/*=========================================================================
 *
 *  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 itkDiscreteGradientMagnitudeGaussianImageFunction_h
#define itkDiscreteGradientMagnitudeGaussianImageFunction_h

#include "itkNeighborhoodOperatorImageFunction.h"
#include "itkGaussianDerivativeOperator.h"

namespace itk
{
/**
 * \class DiscreteGradientMagnitudeGaussianImageFunction
 * \brief Compute the discrete gradient magnitude gaussian of an the image
 *        at a specific location in space, i.e. point, index or continuous
 *        index. This class computes a single derivative given the order in
 *        each direction (by default zero).
 * This class is templated over the input image type.
 *
 * The Initialize() method must be called after setting the parameters and before
 * evaluating the function.
 *
 * \author Ivan Macia, Vicomtech, Spain, https://www.vicomtech.org/en
 *
 * This implementation was taken from the Insight Journal paper:
 * https://www.insight-journal.org/browse/publication/179
 *
 * \sa NeighborhoodOperator
 * \sa ImageFunction
 * \ingroup ITKReview
 */
template <typename TInputImage, typename TOutput = double>
class ITK_TEMPLATE_EXPORT DiscreteGradientMagnitudeGaussianImageFunction
  : public ImageFunction<TInputImage, TOutput, TOutput>
{
public:
  /**Standard "Self" type alias */
  using Self = DiscreteGradientMagnitudeGaussianImageFunction;

  /** Standard "Superclass" type alias */
  using Superclass = ImageFunction<TInputImage, TOutput, TOutput>;

  /** Smart pointer type alias support */
  using Pointer = SmartPointer<Self>;
  using ConstPointer = SmartPointer<const Self>;

  /** Method for creation through the object factory */
  itkNewMacro(Self);

  /** \see LightObject::GetNameOfClass() */
  itkOverrideGetNameOfClassMacro(DiscreteGradientMagnitudeGaussianImageFunction);

  /** Image dependent types */
  using typename Superclass::InputImageType;
  using typename Superclass::InputPixelType;
  using typename Superclass::IndexType;
  using typename Superclass::IndexValueType;
  using typename Superclass::ContinuousIndexType;
  using typename Superclass::PointType;

  /** Dimension of the underlying image */
  static constexpr unsigned int ImageDimension2 = InputImageType::ImageDimension;

  /** Output type */
  using typename Superclass::OutputType;

  /** Arrays for native types */
  using VarianceArrayType = FixedArray<double, Self::ImageDimension2>;
  using OrderArrayType = FixedArray<unsigned int, Self::ImageDimension2>;

  using GaussianDerivativeOperatorType = itk::GaussianDerivativeOperator<TOutput, Self::ImageDimension2>;

  /** Array to store gaussian derivative operators one for each dimension */
  using GaussianDerivativeOperatorArrayType = FixedArray<GaussianDerivativeOperatorType, 2 * Self::ImageDimension2>;

  /** Precomputed N-dimensional derivative kernel */
  using KernelType = Neighborhood<TOutput, Self::ImageDimension2>;

  /** Array to store precomputed N-dimensional kernels for the gradient
    components */
  using KernelArrayType = FixedArray<KernelType, Self::ImageDimension2>;

  /** Image function that performs convolution with the neighborhood operator */
  using OperatorImageFunctionType = NeighborhoodOperatorImageFunction<InputImageType, TOutput>;
  using OperatorImageFunctionPointer = typename OperatorImageFunctionType::Pointer;

  using InterpolationModeEnum = itk::GaussianDerivativeOperatorEnums::InterpolationMode;
#if !defined(ITK_LEGACY_REMOVE)
  /**Exposes enums values for backwards compatibility*/
  static constexpr InterpolationModeEnum NearestNeighbourInterpolation =
    InterpolationModeEnum::NearestNeighbourInterpolation;
  static constexpr InterpolationModeEnum LinearInterpolation = InterpolationModeEnum::LinearInterpolation;
#endif

public:
  /** Evaluate the function in the given dimension at specified point */
  OutputType
  Evaluate(const PointType & point) const override;

  /** Evaluate the function at specified Index position */
  OutputType
  EvaluateAtIndex(const IndexType & index) const override;

  /** Evaluate the function at specified ContinuousIndex position */
  OutputType
  EvaluateAtContinuousIndex(const ContinuousIndexType & index) const override;

