1 /*=========================================================================
2 Program: vv http://www.creatis.insa-lyon.fr/rio/vv
5 - University of LYON http://www.universite-lyon.fr/
6 - Léon Bérard cancer center http://oncora1.lyon.fnclcc.fr
7 - CREATIS CNRS laboratory http://www.creatis.insa-lyon.fr
9 This software is distributed WITHOUT ANY WARRANTY; without even
10 the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
11 PURPOSE. See the copyright notices for more information.
13 It is distributed under dual licence
15 - BSD See included LICENSE.txt file
16 - CeCILL-B http://www.cecill.info/licences/Licence_CeCILL-B_V1-en.html
17 ======================================================================-====*/
20 #include "clitkCommon.h"
24 #include "itkImageFileReader.h"
25 #include "itkImageSeriesReader.h"
26 #include "itkImageFileWriter.h"
27 #include "itkRecursiveGaussianImageFilter.h"
28 #include "itkResampleImageFilter.h"
29 #include "itkAffineTransform.h"
30 #include "itkNearestNeighborInterpolateImageFunction.h"
31 #include "itkLinearInterpolateImageFunction.h"
32 #include "itkBSplineInterpolateImageFunction.h"
33 #include "itkBSplineInterpolateImageFunctionWithLUT.h"
34 #include "itkCommand.h"
39 //--------------------------------------------------------------------
40 template <class TInputImage, class TOutputImage>
41 ResampleImageWithOptionsFilter<TInputImage, TOutputImage>::
42 ResampleImageWithOptionsFilter():itk::ImageToImageFilter<TInputImage, TOutputImage>()
44 static const unsigned int dim = InputImageType::ImageDimension;
45 this->SetNumberOfRequiredInputs(1);
46 m_OutputIsoSpacing = -1;
47 m_InterpolationType = NearestNeighbor;
48 m_GaussianFilteringEnabled = true;
50 m_BLUTSamplingFactor = 20;
51 m_LastDimensionIsTime = false;
52 m_Transform = TransformType::New();
53 if (dim == 4) m_LastDimensionIsTime = true; // by default 4D is 3D+t
54 for(unsigned int i=0; i<dim; i++) {
56 m_OutputSpacing[i] = -1;
57 m_GaussianSigma[i] = -1;
59 m_VerboseOptions = false;
61 //--------------------------------------------------------------------
64 //--------------------------------------------------------------------
65 template <class TInputImage, class TOutputImage>
67 ResampleImageWithOptionsFilter<TInputImage, TOutputImage>::
68 SetInput(const InputImageType * image)
70 // Process object is not const-correct so the const casting is required.
71 this->SetNthInput(0, const_cast<InputImageType *>(image));
73 //--------------------------------------------------------------------
76 //--------------------------------------------------------------------
77 template <class TInputImage, class TOutputImage>
79 ResampleImageWithOptionsFilter<TInputImage, TOutputImage>::
80 GenerateInputRequestedRegion()
82 // call the superclass's implementation of this method
83 Superclass::GenerateInputRequestedRegion();
85 // get pointers to the input and output
86 InputImagePointer inputPtr =
87 const_cast< TInputImage *>( this->GetInput() );
89 // Request the entire input image
90 InputImageRegionType inputRegion;
91 inputRegion = inputPtr->GetLargestPossibleRegion();
92 inputPtr->SetRequestedRegion(inputRegion);
94 //--------------------------------------------------------------------
97 //--------------------------------------------------------------------
98 template <class TInputImage, class TOutputImage>
100 ResampleImageWithOptionsFilter<TInputImage, TOutputImage>::
