This class implements Implicit Shape Model algorithm described in "Hough Transforms and 3D SURF for robust three dimensional classification" by Jan Knopp, Mukta Prasad, Geert Willems, Radu Timofte, and Luc Van Gool. More...

#include <pcl/recognition/implicit_shape_model.h>

Collaboration diagram for pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >:

Classes
struct	LocationInfo
	This structure stores the information about the keypoint. More...

struct	TermCriteria
	This structure is used for determining the end of the k-means clustering process. More...

struct	VisualWordStat
	Structure for storing the visual word. More...

Public Types
using	ISMModelPtr = pcl::features::ISMModel::Ptr

using	Feature = pcl::Feature< PointT, pcl::Histogram< FeatureSize > >

using	FeaturePtr = typename Feature::Ptr

Public Member Functions
	ImplicitShapeModelEstimation ()
	Simple constructor that initializes everything. More...

virtual	~ImplicitShapeModelEstimation ()
	Simple destructor. More...

std::vector< typename pcl::PointCloud< PointT >::Ptr >	getTrainingClouds ()
	This method simply returns the clouds that were set as the training clouds. More...

void	setTrainingClouds (const std::vector< typename pcl::PointCloud< PointT >::Ptr > &training_clouds)
	Allows to set clouds for training the ISM model. More...

std::vector< unsigned int >	getTrainingClasses ()
	Returns the array of classes that indicates which class the corresponding training cloud belongs. More...

void	setTrainingClasses (const std::vector< unsigned int > &training_classes)
	Allows to set the class labels for the corresponding training clouds. More...

std::vector< typename pcl::PointCloud< NormalT >::Ptr >	getTrainingNormals ()
	This method returns the corresponding cloud of normals for every training point cloud. More...

void	setTrainingNormals (const std::vector< typename pcl::PointCloud< NormalT >::Ptr > &training_normals)
	Allows to set normals for the training clouds that were passed through setTrainingClouds method. More...

float	getSamplingSize ()
	Returns the sampling size used for cloud simplification. More...

void	setSamplingSize (float sampling_size)
	Changes the sampling size used for cloud simplification. More...

FeaturePtr	getFeatureEstimator ()
	Returns the current feature estimator used for extraction of the descriptors. More...

void	setFeatureEstimator (FeaturePtr feature)
	Changes the feature estimator. More...

unsigned int	getNumberOfClusters ()
	Returns the number of clusters used for descriptor clustering. More...

void	setNumberOfClusters (unsigned int num_of_clusters)
	Changes the number of clusters. More...

std::vector< float >	getSigmaDists ()
	Returns the array of sigma values. More...

void	setSigmaDists (const std::vector< float > &training_sigmas)
	This method allows to set the value of sigma used for calculating the learned weights for every single class. More...

bool	getNVotState ()
	Returns the state of Nvot coeff from [Knopp et al., 2010, (4)], if set to false then coeff is taken as 1.0. More...

void	setNVotState (bool state)
	Changes the state of the Nvot coeff from [Knopp et al., 2010, (4)]. More...

bool	trainISM (ISMModelPtr &trained_model)
	This method performs training and forms a visual vocabulary. More...

pcl::features::ISMVoteList< PointT >::Ptr	findObjects (ISMModelPtr model, typename pcl::PointCloud< PointT >::Ptr in_cloud, typename pcl::PointCloud< Normal >::Ptr in_normals, int in_class_of_interest)
	This function is searching for the class of interest in a given cloud and returns the list of votes. More...

Protected Member Functions
bool	extractDescriptors (std::vector< pcl::Histogram< FeatureSize > > &histograms, std::vector< LocationInfo, Eigen::aligned_allocator< LocationInfo > > &locations)
	Extracts the descriptors from the input clouds. More...

bool	clusterDescriptors (std::vector< pcl::Histogram< FeatureSize > > &histograms, Eigen::MatrixXi &labels, Eigen::MatrixXf &clusters_centers)
	This method performs descriptor clustering. More...

void	calculateSigmas (std::vector< float > &sigmas)
	This method calculates the value of sigma used for calculating the learned weights for every single class. More...

