Class implementing a 3D correspondence grouping algorithm that can deal with multiple instances of a model template found into a given scene. More...
#include <pcl/recognition/cg/hough_3d.h>
Public Member Functions |
|
| Hough3DGrouping () | |
| Constructor. More... |
|
| void | setInputCloud (const PointCloudConstPtr &cloud) override |
| Provide a pointer to the input dataset. More... |
|
| void | setInputRf (const ModelRfCloudConstPtr &input_rf) |
| Provide a pointer to the input dataset's reference frames. More... |
|
| ModelRfCloudConstPtr | getInputRf () const |
| Getter for the input dataset's reference frames. More... |
|
| void | setSceneCloud (const SceneCloudConstPtr &scene) override |
| Provide a pointer to the scene dataset (i.e. More... |
|
| void | setSceneRf (const SceneRfCloudConstPtr &scene_rf) |
| Provide a pointer to the scene dataset's reference frames. More... |
|
| SceneRfCloudConstPtr | getSceneRf () const |
| Getter for the scene dataset's reference frames. More... |
|
| void | setModelSceneCorrespondences (const CorrespondencesConstPtr &corrs) override |
| Provide a pointer to the precomputed correspondences between points in the input dataset and points in the scene dataset. More... |
|
| void | setHoughThreshold (double threshold) |
| Sets the minimum number of votes in the Hough space needed to infer the presence of a model instance into the scene cloud. More... |
|
| double | getHoughThreshold () const |
| Gets the minimum number of votes in the Hough space needed to infer the presence of a model instance into the scene cloud. More... |
|
| void | setHoughBinSize (double bin_size) |
| Sets the size of each bin into the Hough space. More... |
|
| double | getHoughBinSize () const |
| Gets the size of each bin into the Hough space. More... |
|
| void | setUseInterpolation (bool use_interpolation) |
| Sets whether the vote casting procedure interpolates the score between neighboring bins of the Hough space or not. More... |
|
| bool | getUseInterpolation () const |
| Gets whether the vote casting procedure interpolates the score between neighboring bins of the Hough space or not. More... |
|
| void | setUseDistanceWeight (bool use_distance_weight) |
| Sets whether the vote casting procedure uses the correspondence's distance as a score. More... |
|
| bool | getUseDistanceWeight () const |
| Gets whether the vote casting procedure uses the correspondence's distance as a score. More... |
|
| void | setLocalRfNormalsSearchRadius (float local_rf_normals_search_radius) |
| If the Local reference frame has not been set for either the model cloud or the scene cloud, this algorithm makes the computation itself but needs a suitable search radius to compute the normals in order to subsequently compute the RF (if not set a default 15 nearest neighbors search is performed). More... |
|
| float | getLocalRfNormalsSearchRadius () const |
| If the Local reference frame has not been set for either the model cloud or the scene cloud, this algorithm makes the computation itself but needs a suitable search radius to compute the normals in order to subsequently compute the RF (if not set a default 15 nearest neighbors search is performed). More... |
|
| void | setLocalRfSearchRadius (float local_rf_search_radius) |
| If the Local reference frame has not been set for either the model cloud or the scene cloud, this algorithm makes the computation itself but needs a suitable search radius to do so. More... |
|
| float | getLocalRfSearchRadius () const |
| If the Local reference frame has not been set for either the model cloud or the scene cloud, this algorithm makes the computation itself but needs a suitable search radius to do so. More... |
|
| bool | train () |
| Call this function after setting the input, the input_rf and the hough_bin_size parameters to perform an off line training of the algorithm. More... |
|
| bool | recognize (std::vector< Eigen::Matrix4f, Eigen::aligned_allocator< Eigen::Matrix4f > > &transformations) |
| The main function, recognizes instances of the model into the scene set by the user. More... |
|
| bool | recognize (std::vector< Eigen::Matrix4f, Eigen::aligned_allocator< Eigen::Matrix4f > > &transformations, std::vector< pcl::Correspondences > &clustered_corrs) |
| The main function, recognizes instances of the model into the scene set by the user. More... |
|
 Public Member Functions inherited from pcl::CorrespondenceGrouping< PointModelT, PointSceneT > |
|
| CorrespondenceGrouping () | |
| Empty constructor. More... |
|
| ~CorrespondenceGrouping () | |
| destructor. More... |
|
| SceneCloudConstPtr | getSceneCloud () const |
| Getter for the scene dataset. More... |
|
| CorrespondencesConstPtr | getModelSceneCorrespondences () const |
| Getter for the precomputed correspondences between points in the input dataset and points in the scene dataset. More... |
|
| std::vector< double > | getCharacteristicScales () const |
| Getter for the vector of characteristic scales associated to each cluster. More... |
|
| void | cluster (std::vector< Correspondences > &clustered_corrs) |
| Clusters the input correspondences belonging to different model instances. More... |
|
| Public Member Functions inherited from pcl::PCLBase< PointModelT > |
|