101 GenerateOutputInformation()
103 static const unsigned int dim = InputImageType::ImageDimension;
106 if (!std::numeric_limits<InputImagePixelType>::is_signed) {
107 if ((m_InterpolationType == BSpline) ||
108 (m_InterpolationType == B_LUT)) {
109 std::cerr << "Warning : input pixel type is not signed, use bspline interpolation at your own risk ..." << std::endl;
114 InputImagePointer input = dynamic_cast<InputImageType*>(itk::ProcessObject::GetInput(0));
116 // Perform default implementation
117 Superclass::GenerateOutputInformation();
120 InputImageSpacingType inputSpacing = input->GetSpacing();
121 InputImageSizeType inputSize = input->GetLargestPossibleRegion().GetSize();
123 if (m_OutputIsoSpacing != -1) { // apply isoSpacing
124 for(unsigned int i=0; i<dim; i++) {
125 m_OutputSpacing[i] = m_OutputIsoSpacing;
126 m_OutputSize[i] = (int)lrint(inputSize[i]*inputSpacing[i]/m_OutputSpacing[i]);
129 if (m_OutputSpacing[0] != -1) { // apply spacing, compute size
130 for(unsigned int i=0; i<dim; i++) {
131 m_OutputSize[i] = (int)lrint(inputSize[i]*inputSpacing[i]/m_OutputSpacing[i]);
134 if (m_OutputSize[0] != 0) { // apply size, compute spacing
135 for(unsigned int i=0; i<dim; i++) {
136 m_OutputSpacing[i] = (double)inputSize[i]*inputSpacing[i]/(double)m_OutputSize[i];
138 } else { // copy input size/spacing ... (no resampling)
139 m_OutputSize = inputSize;
140 m_OutputSpacing = inputSpacing;
145 // Special case for temporal image 2D+t or 3D+t
146 if (m_LastDimensionIsTime) {
148 m_OutputSize[l] = inputSize[l];
149 m_OutputSpacing[l] = inputSpacing[l];
153 OutputImagePointer outputImage = this->GetOutput(0);
154 OutputImageRegionType region;
155 region.SetSize(m_OutputSize);
156 region.SetIndex(input->GetLargestPossibleRegion().GetIndex());
157 DD(input->GetLargestPossibleRegion().GetIndex());
158 outputImage->SetLargestPossibleRegion(region);
159 outputImage->SetSpacing(m_OutputSpacing);
161 // Init Gaussian sigma
162 if (m_GaussianSigma[0] != -1) { // Gaussian filter set by user
163 m_GaussianFilteringEnabled = true;
165 if (m_GaussianFilteringEnabled) { // Automated sigma when downsample
166 for(unsigned int i=0; i<dim; i++) {
167 if (m_OutputSpacing[i] > inputSpacing[i]) { // downsample
168 m_GaussianSigma[i] = 0.5*m_OutputSpacing[i];// / inputSpacing[i]);
169 } else m_GaussianSigma[i] = 0; // will be ignore after
173 if (m_GaussianFilteringEnabled && m_LastDimensionIsTime) {
174 m_GaussianSigma[dim-1] = 0;
177 //--------------------------------------------------------------------
180 //--------------------------------------------------------------------
181 template <class TInputImage, class TOutputImage>
183 ResampleImageWithOptionsFilter<TInputImage, TOutputImage>::
188 InputImagePointer input = dynamic_cast<InputImageType*>(itk::ProcessObject::GetInput(0));
189 static const unsigned int dim = InputImageType::ImageDimension;
191 // Create main Resample Image Filter
192 typedef itk::ResampleImageFilter<InputImageType,OutputImageType> FilterType;
193 typename FilterType::Pointer filter = FilterType::New();
194 filter->GraftOutput(this->GetOutput());
195 // this->GetOutput()->Print(std::cout);
196 // this->GetOutput()->SetBufferedRegion(this->GetOutput()->GetLargestPossibleRegion());
197 // this->GetOutput()->Print(std::cout);
199 // Print options if needed
200 if (m_VerboseOptions) {
