transformation_validation_euclidean.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_validation.h>
#include <pcl/search/kdtree.h>
#include <pcl/memory.h>
#include <pcl/pcl_macros.h>
#include <pcl/point_representation.h>
 
namespace pcl {
namespace registration {
/** \brief TransformationValidationEuclidean computes an L2SQR norm between a source and
 * target dataset.
 *
 * To prevent points with bad correspondences to contribute to the overall score, the
 * class also accepts a maximum_range parameter given via \ref setMaxRange that is used
 * as a cutoff value for nearest neighbor distance comparisons.
 *
 * The output score is normalized with respect to the number of valid correspondences
 * found.
 *
 * Usage example:
 * \code
 * pcl::TransformationValidationEuclidean<pcl::PointXYZ, pcl::PointXYZ> tve;
 * tve.setMaxRange (0.01);  // 1cm
 * double score = tve.validateTransformation (source, target, transformation);
 * \endcode
 *
 * \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 Scalar = float>
class TransformationValidationEuclidean {
public:
  using Matrix4 =
      typename TransformationValidation<PointSource, PointTarget, Scalar>::Matrix4;
 
  using Ptr = shared_ptr<TransformationValidation<PointSource, PointTarget, Scalar>>;
  using ConstPtr =
      shared_ptr<const TransformationValidation<PointSource, PointTarget, Scalar>>;
 
  using KdTree = pcl::search::KdTree<PointTarget>;
  using KdTreePtr = typename KdTree::Ptr;
 
  using PointRepresentationConstPtr = typename KdTree::PointRepresentationConstPtr;
 
  using PointCloudSourceConstPtr =
      typename TransformationValidation<PointSource,
                                        PointTarget>::PointCloudSourceConstPtr;
  using PointCloudTargetConstPtr =
      typename TransformationValidation<PointSource,
                                        PointTarget>::PointCloudTargetConstPtr;
 
  /** \brief Constructor.
   * Sets the \a max_range parameter to double::max, \a threshold_ to NaN
   * and initializes the internal search \a tree to a FLANN kd-tree.
   */
  TransformationValidationEuclidean()
  : max_range_(std::numeric_limits<double>::max())
  , threshold_(std::numeric_limits<double>::quiet_NaN())
  , tree_(new pcl::search::KdTree<PointTarget>)
  {}
 
  virtual ~TransformationValidationEuclidean() = default;
 
  /** \brief Set the maximum allowable distance between a point and its correspondence
   * in the target in order for a correspondence to be considered \a valid. Default:
   * double::max. \param[in] max_range the new maximum allowable distance
   */
  inline void
  setMaxRange(double max_range)
  {
    max_range_ = max_range;
  }
 
  /** \brief Get the maximum allowable distance between a point and its
   * correspondence, as set by the user.
   */
  inline double
  getMaxRange()
  {
    return (max_range_);
  }
 
  /** \brief Provide a pointer to the search object used to find correspondences in
   * the target cloud.
   * \param[in] tree a pointer to the spatial search object.
   * \param[in] force_no_recompute If set to true, this tree will NEVER be
   * recomputed, regardless of calls to setInputTarget. Only use if you are
   * confident that the tree will be set correctly.
   */
  inline void
  setSearchMethodTarget(const KdTreePtr& tree, bool force_no_recompute = false)
  {
    tree_ = tree;
    force_no_recompute_ = force_no_recompute;
  }
 
  /** \brief Set a threshold for which a specific transformation is considered valid.
   *
   * \note Since we're using MSE (Mean Squared Error) as a metric, the threshold
   * represents the mean Euclidean distance threshold over all nearest neighbors
   * up to max_range.
   *
   * \param[in] threshold the threshold for which a transformation is vali
   */
  inline void
  setThreshold(double threshold)
  {
    threshold_ = threshold;
  }
 
  /** \brief Get the threshold for which a specific transformation is valid. */
  inline double
  getThreshold()
  {
    return (threshold_);
  }
 
  /** \brief Validate the given transformation with respect to the input cloud data, and
   * return a score.
   *
   * \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
   *
   * \return the score or confidence measure for the given
   * transformation_matrix with respect to the input data
   */
  double
  validateTransformation(const PointCloudSourceConstPtr& cloud_src,
                         const PointCloudTargetConstPtr& cloud_tgt,
                         const Matrix4& transformation_matrix) const;
 
  /** \brief Comparator function for deciding which score is better after running the
   * validation on multiple transforms.
   *
   * \param[in] score1 the first value
   * \param[in] score2 the second value
   *
   * \return true if score1 is better than score2
   */
  virtual bool
  operator()(const double& score1, const double& score2) const
  {
    return (score1 < score2);
  }
 
  /** \brief Check if the score is valid for a specific transformation.
   *
   * \param[in] cloud_src the source point cloud dataset
   * \param[in] cloud_tgt the target point cloud dataset
   * \param[out] transformation_matrix the transformation matrix
   *
   * \return true if the transformation is valid, false otherwise.
   */
  virtual bool
  isValid(const PointCloudSourceConstPtr& cloud_src,
          const PointCloudTargetConstPtr& cloud_tgt,
          const Matrix4& transformation_matrix) const
  {
    if (std::isnan(threshold_)) {
      PCL_ERROR("[pcl::TransformationValidationEuclidean::isValid] Threshold not set! "
                "Please use setThreshold () before continuing.\n");
      return (false);
    }
 
    return (validateTransformation(cloud_src, cloud_tgt, transformation_matrix) <
            threshold_);
  }
 
protected:
  /** \brief The maximum allowable distance between a point and its correspondence in
   * the target in order for a correspondence to be considered \a valid. Default:
   * double::max.
   */
  double max_range_;
 
  /** \brief The threshold for which a specific transformation is valid.
   * Set to NaN by default, as we must require the user to set it.
   */
  double threshold_;
 
  /** \brief A pointer to the spatial search object. */
  KdTreePtr tree_;
 
  /** \brief A flag which, if set, means the tree operating on the target cloud
   * will never be recomputed*/
  bool force_no_recompute_{false};
 
  /** \brief Internal point representation uses only 3D coordinates for L2 */
  class MyPointRepresentation : public pcl::PointRepresentation<PointTarget> {
    using pcl::PointRepresentation<PointTarget>::nr_dimensions_;
    using pcl::PointRepresentation<PointTarget>::trivial_;
 
  public:
    using Ptr = shared_ptr<MyPointRepresentation>;
    using ConstPtr = shared_ptr<const MyPointRepresentation>;
 
    MyPointRepresentation()
    {
      nr_dimensions_ = 3;
      trivial_ = true;
    }
 
    /** \brief Empty destructor */
    virtual ~MyPointRepresentation() = default;
 
    virtual void
    copyToFloatArray(const PointTarget& p, float* out) const
    {
      out[0] = p.x;
      out[1] = p.y;
      out[2] = p.z;
    }
  };
 
public:
  PCL_MAKE_ALIGNED_OPERATOR_NEW
};
} // namespace registration
} // namespace pcl
 
#include <pcl/registration/impl/transformation_validation_euclidean.hpp>