icl::math::RansacFitter< DataPoint, Model > Class Template Reference

Generic RANSAC (RAndom SAmpling Consensus) Implementation. More...

#include <RansacFitter.h>

List of all members.

Classes
struct	Result
	result structure More...
Public Types
typedef std::vector< DataPoint >	DataSet
	DataSet type (just a set of DataPoint instances)
typedef utils::Function< Model, const DataSet & >	ModelFitting
	Function for the fitting module (gets a dataset and returns the fitted model)
typedef utils::Function < icl64f, const Model &, const DataPoint & >	PointError
	Error function for single points.
Public Member Functions
	RansacFitter ()
	empty constructor (creates a dummy instance)
	RansacFitter (int minPointsForModel, int iterations, ModelFitting fitting, PointError err, icl64f maxModelDistance, int minClosePointsForGoodModel, icl64f minErrorExit=0)
	constructor with given parameters
const Result &	fit (const DataSet &allPoints)
	fitting function (actual RANSAC algorithm)
Private Member Functions
void	find_random_consensus_set (DataSet &currConsensusSet, const DataSet &allPoints, std::vector< int > &usedIndices)
	internal utility method
Static Private Member Functions
static bool	find_in (const std::vector< int > &v, int i, int n)
	internal utility method
Private Attributes
int	m_minPointsForModel
	minimum points that are used to create a coarse model
int	m_iterations
	number of iterations
icl64f	m_maxModelDistance
	maximum distance of a point to the model to become an inlier
int	m_minClosePointsForGoodModel
	minimum amount of inliers for a 'good' model
ModelFitting	m_fitting
	fitting function
PointError	m_err
	point-model error function
icl64f	m_minErrorExit
	min error criterion for early exit
Result	m_result
	internal result buffer

---

Detailed Description

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>
class icl::math::RansacFitter< DataPoint, Model >

Generic RANSAC (RAndom SAmpling Consensus) Implementation.

The RansacFitter provides a generic framework, for RANSAC based model fitting.

Algorithms

The RANSAC Algorithm is well described on Wikipedia

See also:

http://de.wikipedia.org/wiki/RANSAC-Algorithmus

Example

An example is given in the ICL Manual

Template Parameters

The two tempalte parameters are kept very general. Therefore, there are just a few restrictions for the DataPoint and Model classes.

• default constructable
• copyable

---

Member Typedef Documentation

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

typedef std::vector<DataPoint> icl::math::RansacFitter< DataPoint, Model >::DataSet

DataSet type (just a set of DataPoint instances)

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

typedef utils::Function<Model,const DataSet&> icl::math::RansacFitter< DataPoint, Model >::ModelFitting

Function for the fitting module (gets a dataset and returns the fitted model)

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

typedef utils::Function<icl64f,const Model&,const DataPoint&> icl::math::RansacFitter< DataPoint, Model >::PointError

Error function for single points.

---

Constructor & Destructor Documentation

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

icl::math::RansacFitter< DataPoint, Model >::RansacFitter

(

)

[inline]

empty constructor (creates a dummy instance)

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

icl::math::RansacFitter< DataPoint, Model >::RansacFitter	(	int	minPointsForModel,
		int	iterations,
		ModelFitting	fitting,
		PointError	err,
		icl64f	maxModelDistance,
		int	minClosePointsForGoodModel,
		icl64f	minErrorExit = 0
	)		[inline]

constructor with given parameters

The parameters functionality is documented with the analogously named member variables

---

Member Function Documentation

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

static bool icl::math::RansacFitter< DataPoint, Model >::find_in	(	const std::vector< int > &	v,
		int	i,
		int	n
	)		[inline, static, private]

internal utility method

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

void icl::math::RansacFitter< DataPoint, Model >::find_random_consensus_set	(	DataSet &	currConsensusSet,
		const DataSet &	allPoints,
		std::vector< int > &	usedIndices
	)		[inline, private]

internal utility method

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

const Result& icl::math::RansacFitter< DataPoint, Model >::fit

(

const DataSet &

allPoints

)

[inline]

fitting function (actual RANSAC algorithm)

---

Member Data Documentation

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

PointError icl::math::RansacFitter< DataPoint, Model >::m_err [private]

point-model error function

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

ModelFitting icl::math::RansacFitter< DataPoint, Model >::m_fitting [private]

fitting function

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

int icl::math::RansacFitter< DataPoint, Model >::m_iterations [private]

number of iterations

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

icl64f icl::math::RansacFitter< DataPoint, Model >::m_maxModelDistance [private]

maximum distance of a point to the model to become an inlier

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

int icl::math::RansacFitter< DataPoint, Model >::m_minClosePointsForGoodModel [private]

minimum amount of inliers for a 'good' model

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

icl64f icl::math::RansacFitter< DataPoint, Model >::m_minErrorExit [private]

min error criterion for early exit

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

int icl::math::RansacFitter< DataPoint, Model >::m_minPointsForModel [private]

minimum points that are used to create a coarse model

template<class DataPoint = std::vector<float>, class Model = std::vector<float>>

Result icl::math::RansacFitter< DataPoint, Model >::m_result [private]

internal result buffer

---

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

• /home/jenkins/workspace/icl-manual-trunk/icl-manual/ICLMath/src/ICLMath/RansacFitter.h