void	calculateWeights (const std::vector< LocationInfo, Eigen::aligned_allocator< LocationInfo > > &locations, const Eigen::MatrixXi &labels, std::vector< float > &sigmas, std::vector< std::vector< unsigned int > > &clusters, std::vector< std::vector< float > > &statistical_weights, std::vector< float > &learned_weights)
	This function forms a visual vocabulary and evaluates weights described in [Knopp et al., 2010, (5)]. More...

void	simplifyCloud (typename pcl::PointCloud< PointT >::ConstPtr in_point_cloud, typename pcl::PointCloud< NormalT >::ConstPtr in_normal_cloud, typename pcl::PointCloud< PointT >::Ptr out_sampled_point_cloud, typename pcl::PointCloud< NormalT >::Ptr out_sampled_normal_cloud)
	Simplifies the cloud using voxel grid principles. More...

void	shiftCloud (typename pcl::PointCloud< PointT >::Ptr in_cloud, Eigen::Vector3f shift_point)
	This method simply shifts the clouds points relative to the passed point. More...

Eigen::Matrix3f	alignYCoordWithNormal (const NormalT &in_normal)
	This method simply computes the rotation matrix, so that the given normal would match the Y axis after the transformation. More...

void	applyTransform (Eigen::Vector3f &io_vec, const Eigen::Matrix3f &in_transform)
	This method applies transform set in in_transform to vector io_vector. More...

void	estimateFeatures (typename pcl::PointCloud< PointT >::Ptr sampled_point_cloud, typename pcl::PointCloud< NormalT >::Ptr normal_cloud, typename pcl::PointCloud< pcl::Histogram< FeatureSize > >::Ptr feature_cloud)
	This method estimates features for the given point cloud. More...

double	computeKMeansClustering (const Eigen::MatrixXf &points_to_cluster, int number_of_clusters, Eigen::MatrixXi &io_labels, TermCriteria criteria, int attempts, int flags, Eigen::MatrixXf &cluster_centers)
	Performs K-means clustering. More...

void	generateCentersPP (const Eigen::MatrixXf &data, Eigen::MatrixXf &out_centers, int number_of_clusters, int trials)
	Generates centers for clusters as described in Arthur, David and Sergei Vassilvitski (2007) k-means++: The Advantages of Careful Seeding. More...

void	generateRandomCenter (const std::vector< Eigen::Vector2f, Eigen::aligned_allocator< Eigen::Vector2f > > &boxes, Eigen::VectorXf &center)
	Generates random center for cluster. More...

float	computeDistance (Eigen::VectorXf &vec_1, Eigen::VectorXf &vec_2)
	Computes the square distance between two vectors. More...

ImplicitShapeModelEstimation &	operator= (const ImplicitShapeModelEstimation &)
	Forbids the assignment operator. More...

Protected Attributes
std::vector< typename pcl::PointCloud< PointT >::Ptr >	training_clouds_ {}
	Stores the clouds used for training. More...

std::vector< unsigned int >	training_classes_ {}
	Stores the class number for each cloud from training_clouds_. More...

std::vector< typename pcl::PointCloud< NormalT >::Ptr >	training_normals_ {}
	Stores the normals for each training cloud. More...

std::vector< float >	training_sigmas_ {}
	This array stores the sigma values for each training class. More...

float	sampling_size_ {0.1f}
	This value is used for the simplification. More...

Feature::Ptr	feature_estimator_
	Stores the feature estimator. More...

unsigned int	number_of_clusters_ {184}
	Number of clusters, is used for clustering descriptors during the training. More...

bool	n_vot_ON_ {true}
	If set to false then Nvot coeff from [Knopp et al., 2010, (4)] is equal 1.0. More...

Static Protected Attributes
static const int	PP_CENTERS = 2
	This const value is used for indicating that for k-means clustering centers must be generated as described in Arthur, David and Sergei Vassilvitski (2007) k-means++: The Advantages of Careful Seeding. More...

static const int	USE_INITIAL_LABELS = 1
	This const value is used for indicating that input labels must be taken as the initial approximation for k-means clustering. More...

Detailed Description

template<int FeatureSize, typename PointT, typename NormalT = pcl::Normal>
class pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >

This class implements Implicit Shape Model algorithm described in "Hough Transforms and 3D SURF for robust three dimensional classification" by Jan Knopp, Mukta Prasad, Geert Willems, Radu Timofte, and Luc Van Gool.

