pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar > Class Template Reference

FastRobustIterativeClosestPoint implements the FRICP variant described in "Fast and Robust Iterative Closest Point", Zhang et al., 2021. More...

#include <pcl/registration/fricp.h>

Inheritance diagram for pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >:

Collaboration diagram for pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >:

Public Types
enum class	RobustFunction { NONE , WELSCH }
using	Ptr = shared_ptr< FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar > >
using	ConstPtr = shared_ptr< const FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar > >
using	Base = IterativeClosestPoint< PointSource, PointTarget, Scalar >
using	Matrix4 = typename Registration< PointSource, PointTarget, Scalar >::Matrix4
using	PointCloudSource = typename Registration< PointSource, PointTarget, Scalar >::PointCloudSource
using	PointCloudTarget = typename Registration< PointSource, PointTarget, Scalar >::PointCloudTarget
Public Types inherited from pcl::IterativeClosestPoint< PointSource, PointTarget, Scalar >
using	PointCloudSource = typename Registration< PointSource, PointTarget, Scalar >::PointCloudSource
using	PointCloudSourcePtr = typename PointCloudSource::Ptr
using	PointCloudSourceConstPtr = typename PointCloudSource::ConstPtr
using	PointCloudTarget = typename Registration< PointSource, PointTarget, Scalar >::PointCloudTarget
using	PointCloudTargetPtr = typename PointCloudTarget::Ptr
using	PointCloudTargetConstPtr = typename PointCloudTarget::ConstPtr
using	PointIndicesPtr = PointIndices::Ptr
using	PointIndicesConstPtr = PointIndices::ConstPtr
using	Ptr = shared_ptr< IterativeClosestPoint< PointSource, PointTarget, Scalar > >
using	ConstPtr = shared_ptr< const IterativeClosestPoint< PointSource, PointTarget, Scalar > >
using	Matrix4 = typename Registration< PointSource, PointTarget, Scalar >::Matrix4
Public Types inherited from pcl::Registration< PointSource, PointTarget, Scalar >
using	Matrix4 = Eigen::Matrix< Scalar, 4, 4 >
using	Ptr = shared_ptr< Registration< PointSource, PointTarget, Scalar > >
using	ConstPtr = shared_ptr< const Registration< PointSource, PointTarget, Scalar > >
using	CorrespondenceRejectorPtr = pcl::registration::CorrespondenceRejector::Ptr
using	KdTree = pcl::search::KdTree< PointTarget >
using	KdTreePtr = typename KdTree::Ptr
using	KdTreeReciprocal = pcl::search::KdTree< PointSource >
using	KdTreeReciprocalPtr = typename KdTreeReciprocal::Ptr
using	PointCloudSource = pcl::PointCloud< PointSource >
using	PointCloudSourcePtr = typename PointCloudSource::Ptr
using	PointCloudSourceConstPtr = typename PointCloudSource::ConstPtr
using	PointCloudTarget = pcl::PointCloud< PointTarget >
using	PointCloudTargetPtr = typename PointCloudTarget::Ptr
using	PointCloudTargetConstPtr = typename PointCloudTarget::ConstPtr
using	PointRepresentationConstPtr = typename KdTree::PointRepresentationConstPtr
using	TransformationEstimation = typename pcl::registration::TransformationEstimation< PointSource, PointTarget, Scalar >
using	TransformationEstimationPtr = typename TransformationEstimation::Ptr
using	TransformationEstimationConstPtr = typename TransformationEstimation::ConstPtr
using	CorrespondenceEstimation = pcl::registration::CorrespondenceEstimationBase< PointSource, PointTarget, Scalar >
using	CorrespondenceEstimationPtr = typename CorrespondenceEstimation::Ptr
using	CorrespondenceEstimationConstPtr = typename CorrespondenceEstimation::ConstPtr
using	UpdateVisualizerCallbackSignature = void(const pcl::PointCloud< PointSource > &, const pcl::Indices &, const pcl::PointCloud< PointTarget > &, const pcl::Indices &)
	The callback signature to the function updating intermediate source point cloud position during it's registration to the target point cloud.
Public Types inherited from pcl::PCLBase< PointSource >
using	PointCloud = pcl::PointCloud< PointSource >
using	PointCloudPtr = typename PointCloud::Ptr
using	PointCloudConstPtr = typename PointCloud::ConstPtr
using	PointIndicesPtr = PointIndices::Ptr
using	PointIndicesConstPtr = PointIndices::ConstPtr