| PCLBase () | |
| Empty constructor. More... |
|
| PCLBase (const PCLBase &base) | |
| Copy constructor. More... |
|
| virtual | ~PCLBase ()=default |
| Destructor. More... |
|
| const PointCloudConstPtr | getInputCloud () const |
| Get a pointer to the input point cloud dataset. More... |
|
| virtual void | setIndices (const IndicesPtr &indices) |
| Provide a pointer to the vector of indices that represents the input data. More... |
|
| virtual void | setIndices (const IndicesConstPtr &indices) |
| Provide a pointer to the vector of indices that represents the input data. More... |
|
| virtual void | setIndices (const PointIndicesConstPtr &indices) |
| Provide a pointer to the vector of indices that represents the input data. More... |
|
| 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. More... |
|
| IndicesPtr | getIndices () |
| Get a pointer to the vector of indices used. More... |
|
| const IndicesConstPtr | getIndices () const |
| Get a pointer to the vector of indices used. More... |
|
| const PointModelT & | operator[] (std::size_t pos) const |
| Override PointCloud operator[] to shorten code. More... |
|
Protected Member Functions |
|
| void | clusterCorrespondences (std::vector< Correspondences > &model_instances) override |
| Cluster the input correspondences in order to distinguish between different instances of the model into the scene. More... |
|
| bool | houghVoting () |
| The Hough space voting procedure. More... |
|
| template<typename PointType , typename PointRfType > | |
| void | computeRf (const typename pcl::PointCloud< PointType >::ConstPtr &input, pcl::PointCloud< PointRfType > &rf) |
| Computes the reference frame for an input cloud. More... |
|
| Protected Member Functions inherited from pcl::CorrespondenceGrouping< PointModelT, PointSceneT > |
|
| bool | initCompute () |
| This method should get called before starting the actual computation. More... |
|
| bool | deinitCompute () |
| This method should get called after finishing the actual computation. More... |
|
| Protected Member Functions inherited from pcl::PCLBase< PointModelT > |
|
| bool | initCompute () |
| This method should get called before starting the actual computation. More... |
|
| bool | deinitCompute () |
| This method should get called after finishing the actual computation. More... |
|
Protected Attributes |
|
| ModelRfCloudConstPtr | input_rf_ |
| The input Rf cloud. More... |
|
| SceneRfCloudConstPtr | scene_rf_ |
| The scene Rf cloud. More... |
|
| bool | needs_training_ |
| If the training of the Hough space is needed; set on change of either the input cloud or the input_rf. More... |
|
| std::vector< Eigen::Vector3f, Eigen::aligned_allocator< Eigen::Vector3f > > | model_votes_ |
| The result of the training. More... |
|
| double | hough_threshold_ |
| The minimum number of votes in the Hough space needed to infer the presence of a model instance into the scene cloud. More... |
|
| double | hough_bin_size_ |
| The size of each bin of the hough space. More... |
|
| bool | use_interpolation_ |
| Use the interpolation between neighboring Hough bins when casting votes. More... |
|
| bool | use_distance_weight_ |
| Use the weighted correspondence distance when casting votes. More... |
|
| float | local_rf_normals_search_radius_ |
| Normals search radius for the potential Rf calculation. More... |
|
| float | local_rf_search_radius_ |
| Search radius for the potential Rf calculation. More... |
|
| pcl::recognition::HoughSpace3D::Ptr | hough_space_ |
| The Hough space. More... |
|
| std::vector< Eigen::Matrix4f, Eigen::aligned_allocator< Eigen::Matrix4f > > | found_transformations_ |
| Transformations found by clusterCorrespondences method. More... |
|
| bool | hough_space_initialized_ |
| Whether the Hough space already contains the correct votes for the current input parameters and so the cluster and recognize calls don't need to recompute each value. More... |
|
| Protected Attributes inherited from pcl::CorrespondenceGrouping< PointModelT, PointSceneT > |
|
| SceneCloudConstPtr | scene_ |
| The scene cloud. More... |
|
| CorrespondencesConstPtr | model_scene_corrs_ |
| The correspondences between points in the input and the scene datasets. More... |
|
| std::vector< double > | corr_group_scale_ |
| characteristic scale associated to each correspondence subset; if the cg algorithm can not handle scale invariance, the size of the vector will be 0. More... |
|
| Protected Attributes inherited from pcl::PCLBase< PointModelT > |
|
| PointCloudConstPtr | input_ |
| The input point cloud dataset. More... |
|
| IndicesPtr | indices_ |
| A pointer to the vector of point indices to use. More... |
|
| bool | use_indices_ |
| Set to true if point indices are used. More... |
|
| bool | fake_indices_ |
| If no set of indices are given, we construct a set of fake indices that mimic the input PointCloud. More... |
|
Detailed Description
template<typename PointModelT, typename PointSceneT, typename PointModelRfT = pcl::ReferenceFrame, typename PointSceneRfT = pcl::ReferenceFrame>
class pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >
Class implementing a 3D correspondence grouping algorithm that can deal with multiple instances of a model template found into a given scene.
