[Insight-users] Anisotropic Diffusion Filtering

[lina septiana]

Hi Pietro,


Actually the good combination depend on what the application you'll apply in. But maybe firstly, I suggest you to understand  the variable in the basic anisotropic diffusion by changing the integral constant, Kappa (threshold for gradient and noise level),and  iteration in the combination narrow and wide 1D signal, you will know how these parameters give the impact to the narrow and wide form signal.
After you understand the characteristic of it, you may be able to design the appropriate anisotropic for your image. Then implement other combination anisotropic filtering.
Below the link for the basic form anisotropic filter (perona malik) and biased anisotropic filtering (nordstrom) link for your reference.

http://www.eecs.berkeley.edu/Pubs/TechRpts/1989/CSD-89-514.pdf

http://www.cs.berkeley.edu/~malik/papers/MP-aniso.pdf




________________________________
 From: Pietro Nardelli <pie.nardelli at gmail.com>
To: lina septiana <livangellista at yahoo.com> 
Sent: Tuesday, May 15, 2012 10:38 PM
Subject: Re: [Insight-users] Anisotropic Diffusion Filtering
 
Hi Lina,

thank you for your reply. I've only tried the Curvature Flow
anisotropic filter so far. Can you suggest some paper about the
combination of anisotropic diffusion with other algorithms?If i
understand correctly it's quite difficult to find an optimal method
for al the cases.

Thank you again,

Pietro

2012/5/15 lina septiana <livangellista at yahoo.com>:
> Hi Pietro,
>
> Anisotropic filter is one the good filter for image, but if you used the
> traditional anisotropic difussion, it may obviously distorted the edge.
> You can try another anisotropic diffusion such as Nordstorm (added with the
> bias) which may correcting the edge distortion but may be will increase the
> noise.
> or try to combine anisotropic diffusion with other algorithm, there're many
> papers explain about it.
>
>
> Lina
>
> ________________________________
> From: Pietro Nardelli <pie.nardelli at gmail.com>
> To: insight-users at itk.org
> Sent: Tuesday, May 15, 2012 7:48 PM
> Subject: [Insight-users] Anisotropic Diffusion Filtering
>
> Hi itk users,
>
> I need some advice. I'm trying to segment the airway of a lung from CT
> Dicom datasets and the threshold connected library seems to be very
> appropriate. I followed the example of the guide and I implemented it
> into Slicer 4.1 and I can get good results. I have a doubt though: in
> most cases I can obtain better results setting to 0 the number of
> iterations of the smoothing variable, but sometimes a good smoothing
> might be very helpful. So, I was wondering whether using another
> anisotropic diffusion filter in place of the curvature flow presented
> in the example could allow to obtain better results. Have you some
> advice about the best filter I can use for CT Dicom images?
>
> Thank you for your help,
>
> Pietro
> _____________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Kitware offers ITK Training Courses, for more information visit:
> http://www.kitware.com/products/protraining.php
>
> Please keep messages on-topic and check the ITK FAQ at:
> http://www.itk.org/Wiki/ITK_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://www.itk.org/mailman/listinfo/insight-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20120515/cec01adf/attachment.htm>