[Insight-users] Graft error: cannot cast: is it a mini pipeline problem?

[lynx.abraxas at freenet.de]

Dear Richard Beare,


Thanks  for  Your  hints. Pointing me the HMinima filter broght up the idea to
not let it transform between two different data formats (eg short and char). I
checked this and found out if the input and output format are the same (either
signed or unsigned short) it works.  I'm  not  sure  if  that's  a  bug  or  a
requirement. If it is a requirement it should be mentioned in the docs. As far
as I understand the input format  has  not  to  match  the  output  format  in
general.
Thanks  alsor  for  describing  how to debug filter. Why do I not have to call
update on each filter if update the writer? Is the  writer  update()  updating
all filters in the pipeline as well?

Regards,
Lynx


On 26/07/09 17:30:50, Richard Beare wrote:
> Hi,
> 
> On closer examination it looks to me like there is something wrong
> with the HMinima filter. The class has two template parameters - input
> and output image types, but the internal erode filter is instantiated
> using only the input type. This looks like a bug to me.
> 
> On Sat, Jul 25, 2009 at 8:44 PM, Richard Beare<richard.beare at gmail.com> wrote:
> > The error seems to be coming from the HMinima filter (determined by
> > updating each filter in your pipeline individually and asking the
> > debugger where the crash happens), but I can't see any obvious problem
> > with it. Perhaps there is some problem with reconstruction by erosion
> > filter, used by the HMinima filter.
> >
> >
> > On Sat, Jul 25, 2009 at 5:38 PM, <lynx.abraxas at freenet.de> wrote:
> >> Hello!
> >>
> >>
> >> Looking  at the other threads about Graft() I wonder if my "cannot cast" error
> >> has something to do with some minipipeline somewhere in the filters I use.  As
> >> far  as  I  understand  MorphologicalWatershedImageFilter  has a mini pipeline
> >> where as MorphologicalWatershedFromMarkersImageFilter has not.
> >> Another idea I have: Could the filters I connect in my program  be  the  wrong
> >> ones  and they don't fit to each other? Contrary they are pretty much the same
> >> as in segGel.cxx (from the watershed insight publication).
> >> How can I find out what PKN3itk10DataObjectE and  PKN3itk5ImageIsLj2EEE  stand
> >> for?
> >>
> >> Many thanks for any hints or help
> >> Lynx
> >>
> >> On 23/07/09 23:31:21, lynx.abraxas at freenet.de wrote:
> >>> Hello!
> >>>
> >>>
> >>> I've written the little test program below but it only works if I use MorphologicalWatershedImageFilter directly but not if I use MorphologicalWatershedFromMarkersImageFilter and hm, rm and cc as discribed in the publication. I used segGel.cxx as an example.
> >>> What is wrong with my test in long form?
> >>> What is causing:
> >>> terminate called after throwing an instance of 'itk::ExceptionObject'
> >>>   what():  /usr/local/include/InsightToolkit/Common/itkImage.txx:135:
> >>> itk::ERROR: Image(0x85dd268): itk::Image::Graft() cannot cast PKN3itk10DataObjectE to PKN3itk5ImageIsLj2EEE
> >>>
> >>> Many thanks for any hints or help
> >>> Lynx