cnc.h

 
#pragma once
 
#include "../defines.h"
#include "../concepts.h"
#include "cncFormulaEigen.h"
 
#include "../Filters/weightFilter.h"
#include "../Filters/weightKernel.h"
 
#define CNC_REQUIREMENTS ProvidesNormal<P>
 
namespace Ponca
{
 
    namespace internal
    {
        template <class DataPoint>
        struct Triangle
        {
        public:
            using Scalar     = typename DataPoint::Scalar;
            using VectorType = typename DataPoint::VectorType;
            using MatrixType = typename DataPoint::MatrixType;
 
        protected:
            std::array<VectorType, 3> m_points;
            std::array<VectorType, 3> m_normals;
 
        public:
            Triangle(DataPoint pointA, DataPoint pointB, DataPoint pointC)
            {
                m_points  = {pointA.pos(), pointB.pos(), pointC.pos()};
                m_normals = {pointA.normal(), pointB.normal(), pointC.normal()};
            }
 
            Triangle(const std::array<VectorType, 3>& points, const std::array<VectorType, 3>& normals)
            {
                m_points  = points;
                m_normals = normals;
            }
 
            PONCA_MULTIARCH [[nodiscard]] VectorType& getPos(const int index) { return m_points[index]; }
 
            PONCA_MULTIARCH [[nodiscard]] bool operator==(const Triangle& other) const
            {
                return (m_points[0] == other.m_points[0]) && (m_points[1] == other.m_points[1]) &&
                       (m_points[2] == other.m_points[2]);
            }
 
            PONCA_MULTIARCH [[nodiscard]] bool operator!=(const Triangle& other) const { return !((*this) == other); }
 
#define DEFINE_CNC_FUNC(CNC_FUNC, RETURN_TYPE)                                                                        \
    template <bool differentOrder = false>                                                                            \
    inline RETURN_TYPE CNC_FUNC()                                                                                     \
    {                                                                                                                 \
        return CNCEigen<DataPoint>::CNC_FUNC(m_points[0], m_points[2 - differentOrder], m_points[1 + differentOrder], \
                                             m_normals[0], m_normals[2 - differentOrder],                             \
                                             m_normals[1 + differentOrder]);                                          \
    }
 
            DEFINE_CNC_FUNC(mu0InterpolatedU, Scalar)
            DEFINE_CNC_FUNC(mu1InterpolatedU, Scalar)
            DEFINE_CNC_FUNC(mu2InterpolatedU, Scalar)
            DEFINE_CNC_FUNC(muXYInterpolatedU, MatrixType)
        };
    } // namespace internal
 
    enum TriangleGenerationMethod
    {
        UniformGeneration,
        HexagramGeneration,
        IndependentGeneration,
        AvgHexagramGeneration
    };
 
    template <class P, TriangleGenerationMethod _method = UniformGeneration>
        requires CNC_REQUIREMENTS
    class CNC : ComputeObject<CNC<P, _method>>
    {
    public:
        using DataPoint   = P;
        using MatrixType  = typename DataPoint::MatrixType;
        using Scalar      = typename DataPoint::Scalar;
        using VectorType  = typename DataPoint::VectorType;
        using DenseVector = Eigen::VectorXd;
        using DenseMatrix = Eigen::MatrixXd;
        using NeighborFilter =
            FilterWithAttributes<DataPoint, VectorType, DistWeightFilter<DataPoint, ConstantWeightKernel<Scalar>>,
                                 internal::ConvertDataPointToNormal<DataPoint>>;
 
    protected:
        // Basis
        NeighborFilter m_nFilter;
 
        // Triangles used for the computation
        int m_nb_vt{0}; // Number of valid triangles
        std::vector<internal::Triangle<DataPoint>> m_triangles;
 
        // Results of the fit
        Scalar m_A{0};   // Area
        Scalar m_H{0};   // Mean Curvatures
        Scalar m_G{0};   // Gaussian Curvatures
        Scalar m_T11{0}; // T11
        Scalar m_T12{0}; // T12
        Scalar m_T13{0}; // T13
        Scalar m_T22{0}; // T22
        Scalar m_T23{0}; // T23
        Scalar m_T33{0}; // T33
 
        Scalar m_k1{0};
        Scalar m_k2{0};
 
