#include <old_prediction_non_normalized.h>

Inheritance diagram for pose_prediction_ism::PaperPredictionNonNormalized:

Public Member Functions
	PaperPredictionNonNormalized (std::string dbfileName)

AttributedPointCloud	predictUnfoundPoses (ISM::PosePtr &referencePose, std::string patternName, double numberOfSpecifiers)
	RecognizerPredictionISM::calcUnfoundPoses Processes the following steps for each object in the objectpattern set. More...

	~PaperPredictionNonNormalized ()

Public Member Functions inherited from pose_prediction_ism::PosePredictor
void	clearPointCloud ()

void	disableRandom ()

void	enableRandom (double sphere_radius, double max_projection_angle)

AttributedPointCloud	getAttributedPointCloud () const

std::string	getMarkerNameSpace () const

PredictorType	getPredictorType () const

void	setFoundObjects (const FoundObjects &value)

void	setPredictionGenerationFactor (double prediction_generation_factor_)

void	traverseISM (std::string pattern_name, unsigned int level)

Private Member Functions
void	addPointToPointCloudWithNumber (ISM::PosePtr poseToAdd, IsmObject object, unsigned int numberOfHypotheses)

void	calculateRecursiveUnfoundPoses (ISM::PosePtr &referencePose, std::string patternName, unsigned int weight)

void	createAttributedPointCloud (ISM::PosePtr reference_pose_ptr, double percentage_of_records_for_prediction)

bool	nonReferenceObjectProcessing (IsmObject object, ISM::VoteSpecifierPtr specifier, ISM::PosePtr &referencePose, unsigned int weight, unsigned int specifiersSize)
	nonReferenceObjectProcessing Processes a non reference Object Processes the following steps. More...

bool	referenceObjectProcessing (std::string objectType, ISM::VoteSpecifierPtr specifier, ISM::PosePtr &referencePose, unsigned int weight, unsigned int specifiersSize)
	referenceObjectProcessing Processes a reference object. Processes the following steps. More...

Additional Inherited Members
Protected Types inherited from pose_prediction_ism::PosePredictor
typedef std::pair< std::string, std::string >	IsmObject

typedef std::vector< IsmObject >	IsmObjects

typedef std::set< IsmObject >	IsmObjectSet

typedef std::map< IsmObject, unsigned int >	SizeMap

typedef boost::variate_generator< boost::mt19937, boost::uniform_real<> >	UniformDistributionGenerator

Protected Member Functions inherited from pose_prediction_ism::PosePredictor
void	addPointToPointCloud (ISM::PosePtr pose_to_add, std::string type, std::string identifier)
	RecognizerPredictionISM::addPointToPointCloud Adds an ISM Pose with object type and identifier as attributed point to the attributed point cloud if the number of hypotheses is higher than the threshold. More...

IsmObjectSet	getObjectTypesAndIdsBelongingToPattern (std::string type)

bool	isFoundObject (IsmObject object)

bool	isPointCloudEmpty ()

bool	isReferenceObject (IsmObject object)

	PosePredictor (std::string database_filename, std::string name_space, PredictorType predictor_type)

ISM::PosePtr	predictPose (ISM::PosePtr reference_pose_ptr, IsmObjects path_to_object)

	~PosePredictor ()

Protected Attributes inherited from pose_prediction_ism::PosePredictor
AttributedPointCloud	attributed_point_cloud_

double	average_votes_ = 0

double	prediction_generation_factor_ = 1

SizeMap	specifiers_size_map_

UniformDistributionGenerator *	udg_dist_

UniformDistributionGenerator *	udg_rot_

ISM::PatternToObjectToVoteMap	votes_

Detailed Description

Definition at line 22 of file old_prediction_non_normalized.h.

Constructor & Destructor Documentation

pose_prediction_ism::PaperPredictionNonNormalized::PaperPredictionNonNormalized ( std::string dbfileName )

Definition at line 22 of file old_prediction_non_normalized.cpp.

pose_prediction_ism::PaperPredictionNonNormalized::~PaperPredictionNonNormalized ( )

Definition at line 26 of file old_prediction_non_normalized.cpp.