201 std::cout << "Output Spacing = " << m_OutputSpacing << std::endl
202 << "Output Size = " << m_OutputSize << std::endl
203 << "Gaussian = " << m_GaussianFilteringEnabled << std::endl;
204 if (m_GaussianFilteringEnabled)
205 std::cout << "Sigma = " << m_GaussianSigma << std::endl;
206 std::cout << "Interpol = ";
207 switch (m_InterpolationType) {
208 case NearestNeighbor:
209 std::cout << "NearestNeighbor" << std::endl;
212 std::cout << "Linear" << std::endl;
215 std::cout << "BSpline " << m_BSplineOrder << std::endl;
218 std::cout << "B-LUT " << m_BSplineOrder << " " << m_BLUTSamplingFactor << std::endl;
221 std::cout << "Threads = " << this->GetNumberOfThreads() << std::endl;
222 std::cout << "LastDimIsTime = " << m_LastDimensionIsTime << std::endl;
225 // Instance of the transform object to be passed to the resample
226 // filter. By default, identity transform is applied
227 filter->SetTransform(m_Transform);
228 filter->SetSize(m_OutputSize);
229 filter->SetOutputSpacing(m_OutputSpacing);
230 filter->SetOutputOrigin(input->GetOrigin());
231 filter->SetDefaultPixelValue(m_DefaultPixelValue);
232 filter->SetNumberOfThreads(this->GetNumberOfThreads());
234 // Select interpolator
235 switch (m_InterpolationType) {
236 case NearestNeighbor: {
237 typedef itk::NearestNeighborInterpolateImageFunction<InputImageType, double> InterpolatorType;
238 typename InterpolatorType::Pointer interpolator = InterpolatorType::New();
239 filter->SetInterpolator(interpolator);
243 typedef itk::LinearInterpolateImageFunction<InputImageType, double> InterpolatorType;
244 typename InterpolatorType::Pointer interpolator = InterpolatorType::New();
245 filter->SetInterpolator(interpolator);
249 typedef itk::BSplineInterpolateImageFunction<InputImageType, double> InterpolatorType;
250 typename InterpolatorType::Pointer interpolator = InterpolatorType::New();
251 interpolator->SetSplineOrder(m_BSplineOrder);
252 filter->SetInterpolator(interpolator);
256 typedef itk::BSplineInterpolateImageFunctionWithLUT<InputImageType, double> InterpolatorType;
257 typename InterpolatorType::Pointer interpolator = InterpolatorType::New();
258 interpolator->SetSplineOrder(m_BSplineOrder);
259 interpolator->SetLUTSamplingFactor(m_BLUTSamplingFactor);
260 filter->SetInterpolator(interpolator);
265 // Initial Gaussian blurring if needed
266 typedef itk::RecursiveGaussianImageFilter<InputImageType, InputImageType> GaussianFilterType;
267 std::vector<typename GaussianFilterType::Pointer> gaussianFilters;
268 if (m_GaussianFilteringEnabled) {
269 for(unsigned int i=0; i<dim; i++) {
270 if (m_GaussianSigma[i] != 0) {
271 gaussianFilters.push_back(GaussianFilterType::New());
272 gaussianFilters[i]->SetDirection(i);
273 gaussianFilters[i]->SetOrder(GaussianFilterType::ZeroOrder);
274 gaussianFilters[i]->SetNormalizeAcrossScale(false);
275 gaussianFilters[i]->SetSigma(m_GaussianSigma[i]); // in millimeter !
276 if (gaussianFilters.size() == 1) { // first
277 gaussianFilters[0]->SetInput(input);
279 gaussianFilters[i]->SetInput(gaussianFilters[i-1]->GetOutput());
283 if (gaussianFilters.size() > 0) {
284 filter->SetInput(gaussianFilters[gaussianFilters.size()-1]->GetOutput());
285 } else filter->SetInput(input);
286 } else filter->SetInput(input);
292 // DD("before Graft");
293 this->GraftOutput(filter->GetOutput());
294 // DD("after Graft");
296 //--------------------------------------------------------------------