateGenerator and make it thread safe with a lock so that 2 threads can't get the same seed.<br>
<br>
</blockquote>
<br>
This is fixed in the main repository:<br>
<br>
<a href="http://mima2.jouy.inra.fr/darcs/contrib-itk/noise/" target="_blank">http://mima2.jouy.inra.fr/darcs/contrib-itk/noise/</a><br>
<br>
As well as the problem you have reported in median.cxx.<br>
Please let me know if there is something else wrong!<br>
<br>
Gaëtan<br>
<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Thanks for the report!<br>
<br>
Gaëtan<br>
<br>
<br>
Le 25 mars 10 à 22:47, Oleksandr Dzyubak a écrit :<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Dear ITK users,<br>
<br>
Did anybody tested the noise simulation codes from the submission<br>
"Noise simulation" <a href="http://hdl.handle.net/10380/3158" target="_blank">http://hdl.handle.net/10380/3158</a> ?<br>
The most needed (for me at least) noise generators<br>
(Gauss and salt and pepper)<br>
produce corrupted images with periodic line patterns<br>
(see attached images). To generate those images, I followed<br>
the parameter recommendations taken from the submission.<br>
<br>
./gauss cthead1.tif cthead1_gauss_s22_8_m0.png 22.8 0<br>
./sp cthead1.tif cthead1_sp_0016.png 0.016<br>
<br>
Did anybody get similar results?<br>
<br>
Alex<br>
<br>
<br>
<br>
Oleksandr Dzyubak wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Hi Gaëtan,<br>
<br>
Nice and long awaited work.<br>
I found a tiny typo in your code that prevents the submitted codes from<br>
being compiled.<br>
At least on my Linux box. The typo is sitting in your median.cxx and the<br>
fix is below.<br>
<br>
// Part that gives the errors.<br>
/*<br>
FilterType::RadiusType r;<br>
r.Fill(2);<br>
// filter->SetRadius(r);<br>
*/<br>
<br>
// this fix solves the problem.<br>
IType::SizeType r;<br>
r[0] = 1;<br>
r[1] = 1;<br>
filter->SetRadius(r);<br>
<br>
Could you please add the fix and upload the code?<br>
<br>
Thanks,<br>
Alex<br>
<br>
</blockquote>
<br>
<cthead1_gauss_s22_8_m0.png><cthead1_sp_0016.png><br>
</blockquote>
<br>
-- <br>
Gaëtan Lehmann<br>
Biologie du Développement et de la Reproduction<br>
INRA de Jouy-en-Josas (France)<br>
tel: +33 1 34 65 29 66 fax: 01 34 65 29 09<br>
<a href="http://voxel.jouy.inra.fr" target="_blank">http://voxel.jouy.inra.fr</a> <a href="http://www.itk.org" target="_blank">http://www.itk.org</a><br>
<a href="http://www.mandriva.org" target="_blank">http://www.mandriva.org</a> <a href="http://www.bepo.fr" target="_blank">http://www.bepo.fr</a><br>
<br>
_____________________________________<br>
Powered by <a href="http://www.kitware.com" target="_blank">www.kitware.com</a><br>
<br>
Visit other Kitware open-source projects at<br>
<a href="http://www.kitware.com/opensource/opensource.html" target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
<br>
Kitware offers ITK Training Courses, for more information visit:<br>
<a href="http://www.kitware.com/products/protraining.html" target="_blank">http://www.kitware.com/products/protraining.html</a><br>
<br>
Please keep messages on-topic and check the ITK FAQ at:<br>
<a href="http://www.itk.org/Wiki/ITK_FAQ" target="_blank">http://www.itk.org/Wiki/ITK_FAQ</a><br>
<br>
Follow this link to subscribe/unsubscribe:<br>
<a href="http://www.itk.org/mailman/listinfo/insight-users" target="_blank">http://www.itk.org/mailman/listinfo/insight-users</a><br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
-- <br>
Gaëtan Lehmann<br>
Biologie du Développement et de la Reproduction<br>
INRA de Jouy-en-Josas (France)<br>
tel: +33 1 34 65 29 66 fax: 01 34 65 29 09<br>
<a href="http://voxel.jouy.inra.fr" target="_blank">http://voxel.jouy.inra.fr</a> <a href="http://www.itk.org" target="_blank">http://www.itk.org</a><br>
<a href="http://www.mandriva.org" target="_blank">http://www.mandriva.org</a> <a href="http://www.bepo.fr" target="_blank">http://www.bepo.fr</a><br>
<br>
</div></div><br>_____________________________________<br>
Powered by <a href="http://www.kitware.com" target="_blank">www.kitware.com</a><br>
<br>
Visit other Kitware open-source projects at<br>
<a href="http://www.kitware.com/opensource/opensource.html" target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
<br>
Kitware offers ITK Training Courses, for more information visit:<br>
<a href="http://www.kitware.com/products/protraining.html" target="_blank">http://www.kitware.com/products/protraining.html</a><br>
<br>
Please keep messages on-topic and check the ITK FAQ at:<br>
<a href="http://www.itk.org/Wiki/ITK_FAQ" target="_blank">http://www.itk.org/Wiki/ITK_FAQ</a><br>
<br>
Follow this link to subscribe/unsubscribe:<br>
<a href="http://www.itk.org/mailman/listinfo/insight-users" target="_blank">http://www.itk.org/mailman/listinfo/insight-users</a><br>
<br></blockquote></div><br>