[Insight-users] Watershed Filter (flat regions)

Richard Beare richard.beare at gmail.com
Thu Apr 26 18:51:48 EDT 2007


Hi,

Flat regions alone shouldn't cause the problem you mention with the
"standard" itk watershed. One of the examples in the report you
mentioned is an artificial image with only flat regions and that
didn't have any problems executing. The issue with flat regions is
that the standard watershed in itk may not give you the expected
result.

What image type are you using for your label image? Could there be
some sort of overflow problem?

Have a try with the morphological watershed and see if it runs - it
doesn't use the hash facilities directly, although the connected
component labelling step may reference it.

On 4/27/07, Langkammer,  Christian (SCR US EXT)
<christian.langkammer.ext at siemens.com> wrote:
> Hello,
>
> I am using the WatershedFilterType to segment CT images.
> When big flat regions with the same pixel values (e.g. masked out
> organs) are used as input a division through zero occurs in
> itk_hashtable.h:640.
> Does someone know details about that and could it be due to the problems
> addressed in [1]?
>
> Adding some noise to this regions solves the problem but it will not
> affect the performance positive...
>
> Is there a calculable upper limit for the size of connected input
> regions with the same intensity?
>
> Thanks and best regards,
> Christian
>
>
> [1]
> http://insight-journal.org/dspace/bitstream/1926/202/2/The_watershed_tra
> nsform_in_ITK_-_discussion_and_new_developments.pdf
>
>
> _______________________________________________
> 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