[Insight-users] Problem grafting a C++ array as an ImageToImageFilter output

[Matt McCormick]

Hi Ramon,

The approach of using SetImportPointer on the Output is not guaranteed
to work.  The assumption is that the ImageToImageFilter is creating
its Output.  It is more reliable to copy the produced Output to the
buffer.

HTH,
Matt

On Fri, May 24, 2013 at 7:16 PM, Ramón Casero Cañas <rcasero at gmail.com> wrote:
> Hi all,
>
> I think there may be some problem with some of the itk::ImageToImageFilter
> when one tries to graft a C++ array to be used as the filter's output, or at
> least with the way I do it.
>
> This was kindly reported by Peter Thalmann in [1], and I'm trying to help
> him. This is a problem that I had also noticed with a couple of other
> filters.
>
> I have attached a minimal example of a C++ MEX file that creates a function
> that can be run from Matlab, with the CMake files to compile it. (I have
> tried this with ITK 4.3.1).
>
> Save everything to directory test, and then build the example from the shell
>
> cd test
> mkdir bin
> cd bin
> cmake ..
> make install
>
> To run the example from Matlab,
>
> cd test
> % create a test binary square with a little hole
> im = zeros(15, 15, 'uint8');
> im(3:13, 3:13) = 1;
> im(7:8, 7) = 0;
> im2 = itk_test(im);
>
> This runs filter itk::VotingBinaryIterativeHoleFillingImageFilter on the
> image. The program outputs to the Matlab shell both the content of
> filter->GetOutput(), and the content of the Matlab output array. As we can
> see, the output is not being saved to the array
>
> Filter result, reading from the filter
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> Filter result, reading from the Matlab output array
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
>
>
> But this approach works for other filters. If we choose the Median filter
> instead: in ItkTestSimpleImFilter.cpp, ucomment
>
>   // typedef itk::MedianImageFilter<ImageType, ImageType>
>   //   FilterType;
>
> and comment out
>
>   typedef itk::VotingBinaryIterativeHoleFillingImageFilter<ImageType>
>     FilterType;
>
> [...]
>
>   // filter parameters only for the
> VotingBinaryIterativeHoleFillingImageFilter
>   filter->SetMaximumNumberOfIterations(4);
>   filter->SetBackgroundValue(0);
>   filter->SetForegroundValue(1);
>   filter->SetMajorityThreshold(2);
>
> then we can see that the array is actually being used as the filter's output
>
>>> im2 = itk_test(im);
> Filter result, reading from the filter
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> Filter result, reading from the Matlab output array
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0
> 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
>
> I have been looking at itkVotingBinaryIterativeHoleFillingImageFilter.hxx,
> and I was wondering whether the problem is in
> VotingBinaryIterativeHoleFillingImageFilter< TInputImage >
> ::GenerateData().
>
> The filter works iteratively, doing
>
>   while ( m_CurrentNumberOfIterations < m_MaximumNumberOfIterations )
>     {
>     filter->SetInput(input);
>     filter->Update();
>
>     m_CurrentNumberOfIterations++;
>     progress.CompletedPixel();   // not really a pixel but an iteration
>     this->InvokeEvent( IterationEvent() );
>
>     const unsigned int numberOfPixelsChangedInThisIteration =
>       filter->GetNumberOfPixelsChanged();
>     m_NumberOfPixelsChanged += numberOfPixelsChangedInThisIteration;
>
>     output = filter->GetOutput();
>     output->DisconnectPipeline();
>     input = output;
>     if ( numberOfPixelsChangedInThisIteration == 0 )
>       {
>       break;
>       }
>     }
>   this->GraftOutput(output);
>
>
> Here, output gets a DisconnectPipeline(). I have tried commenting that like
> out, rebuilding and reinstalling ITK, but it doesn't seem to make a
> difference.
>
> [1]
> https://groups.google.com/forum/?fromgroups#!topic/gerardus-users/pLH0iR0H74o
>
> Best regards,
>
> Ramon.
>
>
>
>
> --
> Dr. Ramón Casero Cañas
>
> Oxford e-Research Centre (OeRC)
> University of Oxford
> 7 Keble Rd
> Oxford OX1 3QG
>
> tlf     +44 (0) 1865 610739
> web     http://www.cs.ox.ac.uk/people/Ramon.CaseroCanas
> photos  http://www.flickr.com/photos/rcasero/
>
> _____________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Kitware offers ITK Training Courses, for more information visit:
> http://www.kitware.com/products/protraining.php
>
> Please keep messages on-topic and check the ITK FAQ at:
> http://www.itk.org/Wiki/ITK_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://www.itk.org/mailman/listinfo/insight-users
>