[vtkusers] Surfacing contours

Tim Hutton T.Hutton at eastman.ucl.ac.uk
Tue Sep 24 05:46:16 EDT 2002


Hi John,

I keep hearing about people with this problem, I thought I should throw in 
my 2 (euro-)cents.

When you've gone from a volume image to contours you have thrown away 
information. Trying to get it back with surfacing algorithms is not a good 
approach - most surfacing algorithms are notoriously bad at highly 
non-isotropic sampling (sampling is much denser in one direction).

The easy answer is to work from the original volume image if at all 
possible. Marching cubes is 100% reliable. Failing this (for example if the 
contours are manually extracted) fake it by filling-in the contours of the 
2D images (a simple polygon paint algorithm) and building a binary volume 
image, and *then* run marching cubes.

Hope this helps!

Tim.

At 19:16 23/09/2002 -0400, you wrote:

>I have a series of contours extracted from MRI images outside of vtk. I'm 
>trying
>to create a polygonal surface from the points (imported as
>vtkUnstructuredPoints, ignoring the connectivity data of the contours) but 
>with
>most of the classes I've tried the surface tends to want to close the face of
>the contour points, making seperate objects, instead of bridging to the next
>contour's points.  I'm at the limit of resolution of the MRI scanner so I 
>can't
>add contours or additional points per contour.
>
>I'm thinking I need an approach that can make use of the connectivity in the
>contours but have not seen a class like this in vtk...does such a beast exist?
>The other approach which I'm dreading is to try to interpolate contours to 
>help
>fill in the gaps between the image slices...but really I don't have a way 
>to do
>this and would be starting from scratch.
>
>So far I have tried with marginal success:
>
>vtkSurfaceReconstruction --- does some really weird looking stuff - worst 
>of the
>classes I tried
>vtkPowerCrust --- looks better but lots of holes in surface, appears to be
>closing contour faces (docs warn about needing point density, which I don't
>have)
>vtkDelauney3D ---- best looking but still a few holes
>
>vtkShepardMethod --- just starting to learn about this one...comments???
>
>Any ideas REALLY appreciated
>
>....john
>
>_______________________________________________
>This is the private VTK discussion list.
>Please keep messages on-topic. Check the FAQ at: 
><http://public.kitware.com/cgi-bin/vtkfaq>
>Follow this link to subscribe/unsubscribe:
>http://public.kitware.com/mailman/listinfo/vtkusers





More information about the vtkusers mailing list