Class to maintain data and optimization model for 2d homography estimation. More...

#include <rrel_homography2d_est.h>

Inheritance diagram for rrel_homography2d_est:

Public Types
enum	scale_t { NONE = 0, SINGLE = 1, MULTIPLE = 2 }
	Type of scale information this problem provides. More...
Public Member Functions
	rrel_homography2d_est (const vcl_vector< vgl_homg_point_2d< double > > &from_pts, const vcl_vector< vgl_homg_point_2d< double > > &to_pts, unsigned int homog_dof=8)
	Constructor from vgl_homg_point_2d's.
	rrel_homography2d_est (const vcl_vector< vnl_vector< double > > &from_pts, const vcl_vector< vnl_vector< double > > &to_pts, unsigned int homog_dof=8)
	Constructor from vnl_vectors.
virtual	~rrel_homography2d_est ()
	Destructor.
unsigned int	num_samples () const
	Total number of correspondences.
unsigned int	residual_dof () const
	The degrees of freedom in the residual.
bool	fit_from_minimal_set (const vcl_vector< int > &point_indices, vnl_vector< double > &params) const
	Generate a parameter estimate from a minimal sample.
void	compute_residuals (const vnl_vector< double > &params, vcl_vector< double > &residuals) const
	Compute unsigned fit residuals relative to the parameter estimate.
bool	weighted_least_squares_fit (vnl_vector< double > &params, vnl_matrix< double > &norm_covar, const vcl_vector< double > *weights=0) const
	Weighted least squares parameter estimate. The normalized covariance is not yet filled in.
virtual void	homog_to_params (const vnl_matrix< double > &m, vnl_vector< double > &p) const
	Convert a homography to a linear parameter list (for estimation).
virtual void	params_to_homog (const vnl_vector< double > &p, vnl_matrix< double > &m) const
	Convert a linear parameter list (from estimation) to a homography.
void	print_points () const
	Print information as a test utility.
unsigned int	param_dof () const
	The degrees of freedom in the parameter set.
unsigned int	num_samples_to_instantiate () const
	Minimum number of samples required to uniquely instantiate a fit.
virtual unsigned int	num_unique_samples () const
	The number of "unique" samples.
virtual void	compute_weights (const vcl_vector< double > &residuals, const rrel_wls_obj *obj, double scale, vcl_vector< double > &weights) const
	Compute the weights for the given residuals.
virtual scale_t	scale_type () const
	Type of scale information the problem provides.
virtual const vcl_vector < double > &	prior_multiple_scales () const
	The prior scale vector, if available.
virtual double	prior_scale () const
	The prior scale, if available.
virtual void	set_prior_multiple_scales (const vcl_vector< double > &scales)
	Sets the scales for heteroscedastic data.
virtual void	set_prior_scale (double scale)
	Sets the scale for homoscedastic data.
virtual void	set_no_prior_scale ()
	Removes the scale information for the problem.
void	set_similarity_weights (const vcl_vector< double > &wgts)
	Set similarity weights.
const vcl_vector< double > &	similarity_weights () const
	Get similarity weights.
Protected Member Functions
void	normalize (const vcl_vector< vnl_vector< double > > &pts, const vcl_vector< double > &wgts, vcl_vector< vnl_vector< double > > &norm_pts, vnl_matrix< double > &norm_matrix) const
void	set_param_dof (unsigned int dof)
	Set the degrees of freedom.
void	set_num_samples_for_fit (unsigned int num_samp)
	Set the number of samples needed for a unique fit.
void	set_scale_type (scale_t t)
	Set the type of prior scale.
Protected Attributes
vcl_vector< vnl_vector< double > >	from_pts_
vcl_vector< vnl_vector< double > >	to_pts_
unsigned int	homog_dof_
unsigned int	min_num_pts_

Detailed Description

Class to maintain data and optimization model for 2d homography estimation.

This class assumes each point has a unique correspondence, even though it may be incorrect. This is the usual assumption used in 2d homography estimation. It probably isn't the best thing to do in practice, though, because correspondences are hard to find without knowing the transformation and robust estimation can pick out the correct correspondences even when they aren't unique.

The corresponding data points are provided as a vectors of vgl_homg_point_2d. Corresponding points are assumed to share the same index in the two vectors.

