[Insight-users] smoothing an Histogram

[Luis Ibanez]

Hi Raquel,

The RecursiveGaussianImageFilter is N-Dimensional,
you can run it in a 1-D image.


Just do:


   typedef itk::RecursiveGaussianImageFilter<
     ImageType1D, ImageType1D >  SmoothingFilterType;

   SmoothingFilterType::Pointer smoothingFilter =
       SmoothingFilterType::New();

   smoothingFilter->SetSigma( 1.5 ); // 1.5mm (change this value)

   smoothingFilter->SetInput( line );
   smoothingFilter->Update();

   writer->SetInput( smoothingFilter->GetOutput() );



FYI:
By default the filter performs smoothing
but you could also set it to compute first
or second derivatives (if you need to).



   Regards,



       Luis



-----------------
Raquel Itk wrote:
> Hi Luis,
> I convert the Histogram into a 1DImage in this way:
> 
> typedef unsigned short   PixelType1D;
> typedef itk::GDCMImageIO ImageIOType;
> typedef itk::Image< PixelType1D, 1 > ImageType1D;
> typedef itk::Image< PixelType1D, 1 > ImageType1D_2;
> typedef itk::ImageFileReader< ImageType1D > ReaderType1D;
> typedef itk::ImageRegionIterator< ImageType1D>       IteratorType;
> 
> ReaderType::Pointer readerHis = ReaderType::New();
> ImageIOType::Pointer gdcmImageIOHis= ImageIOType::New();
> HistogramGeneratorType::Pointer histogramGenerator = 
> HistogramGeneratorType::New();
> ReaderType1D::Pointer reader1D = ReaderType1D::New();
> ImageType1D::Pointer line = ImageType1D::New();
> ImageType1D::SizeType size1D;
> ImageType1D::IndexType pixelValue;
> ImageType1D::IndexType start1D;
> ImageType1D::RegionType region1D;
> 
> 
> 
> void createImage(int *valoresHisto){
> 
>     size1D.Fill( 2286 );
>     start1D.Fill( 0 );
>     region1D.SetSize( size1D );
>     region1D.SetIndex( start1D );
>     line->SetRegions( region1D );
>     line->Allocate();
> 
>     IteratorType lineIt(line, line->GetRequestedRegion());
> 
>     int i = 0;
>     for (lineIt.GoToBegin(); !lineIt.IsAtEnd();
>     ++lineIt)
>     {
>         lineIt.Set( valoresHisto[i]);
> 
>         i=i+1;
>     }
> 
> 
>     typedef itk::ImageFileWriter< ImageType1D > WriterType1D;
>     WriterType1D::Pointer writer1D = WriterType1D::New();
>     writer1D->SetFileName( "Image1D.png" );
> 
>     writer1D->SetInput( line );
>     try
>     {
>         writer1D->Update();
>     }
>     catch (itk::ExceptionObject &e)
>     {
>         std::cerr << e << std::endl;
> 
>     }
> }
> 
> Now I want to apply SmoothingRecursiveGaussianImageFilter2.cxx, but I 
> have a problem because if need a 2d Image. How can I resolve it?
> 
> Thanks
> 
> Raquel
> 
> 
> 2009/7/4 Luis Ibanez <luis.ibanez at kitware.com 
> <mailto:luis.ibanez at kitware.com>>
> 
> 
>     Hi Raquel,
> 
>     You can convert the Histogram into a 1D Image,
>     and use any of the ITK Smoothing filters on
>     this 1D image, then convert the image back
>     to a Histogram.
> 
>     See the class
>     Insight/Code/Numerics/Statistics/itkHistogramToImageFilter.h
> 
>     and its related classes
> 
>     itkHistogramToEntropyImageFilter.h
>     itkHistogramToIntensityImageFilter.h
>     itkHistogramToLogProbabilityImageFilter.h
>     itkHistogramToProbabilityImageFilter.h
> 
>     Then look at the section of Smoothing image filters
>     in the ITK Software Guide
> 
> 
>      http://www.itk.org/ItkSoftwareGuide.pdf
> 
> 
> 
>       Regards,
> 
> 
>           Luis
> 
> 
> 
>     --------------------
>     Raquel Itk wrote:
> 
>         Hi,
> 
>         I have computed the histogram of an image with
>         Examples/Statistics/ImageHistogram2.cxx, now I want to smooth
>         the array , where I have saved the values of an Histogram (
>         valoresHisto[i]) because I have a lot of peaks. I would like to
>         use a Gaussian method, I have been read about Gaussian but only
>         for Images.
>          
>         while( itr != end )
> 
>         {
> 
>         valoresHisto[i]=itr.GetFrequency();
> 
>         ++itr;
> 
>         ++binNumber;
> 
>         ++i;
> 
>         }
> 
>         return(valoresHisto);
> 
> 
>         Thanks
> 
> 
>         Raquel
> 
> 
>         ------------------------------------------------------------------------
> 
>         _____________________________________
>         Powered by www.kitware.com <http://www.kitware.com>
> 
>         Visit other Kitware open-source projects at
>         http://www.kitware.com/opensource/opensource.html
> 
>         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
> 
>