transformation_estimation_lm.h

/*
 * Software License Agreement (BSD License)
 *
 *  Point Cloud Library (PCL) - www.pointclouds.org
 *  Copyright (c) 2010-2011, Willow Garage, Inc.
 *  Copyright (c) 2012-, Open Perception, Inc.
 *
 *  All rights reserved.
 *
 *  Redistribution and use in source and binary forms, with or without
 *  modification, are permitted provided that the following conditions
 *  are met:
 *
 *   * Redistributions of source code must retain the above copyright
 *     notice, this list of conditions and the following disclaimer.
 *   * Redistributions in binary form must reproduce the above
 *     copyright notice, this list of conditions and the following
 *     disclaimer in the documentation and/or other materials provided
 *     with the distribution.
 *   * Neither the name of the copyright holder(s) nor the names of its
 *     contributors may be used to endorse or promote products derived
 *     from this software without specific prior written permission.
 *
 *  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 *  "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 *  LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
 *  FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
 *  COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
 *  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
 *  BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 *  LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
 *  CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 *  LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
 *  ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
 *  POSSIBILITY OF SUCH DAMAGE.
 *
 * $Id$
 *
 */
 
#pragma once
 
#include <pcl/registration/transformation_estimation.h>
#include <pcl/registration/warp_point_rigid.h>
#include <pcl/memory.h>
 
namespace pcl {
namespace registration {
/** @b TransformationEstimationLM implements Levenberg Marquardt-based
 * estimation of the transformation aligning the given correspondences.
 *
 * \note The class is templated on the source and target point types as well as on the
 * output scalar of the transformation matrix (i.e., float or double). Default: float.
 * \author Radu B. Rusu
 * \ingroup registration
 */
template <typename PointSource, typename PointTarget, typename MatScalar = float>
class TransformationEstimationLM
: public TransformationEstimation<PointSource, PointTarget, MatScalar> {
  using PointCloudSource = pcl::PointCloud<PointSource>;
  using PointCloudSourcePtr = typename PointCloudSource::Ptr;
  using PointCloudSourceConstPtr = typename PointCloudSource::ConstPtr;
 
  using PointCloudTarget = pcl::PointCloud<PointTarget>;
 
  using PointIndicesPtr = PointIndices::Ptr;
  using PointIndicesConstPtr = PointIndices::ConstPtr;
 
public:
  using Ptr =
      shared_ptr<TransformationEstimationLM<PointSource, PointTarget, MatScalar>>;
  using ConstPtr =
      shared_ptr<const TransformationEstimationLM<PointSource, PointTarget, MatScalar>>;
 
  using VectorX = Eigen::Matrix<MatScalar, Eigen::Dynamic, 1>;
  using Vector4 = Eigen::Matrix<MatScalar, 4, 1>;
  using Matrix4 =
      typename TransformationEstimation<PointSource, PointTarget, MatScalar>::Matrix4;
 
  /** \brief Constructor. */
  TransformationEstimationLM();
 
  /** \brief Copy constructor.
   * \param[in] src the TransformationEstimationLM object to copy into this
   */
  TransformationEstimationLM(const TransformationEstimationLM& src)
  : tmp_src_(src.tmp_src_)
  , tmp_tgt_(src.tmp_tgt_)
  , tmp_idx_src_(src.tmp_idx_src_)
  , tmp_idx_tgt_(src.tmp_idx_tgt_)
  , warp_point_(src.warp_point_){};
 
  /** \brief Copy operator.
   * \param[in] src the TransformationEstimationLM object to copy into this
   */
  TransformationEstimationLM&
  operator=(const TransformationEstimationLM& src)
  {
    tmp_src_ = src.tmp_src_;
    tmp_tgt_ = src.tmp_tgt_;
    tmp_idx_src_ = src.tmp_idx_src_;
    tmp_idx_tgt_ = src.tmp_idx_tgt_;
    warp_point_ = src.warp_point_;
    return (*this);
  }
 
  /** \brief Destructor. */
  ~TransformationEstimationLM() override = default;
 
  /** \brief Estimate a rigid rotation transformation between a source and a target
   * point cloud using LM. \param[in] cloud_src the source point cloud dataset
   * \param[in] cloud_tgt the target point cloud dataset
   * \param[out] transformation_matrix the resultant transformation matrix
   */
  inline void
  estimateRigidTransformation(const pcl::PointCloud<PointSource>& cloud_src,
                              const pcl::PointCloud<PointTarget>& cloud_tgt,
                              Matrix4& transformation_matrix) const override;
 
