aliceVision::robustEstimation::PointFittingKernel< SolverT_, ErrorT_, ModelT_ > Class Template Reference

This is one example (targeted at solvers that operate on correspondences between two views) that shows the "kernel" part of a robust fitting problem: More...

#include <PointFittingKernel.hpp>

Inheritance diagram for aliceVision::robustEstimation::PointFittingKernel< SolverT_, ErrorT_, ModelT_ >:

Public Types
using	SolverT = SolverT_
using	ErrorT = ErrorT_
using	ModelT = ModelT_

Public Member Functions
	PointFittingKernel (const Mat &x1, const Mat &x2)
std::size_t	getMinimumNbRequiredSamples () const
	Return the minimum number of required samples. More...
std::size_t	getMaximumNbModels () const
	Return the maximum number of models. More...
virtual void	fit (const std::vector< std::size_t > &samples, std::vector< ModelT > &models) const
	Extract required sample and fit model(s) to the sample. More...
virtual double	error (std::size_t sample, const ModelT &model) const
	Return the error associated to the model and a sample point. More...
virtual void	errors (const ModelT &model, std::vector< double > &errors) const
	Return the errors associated to the model and each sample point. More...
std::size_t	nbSamples () const
	get the number of putative points More...

Protected Attributes
const Mat &	_x1
	left corresponding data
const Mat &	_x2
	right corresponding data
const SolverT	_kernelSolver {}
	two view solver
const ErrorT	_errorEstimator {}
	solver error estimation

Detailed Description

template<typename SolverT_, typename ErrorT_, typename ModelT_ = Mat3Model>
class aliceVision::robustEstimation::PointFittingKernel< SolverT_, ErrorT_, ModelT_ >

This is one example (targeted at solvers that operate on correspondences between two views) that shows the "kernel" part of a robust fitting problem:

1. The model; Mat3Model in the case of the F or H matrix.
2. The minimum number of samples needed to fit; 7 or 8 (or 4).
3. A way to convert samples to a model.
4. A way to convert a sample and a model to an error.

Of particular note is that the kernel does not expose what the samples are. All the robust fitting algorithm sees is that there is some number of samples; it is able to fit subsets of them (via the kernel) and check their error, but can never access the samples themselves.

The Kernel objects must follow the following concept so that the robust fitting algorithm can fit this type of relation: ModelT

1. kernel.getMaximumNbModels()
2. kernel.getMinimumNbRequiredSamples()
3. kernel.fit(std::vector<std::size_t>, std::vector<ModelT>&)
4. kernel.error(std::size_t, ModelT) -> error

The fit routine must not clear existing entries in the vector of models; it should append new solutions to the end.

Member Function Documentation

error()

template<typename SolverT_ , typename ErrorT_ , typename ModelT_ = Mat3Model>

virtual double aliceVision::robustEstimation::PointFittingKernel< SolverT_, ErrorT_, ModelT_ >::error ( std::size_t  sample, const ModelT &  model  ) const

inlinevirtual

Return the error associated to the model and a sample point.

Parameters

[in]	sample
[in]	model

Returns

error value

Reimplemented in aliceVision::robustEstimation::PointFittingRansacKernel< SolverT_, ErrorT_, robustEstimation::Mat3Model, robustEstimation::UndefinedSolver< robustEstimation::Mat3Model > >, aliceVision::robustEstimation::PointFittingRansacKernel< SolverT_, ErrorT_, robustEstimation::Mat34Model, robustEstimation::UndefinedSolver< robustEstimation::Mat34Model > >, aliceVision::robustEstimation::PointFittingRansacKernel< SolverT_, ErrorT_, ModelT_, SolverLsT_ >, and aliceVision::multiview::relativePose::EssentialKernel< SolverT, ErrorT, ModelT >.

errors()

template<typename SolverT_ , typename ErrorT_ , typename ModelT_ = Mat3Model>

virtual void aliceVision::robustEstimation::PointFittingKernel< SolverT_, ErrorT_, ModelT_ >::errors ( const ModelT &  model, std::vector< double > &  errors  ) const

inlinevirtual

Return the errors associated to the model and each sample point.

Parameters

[in]	model
[out]	errors

Reimplemented in aliceVision::robustEstimation::PointFittingRansacKernel< SolverT_, ErrorT_, robustEstimation::Mat3Model, robustEstimation::UndefinedSolver< robustEstimation::Mat3Model > >, aliceVision::robustEstimation::PointFittingRansacKernel< SolverT_, ErrorT_, robustEstimation::Mat34Model, robustEstimation::UndefinedSolver< robustEstimation::Mat34Model > >, and aliceVision::robustEstimation::PointFittingRansacKernel< SolverT_, ErrorT_, ModelT_, SolverLsT_ >.

fit()

template<typename SolverT_ , typename ErrorT_ , typename ModelT_ = Mat3Model>

virtual void aliceVision::robustEstimation::PointFittingKernel< SolverT_, ErrorT_, ModelT_ >::fit ( const std::vector< std::size_t > &  samples, std::vector< ModelT > &  models  ) const

inlinevirtual

Extract required sample and fit model(s) to the sample.

Parameters

[in]	samples
[out]	models

Reimplemented in aliceVision::robustEstimation::PointFittingRansacKernel< SolverT_, ErrorT_, ModelT_, SolverLsT_ >, and aliceVision::multiview::relativePose::EssentialKernel< SolverT, ErrorT, ModelT >.

getMaximumNbModels()

template<typename SolverT_ , typename ErrorT_ , typename ModelT_ = Mat3Model>

std::size_t aliceVision::robustEstimation::PointFittingKernel< SolverT_, ErrorT_, ModelT_ >::getMaximumNbModels ( ) const

inline

Return the maximum number of models.

Returns

maximum number of models

getMinimumNbRequiredSamples()

template<typename SolverT_ , typename ErrorT_ , typename ModelT_ = Mat3Model>

std::size_t aliceVision::robustEstimation::PointFittingKernel< SolverT_, ErrorT_, ModelT_ >::getMinimumNbRequiredSamples ( ) const

inline

Return the minimum number of required samples.

Returns

minimum number of required samples

nbSamples()

template<typename SolverT_ , typename ErrorT_ , typename ModelT_ = Mat3Model>

std::size_t aliceVision::robustEstimation::PointFittingKernel< SolverT_, ErrorT_, ModelT_ >::nbSamples ( ) const

inline

get the number of putative points

Returns

number of putative points

---

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

• /home/docs/checkouts/readthedocs.org/user_builds/alicevision/checkouts/latest/src/aliceVision/robustEstimation/PointFittingKernel.hpp