[Insight-users] ITK -- Connected Components filter on itk::mesh data

Richard Beare richard.beare at gmail.com
Fri Mar 21 19:30:32 EDT 2008


Hi,

Creating connected components is basically a graph search. For images
the graph edges are defined by the choice of neighborhood - i.e face
or fully connected, and the implementation complexity is to do with
arranging for the search to happen efficiently given the particular
constraints of an image based graph. Briefly, the process is organised
so that voxels are visited in raster order, to maximize cache/memory
performance and the number of neighborhood visits is minimized because
these are comparatively expensive. In the current itk filter the image
based neighborhood visits are eliminated entirely due to a runlength
encoding process.

I haven't dealt with itk mesh structures yet, but imagine that they
form a graph. You'll need to be able add some extra information to the
nodes that you want to form into components - namely a label value.

The algorithm is then a simple graph flooding:

Initialise the label at each node to zero.

Visit each node in the mesh.
If the label value is zero, reset it to the nextLabel value.
Recursively visit all unlabelled nodes attached to this one and set
their labels.
Increment nextLabel

On Fri, Mar 21, 2008 at 10:52 PM, Andre Phillion <andre.phillion at epfl.ch> wrote:
>
>
>
>
> Hello
>
>
>
> I am a relatively new ITK user, who would like help using the segmentation
> filters for a materials science application (modeling the solidification of
> aluminum alloys).
>
> I have previously used the itkConnectedComponents filter on voxel image
> data, as you can see in the attached file.
>
>
>
> Is it possible to use the itkConnectedComponents filter on itk::mesh data?
> The mesh data contains nodes, 3D tetrahedral element connectivity, and a
> binary value (0 if liquid, and 1 if solid).
>
> I am trying to figure out the cluster size of tetrahedral elements which are
> solid.
>
>
>
> Does anyone know if this implementation is similar to image data, or
> substantially different?
>
>
>
> Thanks for your help,
>
>
>
> Regards,
>
>
>
> André
>
>
>
> Dr. André Phillion
>
> Laboratoire de Simulation des Matériaux
>  Ecole Polytechnique Fédérale de Lausanne
>
> MXG-318, Station 12, CH-1015, Switzerland
>
> Tel / Fax: +41(21)693 3942 / 5890
>
>
>
>
> _______________________________________________
>  Insight-users mailing list
>  Insight-users at itk.org
>  http://www.itk.org/mailman/listinfo/insight-users
>
>


More information about the Insight-users mailing list