It has two main member functions. One for training, using the data for which we know which class it belongs to. And second for investigating a cloud for the presence of the class of interest. Implementation of the ISM algorithm described in "Hough Transforms and 3D SURF for robust three dimensional classification" by Jan Knopp, Mukta Prasad, Geert Willems, Radu Timofte, and Luc Van Gool

Authors: Roman Shapovalov, Alexander Velizhev, Sergey Ushakov

Definition at line 241 of file implicit_shape_model.h.

Member Typedef Documentation

◆ Feature

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

using pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::Feature = pcl::Feature<PointT, pcl::Histogram<FeatureSize> >

Definition at line 246 of file implicit_shape_model.h.

◆ FeaturePtr

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

using pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::FeaturePtr = typename Feature::Ptr

Definition at line 247 of file implicit_shape_model.h.

◆ ISMModelPtr

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

using pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::ISMModelPtr = pcl::features::ISMModel::Ptr

Definition at line 245 of file implicit_shape_model.h.

Constructor & Destructor Documentation

◆ ImplicitShapeModelEstimation()

template<int FeatureSize, typename PointT , typename NormalT >

pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::ImplicitShapeModelEstimation ( )

default

Simple constructor that initializes everything.

◆ ~ImplicitShapeModelEstimation()

template<int FeatureSize, typename PointT , typename NormalT >

pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::~ImplicitShapeModelEstimation

virtual

Simple destructor.

Definition at line 534 of file implicit_shape_model.hpp.

Member Function Documentation

◆ alignYCoordWithNormal()

template<int FeatureSize, typename PointT , typename NormalT >

Eigen::Matrix3f pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::alignYCoordWithNormal ( const NormalT & in_normal )

protected

This method simply computes the rotation matrix, so that the given normal would match the Y axis after the transformation.

This is done because the algorithm needs to be invariant to the affine transformations.

Parameters

[in] in_normal normal for which the rotation matrix need to be computed

Definition at line 1164 of file implicit_shape_model.hpp.

References pcl::B.

◆ applyTransform()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::applyTransform	(	Eigen::Vector3f &	io_vec,
		const Eigen::Matrix3f &	in_transform
	)

protected

This method applies transform set in in_transform to vector io_vector.

Parameters

[in]	io_vec	vector that need to be transformed
[in]	in_transform	matrix that contains the transformation

Definition at line 1195 of file implicit_shape_model.hpp.

◆ calculateSigmas()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::calculateSigmas ( std::vector< float > & sigmas )

protected

This method calculates the value of sigma used for calculating the learned weights for every single class.

Parameters

[out] sigmas computed sigmas.

Definition at line 905 of file implicit_shape_model.hpp.

◆ calculateWeights()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::calculateWeights	(	const std::vector< LocationInfo, Eigen::aligned_allocator< LocationInfo > > &	locations,
		const Eigen::MatrixXi &	labels,
		std::vector< float > &	sigmas,
		std::vector< std::vector< unsigned int > > &	clusters,
		std::vector< std::vector< float > > &	statistical_weights,
		std::vector< float > &	learned_weights
	)

protected

This function forms a visual vocabulary and evaluates weights described in [Knopp et al., 2010, (5)].

Parameters

[in]	locations	array containing description of each keypoint: its position, which cloud belongs and expected direction to center
[in]	labels	labels that were obtained during k-means clustering
[in]	sigmas	array of sigmas for each class
[in]	clusters	clusters that were obtained during k-means clustering
[out]	statistical_weights	stores the computed statistical weights
[out]	learned_weights	stores the computed learned weights

Definition at line 956 of file implicit_shape_model.hpp.

◆ clusterDescriptors()

template<int FeatureSize, typename PointT , typename NormalT >

bool pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::clusterDescriptors	(	std::vector< pcl::Histogram< FeatureSize > > &	histograms,
		Eigen::MatrixXi &	labels,
		Eigen::MatrixXf &	clusters_centers
	)

protected

This method performs descriptor clustering.