Public Member Functions
	FastRobustIterativeClosestPoint ()
void	setRobustFunction (RobustFunction f)
RobustFunction	getRobustFunction () const
void	setUseAndersonAcceleration (bool enabled)
	Enable or disable Anderson acceleration in the FRICP optimization loop.
bool	getUseAndersonAcceleration () const
void	setAndersonHistorySize (std::size_t history)
	Set the history size used by Anderson acceleration.
std::size_t	getAndersonHistorySize () const
void	setDynamicWelschBeginRatio (double ratio)
	Set the initial Welsch scale ratio used in dynamic robust weighting.
void	setDynamicWelschEndRatio (double ratio)
	Set the final Welsch scale ratio used in dynamic robust weighting.
void	setDynamicWelschDecay (double ratio)
	Set the multiplicative decay applied to the dynamic Welsch scale.
Public Member Functions inherited from pcl::IterativeClosestPoint< PointSource, PointTarget, Scalar >
	IterativeClosestPoint ()
	Empty constructor.
	IterativeClosestPoint (const IterativeClosestPoint &)=delete
	Due to convergence_criteria_ holding references to the class members, it is tricky to correctly implement its copy and move operations correctly.
	IterativeClosestPoint (IterativeClosestPoint &&)=delete
IterativeClosestPoint &	operator= (const IterativeClosestPoint &)=delete
IterativeClosestPoint &	operator= (IterativeClosestPoint &&)=delete
	~IterativeClosestPoint () override=default
	Empty destructor.
pcl::registration::DefaultConvergenceCriteria< Scalar >::Ptr	getConvergeCriteria ()
	Returns a pointer to the DefaultConvergenceCriteria used by the IterativeClosestPoint class.
void	setInputSource (const PointCloudSourceConstPtr &cloud) override
	Provide a pointer to the input source (e.g., the point cloud that we want to align to the target)
void	setInputTarget (const PointCloudTargetConstPtr &cloud) override
	Provide a pointer to the input target (e.g., the point cloud that we want to align the input source to)
void	setUseReciprocalCorrespondences (bool use_reciprocal_correspondence)
	Set whether to use reciprocal correspondence or not.
bool	getUseReciprocalCorrespondences () const
	Obtain whether reciprocal correspondence are used or not.
void	setNumberOfThreads (unsigned int nr_threads)
	Set the number of threads to use.
Public Member Functions inherited from pcl::Registration< PointSource, PointTarget, Scalar >
	Registration ()
	Empty constructor.
	~Registration () override=default
	destructor.
void	setTransformationEstimation (const TransformationEstimationPtr &te)
	Provide a pointer to the transformation estimation object.
void	setCorrespondenceEstimation (const CorrespondenceEstimationPtr &ce)
	Provide a pointer to the correspondence estimation object.
PointCloudSourceConstPtr const	getInputSource ()
	Get a pointer to the input point cloud dataset target.
PointCloudTargetConstPtr const	getInputTarget ()
	Get a pointer to the input point cloud dataset target.
void	setSearchMethodTarget (const KdTreePtr &tree, bool force_no_recompute=false)
	Provide a pointer to the search object used to find correspondences in the target cloud.
KdTreePtr	getSearchMethodTarget () const
	Get a pointer to the search method used to find correspondences in the target cloud.
void	setSearchMethodSource (const KdTreeReciprocalPtr &tree, bool force_no_recompute=false)
	Provide a pointer to the search object used to find correspondences in the source cloud (usually used by reciprocal correspondence finding).
KdTreeReciprocalPtr	getSearchMethodSource () const
	Get a pointer to the search method used to find correspondences in the source cloud.
Matrix4	getFinalTransformation ()
	Get the final transformation matrix estimated by the registration method.
Matrix4	getLastIncrementalTransformation ()
	Get the last incremental transformation matrix estimated by the registration method.
void	setMaximumIterations (int nr_iterations)
	Set the maximum number of iterations the internal optimization should run for.
int	getMaximumIterations ()
	Get the maximum number of iterations the internal optimization should run for, as set by the user.
void	setRANSACIterations (int ransac_iterations)
	Set the number of iterations RANSAC should run for.
double	getRANSACIterations ()
	Get the number of iterations RANSAC should run for, as set by the user.
void	setRANSACOutlierRejectionThreshold (double inlier_threshold)
	Set the inlier distance threshold for the internal RANSAC outlier rejection loop.
double	getRANSACOutlierRejectionThreshold ()
	Get the inlier distance threshold for the internal outlier rejection loop as set by the user.
void	setMaxCorrespondenceDistance (double distance_threshold)
	Set the maximum distance threshold between two correspondent points in source <-> target.
double	getMaxCorrespondenceDistance ()
	Get the maximum distance threshold between two correspondent points in source <-> target.
void	setTransformationEpsilon (double epsilon)
	Set the transformation epsilon (maximum allowable translation squared difference between two consecutive transformations) in order for an optimization to be considered as having converged to the final solution.
double	getTransformationEpsilon ()
	Get the transformation epsilon (maximum allowable translation squared difference between two consecutive transformations) as set by the user.
void	setTransformationRotationEpsilon (double epsilon)
	Set the transformation rotation epsilon (maximum allowable rotation difference between two consecutive transformations) in order for an optimization to be considered as having converged to the final solution.
double	getTransformationRotationEpsilon ()
	Get the transformation rotation epsilon (maximum allowable difference between two consecutive transformations) as set by the user (epsilon is the cos(angle) in a axis-angle representation).
void	setEuclideanFitnessEpsilon (double epsilon)
	Set the maximum allowed Euclidean error between two consecutive steps in the ICP loop, before the algorithm is considered to have converged.
double	getEuclideanFitnessEpsilon ()
	Get the maximum allowed distance error before the algorithm will be considered to have converged, as set by the user.
void	setPointRepresentation (const PointRepresentationConstPtr &point_representation)
	Provide a boost shared pointer to the PointRepresentation to be used when comparing points.
bool	registerVisualizationCallback (std::function< UpdateVisualizerCallbackSignature > &visualizerCallback)
	Register the user callback function which will be called from registration thread in order to update point cloud obtained after each iteration.
double	getFitnessScore (double max_range=std::numeric_limits< double >::max(), bool use_indices=false)
	Obtain the Euclidean fitness score (e.g., mean of squared distances from the source to the target)
double	getFitnessScore (const std::vector< float > &distances_a, const std::vector< float > &distances_b)
	Obtain the Euclidean fitness score (e.g., mean of squared distances from the source to the target) from two sets of correspondence distances (distances between source and target points)
bool	hasConverged () const
	Return the state of convergence after the last align run.
void	align (PointCloudSource &output)
	Call the registration algorithm which estimates the transformation and returns the transformed source (input) as output.
void	align (PointCloudSource &output, const Matrix4 &guess)
	Call the registration algorithm which estimates the transformation and returns the transformed source (input) as output.
const std::string &	getClassName () const
	Abstract class get name method.
bool	initCompute ()
	Internal computation initialization.
bool	initComputeReciprocal ()
	Internal computation when reciprocal lookup is needed.
void	addCorrespondenceRejector (const CorrespondenceRejectorPtr &rejector)
	Add a new correspondence rejector to the list.
std::vector< CorrespondenceRejectorPtr >	getCorrespondenceRejectors ()
	Get the list of correspondence rejectors.
bool	removeCorrespondenceRejector (unsigned int i)
	Remove the i-th correspondence rejector in the list.
void	clearCorrespondenceRejectors ()
	Clear the list of correspondence rejectors.
Public Member Functions inherited from pcl::PCLBase< PointSource >
	PCLBase ()
	Empty constructor.
	PCLBase (const PCLBase &base)
	Copy constructor.
virtual	~PCLBase ()=default
	Destructor.
virtual void	setInputCloud (const PointCloudConstPtr &cloud)
	Provide a pointer to the input dataset.
PointCloudConstPtr const	getInputCloud () const
	Get a pointer to the input point cloud dataset.
virtual void	setIndices (const IndicesPtr &indices)
	Provide a pointer to the vector of indices that represents the input data.
virtual void	setIndices (const IndicesConstPtr &indices)
	Provide a pointer to the vector of indices that represents the input data.
virtual void	setIndices (const PointIndicesConstPtr &indices)
	Provide a pointer to the vector of indices that represents the input data.
virtual void	setIndices (std::size_t row_start, std::size_t col_start, std::size_t nb_rows, std::size_t nb_cols)
	Set the indices for the points laying within an interest region of the point cloud.
IndicesPtr	getIndices ()
	Get a pointer to the vector of indices used.
IndicesConstPtr const	getIndices () const
	Get a pointer to the vector of indices used.
const PointSource &	operator[] (std::size_t pos) const
	Override PointCloud operator[] to shorten code.

