From fef5d95e943e493bd95fef9395e1b5ac3b7e00d9 Mon Sep 17 00:00:00 2001
From: "nissim.hadar" <nissim.hadar@gmail.com>
Date: Mon, 20 Nov 2017 12:33:38 -0800
Subject: [PATCH] First fully-coded registration.

---
 tools/auto-tester/src/ITKImageComparer.cpp | 75 +++++++++++++++++++++-
 1 file changed, 73 insertions(+), 2 deletions(-)

diff --git a/tools/auto-tester/src/ITKImageComparer.cpp b/tools/auto-tester/src/ITKImageComparer.cpp
index b7e1ee2423..e0ebcb8c16 100644
--- a/tools/auto-tester/src/ITKImageComparer.cpp
+++ b/tools/auto-tester/src/ITKImageComparer.cpp
@@ -13,6 +13,10 @@
 #include <itkRGBPixel.h>
 #include <itkImage.h>
 #include <itkRGBToLuminanceImageFilter.h>
+#include <itkTranslationTransform.h>
+#include <itkRegularStepGradientDescentOptimizerv4.h>
+#include <itkMeanSquaresImageToImageMetricv4.h>
+#include <itkImageRegistrationMethodv4.h>
 
 ITKImageComparer::ITKImageComparer() {
     // These are smart pointers that do not need to be deleted
@@ -25,7 +29,7 @@ float ITKImageComparer::compareImages(QString resultImageFilename, QString expec
     expectedImageReader->SetFileName(expectedImageFilename.toStdString().c_str());
 
     // Images are converted to monochrome for comparison
-    using MonochromePixelType = unsigned char;
+    using MonochromePixelType = double;
     using MonochromeImageType = itk::Image<MonochromePixelType, Dimension>;
     using FilterType = itk::RGBToLuminanceImageFilter<RGBImageType, MonochromeImageType>;
     
@@ -35,5 +39,72 @@ float ITKImageComparer::compareImages(QString resultImageFilename, QString expec
     resultImageToMonochrome->SetInput(resultImageReader->GetOutput());
     expectedImageToMonochrome->SetInput(expectedImageReader->GetOutput());
     
-    return 0.0;
+    // Setup registration components
+    // The transform that will map the fixed image space into the moving image space
+    using TransformType =  itk::TranslationTransform<double, Dimension>;
+
+    // The optimizer explores the parameter space of the transform in search of optimal values of the metric
+    using OptimizerType =  itk::RegularStepGradientDescentOptimizerv4<double>;
+
+    // The metric will compare how well the two images match each other
+    using MetricType =  itk::MeanSquaresImageToImageMetricv4<MonochromeImageType, MonochromeImageType>;
+
+    //The registration method type is instantiated using the types of the fixed and moving images as well
+    // as the output transform type.This class is responsible for interconnecting the components 
+    using RegistrationType = itk::ImageRegistrationMethodv4<MonochromeImageType, MonochromeImageType, TransformType>;
+
+    // Create registration components (smart pointers, so no need to delete)
+    MetricType::Pointer metric = MetricType::New();
+    OptimizerType::Pointer optimizer = OptimizerType::New();
+    RegistrationType::Pointer registration = RegistrationType::New();
+
+    // The comparison metric needs an interpreter.
+    // The same interpolator is used for both images
+    using LinearInterpolatorType = itk::LinearInterpolateImageFunction<MonochromeImageType, double>;
+    LinearInterpolatorType::Pointer interpolator = LinearInterpolatorType::New();
+    metric->SetFixedInterpolator(interpolator);
+    metric->SetMovingInterpolator(interpolator);
+
+    // Connect components
+    registration->SetMetric(metric);
+    registration->SetOptimizer(optimizer);
+    registration->SetFixedImage(resultImageToMonochrome->GetOutput());
+    registration->SetMovingImage(expectedImageToMonochrome->GetOutput());
+
+    // Initialization
+    TransformType::Pointer movingInitialTransform = TransformType::New();
+    TransformType::ParametersType initialParameters(movingInitialTransform->GetNumberOfParameters());
+
+    initialParameters[0] = 0.0; // Initial offset in mm along X
+    initialParameters[1] = 0.0; // Initial offset in mm along Y
+
+    movingInitialTransform->SetParameters(initialParameters);
+    registration->SetMovingInitialTransform(movingInitialTransform);
+
+    // Set optimizer parameters
+    optimizer->SetLearningRate(4);
+    optimizer->SetMinimumStepLength(0.001);
+    optimizer->SetRelaxationFactor(0.5);
+    optimizer->SetNumberOfIterations(200);
+
+    // Define registration criteria
+    const unsigned int numberOfLevels = 1;
+    RegistrationType::ShrinkFactorsArrayType shrinkFactorsPerLevel;
+    shrinkFactorsPerLevel.SetSize(1);
+    shrinkFactorsPerLevel[0] = 1;
+    RegistrationType::SmoothingSigmasArrayType smoothingSigmasPerLevel;
+    smoothingSigmasPerLevel.SetSize(1);
+    smoothingSigmasPerLevel[0] = 0;
+    registration->SetNumberOfLevels(numberOfLevels);
+    registration->SetSmoothingSigmasPerLevel(smoothingSigmasPerLevel);
+    registration->SetShrinkFactorsPerLevel(shrinkFactorsPerLevel);
+
+    // Activate the registration project
+    try
+    {
+        registration->Update();
+        return 0.0;
+    } catch (itk::ExceptionObject & err) {
+        return EXIT_FAILURE;
+    }
 }