Member Function Documentation

void pose_prediction_ism::PaperPredictionNonNormalized::addPointToPointCloudWithNumber	(	ISM::PosePtr	poseToAdd,
		IsmObject	object,
		unsigned int	numberOfHypotheses
	)

private

Definition at line 146 of file old_prediction_non_normalized.cpp.

void pose_prediction_ism::PaperPredictionNonNormalized::calculateRecursiveUnfoundPoses	(	ISM::PosePtr &	referencePose,
		std::string	patternName,
		unsigned int	weight
	)

private

Definition at line 43 of file old_prediction_non_normalized.cpp.

void pose_prediction_ism::PaperPredictionNonNormalized::createAttributedPointCloud	(	ISM::PosePtr	reference_pose_ptr,
		double	percentage_of_records_for_prediction
	)

privatevirtual

Implements pose_prediction_ism::PosePredictor.

Definition at line 142 of file old_prediction_non_normalized.cpp.

bool pose_prediction_ism::PaperPredictionNonNormalized::nonReferenceObjectProcessing	(	IsmObject	object,
		ISM::VoteSpecifierPtr	specifier,
		ISM::PosePtr &	referencePose,
		unsigned int	weight,
		unsigned int	specifiersSize
	)

private

nonReferenceObjectProcessing Processes a non reference Object Processes the following steps.

Calculate absolute pose of unknown object.
Calculate full pose for next call of this function.
Check whether this non-reference object is identical to the reference of the ism
in which it is object. If this non-reference object is equal to the reference of the ism in which it resides.
- Every single vote for a hypotheses in specifier has to be repreated for all
  redundancies in lower parts of the tree.
- As all votes are identical, just skip all further votes and insert points
  for all of them at once.
- Stop going through all further votes.
If this non-reference object is not equal to the reference of the ism in which it resides.
- Insert absolute unkown object pose into tesselated sphere quad just for one vote,
  but considering all redundancies of the isms in lover parts of the tree

Parameters

object
specifier
referencePose
weight
specifiersSize

Definition at line 90 of file old_prediction_non_normalized.cpp.

AttributedPointCloud pose_prediction_ism::PaperPredictionNonNormalized::predictUnfoundPoses	(	ISM::PosePtr &	referencePose,
		std::string	patternName,
		double	numberOfSpecifiers
	)

virtual

RecognizerPredictionISM::calcUnfoundPoses Processes the following steps for each object in the objectpattern set.

Look whether we have an unfound object (reference object can never be
found as they do not exist). Else ignore vote, as we already know itsattributedPoint.pose.
If the object was not found
- Get all votes that fit to combination of this reference
  (to reference and non-reference objects) and unfound object.
- If we have a reference object
  - search under it for unknown object.
- If we have a non-reference object
  - we have not found and save its position.
- Save that we found this object as missing in our currently considered scene hypotheses.

Parameters

referencePose
patternName
weight
depth

Implements pose_prediction_ism::PosePredictor.

Definition at line 31 of file old_prediction_non_normalized.cpp.

bool pose_prediction_ism::PaperPredictionNonNormalized::referenceObjectProcessing	(	std::string	objectType,
		ISM::VoteSpecifierPtr	specifier,
		ISM::PosePtr &	referencePose,
		unsigned int	weight,
		unsigned int	specifiersSize
	)

private

referenceObjectProcessing Processes a reference object. Processes the following steps.

Calculate absolute pose of unknown object.
Calculate full pose for next call of this function.
Check whether this reference object is identical to
the reference of the ism in which it is object. If this reference object is equal to the reference of the ism in which it resides.
- Every single vote for a hypotheses in specifier has to be repreated
  for all redundancies in lower parts of the tree.
- One level higher in the scene model hierarchy, object pose hypotheses
  should be replicated for every vote in specifiers as well as every time
  this function would have been called. Reference object in this ism gets
  reference of next higher ism.
- If position of this object in scene is equal to pose of reference, then
  it has been chosen as reference and its pose will be equal to the reference
  through all votes. So we do not need to process any other votes here.
If this reference object is not equal to the reference of the ism in which it resides.
- One level higher in the scene model hierarchy, object pose hypotheses
  should be replicated for every time this function would have been called.

Parameters

objectType	The object type
specifier	The vote specifier
referencePose	The reference pose of this function call
weight	The weight of this call
specifiersSize	The size of the specifier container.

Returns: Returns true, if we do not have to process further

Definition at line 112 of file old_prediction_non_normalized.cpp.

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

Public Member Functions

Private Member Functions

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation