[Insight-users] 3D thinning algorithm (skeletonization)

[Oleksandr Dzyubak]

Hi Richard,

Many thanks for your advice. I downloaded the submission
http://hdl.handle.net/1926/304 and gave it a try.

My setup: Debian Lenny Linux with gcc version 4.3.2 (Debian 4.3.2-1.1) .

1) GCC gave me a lot of warnings concerning the data type usage
and compatibility in templates. I have already seen that such an issue
happens quite often if you switched from gcc 3.95 (or 4.1) to gcc 4.2 
and higher.
So it could be fixed.

2) In general both algorithms extract pretty much the same skeleton
and time performance 3D thinning vs DigitalTopology
is the same: 3 minutes for the image 540x540x541.

3) However, at least on my images, DigitalTopology generates
more loops and some branching points are incorrect.
It should be investigated more accurately though.

Needless to say, that my comparison is highly unfair since
DigitalTopology classes include some "tunning parameters"
whereas I just used defaults.

And you are right in a sense that classes from
the Digital Topology submission have a great potential and flexibility.
But I must admit that my background in Digital Topology
is not as good as Profs. Kovalevsky,  Bertrand, or Malandain.
Thus, I have to refresh my knowledge on this topic before
I start making any changes to the classes.

Concerning the link you provided.

http://voxel.jouy.inra.fr/darcs/contrib-itk/skeletonUtils

My Internet Provider says that "this site was blocked."
Is this one of the consequences of the volcano actions?
Just joking... Anyway, I will try to go to some cafe with Internet Service
and I will try to get to your site from there.

Regards,

Alex


Richard Beare wrote:
> This doesn't answer your question, but there is some alternative code
> you might be interested in.
> http://hdl.handle.net/1926/304 describes a 3D skeletonization class,
> and I ported it to a more ITK conforming style, available at:
>
> darcs get http://voxel.jouy.inra.fr/darcs/contrib-itk/skeletonUtils
>
> On Wed, Apr 28, 2010 at 5:05 AM, Oleksandr Dzyubak <adzyubak at gmail.com> wrote:
>   
>> Dear colleagues,
>>
>> This email was inspired by my recent upgrade to ITK 3.18
>> that caused some my algorithms stopped working
>> whereas the other started working incorrectly.
>>
>> Ultimately I figured out that partly the reason for this is that
>> whenever the community and developers find some
>> work around or  the bug fix, I just overwrite the headers and templates
>> and still keep "old" release. Bad practice.
>>
>> That is exactly what happened to the "BinaryThinning" algorithm.
>> Couple years ago I tested the "Implementation of a 3D thinning algorithm"
>> by Hanno Homann (http://hdl.handle.net/1926/1292).
>> It worked so great that I just carelessly overwrote
>> the release in-stock "BinaryThinning" algorithm
>> and happily used that algorithm for couple years.
>>
>> Since I did not have any doubts that this algorithm is now
>> a native part of the ITK library,  I was  surprised very much
>> when found that my programs do not work anymore after
>> I made upgrade to ITK-3.18.  And it took a  while before I understood why.
>>
>> Surprisingly the  algorithm by Hanno Homman did not get into 3.18.
>> Which is more, still there is no even a review /request for that algorithm.
>>
>> How do the algorithms get into the ITK library?
>> Is that the review which initializes the process?
>> Can I promote the algorithm by writing a review?
>>
>> Regards,
>>
>> Alex
>>
>> _____________________________________
>> 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.html
>>
>> 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
>>
>>