Examples/Statistics/BayesianClassifier.cxx

/*=========================================================================
 *
 *  Copyright NumFOCUS
 *
 *  Licensed under the Apache License, Version 2.0 (the "License");
 *  you may not use this file except in compliance with the License.
 *  You may obtain a copy of the License at
 *
 *         https://www.apache.org/licenses/LICENSE-2.0.txt
 *
 *  Unless required by applicable law or agreed to in writing, software
 *  distributed under the License is distributed on an "AS IS" BASIS,
 *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 *  See the License for the specific language governing permissions and
 *  limitations under the License.
 *
 *=========================================================================*/
 
// This example demonstrates usage of the itk::BayesianClassifierImageFilter
// The input to this example is an itk::VectorImage that represents pixel
// memberships to 'n' classes.
//
// This image is conveniently generated by the
// BayesianClassifierInitializer.cxx example.
//
// The output of the filter is a label map (an image of unsigned char's) with
// pixel values indicating the classes they correspond to. Pixels with
// intensity 0 belong to the 0th class, 1 belong to the 1st class etc. The
// classification is done by applying a Maximum decision rule to the posterior
// image.
//
// The filter allows you to specify a prior image as well, (although this is
// not done in this example). The prior image, if specified will be a
// itk::VectorImage with as many components as the number of classes. The
// posterior image is then generated by multiplying the prior image with the
// membership image. If the prior image is not specified, the posterior image
// is the same as the membership image.
//
// The filter optionally accepts a smoothingIterations argument. See the
// itk::BayesianClassifierImageFilter for details on how this affects the
// classification. The philosophy is that the filter allows you to iteratively
// smooth the posteriors prior to applying the decision rule. It is hoped
// that this would yield a better classification. The user will need to plug
// in his own smoothing filter. In this case, we specify a
// GradientAnisotropicDiffusionImageFilter.
//
// Example args:
//   Memberships.mhd Labelmap.png  3
 
#include "itkImage.h"
#include "itkImageFileReader.h"
#include "itkBayesianClassifierImageFilter.h"
#include "itkImageFileWriter.h"
#include "itkGradientAnisotropicDiffusionImageFilter.h"
#include "itkRescaleIntensityImageFilter.h"
 
int
main(int argc, char * argv[])
{
 
  if (argc < 3)
  {
    std::cerr << "Usage: " << std::endl;
    std::cerr << argv[0]
              << " inputImageFile outputImageFile [smoothingIterations]"
              << std::endl;
    return EXIT_FAILURE;
  }
 
  try
  {
    constexpr unsigned int Dimension = 2;
    using InputPixelType = float;
    using InputImageType = itk::VectorImage<InputPixelType, Dimension>;
 
    auto input = itk::ReadImage<InputImageType>(argv[1]);
 
    using LabelType = unsigned char;
    using PriorType = float;
    using PosteriorType = float;
 
    using ClassifierFilterType =
      itk::BayesianClassifierImageFilter<InputImageType,
                                         LabelType,
                                         PosteriorType,
                                         PriorType>;
 
    auto filter = ClassifierFilterType::New();
 
    filter->SetInput(input);
 
    if (argc > 3)
    {
      filter->SetNumberOfSmoothingIterations(std::stoi(argv[3]));
      using ExtractedComponentImageType =
        ClassifierFilterType::ExtractedComponentImageType;
      using SmoothingFilterType =
        itk::GradientAnisotropicDiffusionImageFilter<
          ExtractedComponentImageType,
          ExtractedComponentImageType>;
      auto smoother = SmoothingFilterType::New();
      smoother->SetNumberOfIterations(1);
      smoother->SetTimeStep(0.125);
      smoother->SetConductanceParameter(3);
      filter->SetSmoothingFilter(smoother);
    }
 
    // SET FILTER'S PRIOR PARAMETERS
    // do nothing here to default to uniform priors
    // otherwise set the priors to some user provided values
 
    // Rescale the label map to the dynamic range of the datatype and write it
    using ClassifierOutputImageType = ClassifierFilterType::OutputImageType;
    using OutputImageType = itk::Image<unsigned char, Dimension>;
    using RescalerType =
      itk::RescaleIntensityImageFilter<ClassifierOutputImageType,
                                       OutputImageType>;
    auto rescaler = RescalerType::New();
    rescaler->SetInput(filter->GetOutput());
    rescaler->SetOutputMinimum(0);
    rescaler->SetOutputMaximum(255);
 
    // Write labelmap to file
    itk::WriteImage(rescaler->GetOutput(), argv[2]);
 
    // Testing print
    filter->Print(std::cout);
    std::cout << "Test passed." << std::endl;
  }
  catch (const itk::ExceptionObject & excp)
  {
    std::cerr << "ITK exception caught:\n" << excp << std::endl;
    return EXIT_FAILURE;
  }
  return EXIT_SUCCESS;
}