Protected Member Functions
void	computeTransformation (PointCloudSource &output, const Matrix4 &guess) override
	Abstract transformation computation method with initial guess.
Protected Member Functions inherited from pcl::IterativeClosestPoint< PointSource, PointTarget, Scalar >
virtual void	transformCloud (const PointCloudSource &input, PointCloudSource &output, const Matrix4 &transform)
	Apply a rigid transform to a given dataset.
void	computeTransformation (PointCloudSource &output, const Matrix4 &guess) override
	Rigid transformation computation method with initial guess.
virtual void	determineRequiredBlobData ()
	Looks at the Estimators and Rejectors and determines whether their blob-setter methods need to be called.
Protected Member Functions inherited from pcl::Registration< PointSource, PointTarget, Scalar >
bool	searchForNeighbors (const PointCloudSource &cloud, int index, pcl::Indices &indices, std::vector< float > &distances)
	Search for the closest nearest neighbor of a given point.
Protected Member Functions inherited from pcl::PCLBase< PointSource >
bool	initCompute ()
	This method should get called before starting the actual computation.
bool	deinitCompute ()
	This method should get called after finishing the actual computation.

Additional Inherited Members
Public Attributes inherited from pcl::IterativeClosestPoint< PointSource, PointTarget, Scalar >
pcl::registration::DefaultConvergenceCriteria< Scalar >::Ptr	convergence_criteria_
Protected Attributes inherited from pcl::IterativeClosestPoint< PointSource, PointTarget, Scalar >
std::size_t	x_idx_offset_ {0}
	XYZ fields offset.
std::size_t	y_idx_offset_ {0}
std::size_t	z_idx_offset_ {0}
std::size_t	nx_idx_offset_ {0}
	Normal fields offset.
std::size_t	ny_idx_offset_ {0}
std::size_t	nz_idx_offset_ {0}
bool	use_reciprocal_correspondence_ {false}
	The correspondence type used for correspondence estimation.
bool	source_has_normals_ {false}
	Internal check whether source dataset has normals or not.
bool	target_has_normals_ {false}
	Internal check whether target dataset has normals or not.
bool	need_source_blob_
	Checks for whether estimators and rejectors need various data.
bool	need_target_blob_
Protected Attributes inherited from pcl::Registration< PointSource, PointTarget, Scalar >
std::string	reg_name_
	The registration method name.
KdTreePtr	tree_
	A pointer to the spatial search object.
KdTreeReciprocalPtr	tree_reciprocal_
	A pointer to the spatial search object of the source.
int	nr_iterations_ {0}
	The number of iterations the internal optimization ran for (used internally).
int	max_iterations_ {10}
	The maximum number of iterations the internal optimization should run for.
int	ransac_iterations_ {0}
	The number of iterations RANSAC should run for.
PointCloudTargetConstPtr	target_
	The input point cloud dataset target.
Matrix4	final_transformation_
	The final transformation matrix estimated by the registration method after N iterations.
Matrix4	transformation_
	The transformation matrix estimated by the registration method.
Matrix4	previous_transformation_
	The previous transformation matrix estimated by the registration method (used internally).
double	transformation_epsilon_ {0.0}
	The maximum difference between two consecutive transformations in order to consider convergence (user defined).
double	transformation_rotation_epsilon_ {0.0}
	The maximum rotation difference between two consecutive transformations in order to consider convergence (user defined).
double	euclidean_fitness_epsilon_
	The maximum allowed Euclidean error between two consecutive steps in the ICP loop, before the algorithm is considered to have converged.
double	corr_dist_threshold_
	The maximum distance threshold between two correspondent points in source <-> target.
double	inlier_threshold_ {0.05}
	The inlier distance threshold for the internal RANSAC outlier rejection loop.
bool	converged_ {false}
	Holds internal convergence state, given user parameters.
unsigned int	min_number_correspondences_ {3}
	The minimum number of correspondences that the algorithm needs before attempting to estimate the transformation.
CorrespondencesPtr	correspondences_
	The set of correspondences determined at this ICP step.
TransformationEstimationPtr	transformation_estimation_
	A TransformationEstimation object, used to calculate the 4x4 rigid transformation.
CorrespondenceEstimationPtr	correspondence_estimation_
	A CorrespondenceEstimation object, used to estimate correspondences between the source and the target cloud.
std::vector< CorrespondenceRejectorPtr >	correspondence_rejectors_
	The list of correspondence rejectors to use.
bool	target_cloud_updated_ {true}
	Variable that stores whether we have a new target cloud, meaning we need to pre-process it again.
bool	source_cloud_updated_ {true}
	Variable that stores whether we have a new source cloud, meaning we need to pre-process it again.
bool	force_no_recompute_ {false}
	A flag which, if set, means the tree operating on the target cloud will never be recomputed.
bool	force_no_recompute_reciprocal_ {false}
	A flag which, if set, means the tree operating on the source cloud will never be recomputed.
std::function< UpdateVisualizerCallbackSignature >	update_visualizer_
	Callback function to update intermediate source point cloud position during it's registration to the target point cloud.
Protected Attributes inherited from pcl::PCLBase< PointSource >
PointCloudConstPtr	input_
	The input point cloud dataset.
IndicesPtr	indices_
	A pointer to the vector of point indices to use.
bool	use_indices_
	Set to true if point indices are used.
bool	fake_indices_
	If no set of indices are given, we construct a set of fake indices that mimic the input PointCloud.

Detailed Description

template<typename PointSource, typename PointTarget, typename Scalar = float>
class pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >

FastRobustIterativeClosestPoint implements the FRICP variant described in "Fast and Robust Iterative Closest Point", Zhang et al., 2021.

The solver relies on Welsch reweighting for robustness and optional Anderson acceleration for faster convergence.

Note

FRICP uses its own robust correspondence mechanism based on the Welsch function. The following inherited ICP settings have no effect:

• setMaxCorrespondenceDistance() 閳?FRICP controls outlier rejection via the Welsch scale parameter instead.
• Correspondence estimators, rejectors, and reciprocal correspondences are not used by this class.
• setEuclideanFitnessEpsilon() and setRotationEpsilon() are not evaluated.

setMaximumIterations() controls the number of inner-loop iterations per Welsch scale stage. The outer loop may run multiple stages as the scale decays, so the total number of iterations can exceed this value.

Input point clouds must not contain NaN or Inf values. Non-finite points are not handled and will produce incorrect results.

pcl::FastRobustIterativeClosestPoint<PointT, PointT> reg;
reg.setInputSource (src); // src and tgt are clouds that must be created before
reg.setInputTarget (tgt);
// parameters may have to be tuned, depending on the point clouds
reg.setMaximumIterations (60);
reg.setTransformationEpsilon (1e-8);
pcl::PointCloud<PointT> output;
reg.align (output);

Definition at line 61 of file fricp.h.

Member Typedef Documentation

Base

template<typename PointSource , typename PointTarget , typename Scalar = float>

using pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::Base = IterativeClosestPoint<PointSource, PointTarget, Scalar>

Definition at line 69 of file fricp.h.

ConstPtr

template<typename PointSource , typename PointTarget , typename Scalar = float>

using pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::ConstPtr = shared_ptr< const FastRobustIterativeClosestPoint<PointSource, PointTarget, Scalar> >

Definition at line 66 of file fricp.h.

Matrix4

template<typename PointSource , typename PointTarget , typename Scalar = float>

using pcl::IterativeClosestPoint< PointSource, PointTarget, Scalar >::Matrix4 = typename Registration<PointSource, PointTarget, Scalar>::Matrix4

Definition at line 143 of file icp.h.

PointCloudSource

template<typename PointSource , typename PointTarget , typename Scalar = float>

using pcl::IterativeClosestPoint< PointSource, PointTarget, Scalar >::PointCloudSource = typename Registration<PointSource, PointTarget, Scalar>::PointCloudSource

Definition at line 100 of file icp.h.

PointCloudTarget

template<typename PointSource , typename PointTarget , typename Scalar = float>

using pcl::IterativeClosestPoint< PointSource, PointTarget, Scalar >::PointCloudTarget = typename Registration<PointSource, PointTarget, Scalar>::PointCloudTarget

Definition at line 105 of file icp.h.