Parameters

[in]	histograms	descriptors to cluster
[out]	labels	it contains labels for each descriptor
[out]	clusters_centers	stores the centers of clusters

Definition at line 879 of file implicit_shape_model.hpp.

◆ computeDistance()

template<int FeatureSize, typename PointT , typename NormalT >

float pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::computeDistance	(	Eigen::VectorXf &	vec_1,
		Eigen::VectorXf &	vec_2
	)

protected

Computes the square distance between two vectors.

Parameters

[in]	vec_1	first vector
[in]	vec_2	second vector

Definition at line 1485 of file implicit_shape_model.hpp.

References pcl::geometry::distance().

◆ computeKMeansClustering()

template<int FeatureSize, typename PointT , typename NormalT >

double pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::computeKMeansClustering	(	const Eigen::MatrixXf &	points_to_cluster,
		int	number_of_clusters,
		Eigen::MatrixXi &	io_labels,
		TermCriteria	criteria,
		int	attempts,
		int	flags,
		Eigen::MatrixXf &	cluster_centers
	)

protected

Performs K-means clustering.

Parameters

[in]	points_to_cluster	points to cluster
[in]	number_of_clusters	desired number of clusters
[out]	io_labels	output parameter, which stores the label for each point
[in]	criteria	defines when the computational process need to be finished. For example if the desired accuracy is achieved or the iteration number exceeds given value
[in]	attempts	number of attempts to compute clustering
[in]	flags	if set to USE_INITIAL_LABELS then initial approximation of labels is taken from io_labels
[out]	cluster_centers	it will store the cluster centers

Definition at line 1227 of file implicit_shape_model.hpp.

References pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::TermCriteria::epsilon_, pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::TermCriteria::max_count_, and pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::TermCriteria::type_.

◆ estimateFeatures()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::estimateFeatures	(	typename pcl::PointCloud< PointT >::Ptr	sampled_point_cloud,
		typename pcl::PointCloud< NormalT >::Ptr	normal_cloud,
		typename pcl::PointCloud< pcl::Histogram< FeatureSize > >::Ptr	feature_cloud
	)

protected

This method estimates features for the given point cloud.

Parameters

[in]	sampled_point_cloud	sampled point cloud for which the features must be computed
[in]	normal_cloud	normals for the original point cloud
[out]	feature_cloud	it will store the computed histograms (features) for the given cloud

Definition at line 1202 of file implicit_shape_model.hpp.

References pcl::PointCloud< PointT >::makeShared(), and pcl::FeatureFromNormals< PointInT, PointNT, PointOutT >::setInputNormals().

◆ extractDescriptors()

template<int FeatureSize, typename PointT , typename NormalT >

bool pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::extractDescriptors	(	std::vector< pcl::Histogram< FeatureSize > > &	histograms,
		std::vector< LocationInfo, Eigen::aligned_allocator< LocationInfo > > &	locations
	)

protected

Extracts the descriptors from the input clouds.

Parameters

[out]	histograms	it will store the descriptors for each key point
[out]	locations	it will contain the comprehensive information (such as direction, initial keypoint) for the corresponding descriptors

Definition at line 818 of file implicit_shape_model.hpp.

References pcl::PointCloud< PointT >::begin(), pcl::geometry::distance(), and pcl::PointCloud< PointT >::points.

◆ findObjects()

template<int FeatureSize, typename PointT , typename NormalT >

pcl::features::ISMVoteList< PointT >::Ptr pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::findObjects	(	ISMModelPtr	model,
		typename pcl::PointCloud< PointT >::Ptr	in_cloud,
		typename pcl::PointCloud< Normal >::Ptr	in_normals,
		int	in_class_of_interest
	)

This function is searching for the class of interest in a given cloud and returns the list of votes.

Parameters

[in]	model	trained model which will be used for searching the objects
[in]	in_cloud	input cloud that need to be investigated
[in]	in_normals	cloud of normals corresponding to the input cloud
[in]	in_class_of_interest	class which we are looking for

Definition at line 723 of file implicit_shape_model.hpp.

References pcl::features::ISMVoteList< PointT >::addVote(), pcl::PointCloud< PointT >::points, pcl::PointCloud< PointT >::size(), and pcl::InterestPoint::strength.