  /** Set/Get the variance for the discrete Gaussian kernel.
   * Sets the variance for individual dimensions. The default is 0.0 in each dimension.
   * If UseImageSpacing is true, the units are the physical units of your image.
   * If UseImageSpacing is false then the units are pixels */
  itkSetMacro(Variance, VarianceArrayType);
  itkGetConstMacro(Variance, const VarianceArrayType);
  itkSetVectorMacro(Variance, double, VarianceArrayType::Length);

  /** Convenience method for setting the variance for all dimensions */
  virtual void
  SetVariance(double variance)
  {
    m_Variance.Fill(variance);
    this->Modified();
  }

  /** Convenience method for setting the variance through the standard deviation
   */
  void
  SetSigma(const double sigma)
  {
    SetVariance(sigma * sigma);
  }

  /** Set/Get the desired maximum error of the gaussian approximation.  Maximum
   * error is the difference between the area under the discrete Gaussian curve
   * and the area under the continuous Gaussian. Maximum error affects the
   * Gaussian operator size. The value is clamped between 0.00001 and
   * 0.99999. */
  itkSetClampMacro(MaximumError, double, 0.00001, 0.99999);
  itkGetConstMacro(MaximumError, double);

  /** Set/Get the flag for calculating scale-space normalized derivatives.
   * Normalized derivatives are obtained multiplying by the scale
   * parameter t. */
  itkSetMacro(NormalizeAcrossScale, bool);
  itkGetConstMacro(NormalizeAcrossScale, bool);
  itkBooleanMacro(NormalizeAcrossScale);

  /** Set/Get the flag for using image spacing when calculating derivatives. */
  itkSetMacro(UseImageSpacing, bool);
  itkGetConstMacro(UseImageSpacing, bool);
  itkBooleanMacro(UseImageSpacing);

  /** Set/Get a limit for growth of the kernel. Small maximum error values with
   *  large variances will yield very large kernel sizes. This value can be
   *  used to truncate a kernel in such instances. A warning will be given on
   *  truncation of the kernel. */
  itkSetMacro(MaximumKernelWidth, unsigned int);
  itkGetConstMacro(MaximumKernelWidth, unsigned int);

  /** Set/Get the interpolation mode. */
  itkSetEnumMacro(InterpolationMode, InterpolationModeEnum);
  itkGetEnumMacro(InterpolationMode, InterpolationModeEnum);

  /** Set the input image.
   * \warning this method caches BufferedRegion information.
   * If the BufferedRegion has changed, user must call
   * SetInputImage again to update cached values. */
  void
  SetInputImage(const InputImageType * ptr) override;

  /** Initialize the Gaussian kernel. Call this method before evaluating the function.
   * This method MUST be called after any changes to function parameters. */
  virtual void
  Initialize()
  {
    RecomputeGaussianKernel();
  }

protected:
  DiscreteGradientMagnitudeGaussianImageFunction();
  DiscreteGradientMagnitudeGaussianImageFunction(const Self &) {}

  ~DiscreteGradientMagnitudeGaussianImageFunction() override = default;

  void
  operator=(const Self &)
  {}
  void
  PrintSelf(std::ostream & os, Indent indent) const override;

  void
  RecomputeGaussianKernel();

  // void RecomputeContinuousGaussianKernel(
  //        const double* offset) const;

private:
  /** Desired variance of the discrete Gaussian function */
  VarianceArrayType m_Variance{};

  /** Difference between the areas under the curves of the continuous and
   * discrete Gaussian functions */
  double m_MaximumError{ 0.005 };

  /** Maximum kernel size allowed.  This value is used to truncate a kernel
   *  that has grown too large.  A warning is given when the specified maximum
   *  error causes the kernel to exceed this size */
  unsigned int m_MaximumKernelWidth{ 30 };

  /** Array of derivative operators, one for each dimension and order.
   * First N zero-order operators are stored, then N first-order making
   * 2*N operators altogether where N=ImageDimension */
  GaussianDerivativeOperatorArrayType m_OperatorArray{};

  /** Array of N-dimensional kernels used to calculate gradient components */
  KernelArrayType m_KernelArray{};

  /** OperatorImageFunction */
  OperatorImageFunctionPointer m_OperatorImageFunction{};

  /** Flag for scale-space normalization of derivatives */
  bool m_NormalizeAcrossScale{ true };

  /** Flag to indicate whether to use image spacing */
  bool m_UseImageSpacing{ true };

  /** Interpolation mode */
  InterpolationModeEnum m_InterpolationMode{ InterpolationModeEnum::NearestNeighbourInterpolation };
};
} // namespace itk

#ifndef ITK_MANUAL_INSTANTIATION
#  include "itkDiscreteGradientMagnitudeGaussianImageFunction.hxx"
#endif

#endif