  /** \brief Estimate a rigid rotation transformation between a source and a target
   * point cloud using LM. \param[in] cloud_src the source point cloud dataset
   * \param[in] indices_src the vector of indices describing the points of interest in
   * \a cloud_src
   * \param[in] cloud_tgt the target point cloud dataset
   * \param[out] transformation_matrix the resultant transformation matrix
   */
  inline void
  estimateRigidTransformation(const pcl::PointCloud<PointSource>& cloud_src,
                              const pcl::Indices& indices_src,
                              const pcl::PointCloud<PointTarget>& cloud_tgt,
                              Matrix4& transformation_matrix) const override;
 
  /** \brief Estimate a rigid rotation transformation between a source and a target
   * point cloud using LM. \param[in] cloud_src the source point cloud dataset
   * \param[in] indices_src the vector of indices describing the points of interest in
   * \a cloud_src
   * \param[in] cloud_tgt the target point cloud dataset
   * \param[in] indices_tgt the vector of indices describing the correspondences of the
   * interest points from \a indices_src
   * \param[out] transformation_matrix the resultant transformation matrix
   */
  inline void
  estimateRigidTransformation(const pcl::PointCloud<PointSource>& cloud_src,
                              const pcl::Indices& indices_src,
                              const pcl::PointCloud<PointTarget>& cloud_tgt,
                              const pcl::Indices& indices_tgt,
                              Matrix4& transformation_matrix) const override;
 
  /** \brief Estimate a rigid rotation transformation between a source and a target
   * point cloud using LM. \param[in] cloud_src the source point cloud dataset
   * \param[in] cloud_tgt the target point cloud dataset
   * \param[in] correspondences the vector of correspondences between source and target
   * point cloud \param[out] transformation_matrix the resultant transformation matrix
   */
  inline void
  estimateRigidTransformation(const pcl::PointCloud<PointSource>& cloud_src,
                              const pcl::PointCloud<PointTarget>& cloud_tgt,
                              const pcl::Correspondences& correspondences,
                              Matrix4& transformation_matrix) const override;
 
  /** \brief Set the function we use to warp points. Defaults to rigid 6D warp.
   * \param[in] warp_fcn a shared pointer to an object that warps points
   */
  void
  setWarpFunction(
      const typename WarpPointRigid<PointSource, PointTarget, MatScalar>::Ptr& warp_fcn)
  {
    warp_point_ = warp_fcn;
  }
 
protected:
  /** \brief Compute the distance between a source point and its corresponding target
   * point \param[in] p_src The source point \param[in] p_tgt The target point \return
   * The distance between \a p_src and \a p_tgt
   *
   * \note Older versions of PCL used this method internally for calculating the
   * optimization gradient. Since PCL 1.7, a switch has been made to the
   * computeDistance method using Vector4 types instead. This method is only
   * kept for API compatibility reasons.
   */
  virtual MatScalar
  computeDistance(const PointSource& p_src, const PointTarget& p_tgt) const
  {
    Vector4 s(p_src.x, p_src.y, p_src.z, 0);
    Vector4 t(p_tgt.x, p_tgt.y, p_tgt.z, 0);
    return ((s - t).norm());
  }
 
  /** \brief Compute the distance between a source point and its corresponding target
   * point \param[in] p_src The source point \param[in] p_tgt The target point \return
   * The distance between \a p_src and \a p_tgt
   *
   * \note A different distance function can be defined by creating a subclass of
   * TransformationEstimationLM and overriding this method.
   * (See \a TransformationEstimationPointToPlane)
   */
  virtual MatScalar
  computeDistance(const Vector4& p_src, const PointTarget& p_tgt) const
  {
    Vector4 t(p_tgt.x, p_tgt.y, p_tgt.z, 0);
    return ((p_src - t).norm());
  }
 
  /** \brief Temporary pointer to the source dataset. */
  mutable const PointCloudSource* tmp_src_{nullptr};
 
  /** \brief Temporary pointer to the target dataset. */
  mutable const PointCloudTarget* tmp_tgt_{nullptr};
 
  /** \brief Temporary pointer to the source dataset indices. */
  mutable const pcl::Indices* tmp_idx_src_{nullptr};
 
  /** \brief Temporary pointer to the target dataset indices. */
  mutable const pcl::Indices* tmp_idx_tgt_{nullptr};
 
  /** \brief The parameterized function used to warp the source to the target. */
  typename pcl::registration::WarpPointRigid<PointSource, PointTarget, MatScalar>::Ptr
      warp_point_;
 