◆ generateCentersPP()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::generateCentersPP	(	const Eigen::MatrixXf &	data,
		Eigen::MatrixXf &	out_centers,
		int	number_of_clusters,
		int	trials
	)

protected

Generates centers for clusters as described in Arthur, David and Sergei Vassilvitski (2007) k-means++: The Advantages of Careful Seeding.

Parameters

[in]	data	points to cluster
[out]	out_centers	it will contain generated centers
[in]	number_of_clusters	defines the number of desired cluster centers
[in]	trials	number of trials to generate a center

Definition at line 1393 of file implicit_shape_model.hpp.

◆ generateRandomCenter()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::generateRandomCenter	(	const std::vector< Eigen::Vector2f, Eigen::aligned_allocator< Eigen::Vector2f > > &	boxes,
		Eigen::VectorXf &	center
	)

protected

Generates random center for cluster.

Parameters

[in]	boxes	contains min and max values for each dimension
[out]	center	it will the contain generated center

Definition at line 1469 of file implicit_shape_model.hpp.

◆ getFeatureEstimator()

template<int FeatureSize, typename PointT , typename NormalT >

pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::FeaturePtr pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::getFeatureEstimator

Returns the current feature estimator used for extraction of the descriptors.

Definition at line 610 of file implicit_shape_model.hpp.

◆ getNumberOfClusters()

template<int FeatureSize, typename PointT , typename NormalT >

unsigned int pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::getNumberOfClusters

Returns the number of clusters used for descriptor clustering.

Definition at line 624 of file implicit_shape_model.hpp.

◆ getNVotState()

template<int FeatureSize, typename PointT , typename NormalT >

bool pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::getNVotState

Returns the state of Nvot coeff from [Knopp et al., 2010, (4)], if set to false then coeff is taken as 1.0.

It is just a kind of heuristic. The default behavior is as in the article. So you can ignore this if you want.

Definition at line 655 of file implicit_shape_model.hpp.

◆ getSamplingSize()

template<int FeatureSize, typename PointT , typename NormalT >

float pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::getSamplingSize

Returns the sampling size used for cloud simplification.

Definition at line 595 of file implicit_shape_model.hpp.

◆ getSigmaDists()

template<int FeatureSize, typename PointT , typename NormalT >

std::vector< float > pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::getSigmaDists

Returns the array of sigma values.

Definition at line 639 of file implicit_shape_model.hpp.

◆ getTrainingClasses()

template<int FeatureSize, typename PointT , typename NormalT >

std::vector< unsigned int > pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::getTrainingClasses

Returns the array of classes that indicates which class the corresponding training cloud belongs.

Definition at line 562 of file implicit_shape_model.hpp.

◆ getTrainingClouds()

template<int FeatureSize, typename PointT , typename NormalT >

std::vector< typename pcl::PointCloud< PointT >::Ptr > pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::getTrainingClouds

This method simply returns the clouds that were set as the training clouds.

Definition at line 545 of file implicit_shape_model.hpp.

◆ getTrainingNormals()

template<int FeatureSize, typename PointT , typename NormalT >

std::vector< typename pcl::PointCloud< NormalT >::Ptr > pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::getTrainingNormals

This method returns the corresponding cloud of normals for every training point cloud.

Definition at line 578 of file implicit_shape_model.hpp.

◆ operator=()

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

ImplicitShapeModelEstimation& pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::operator= ( const ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT > & )

protected

Forbids the assignment operator.

◆ setFeatureEstimator()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::setFeatureEstimator ( FeaturePtr feature )

Changes the feature estimator.

Parameters

[in] feature feature estimator that will be used to extract the descriptors. Note that it must be fully initialized and configured.

Definition at line 617 of file implicit_shape_model.hpp.

◆ setNumberOfClusters()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::setNumberOfClusters ( unsigned int num_of_clusters )

Changes the number of clusters.

Parameters

num_of_clusters desired number of clusters

Definition at line 631 of file implicit_shape_model.hpp.

◆ setNVotState()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::setNVotState ( bool state )

Changes the state of the Nvot coeff from [Knopp et al., 2010, (4)].

Parameters

[in] state desired state, if false then Nvot is taken as 1.0

Definition at line 662 of file implicit_shape_model.hpp.