Several aspects of this class aren't quite up with the "best" techniques in the literature, although the practical significance of this is known to be quite limited. First, the symmetric transfer error is used in computing residuals. Second, the weighted least-squares fit is just a robust version of Hartley's normalized 8-point algorithm. More sophisticated versions could be developed, but this class was written mostly for demonstration purposes.

Definition at line 38 of file rrel_homography2d_est.h.

Member Enumeration Documentation

enum rrel_estimation_problem::scale_t [inherited]

Type of scale information this problem provides.

NONE: problem does not provide any scale information. SINGLE: problem provides a single prior scale (homoscedastic data). MULTIPLE: problem provides a scale for each residual (heteroscedastic data).

Enumerator:

NONE
SINGLE
MULTIPLE

Definition at line 40 of file rrel_estimation_problem.h.

Constructor & Destructor Documentation

rrel_homography2d_est::rrel_homography2d_est	(	const vcl_vector< vgl_homg_point_2d< double > > &	from_pts,
		const vcl_vector< vgl_homg_point_2d< double > > &	to_pts,
		unsigned int	homog_dof = `8`
	)

Constructor from vgl_homg_point_2d's.

By default, we want a full 8-DOF homography

Definition at line 13 of file rrel_homography2d_est.cxx.

rrel_homography2d_est::rrel_homography2d_est	(	const vcl_vector< vnl_vector< double > > &	from_pts,
		const vcl_vector< vnl_vector< double > > &	to_pts,
		unsigned int	homog_dof = `8`
	)

Constructor from vnl_vectors.

By default, we want a full 8-DOF homography

Definition at line 40 of file rrel_homography2d_est.cxx.

rrel_homography2d_est::~rrel_homography2d_est ( ) [virtual]

Destructor.

Definition at line 57 of file rrel_homography2d_est.cxx.

Member Function Documentation

void rrel_homography2d_est::compute_residuals	(	const vnl_vector< double > &	params,
		vcl_vector< double > &	residuals
	)		const `[virtual]`

Compute unsigned fit residuals relative to the parameter estimate.

Implements rrel_estimation_problem.

Definition at line 102 of file rrel_homography2d_est.cxx.

void rrel_estimation_problem::compute_weights	(	const vcl_vector< double > &	residuals,
		const rrel_wls_obj *	obj,
		double	scale,
		vcl_vector< double > &	weights
	)		const `[virtual, inherited]`

Compute the weights for the given residuals.

The residuals are essentially those returned by compute_residuals(). The default behaviour is to apply obj->wgt() to each residual. Some problems, however, many need to augment the weights. Such problems should override this function (but may want to call this to compute the "basic" weights).

Definition at line 33 of file rrel_estimation_problem.cxx.

bool rrel_homography2d_est::fit_from_minimal_set	(	const vcl_vector< int > &	point_indices,
		vnl_vector< double > &	params
	)		const `[virtual]`

Generate a parameter estimate from a minimal sample.

Implements rrel_estimation_problem.

Definition at line 70 of file rrel_homography2d_est.cxx.

void rrel_homography2d_est::homog_to_params	(	const vnl_matrix< double > &	m,
		vnl_vector< double > &	p
	)		const `[virtual]`

Convert a homography to a linear parameter list (for estimation).

Overloaded for specialized reduced-DOF homographies (i.e. affine)

Definition at line 203 of file rrel_homography2d_est.cxx.

void rrel_homography2d_est::normalize	(	const vcl_vector< vnl_vector< double > > &	pts,
		const vcl_vector< double > &	wgts,
		vcl_vector< vnl_vector< double > > &	norm_pts,
		vnl_matrix< double > &	norm_matrix
	)		const `[protected]`

Definition at line 221 of file rrel_homography2d_est.cxx.

unsigned int rrel_homography2d_est::num_samples ( ) const [virtual]

Total number of correspondences.

Implements rrel_estimation_problem.

Definition at line 63 of file rrel_homography2d_est.cxx.

unsigned int rrel_estimation_problem::num_samples_to_instantiate ( ) const [inline, inherited]

Minimum number of samples required to uniquely instantiate a fit.

dof()/num_samples_to_instantiate() gives the number of constraints per sample.