Each correspondence casts a vote for a reference point in a 3D Hough Space. The remaining 3 DOF are taken into account by associating each correspondence with a local Reference Frame. The suggested PointModelRfT is pcl::ReferenceFrame
- Note
-
If you use this code in any academic work, please cite the original paper:
- F. Tombari, L. Di Stefano: Object recognition in 3D scenes with occlusions and clutter by Hough voting. 2010, Fourth Pacific-Rim Symposium on Image and Video Technology
Definition at line 148 of file hough_3d.h.
Member Typedef Documentation
ModelRfCloud
| using pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::ModelRfCloud = pcl::PointCloud<PointModelRfT> |
Definition at line 151 of file hough_3d.h.
ModelRfCloudConstPtr
| using pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::ModelRfCloudConstPtr = typename ModelRfCloud::ConstPtr |
Definition at line 153 of file hough_3d.h.
ModelRfCloudPtr
| using pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::ModelRfCloudPtr = typename ModelRfCloud::Ptr |
Definition at line 152 of file hough_3d.h.
PointCloud
| using pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::PointCloud = pcl::PointCloud<PointModelT> |
Definition at line 159 of file hough_3d.h.
PointCloudConstPtr
| using pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::PointCloudConstPtr = typename PointCloud::ConstPtr |
Definition at line 161 of file hough_3d.h.
PointCloudPtr
| using pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::PointCloudPtr = typename PointCloud::Ptr |
Definition at line 160 of file hough_3d.h.
SceneCloudConstPtr
| using pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::SceneCloudConstPtr = typename pcl::CorrespondenceGrouping<PointModelT, PointSceneT>::SceneCloudConstPtr |
Definition at line 163 of file hough_3d.h.
SceneRfCloud
| using pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::SceneRfCloud = pcl::PointCloud<PointSceneRfT> |
Definition at line 155 of file hough_3d.h.
SceneRfCloudConstPtr
| using pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::SceneRfCloudConstPtr = typename SceneRfCloud::ConstPtr |
Definition at line 157 of file hough_3d.h.
SceneRfCloudPtr
| using pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::SceneRfCloudPtr = typename SceneRfCloud::Ptr |
Definition at line 156 of file hough_3d.h.
Constructor & Destructor Documentation
Hough3DGrouping()
|
inline |
Constructor.
Definition at line 166 of file hough_3d.h.
Member Function Documentation
clusterCorrespondences()
|
overrideprotectedvirtual |
Cluster the input correspondences in order to distinguish between different instances of the model into the scene.
- Parameters
-
[out] model_instances a vector containing the clustered correspondences for each model found on the scene.
- Returns
- true if the clustering had been successful or false if errors have occurred.
Implements pcl::CorrespondenceGrouping< PointModelT, PointSceneT >.
Definition at line 259 of file hough_3d.hpp.
References pcl::copyPointCloud(), pcl::registration::CorrespondenceRejectorSampleConsensus< PointT >::getBestTransformation(), pcl::registration::CorrespondenceRejectorSampleConsensus< PointT >::getRemainingCorrespondences(), pcl::registration::CorrespondenceRejectorSampleConsensus< PointT >::setInlierThreshold(), pcl::registration::CorrespondenceRejectorSampleConsensus< PointT >::setInputSource(), pcl::registration::CorrespondenceRejectorSampleConsensus< PointT >::setInputTarget(), and pcl::registration::CorrespondenceRejectorSampleConsensus< PointT >::setMaximumIterations().
computeRf()
|
protected |
Computes the reference frame for an input cloud.