◆ setSamplingSize()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::setSamplingSize ( float sampling_size )

Changes the sampling size used for cloud simplification.

Parameters

[in] sampling_size desired size of grid bin

Definition at line 602 of file implicit_shape_model.hpp.

◆ setSigmaDists()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::setSigmaDists ( const std::vector< float > & training_sigmas )

This method allows to set the value of sigma used for calculating the learned weights for every single class.

Parameters

[in] training_sigmas new sigmas for every class. If you want these values to be computed automatically, just pass the empty array. The automatic regime calculates the maximum distance between the objects points and takes 10% of this value as recommended in the article. If there are several objects of the same class, then it computes the average maximum distance and takes 10%. Note that each class has its own sigma value.

Definition at line 646 of file implicit_shape_model.hpp.

◆ setTrainingClasses()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::setTrainingClasses ( const std::vector< unsigned int > & training_classes )

Allows to set the class labels for the corresponding training clouds.

Parameters

[in] training_classes array of class labels

Definition at line 569 of file implicit_shape_model.hpp.

◆ setTrainingClouds()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::setTrainingClouds ( const std::vector< typename pcl::PointCloud< PointT >::Ptr > & training_clouds )

Allows to set clouds for training the ISM model.

Parameters

[in] training_clouds array of point clouds for training

Definition at line 552 of file implicit_shape_model.hpp.

◆ setTrainingNormals()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::setTrainingNormals ( const std::vector< typename pcl::PointCloud< NormalT >::Ptr > & training_normals )

Allows to set normals for the training clouds that were passed through setTrainingClouds method.

Parameters

[in] training_normals array of clouds, each cloud is the cloud of normals

Definition at line 585 of file implicit_shape_model.hpp.

◆ shiftCloud()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::shiftCloud	(	typename pcl::PointCloud< PointT >::Ptr	in_cloud,
		Eigen::Vector3f	shift_point
	)

protected

This method simply shifts the clouds points relative to the passed point.

Parameters

[in]	in_cloud	cloud to shift
[in]	shift_point	point relative to which the cloud will be shifted

Definition at line 1150 of file implicit_shape_model.hpp.

References pcl::PointCloud< PointT >::points.

◆ simplifyCloud()

template<int FeatureSize, typename PointT , typename NormalT >

void pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::simplifyCloud	(	typename pcl::PointCloud< PointT >::ConstPtr	in_point_cloud,
		typename pcl::PointCloud< NormalT >::ConstPtr	in_normal_cloud,
		typename pcl::PointCloud< PointT >::Ptr	out_sampled_point_cloud,
		typename pcl::PointCloud< NormalT >::Ptr	out_sampled_normal_cloud
	)

protected

Simplifies the cloud using voxel grid principles.

Parameters

[in]	in_point_cloud	cloud that need to be simplified
[in]	in_normal_cloud	normals of the cloud that need to be simplified
[out]	out_sampled_point_cloud	simplified cloud
[out]	out_sampled_normal_cloud	and the corresponding normals

Definition at line 1086 of file implicit_shape_model.hpp.

References pcl::geometry::distance(), pcl::FilterIndices< PointT >::filter(), pcl::Filter< PointT >::filter(), pcl::VoxelGrid< PointT >::getCentroidIndex(), pcl::PCLBase< PointT >::setIndices(), pcl::PCLBase< PointT >::setInputCloud(), pcl::VoxelGrid< PointT >::setLeafSize(), pcl::VoxelGrid< PointT >::setSaveLeafLayout(), and pcl::PointCloud< PointT >::size().

◆ trainISM()

template<int FeatureSize, typename PointT , typename NormalT >

bool pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::trainISM ( ISMModelPtr & trained_model )

This method performs training and forms a visual vocabulary.

It returns a trained model that can be saved to file for later usage.

Parameters

[out] trained_model trained model

Definition at line 669 of file implicit_shape_model.hpp.

Member Data Documentation

◆ feature_estimator_

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

Feature::Ptr pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::feature_estimator_

protected

Stores the feature estimator.

Definition at line 602 of file implicit_shape_model.h.