Ptr

template<typename PointSource , typename PointTarget , typename Scalar = float>

using pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::Ptr = shared_ptr<FastRobustIterativeClosestPoint<PointSource, PointTarget, Scalar> >

Definition at line 64 of file fricp.h.

Member Enumeration Documentation

RobustFunction

template<typename PointSource , typename PointTarget , typename Scalar = float>

enum class pcl::FastRobustIterativeClosestPoint::RobustFunction

strong

Enumerator
NONE
WELSCH

Definition at line 74 of file fricp.h.

Constructor & Destructor Documentation

FastRobustIterativeClosestPoint()

template<typename PointSource , typename PointTarget , typename Scalar >

pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::FastRobustIterativeClosestPoint

(

)

Definition at line 30 of file fricp.hpp.

Member Function Documentation

computeTransformation()

template<typename PointSource , typename PointTarget , typename Scalar >

void pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::computeTransformation ( PointCloudSource &  output, const Matrix4 &  guess  )

overrideprotectedvirtual

Abstract transformation computation method with initial guess.

Implements pcl::Registration< PointSource, PointTarget, Scalar >.

Definition at line 117 of file fricp.hpp.

References pcl::computeMedian(), pcl::Registration< PointSource, PointTarget, Scalar >::initComputeReciprocal(), pcl::isFinite(), pcl::PointCloud< PointT >::resize(), pcl::search::Search< PointT >::setInputCloud(), and pcl::transformPointCloud().

getAndersonHistorySize()

template<typename PointSource , typename PointTarget , typename Scalar >

std::size_t pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::getAndersonHistorySize

(

)

const

Definition at line 83 of file fricp.hpp.

getRobustFunction()

template<typename PointSource , typename PointTarget , typename Scalar >

FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::RobustFunction pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::getRobustFunction

(

)

const

Definition at line 51 of file fricp.hpp.

getUseAndersonAcceleration()

template<typename PointSource , typename PointTarget , typename Scalar >

bool pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::getUseAndersonAcceleration

(

)

const

Definition at line 67 of file fricp.hpp.

setAndersonHistorySize()

template<typename PointSource , typename PointTarget , typename Scalar >

void pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::setAndersonHistorySize

(

std::size_t

history

)

Set the history size used by Anderson acceleration.

Larger values may improve acceleration quality but can increase instability and memory usage. Values smaller than 1 are clamped to 1.

Definition at line 75 of file fricp.hpp.

setDynamicWelschBeginRatio()

template<typename PointSource , typename PointTarget , typename Scalar >

void pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::setDynamicWelschBeginRatio

(

double

ratio

)

Set the initial Welsch scale ratio used in dynamic robust weighting.

Larger values start with weaker down-weighting of outliers. Values are clamped to a small positive threshold.

Definition at line 91 of file fricp.hpp.

setDynamicWelschDecay()

template<typename PointSource , typename PointTarget , typename Scalar >

void pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::setDynamicWelschDecay

(

double

ratio

)

Set the multiplicative decay applied to the dynamic Welsch scale.

Valid range is (0, 1]. A value of 0 or a very small positive number is silently replaced by the default (0.5). Smaller values reduce the scale faster per outer iteration, while larger values keep it closer to the current value.

Definition at line 107 of file fricp.hpp.

setDynamicWelschEndRatio()

template<typename PointSource , typename PointTarget , typename Scalar >

void pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::setDynamicWelschEndRatio

(

double

ratio

)

Set the final Welsch scale ratio used in dynamic robust weighting.

Smaller values end with stronger outlier suppression. Values are clamped to a small positive threshold.

Definition at line 99 of file fricp.hpp.

setRobustFunction()

template<typename PointSource , typename PointTarget , typename Scalar >

void pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::setRobustFunction

(

RobustFunction

f

)

Definition at line 41 of file fricp.hpp.

setUseAndersonAcceleration()

template<typename PointSource , typename PointTarget , typename Scalar >

void pcl::FastRobustIterativeClosestPoint< PointSource, PointTarget, Scalar >::setUseAndersonAcceleration

(

bool

enabled

)

Enable or disable Anderson acceleration in the FRICP optimization loop.

When enabled, convergence can be faster on some datasets but may become less stable. The default is disabled to keep behavior predictable.

Definition at line 59 of file fricp.hpp.

---

The documentation for this class was generated from the following files:

• pcl/registration/fricp.h
• pcl/registration/impl/fricp.hpp