- Parameters
-
[in] input the input cloud. [out] rf the resulting reference frame.
Definition at line 55 of file hough_3d.hpp.
References pcl::Feature< PointInT, PointOutT >::compute(), pcl::Feature< PointInT, ReferenceFrame >::compute(), pcl::BOARDLocalReferenceFrameEstimation< PointInT, PointNT, PointOutT >::setFindHoles(), pcl::PCLBase< PointInT >::setInputCloud(), pcl::NormalEstimation< PointInT, PointOutT >::setInputCloud(), pcl::FeatureFromNormals< PointInT, PointNT, ReferenceFrame >::setInputNormals(), pcl::Feature< PointInT, PointOutT >::setKSearch(), pcl::Feature< PointInT, ReferenceFrame >::setRadiusSearch(), and pcl::Feature< PointInT, PointOutT >::setRadiusSearch().
getHoughBinSize()
|
inline |
Gets the size of each bin into the Hough space.
- Returns
- the size of each Hough space's bin.
Definition at line 306 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_bin_size_.
getHoughThreshold()
|
inline |
Gets the minimum number of votes in the Hough space needed to infer the presence of a model instance into the scene cloud.
- Returns
- the threshold for the Hough space voting.
Definition at line 285 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_threshold_.
getInputRf()
|
inline |
Getter for the input dataset's reference frames.
Each point in the reference frame cloud should be the reference frame of the correspondent point in the input dataset.
- Returns
- the pointer to the input cloud's reference frames.
Definition at line 212 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::input_rf_.
getLocalRfNormalsSearchRadius()
|
inline |
If the Local reference frame has not been set for either the model cloud or the scene cloud, this algorithm makes the computation itself but needs a suitable search radius to compute the normals in order to subsequently compute the RF (if not set a default 15 nearest neighbors search is performed).
- Returns
- the normals search radius for the local reference frame calculation.
Definition at line 376 of file hough_3d.h.
getLocalRfSearchRadius()
|
inline |
If the Local reference frame has not been set for either the model cloud or the scene cloud, this algorithm makes the computation itself but needs a suitable search radius to do so.
- Attention
- This parameter NEEDS to be set if the reference frames are not precomputed externally, otherwise the recognition results won't be correct.
- Returns
- the search radius for the local reference frame calculation.
Definition at line 404 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::local_rf_search_radius_.
getSceneRf()
|
inline |
Getter for the scene dataset's reference frames.
Each point in the reference frame cloud should be the reference frame of the correspondent point in the scene dataset.
- Returns
- the pointer to the scene cloud's reference frames.
Definition at line 249 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::scene_rf_.
getUseDistanceWeight()
|
inline |
Gets whether the vote casting procedure uses the correspondence's distance as a score.
- Returns
- if the algorithm should use the weighted distance when calculating the Hough voting score.
Definition at line 350 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::use_distance_weight_.
getUseInterpolation()
|
inline |
Gets whether the vote casting procedure interpolates the score between neighboring bins of the Hough space or not.
- Returns
- if the algorithm should interpolate the vote score between neighboring bins.
Definition at line 329 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::use_interpolation_.
houghVoting()
|
protected |
The Hough space voting procedure.
- Returns
- true if the voting had been successful or false if errors have occurred.
Definition at line 138 of file hough_3d.hpp.
References pcl::geometry::distance().
recognize() [1/2]
| bool pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::recognize | ( | std::vector< Eigen::Matrix4f, Eigen::aligned_allocator< Eigen::Matrix4f > > & | transformations | ) |
The main function, recognizes instances of the model into the scene set by the user.
- Parameters
-
[out] transformations a vector containing one transformation matrix for each instance of the model recognized into the scene.
- Returns
- true if the recognition had been successful or false if errors have occurred.
Definition at line 334 of file hough_3d.hpp.
recognize() [2/2]
| bool pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::recognize | ( | std::vector< Eigen::Matrix4f, Eigen::aligned_allocator< Eigen::Matrix4f > > & | transformations, |
| std::vector< pcl::Correspondences > & | clustered_corrs | ||
| ) |
The main function, recognizes instances of the model into the scene set by the user.
