[Insight-users] VectorConfidenceConnected compilation problem
Luis Ibanez
luis . ibanez at kitware . com
Fri, 05 Dec 2003 13:06:29 -0500
Hi Stefan,
It is certainly possible to use the same approach that
is used in the scalar version of ConfidenceConnected.
My only concern with this approach is the fact that
the user sets a value for controlling the thresholds
and the value is overriden by the filter.
But, since the behavior is already in the scalar version
of the filter it seems reasonable to add it to the RGB
version.
This issue has been entered as Bug #417 in the bug tracker
http://www . itk . org/Bug/
you may want to set up your own account in the bug
tracker so you can add comments to the bug entries, vote
for them and receive email updates on their status.
Please let us know if you find any problems getting
a login.
Thanks
Luis
-----------------
Stefan Lindenau wrote:
> Hi Luis,
>
> About your proposal for my second problem:
>
> I am not sure if it is possible but in my eyes the best solution would
> be to enforce that the seed points are lying in the range of intensities
> as it is done in the "normal" ConfidenceConnectedFilter. In this filter
> this is quiet simple since it requires only adjusting the lower and
> upper threshold, but I do not know how this could be done with the
> Mahalanobis distance that is used in the
> VectorConfidenceConnectedImageFilter.
>
> If this is not possible I think the most reasonable solution would be to
> throw an exception.
>
> Stefan
>
> Luis Ibanez wrote:
>
>>
>> Hi Stefan,
>>
>> Thanks for pointing this out.
>>
>> The "return" keyword was actually missing in these two methods.
>> A fix has been commited now to the CVS version.
>>
>> That takes care of the first issue in your message.
>>
>> About the second problem:
>>
>> A multiplier of 2.5 gives more tolerance than a multiplier
>> value of 1.85. In the second case it may simply happen that
>> the seed points do not match the range of intensities (as you
>> just suggested). This results in an empty region and in
>> divisions by zero when computing mean and covariance.
>>
>> We should probably throw an exception when no points are
>> being left in the segmented region...
>>
>> This means that your code will have to catch potential
>> exceptions when you run this segmentation filter.
>>
>> Would find this option to be reasonable ?
>>
>>
>>
>> Please let us know,
>>
>> Thanks
>>
>>
>> Luis
>>
>>
>> -------------------------
>> Stefan Lindenau wrote:
>>
>>> Hi all,
>>>
>>> when I use the GetMean() or the GetCovariance() functions of the
>>> VectorConfidencConnectedImageFilter I get a runtime error (Release
>>> Build) or a compiletime error (Debug Release). My Compiler is VC++ 6.0
>>>
>>> I use the functions in the following manner:
>>>
>>> const ConnectedFilterType::CovarianceMatrixType &covariance =
>>> connFilter->GetCovariance();
>>> const ConnectedFilterType::MeanVectorType &mean =
>>> connFilter->GetMean();
>>>
>>> The compiler in Debug Mode complains about:
>>> C:\Surgical
>>> Simulator\InsightToolkit-1.4.0\Code\BasicFilters\itkVectorConfidenceConnectedImageFilter.txx(319)
>>> : error C4716:
>>> 'itk::VectorConfidenceConnectedImageFilter<itk::Image<itk::RGBPixel<unsigned
>>> char>,3>,itk::Image<unsigned char,3> >::GetMean
>>> ' : must return a value
>>>
>>> which refers to this function:
>>> template <class TInputImage, class TOutputImage>
>>> const typename
>>> VectorConfidenceConnectedImageFilter<TInputImage,TOutputImage>::MeanVectorType
>>> &
>>> VectorConfidenceConnectedImageFilter<TInputImage,TOutputImage>
>>> ::GetMean() const
>>> {
>>> m_ThresholdFunction->GetMean();
>>> }
>>>
>>> if i put a return before the "m_ThresholdFunction->GetMean()" it
>>> compiles.
>>>
>>> But if I use for example 1.85 as Value for the multiplier I get this
>>> as output from the source code I have attached and there is no
>>> segmentation visible in the output Image.
>>>
>>> Mean:
>>>
>>> -1.#IND -1.#IND -1.#IND Covariance:
>>> -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
>>> -1.#IND
>>>
>>> If I use 2.5 for the multiplier everything is fine. My suspicion is
>>> that the mean and variance are developing away from the seed points
>>> and in the end the seed point are not matching the criteria any more.
>>>
>>> Is it all really an ITK Bug or did I something fundamental wrong?