◆ n_vot_ON_

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

bool pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::n_vot_ON_ {true}

protected

If set to false then Nvot coeff from [Knopp et al., 2010, (4)] is equal 1.0.

Definition at line 608 of file implicit_shape_model.h.

◆ number_of_clusters_

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

unsigned int pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::number_of_clusters_ {184}

protected

Number of clusters, is used for clustering descriptors during the training.

Definition at line 605 of file implicit_shape_model.h.

◆ PP_CENTERS

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

const int pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::PP_CENTERS = 2

staticprotected

This const value is used for indicating that for k-means clustering centers must be generated as described in Arthur, David and Sergei Vassilvitski (2007) k-means++: The Advantages of Careful Seeding.

Definition at line 613 of file implicit_shape_model.h.

◆ sampling_size_

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

float pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::sampling_size_ {0.1f}

protected

This value is used for the simplification.

It sets the size of grid bin.

Definition at line 599 of file implicit_shape_model.h.

◆ training_classes_

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

std::vector<unsigned int> pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::training_classes_ {}

protected

Stores the class number for each cloud from training_clouds_.

Definition at line 588 of file implicit_shape_model.h.

◆ training_clouds_

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

std::vector<typename pcl::PointCloud<PointT>::Ptr> pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::training_clouds_ {}

protected

Stores the clouds used for training.

Definition at line 585 of file implicit_shape_model.h.

◆ training_normals_

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

std::vector<typename pcl::PointCloud<NormalT>::Ptr> pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::training_normals_ {}

protected

Stores the normals for each training cloud.

Definition at line 591 of file implicit_shape_model.h.

◆ training_sigmas_

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

std::vector<float> pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::training_sigmas_ {}

protected

This array stores the sigma values for each training class.

If this array has a size equals 0, then sigma values will be calculated automatically.

Definition at line 596 of file implicit_shape_model.h.

◆ USE_INITIAL_LABELS

template<int FeatureSize, typename PointT , typename NormalT = pcl::Normal>

const int pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >::USE_INITIAL_LABELS = 1

staticprotected

This const value is used for indicating that input labels must be taken as the initial approximation for k-means clustering.

Definition at line 617 of file implicit_shape_model.h.

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

pcl/recognition/implicit_shape_model.h
pcl/recognition/impl/implicit_shape_model.hpp

Classes

Public Types

Public Member Functions

Protected Member Functions

Protected Attributes

Static Protected Attributes

Detailed Description

template<int FeatureSize, typename PointT, typename NormalT = pcl::Normal> class pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >

Member Typedef Documentation

◆ Feature

◆ FeaturePtr

◆ ISMModelPtr

Constructor & Destructor Documentation

◆ ImplicitShapeModelEstimation()

◆ ~ImplicitShapeModelEstimation()

Member Function Documentation

◆ alignYCoordWithNormal()

◆ applyTransform()

◆ calculateSigmas()

◆ calculateWeights()

◆ clusterDescriptors()

◆ computeDistance()

◆ computeKMeansClustering()

◆ estimateFeatures()

◆ extractDescriptors()

◆ findObjects()

◆ generateCentersPP()

◆ generateRandomCenter()

◆ getFeatureEstimator()

◆ getNumberOfClusters()

◆ getNVotState()

◆ getSamplingSize()

◆ getSigmaDists()

◆ getTrainingClasses()

◆ getTrainingClouds()

◆ getTrainingNormals()

◆ operator=()

◆ setFeatureEstimator()

◆ setNumberOfClusters()

◆ setNVotState()

◆ setSamplingSize()

◆ setSigmaDists()

◆ setTrainingClasses()

◆ setTrainingClouds()

◆ setTrainingNormals()

◆ shiftCloud()

◆ simplifyCloud()

◆ trainISM()

Member Data Documentation

◆ feature_estimator_

◆ n_vot_ON_

◆ number_of_clusters_

◆ PP_CENTERS

◆ sampling_size_

◆ training_classes_

◆ training_clouds_

◆ training_normals_

◆ training_sigmas_

◆ USE_INITIAL_LABELS

template<int FeatureSize, typename PointT, typename NormalT = pcl::Normal>
class pcl::ism::ImplicitShapeModelEstimation< FeatureSize, PointT, NormalT >