- Parameters
-
[out] transformations a vector containing one transformation matrix for each instance of the model recognized into the scene. [out] clustered_corrs a vector containing the correspondences for each instance of the model found within the input data (the same output of clusterCorrespondences).
- Returns
- true if the recognition had been successful or false if errors have occurred.
Definition at line 343 of file hough_3d.hpp.
setHoughBinSize()
|
inline |
Sets the size of each bin into the Hough space.
- Parameters
-
[in] bin_size the size of each Hough space's bin.
Definition at line 295 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_bin_size_, and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_space_initialized_.
setHoughThreshold()
|
inline |
Sets the minimum number of votes in the Hough space needed to infer the presence of a model instance into the scene cloud.
- Parameters
-
[in] threshold the threshold for the Hough space voting, if set between -1 and 0 the maximum vote in the entire space is automatically calculated and -threshold the maximum value is used as a threshold. This means that a value between -1 and 0 should be used only if at least one instance of the model is always present in the scene, or if this false positive can be filtered later.
Definition at line 275 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_threshold_.
setInputCloud()
|
inlineoverridevirtual |
Provide a pointer to the input dataset.
- Parameters
-
[in] cloud the const boost shared pointer to a PointCloud message.
Reimplemented from pcl::PCLBase< PointModelT >.
Definition at line 183 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_space_initialized_, pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::input_rf_, pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::needs_training_, and pcl::PCLBase< PointT >::setInputCloud().
setInputRf()
|
inline |
Provide a pointer to the input dataset's reference frames.
Each point in the reference frame cloud should be the reference frame of the correspondent point in the input dataset.
- Parameters
-
[in] input_rf the pointer to the input cloud's reference frames.
Definition at line 198 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_space_initialized_, pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::input_rf_, and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::needs_training_.
setLocalRfNormalsSearchRadius()
|
inline |
If the Local reference frame has not been set for either the model cloud or the scene cloud, this algorithm makes the computation itself but needs a suitable search radius to compute the normals in order to subsequently compute the RF (if not set a default 15 nearest neighbors search is performed).
- Parameters
-
[in] local_rf_normals_search_radius the normals search radius for the local reference frame calculation.
Definition at line 362 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_space_initialized_, pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::local_rf_normals_search_radius_, and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::needs_training_.
setLocalRfSearchRadius()
|
inline |
If the Local reference frame has not been set for either the model cloud or the scene cloud, this algorithm makes the computation itself but needs a suitable search radius to do so.
- Attention
- This parameter NEEDS to be set if the reference frames are not precomputed externally, otherwise the recognition results won't be correct.
- Parameters
-
[in] local_rf_search_radius the search radius for the local reference frame calculation.
Definition at line 389 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_space_initialized_, pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::local_rf_search_radius_, and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::needs_training_.
setModelSceneCorrespondences()
|
inlineoverridevirtual |
Provide a pointer to the precomputed correspondences between points in the input dataset and points in the scene dataset.
The correspondences are going to be clustered into different model instances by the algorithm.
- Parameters
-
[in] corrs the correspondences between the model and the scene.
Reimplemented from pcl::CorrespondenceGrouping< PointModelT, PointSceneT >.
Definition at line 261 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_space_initialized_, and pcl::CorrespondenceGrouping< PointModelT, PointSceneT >::model_scene_corrs_.
setSceneCloud()
|
inlineoverridevirtual |
Provide a pointer to the scene dataset (i.e.
the cloud in which the algorithm has to search for instances of the input model)
- Parameters
-
[in] scene the const boost shared pointer to a PointCloud message.
Reimplemented from pcl::CorrespondenceGrouping< PointModelT, PointSceneT >.
Definition at line 222 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_space_initialized_, pcl::CorrespondenceGrouping< PointModelT, PointSceneT >::scene_, and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::scene_rf_.
setSceneRf()
|
inline |
Provide a pointer to the scene dataset's reference frames.
Each point in the reference frame cloud should be the reference frame of the correspondent point in the scene dataset.
- Parameters
-
[in] scene_rf the pointer to the scene cloud's reference frames.
Definition at line 236 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_space_initialized_, and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::scene_rf_.
setUseDistanceWeight()
|
inline |
Sets whether the vote casting procedure uses the correspondence's distance as a score.
- Parameters
-
[in] use_distance_weight the algorithm should use the weighted distance when calculating the Hough voting score.