>>>
>>>
>>> Thanks
>>> Stefan
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>> #include "itkRGBPixel.h"
>>> #include "itkImageFileReader.h"
>>> #include "itkImage.h"
>>> #include "itkVectorConfidenceConnectedImageFilter.h"
>>> #include "itkImageFileWriter.h"
>>>
>>>
>>> //#include "CommandProgress.h"
>>>
>>>
>>> int main(int argc,char **argv) {
>>> if(argc!=4) {
>>> std::cerr<<"Incorrect number of arguments:"<<std::endl;
>>> std::cerr<<"Please use this program in this
>>> way:"<<std::endl<<std::endl;
>>> std::cerr<<argv[0]<<" inputFileName outputFileName
>>> multiplier"<<std::endl;
>>> return -1;
>>> }
>>> /*
>>> Setting the typedefinitions as required from ITK
>>> */
>>> typedef itk::RGBPixel<unsigned char> InputPixelType;
>>> typedef itk::Image<InputPixelType,3> InputImageType;
>>> typedef itk::ImageFileReader<InputImageType> ReaderType;
>>>
>>> typedef unsigned char OutputPixelType;
>>> typedef itk::Image<OutputPixelType,3> OutputImageType;
>>> typedef
>>> itk::VectorConfidenceConnectedImageFilter<InputImageType,OutputImageType>
>>>
>>> ConnectedFilterType;
>>> typedef itk::ImageFileWriter<OutputImageType> WriterType;
>>>
>>> /*
>>> Declaration of necessary pointers - remember ITK uses smartpointers!
>>> */
>>> ReaderType::Pointer reader = ReaderType::New();
>>> ConnectedFilterType::Pointer connFilter =
>>> ConnectedFilterType::New();
>>> WriterType::Pointer writer = WriterType::New();
>>> const InputImageType::IndexType index = {48,196,198};
>>> const InputImageType::IndexType index2 = {29,248,103};
>>>
>>> //CommandProgress::Pointer observer = CommandProgress::New();
>>> /*
>>> Setting the filenames
>>> */
>>> std::string input=argv[1];
>>> std::string output=argv[2];
>>> std::istringstream iss(argv[3]);
>>> float multiplier;
>>> iss >> multiplier;
>>>
>>> /*
>>> Setting the reader up
>>> */
>>> reader->SetFileName(input.c_str());
>>> reader->ReleaseDataFlagOn();
>>> //reader->AddObserver(itk::StartEvent(),observer);
>>> //reader->AddObserver(itk::EndEvent(),observer);
>>>
>>> /*
>>> Setting the confidenceConnectedImageFilter up
>>> */
>>> connFilter->SetInput(reader->GetOutput());
>>> //connFilter->AddObserver(itk::StartEvent(),observer);
>>> //connFilter->AddObserver(itk::EndEvent(),observer);
>>> //connFilter->AddObserver(itk::ProgressEvent(),observer);
>>> connFilter->SetSeed(index);
>>> connFilter->AddSeed(index2);
>>> connFilter->SetInitialNeighborhoodRadius(3);
>>> connFilter->SetMultiplier(multiplier);
>>> connFilter->SetNumberOfIterations(5);
>>> connFilter->SetReplaceValue(255);
>>>
>>> //connFilter->SetSeed
>>> /*
>>> Setting the writer up
>>> */
>>> writer->SetFileName(output.c_str());
>>> //writer->SetInput(reader->GetOutput());
>>> writer->SetInput(connFilter->GetOutput());
>>> //writer->AddObserver(itk::StartEvent(),observer);
>>> //writer->AddObserver(itk::EndEvent(),observer);
>>> try {
>>> writer->Update();
>>> } catch (itk::ExceptionObject &err) {
>>> std::cerr<< "Exception caught:" << std::endl;
>>> std::cerr<< err <<std::endl;
>>> return -1;
>>> }
>>>
>>>
>>> const ConnectedFilterType::CovarianceMatrixType &covariance =
>>> connFilter->GetCovariance();
>>>
>>> const ConnectedFilterType::MeanVectorType &mean =
>>> connFilter->GetMean();
>>>
>>>
>>> std::cout << "Mean: " << std::endl;
>>> // Write the mean and covariance to a file
>>> for(unsigned int ii=0; ii<3; ii++)
>>> {
>>> std::cout << mean[ii] << " ";
>>> }
>>> std::cout << std::endl;
>>>
>>> std::cout << "Covariance: " << std::endl;
>>> for(unsigned int ki=0; ki<3; ki++)
>>> {
>>> for(unsigned int kj=0; kj<3; kj++)
>>> {
>>> std::cout << covariance[ki][kj] << " ";
>>> }
>>> std::cout << std::endl;
>>> }
>>> return 0;
>>> }
>>
>>
>>
>>
>>
>>
>
>
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk . org
> http://www . itk . org/mailman/listinfo/insight-users
>