covariance.h

/*
 Copyright (C) 2021 aniket agarwalla <aniketagarwalla37@gmail.com>
 Copyright (C) 2022 nicolas mellado <nmellado0@gmail.com>
 
 This Source Code Form is subject to the terms of the Mozilla Public
 License, v. 2.0. If a copy of the MPL was not distributed with this
 file, You can obtain one at http://mozilla.org/MPL/2.0/.
*/
 
#pragma once
#include "../../defines.h"
#include "../concepts.h"
 
#include PONCA_MULTIARCH_INCLUDE_STD(cmath)
#include PONCA_MULTIARCH_INCLUDE_CU_STD(limits)
 
#include <Eigen/Dense>
 
#define COVARIANCE_FIT_BASE_REQUIREMENTS ProvidesMeanPosition<T>
#define COVARIANCE_FIT_DER_REQUIREMENTS \
    ProvidesBasketDiffUnitBase<T>&& ProvidesMeanPositionDerivative<T>&& ProvidesPositionCovariance<T>
 
namespace Ponca
{


 
       All the computed features are defined for the 3 eigenvalues \f$ 0 < \lambda_0
       \leq \lambda_1 \leq \lambda_2 \f$.
 
 
       \warning This class is valid for any dimensions, however some statistics can only be computed in 3d.
     */
 
    template <class DataPoint, class _NFilter, typename T>
        requires COVARIANCE_FIT_BASE_REQUIREMENTS
    class CovarianceBase : public T
    {
        PONCA_FITTING_DECLARE_DEFAULT_TYPES
    public:
        using MatrixType = typename DataPoint::MatrixType; 
        /*! \brief Solver used to analyse the covariance matrix*/
        using Solver = Eigen::SelfAdjointEigenSolver<MatrixType>;
 
    protected:
        // computation data
        MatrixType m_cov{MatrixType::Zero()}; 
        Solver m_solver;                      
 
    public:
        PONCA_EXPLICIT_BASKET_CAST_OPERATORS(CovarianceBase, covarianceBase)
        PONCA_FITTING_DECLARE_INIT_ADD_FINALIZE


        */
        PONCA_MULTIARCH [[nodiscard]] inline Scalar surfaceVariation() const;


        */
        PONCA_MULTIARCH [[nodiscard]] inline Scalar planarity() const;


        */
        PONCA_MULTIARCH [[nodiscard]] inline Scalar linearity() const;


        */
        PONCA_MULTIARCH [[nodiscard]] inline Scalar sphericity() const;


        */
        PONCA_MULTIARCH [[nodiscard]] inline Scalar anisotropy() const;


        PONCA_MULTIARCH [[nodiscard]] inline Scalar eigenentropy() const;

        PONCA_MULTIARCH [[nodiscard]] inline Scalar lambda_0() const;


        PONCA_MULTIARCH [[nodiscard]] inline Scalar lambda_1() const;

        /*! \brief The maximum eigenvalue \f$ \lambda_2 \f$.
         */
        PONCA_MULTIARCH [[nodiscard]] inline Scalar lambda_2() const;

        PONCA_MULTIARCH [[nodiscard]] inline const Solver& solver() const { return m_solver; }
    };

    template <class DataPoint, class _NFilter, int DiffType, typename T>
        requires COVARIANCE_FIT_DER_REQUIREMENTS
    class CovarianceDer : public T
    {
        PONCA_FITTING_DECLARE_DEFAULT_TYPES
        PONCA_FITTING_DECLARE_MATRIX_TYPE
        PONCA_FITTING_DECLARE_DEFAULT_DER_TYPES
    protected:
        MatrixType m_dCov[Base::NbDerivatives];
 
    public:
        PONCA_EXPLICIT_BASKET_CAST_OPERATORS_DER(CovarianceDer, covarianceDer)
        PONCA_FITTING_DECLARE_INIT_ADDDER_FINALIZE
    }; // class CovarianceDer
 
#include "covariance.hpp"
 
} // namespace Ponca