/*=========================================================================
 *
 *  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.
 *
 *=========================================================================*/
#ifndef itkCovarianceImageFunction_hxx
#define itkCovarianceImageFunction_hxx

#include "itkMatrix.h"
#include "itkConstNeighborhoodIterator.h"

namespace itk
{

template <typename TInputImage, typename TCoordRep>
CovarianceImageFunction<TInputImage, TCoordRep>::CovarianceImageFunction()

  = default;

template <typename TInputImage, typename TCoordRep>
auto
CovarianceImageFunction<TInputImage, TCoordRep>::EvaluateAtIndex(const IndexType & index) const -> RealType
{
  using PixelType = typename TInputImage::PixelType;
  using PixelComponentType = typename PixelType::ValueType;

  using PixelComponentRealType = typename NumericTraits<PixelComponentType>::RealType;


  if (!this->GetInputImage())
  {
    itkExceptionMacro("No image connected to CovarianceImageFunction");
  }

  const unsigned int VectorDimension = this->GetInputImage()->GetNumberOfComponentsPerPixel();
  RealType           covariance(VectorDimension, VectorDimension);

  if (!this->IsInsideBuffer(index))
  {
    covariance.fill(NumericTraits<PixelComponentRealType>::max());
    return covariance;
  }


  covariance.fill(PixelComponentRealType{});

  using MeanVectorType = vnl_vector<PixelComponentRealType>;
  MeanVectorType mean(VectorDimension);
  mean.fill(PixelComponentRealType{});

  // Create an N-d neighborhood kernel, using a zeroflux boundary condition
  typename InputImageType::SizeType kernelSize;
  kernelSize.Fill(m_NeighborhoodRadius);

  ConstNeighborhoodIterator<InputImageType> it(
    kernelSize, this->GetInputImage(), this->GetInputImage()->GetBufferedRegion());

  // Set the iterator at the desired location
  it.SetLocation(index);

  // Walk the neighborhood
  const unsigned int size = it.Size();
  for (unsigned int i = 0; i < size; ++i)
  {
    const PixelType pixel = it.GetPixel(i);

    for (unsigned int dimx = 0; dimx < VectorDimension; ++dimx)
    {
      mean[dimx] += pixel[dimx];
      for (unsigned int dimy = 0; dimy < VectorDimension; ++dimy)
      {
        covariance[dimx][dimy] +=
          static_cast<PixelComponentRealType>(pixel[dimx]) * static_cast<PixelComponentRealType>(pixel[dimy]);
      }
    }
  }

  const auto rsize = static_cast<PixelComponentRealType>(size);

  mean /= rsize;

  for (unsigned int dimx = 0; dimx < VectorDimension; ++dimx)
  {
    for (unsigned int dimy = 0; dimy < VectorDimension; ++dimy)
    {
      covariance[dimx][dimy] /= rsize;
      covariance[dimx][dimy] -= mean[dimx] * mean[dimy];
    }
  }

  return (covariance);
}

template <typename TInputImage, typename TCoordRep>
void
CovarianceImageFunction<TInputImage, TCoordRep>::PrintSelf(std::ostream & os, Indent indent) const
{
  Superclass::PrintSelf(os, indent);

  os << indent << "NeighborhoodRadius: " << m_NeighborhoodRadius << std::endl;
}
} // end namespace itk

#endif