[Insight-users] Watershed Segmentation of 3D images

[Luis Ibanez]

Hi Nathaniel,

If your image is Binary, using the Watershed image filter
for separating cell may be an overkill. Watersheds are built
based on the notion of hill-descending, which in your case
will only happen abruptly in the boundaries of the cells.

Instead of Watershed, you may want to try Morphological Filters,
or the Voting filters. In particular the  HoleFilling filter
and its Iterative cousin:

http://www.itk.org/Insight/Doxygen/html/classitk_1_1VotingBinaryHoleFillingImageFilter.html
http://www.itk.org/Insight/Doxygen/html/classitk_1_1VotingBinaryIterativeHoleFillingImageFilter.html

Note that depending on whether your cells are background
or foreground you will have to setup the majority rule
accordingly.


If you really want to use the Watershed,
here are two possible options:

You could first pass your binary image through a gradien magnitude
recursive image filter with a sigma that is about 1/4 of the diameter
of the typical cell.

or

You could also run your binary image through a DanielssonDistanceMap
in order to obtain intensity ramps inside the cells. Then run watersheds
on top of those.



    Regards,


       Luis



-----------------------
Nathaniel Brown wrote:

> I am wondering if anyone has any experience using the 
> itkWatershedImageFilter to do segmentation of 3D binary images?  I am 
> trying to separate the cells of an open cell foam.  Any help or 
> suggestions would be much appreciated.
> 
> Thanks,
> Nathaniel
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> 
>