[Insight-users] MeshSpatialObject

[Luis Ibanez]

Hi Gheorhe,

There are two options that comes to mind:

1) You can rasterize your Mesh, and then use the resulting
    binary mask for computing the signed distance map.

OR

2) If you get the contours out of binary masks, in the
    first place, then you can rather forget about the
    contours and directly compute the signed distance
    map by using the original binary masks as input.


Option (1) can be implemented using:

http://www.itk.org/Insight/Doxygen/html/classitk_1_1TriangleMeshToBinaryImageFilter.html
and
http://www.itk.org/Insight/Doxygen/html/classitk_1_1ApproximateSignedDistanceMapImageFilter.html 

or
http://www.itk.org/Insight/Doxygen/html/classitk_1_1ApproximateSignedDistanceMapImageFilter.html
or
http://www.itk.org/Insight/Doxygen/html/classSignedMaurerDistanceMapImageFilter.html


Option (2) can be implemented using any of the three signed
distance functions above.

The Maurer version seems to requie smaller computing times.


     Regards,


        Luis



----------------------------
Gheorghe Postelnicu wrote:
> Hi,
> 
> I would have liked to use a MeshSpatialObject to get the signed
> distance function from a set of contours in 2D. I chose as a starting
> point the example code from MeshSpatialObject.cxx and just adapted it
> by changing the dimension from 3 to 2. However, the code results in a
> segmentation fault.
> 
> I am working with ITK 2.8.1 on Linux.
> Any suggestions?
> 
> Thanks,