Definition at line 339 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_space_initialized_, and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::use_distance_weight_.
setUseInterpolation()
|
inline |
Sets whether the vote casting procedure interpolates the score between neighboring bins of the Hough space or not.
- Parameters
-
[in] use_interpolation the algorithm should interpolate the vote score between neighboring bins.
Definition at line 317 of file hough_3d.h.
References pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::hough_space_initialized_, and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::use_interpolation_.
train()
| bool pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::train |
Call this function after setting the input, the input_rf and the hough_bin_size parameters to perform an off line training of the algorithm.
This might be useful if one wants to perform once and for all a pre-computation of votes that only concern the models, increasing the on-line efficiency of the grouping algorithm. The algorithm is automatically trained on the first invocation of the recognize method or the cluster method if this training function has not been manually invoked.
- Returns
- true if the training had been successful or false if errors have occurred.
Definition at line 85 of file hough_3d.hpp.
Member Data Documentation
found_transformations_
|
protected |
Transformations found by clusterCorrespondences method.
Definition at line 475 of file hough_3d.h.
hough_bin_size_
|
protected |
The size of each bin of the hough space.
Definition at line 457 of file hough_3d.h.
Referenced by pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::getHoughBinSize(), and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setHoughBinSize().
hough_space_
|
protected |
The Hough space.
Definition at line 472 of file hough_3d.h.
hough_space_initialized_
|
protected |
Whether the Hough space already contains the correct votes for the current input parameters and so the cluster and recognize calls don't need to recompute each value.
Reset on the change of any parameter except the hough_threshold.
Definition at line 480 of file hough_3d.h.
Referenced by pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setHoughBinSize(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setInputCloud(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setInputRf(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setLocalRfNormalsSearchRadius(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setLocalRfSearchRadius(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setModelSceneCorrespondences(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setSceneCloud(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setSceneRf(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setUseDistanceWeight(), and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setUseInterpolation().
hough_threshold_
|
protected |
The minimum number of votes in the Hough space needed to infer the presence of a model instance into the scene cloud.
Definition at line 454 of file hough_3d.h.
Referenced by pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::getHoughThreshold(), and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setHoughThreshold().
input_rf_
|
protected |
The input Rf cloud.
Definition at line 442 of file hough_3d.h.
Referenced by pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::getInputRf(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setInputCloud(), and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setInputRf().
local_rf_normals_search_radius_
|
protected |
Normals search radius for the potential Rf calculation.
Definition at line 466 of file hough_3d.h.
Referenced by pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::getLocalRfNormalsSearchRadius(), and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setLocalRfNormalsSearchRadius().
local_rf_search_radius_
|
protected |
Search radius for the potential Rf calculation.
Definition at line 469 of file hough_3d.h.
Referenced by pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::getLocalRfSearchRadius(), and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setLocalRfSearchRadius().
model_votes_
|
protected |
The result of the training.
The vector between each model point and the centroid of the model adjusted by its local reference frame.
Definition at line 451 of file hough_3d.h.
needs_training_
|
protected |
If the training of the Hough space is needed; set on change of either the input cloud or the input_rf.
Definition at line 448 of file hough_3d.h.
Referenced by pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setInputCloud(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setInputRf(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setLocalRfNormalsSearchRadius(), and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setLocalRfSearchRadius().
scene_rf_
|
protected |
The scene Rf cloud.
Definition at line 445 of file hough_3d.h.
Referenced by pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::getSceneRf(), pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setSceneCloud(), and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setSceneRf().
use_distance_weight_
|
protected |
Use the weighted correspondence distance when casting votes.
Definition at line 463 of file hough_3d.h.
Referenced by pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::getUseDistanceWeight(), and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setUseDistanceWeight().
use_interpolation_
|
protected |
Use the interpolation between neighboring Hough bins when casting votes.
Definition at line 460 of file hough_3d.h.
Referenced by pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::getUseInterpolation(), and pcl::Hough3DGrouping< PointModelT, PointSceneT, PointModelRfT, PointSceneRfT >::setUseInterpolation().
The documentation for this class was generated from the following files:
- pcl/recognition/cg/hough_3d.h
- pcl/recognition/impl/cg/hough_3d.hpp
© 2009–2012, Willow Garage, Inc.
© 2012–, Open Perception, Inc.
Licensed under the BSD License.
https://pointclouds.org/documentation/classpcl_1_1_hough3_d_grouping.html
