[Insight-users] Medial Axis Transform (MAT) or a skeleton in 3D image or surface mesh

[Luis Ibanez]

Hi John,

Thanks for pointing this out.

Looking at the web site:
http://www.eastman.ucl.ac.uk/~thutton/powercrust/

It seems that the code is covered by a GPL license
which is very unfortunate...   :-/

GPL is too restrictive for allowing us to include
the code into ITK or VTK.

We could go back to the papers and reimplement
the code trying to use existing ITK classes.
It seems that a good start will be to use the
DanielssonDistanceMap filter in order to build
the distance map from the borders of the object.
This filter also produces a Voronoi map.

The second stage will then require to locate
ridges in the DistanceMap, the trace of such
ridges should correspond to the skeleton of the
figure.  We already have classes for detecting
ridges in 2D by using the eigen analysis of the
Hessian, and constructing a 3D parametric space
out of it.

--

The authors of the algorithm already posted a
number of implementation recommendations:
http://www.cs.utexas.edu/users/amenta/powercrust/implementation.html

and made their papers available online:
http://www.cs.utexas.edu/users/amenta/powercrust/refs.html


If somebody wants to tackle this development,
we will be glad to guide you through the process
and of course will encourage you to contribute
it back to ITK (under the open ITK license).



Any volunteers ?



    Thanks,


      Luis




------------------------
anast.jm at pg.com wrote:

> 
> Luis/Inaxio
> 
> Tim Hutton had ported the PowerCrust Reconstruction filter into vtk 
> (http://www.eastman.ucl.ac.uk/~thutton/powercrust) from the origianl 
> code by Nina Amenta <http://www.cs.utexas.edu/users/amenta>, Sunghee 
> Choi <http://www.cs.utexas.edu/users/sunghee> and Ravi Krishna Kolluri 
> <http://www.cs.utexas.edu/users/rkolluri> 
> (http://www.cs.utexas.edu/users/amenta/powercrust/welcome.html). If I've 
> gotten the story wrong, sorry Tim....
> 
> This approach uses the Medial Axis Transform but I know nothing about it 
> (other than it works pretty well in a number of surfacing cases I 
> tried). Maybe there is some code that could be reused there. I don't 
> know what the status of the licensing is so need to be a bit careful how 
> you use it. Perhaps Tim could be of assistance?  
> 
> ....john
> 
> 
> 
> 	*Luis Ibanez <luis.ibanez at kitware.com>*
> Sent by: insight-users-admin at itk.org
> 
> 04/09/2004 05:20 PM
> 
> 	       
>         To:        Inaxio Aguirre <I.Aguirre at exeter.ac.uk>
>         cc:        insight-users at itk.org, (bcc: John Anast-JM/PGI)
>         Subject:        Re: [Insight-users] Medial Axis Transform (MAT) 
> or a skeleton in 3D image or surface mesh
> 
> 
> 
> 
> 
> Hi Inaxio,
> 
> Nope, nothing yet...
> 
> Any volunteers ?
> 
> For anyone giving a course in Image Processing,
> that will make for a nice course project...     :-)
> 
> 
>   Luis
> 
> -----------------------
> Inaxio Aguirre wrote:
> 
>  > Hello,
>  >
>  > Referring to a previous mail
>  > (http://www.itk.org/pipermail/insight-users/2003-March/002885.html) 
> sent on
>  > March last year I want to know if there is already something available in
>  > itk to compute the Medial Axis Transform (MAT) or a skeleton from a 
> 3D image
>  > or a corresponding polygonal surface mesh.
>  >
>  > Thanks,
>  >     Inaxio
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> 
>