        VectorType m_v1;
        VectorType m_v2;
 
        FIT_RESULT m_eCurrentState{UNDEFINED};
 
    public:
        PONCA_EXPLICIT_CAST_OPERATOR(meanCurvature, CNC<P, _method>)
        PONCA_EXPLICIT_CAST_OPERATOR(curvatureTensor, CNC<P, _method>)
 
        PONCA_FITTING_DECLARE_FINALIZE
 
        PONCA_MULTIARCH inline void init()
        {
            m_eCurrentState = UNDEFINED;
            m_A             = Scalar(0);
            m_H             = Scalar(0);
            m_G             = Scalar(0);
 
            m_k1 = Scalar(0);
            m_k2 = Scalar(0);
 
            m_triangles.clear();
        }
 
        template <typename IteratorBegin, typename IteratorEnd>
        PONCA_MULTIARCH inline FIT_RESULT compute(const IteratorBegin& begin, const IteratorEnd& end);
 
        template <typename PointContainer>
        PONCA_MULTIARCH inline FIT_RESULT compute(const PointContainer& points);
 
        template <typename IndexRange, typename PointContainer>
        PONCA_MULTIARCH inline FIT_RESULT computeWithIds(const IndexRange& ids, const PointContainer& points);
 
        PONCA_MULTIARCH [[nodiscard]] inline size_t getNumTriangles() const { return static_cast<size_t>(m_nb_vt); }
 
        PONCA_FITTING_APIDOC_SETWFUNC
        PONCA_MULTIARCH inline void setNeighborFilter(const NeighborFilter& _nFilter) { m_nFilter = _nFilter; }
 
        PONCA_MULTIARCH [[nodiscard]] std::vector<internal::Triangle<DataPoint>>& getTriangles() { return m_triangles; }
 
        PONCA_MULTIARCH [[nodiscard]] bool operator==(const CNC& other) const
        {
            // We use the matrix to compare the fitting results
            return (m_eCurrentState == other.m_eCurrentState) && (kMean() == other.kMean()) &&
                   (kmin() == other.kmin()) && (kmax() == other.kmax()) && (kminDirection() == other.kminDirection()) &&
                   (kmaxDirection() == other.kmaxDirection()) && (GaussianCurvature() == other.GaussianCurvature()) &&
                   (m_T11 == other.m_T11) && (m_T12 == other.m_T12) && (m_T13 == other.m_T13) &&
                   (m_T22 == other.m_T22) && (m_T23 == other.m_T23) && (m_T33 == other.m_T33);
        }
 
        PONCA_MULTIARCH [[nodiscard]] bool operator!=(const CNC& other) const
        {
            // We use the matrix to compare the fitting results
            return !(this == &other);
        }
 
        PONCA_MULTIARCH [[nodiscard]] bool isApprox(
            const CNC& other, const Scalar& epsilon = Eigen::NumTraits<Scalar>::dummy_precision()) const
        {
            PONCA_MULTIARCH_STD_MATH(abs);
 
            return (m_eCurrentState == other.m_eCurrentState) && (std::abs(kMean() - other.kMean()) < epsilon) &&
                   (std::abs(GaussianCurvature() - other.GaussianCurvature()) < epsilon) &&
                   (std::abs(kmin() - other.kmin()) < epsilon) && (std::abs(kmax() - other.kmax()) < epsilon);
        }
 
        PONCA_MULTIARCH [[nodiscard]] inline bool isStable() const { return m_eCurrentState == STABLE; }
 
        PONCA_MULTIARCH [[nodiscard]] inline Scalar kmin() const { return m_k1; }
 
        PONCA_MULTIARCH [[nodiscard]] inline Scalar kmax() const { return m_k2; }
 
        PONCA_MULTIARCH [[nodiscard]] inline VectorType kminDirection() const { return m_v1; }
 
        PONCA_MULTIARCH [[nodiscard]] inline VectorType kmaxDirection() const { return m_v2; }
 
        PONCA_MULTIARCH [[nodiscard]] inline Scalar kMean() const { return m_H; }
 
        PONCA_MULTIARCH [[nodiscard]] inline Scalar GaussianCurvature() const { return m_G; }
    }; // class CNC
 
} // namespace Ponca
 
#include "cnc.hpp"