  /** Base functor all the models that need non linear optimization must
   * define their own one and implement operator() (const Eigen::VectorXd& x,
   * Eigen::VectorXd& fvec) or operator() (const Eigen::VectorXf& x, Eigen::VectorXf&
   * fvec) depending on the chosen _Scalar
   */
  template <typename _Scalar, int NX = Eigen::Dynamic, int NY = Eigen::Dynamic>
  struct Functor {
    using Scalar = _Scalar;
    enum { InputsAtCompileTime = NX, ValuesAtCompileTime = NY };
    using InputType = Eigen::Matrix<_Scalar, InputsAtCompileTime, 1>;
    using ValueType = Eigen::Matrix<_Scalar, ValuesAtCompileTime, 1>;
    using JacobianType =
        Eigen::Matrix<_Scalar, ValuesAtCompileTime, InputsAtCompileTime>;
 
    /** \brief Empty Constructor. */
    Functor() : m_data_points_(ValuesAtCompileTime) {}
 
    /** \brief Constructor
     * \param[in] m_data_points number of data points to evaluate.
     */
    Functor(int m_data_points) : m_data_points_(m_data_points) {}
 
    /** \brief Destructor. */
    virtual ~Functor() = default;
 
    /** \brief Get the number of values. */
    int
    values() const
    {
      return (m_data_points_);
    }
 
  protected:
    int m_data_points_;
  };
 
  struct OptimizationFunctor : public Functor<MatScalar> {
    using Functor<MatScalar>::values;
 
    /** Functor constructor
     * \param[in] m_data_points the number of data points to evaluate
     * \param[in,out] estimator pointer to the estimator object
     */
    OptimizationFunctor(int m_data_points, const TransformationEstimationLM* estimator)
    : Functor<MatScalar>(m_data_points), estimator_(estimator)
    {}
 
    /** Copy constructor
     * \param[in] src the optimization functor to copy into this
     */
    inline OptimizationFunctor(const OptimizationFunctor& src)
    : Functor<MatScalar>(src.m_data_points_), estimator_()
    {
      *this = src;
    }
 
    /** Copy operator
     * \param[in] src the optimization functor to copy into this
     */
    inline OptimizationFunctor&
    operator=(const OptimizationFunctor& src)
    {
      Functor<MatScalar>::operator=(src);
      estimator_ = src.estimator_;
      return (*this);
    }
 
    /** \brief Destructor. */
    ~OptimizationFunctor() override = default;
 
    /** Fill fvec from x. For the current state vector x fill the f values
     * \param[in] x state vector
     * \param[out] fvec f values vector
     */
    int
    operator()(const VectorX& x, VectorX& fvec) const;
 
    const TransformationEstimationLM<PointSource, PointTarget, MatScalar>* estimator_;
  };
 
  struct OptimizationFunctorWithIndices : public Functor<MatScalar> {
    using Functor<MatScalar>::values;
 
    /** Functor constructor
     * \param[in] m_data_points the number of data points to evaluate
     * \param[in,out] estimator pointer to the estimator object
     */
    OptimizationFunctorWithIndices(int m_data_points,
                                   const TransformationEstimationLM* estimator)
    : Functor<MatScalar>(m_data_points), estimator_(estimator)
    {}
 
    /** Copy constructor
     * \param[in] src the optimization functor to copy into this
     */
    inline OptimizationFunctorWithIndices(const OptimizationFunctorWithIndices& src)
    : Functor<MatScalar>(src.m_data_points_), estimator_()
    {
      *this = src;
    }
 
    /** Copy operator
     * \param[in] src the optimization functor to copy into this
     */
    inline OptimizationFunctorWithIndices&
    operator=(const OptimizationFunctorWithIndices& src)
    {
      Functor<MatScalar>::operator=(src);
      estimator_ = src.estimator_;
      return (*this);
    }
 
    /** \brief Destructor. */
    ~OptimizationFunctorWithIndices() override = default;
 
    /** Fill fvec from x. For the current state vector x fill the f values
     * \param[in] x state vector
     * \param[out] fvec f values vector
     */
    int
    operator()(const VectorX& x, VectorX& fvec) const;
 
    const TransformationEstimationLM<PointSource, PointTarget, MatScalar>* estimator_;
  };
 
public:
  PCL_MAKE_ALIGNED_OPERATOR_NEW
};
} // namespace registration
} // namespace pcl
 
#include <pcl/registration/impl/transformation_estimation_lm.hpp>