Definition at line 61 of file rrel_estimation_problem.h.

virtual unsigned int rrel_estimation_problem::num_unique_samples ( ) const [inline, virtual, inherited]

The number of "unique" samples.

In most problems, this equals num_samples. With estimation problems involving non-unique correspondences, however, the total number of possible correspondences (= num_samples()) generally much greater than the number of "unique samples" (which would be the number of points from one image).

This is used by random sampling techniques to determine the number of samples required to get a given probability of success.

Definition at line 84 of file rrel_estimation_problem.h.

unsigned int rrel_estimation_problem::param_dof ( ) const [inline, inherited]

The degrees of freedom in the parameter set.

Definition at line 57 of file rrel_estimation_problem.h.

void rrel_homography2d_est::params_to_homog	(	const vnl_vector< double > &	p,
		vnl_matrix< double > &	m
	)		const `[virtual]`

Convert a linear parameter list (from estimation) to a homography.

Overloaded for specialized reduced-DOF homographies (i.e. affine)

Definition at line 212 of file rrel_homography2d_est.cxx.

void rrel_homography2d_est::print_points ( ) const

Print information as a test utility.

Definition at line 260 of file rrel_homography2d_est.cxx.

const vcl_vector< double > & rrel_estimation_problem::prior_multiple_scales ( ) const [virtual, inherited]

The prior scale vector, if available.

The call is valid only if scale_type() == MULTIPLE.

Definition at line 59 of file rrel_estimation_problem.cxx.

double rrel_estimation_problem::prior_scale ( ) const [virtual, inherited]

The prior scale, if available.

The call is valid only if scale_type() == SINGLE.

Definition at line 66 of file rrel_estimation_problem.cxx.

unsigned int rrel_homography2d_est::residual_dof ( ) const [inline, virtual]

The degrees of freedom in the residual.

Each coordinate of the correspondence pair has Gaussian error, so the Euclidean distance residual has 4 degrees of freedom.

Reimplemented from rrel_estimation_problem.

Definition at line 63 of file rrel_homography2d_est.h.

virtual scale_t rrel_estimation_problem::scale_type ( ) const [inline, virtual, inherited]

Type of scale information the problem provides.

Definition at line 116 of file rrel_estimation_problem.h.

void rrel_estimation_problem::set_no_prior_scale ( ) [virtual, inherited]

Removes the scale information for the problem.

Side effect: set scale_type() = NONE.

Definition at line 90 of file rrel_estimation_problem.cxx.

void rrel_estimation_problem::set_num_samples_for_fit ( unsigned int num_samp ) [inline, protected, inherited]

Set the number of samples needed for a unique fit.

Definition at line 167 of file rrel_estimation_problem.h.

void rrel_estimation_problem::set_param_dof ( unsigned int dof ) [inline, protected, inherited]

Set the degrees of freedom.

Definition at line 164 of file rrel_estimation_problem.h.

void rrel_estimation_problem::set_prior_multiple_scales ( const vcl_vector< double > & scales ) [virtual, inherited]

Sets the scales for heteroscedastic data.

Side effect: set scale_type() = MULTIPLE.

Definition at line 73 of file rrel_estimation_problem.cxx.

void rrel_estimation_problem::set_prior_scale ( double scale ) [virtual, inherited]

Sets the scale for homoscedastic data.

Side effect: set scale_type() = SINGLE.

Definition at line 83 of file rrel_estimation_problem.cxx.

void rrel_estimation_problem::set_scale_type ( scale_t t ) [inline, protected, inherited]

Set the type of prior scale.

Definition at line 170 of file rrel_estimation_problem.h.

void rrel_estimation_problem::set_similarity_weights ( const vcl_vector< double > & wgts ) [inline, inherited]

Set similarity weights.

Currently it is only used in wgted random sampling search

Definition at line 140 of file rrel_estimation_problem.h.

const vcl_vector<double>& rrel_estimation_problem::similarity_weights ( ) const [inline, inherited]

Get similarity weights.

Definition at line 144 of file rrel_estimation_problem.h.

bool rrel_homography2d_est::weighted_least_squares_fit	(	vnl_vector< double > &	params,
		vnl_matrix< double > &	norm_covar,
		const vcl_vector< double > *	weights = `0`
	)		const `[virtual]`