icl::io::IntrinsicCalibrator Class Reference

#include <IntrinsicCalibrator.h>

Inheritance diagram for icl::io::IntrinsicCalibrator:

Classes
struct	CalibrationData
struct	Result
	Simple struct for handle result of computation of intrinsics. More...

Public Types
typedef math::FixedMatrix< double, 1, 2 >	Pos2D
typedef utils::Array2D< Pos2D >	DetectedGrid

Public Member Functions
	IntrinsicCalibrator (unsigned int boardWidth=6, unsigned int boardHeight=9, unsigned int boardCount=8, unsigned int imageWidth=640, unsigned int imageHeight=480)
	Constructs a new object of this class for computation of intrinc parameters. More...
	~IntrinsicCalibrator ()
	Destructor. More...
Result	calibrate (const math::DynMatrix< icl64f > &impoints, const math::DynMatrix< icl64f > &worldpoints)
	calibrates the camera More...
void	saveIntrinsics (const std::string &filename)
	saves computed/current intrinsics to xml-file More...
void	loadIntrinsics (const std::string &filename)
	loads intrincs from xml-file More...
void	resetData (unsigned int boardWidth, unsigned int boardHeight, unsigned int boardCount, unsigned int imageWidth, unsigned int imageHeight)
	resets data to compute new intrinsics with More...
Result	getResult () const
	return the result of computation More...

Static Public Member Functions
static Result	optimize (const CalibrationData &data)
	computes the calibration for given CalibrationData More...

Private Member Functions
void	rigid_motion (const math::DynMatrix< icl64f > &X, const math::DynMatrix< icl64f > &om, const math::DynMatrix< icl64f > &T, math::DynMatrix< icl64f > &Y, math::DynMatrix< icl64f > &dYdom, math::DynMatrix< icl64f > &dYdT)
	Computes the rigid motion transformation Y = R*X+T. More...
void	project_points2 (const math::DynMatrix< icl64f > &X, const math::DynMatrix< icl64f > &om, const math::DynMatrix< icl64f > &T, const math::DynMatrix< icl64f > &f, const math::DynMatrix< icl64f > &c, const math::DynMatrix< icl64f > &k, const double alpha, math::DynMatrix< icl64f > &xp, math::DynMatrix< icl64f > &dxpdom, math::DynMatrix< icl64f > &dxpdT, math::DynMatrix< icl64f > &dxpdf, math::DynMatrix< icl64f > &dxpdc, math::DynMatrix< icl64f > &dxpdk, math::DynMatrix< icl64f > &dxpdalpha)
	Projects a 3D structure onto the image plane. More...
void	comp_distortion_oulu (const math::DynMatrix< icl64f > xd, const math::DynMatrix< icl64f > k, math::DynMatrix< icl64f > &x)
	Compensates for radial and tangential distortion (Model From Oulu university) More...
void	compute_homography (const math::DynMatrix< icl64f > &m, const math::DynMatrix< icl64f > &M, math::DynMatrix< icl64f > &H)
	Computes the planar homography between the point coordinates on the plane (M) and the image point coordinates (m). More...
void	optimize (const math::DynMatrix< icl64f > &impoints, const math::DynMatrix< icl64f > &worldPoints, double *params)
	Levenberg-Marquardt sparse algorithm optimizes all parameters. More...
void	init_intrinsic_param (const math::DynMatrix< icl64f > &x, const math::DynMatrix< icl64f > &X, math::DynMatrix< icl64f > &fc, math::DynMatrix< icl64f > &cc, math::DynMatrix< icl64f > &kc, double &alpha_c)
	Initialization of the intrinsic parameters. More...
void	comp_ext_calib (const math::DynMatrix< icl64f > &x_kk, const math::DynMatrix< icl64f > &X_kk, const math::DynMatrix< icl64f > &fc, const math::DynMatrix< icl64f > &cc, const math::DynMatrix< icl64f > &kc, const double alpha_c, const double thresh_cond, math::DynMatrix< icl64f > &omckk, math::DynMatrix< icl64f > &Tckk, math::DynMatrix< icl64f > &Rckk)
	Computes the extrinsic parameters. More...
void	normalize_pixel (const math::DynMatrix< icl64f > x_kk, const math::DynMatrix< icl64f > fc, const math::DynMatrix< icl64f > cc, const math::DynMatrix< icl64f > kc, const double alpha_c, math::DynMatrix< icl64f > &xn)
	Computes the normalized coordinates xn given the pixel coordinates x_kk and the intrinsic camera parameters fc, cc and kc. More...
void	mean (const math::DynMatrix< icl64f > &x_k, math::DynMatrix< icl64f > &res)
	Computes the mean of each col of x_k. More...
void	rodrigues (const math::DynMatrix< icl64f > &in, math::DynMatrix< icl64f > &out, math::DynMatrix< icl64f > &dout)
	Computes the rodrigues transformation and derivative. More...
void	compute_extrinsic_init (const math::DynMatrix< icl64f > &x_kk, const math::DynMatrix< icl64f > &X_kk, const math::DynMatrix< icl64f > &fc, const math::DynMatrix< icl64f > &cc, const math::DynMatrix< icl64f > &kc, const double &alpha_c, math::DynMatrix< icl64f > &omckk, math::DynMatrix< icl64f > &Tckk, math::DynMatrix< icl64f > &Rckk)
	Computes extrinsic initial parameters. More...
void	compute_extrinsic_refine (const math::DynMatrix< icl64f > &omc_init, const math::DynMatrix< icl64f > &Tc_init, const math::DynMatrix< icl64f > &x_kk, const math::DynMatrix< icl64f > &X_kk, const math::DynMatrix< icl64f > &fc, const math::DynMatrix< icl64f > &cc, const math::DynMatrix< icl64f > &kc, const double alpha_c, const int MaxIter, double thresh_cond, math::DynMatrix< icl64f > &omckk, math::DynMatrix< icl64f > &Tckk, math::DynMatrix< icl64f > &Rckk, math::DynMatrix< icl64f > &JJ)
	optimizes precomputed extrinsic parameters More...

Private Attributes
Data *	m_data
Result	m_calres
	struct for handling result of computation More...

Additional Inherited Members
Protected Member Functions inherited from icl::utils::Uncopyable
	Uncopyable ()
	Empty base constructor. More...

Detailed Description

This class implements the image undistortion procedure. It is a reimplementation from the matlab toolbox Pass in data where world and image points corresponds to each other in layout or use the optimze call where world points are generated automatically and layout of image points must be rowwise.

Member Typedef Documentation

DetectedGrid

typedef utils::Array2D<Pos2D> icl::io::IntrinsicCalibrator::DetectedGrid

Pos2D

typedef math::FixedMatrix<double,1,2> icl::io::IntrinsicCalibrator::Pos2D

Constructor & Destructor Documentation

IntrinsicCalibrator()

icl::io::IntrinsicCalibrator::IntrinsicCalibrator	(	unsigned int	boardWidth = 6,
		unsigned int	boardHeight = 9,
		unsigned int	boardCount = 8,
		unsigned int	imageWidth = 640,
		unsigned int	imageHeight = 480
	)

Constructs a new object of this class for computation of intrinc parameters.

Parameters

boardWidth	the width of the chessboard
boardHeight	the height of the chessboard
boardCount	the number of board for computations
imageWidth	the imagewidth
imageHeight	the imageheight

~IntrinsicCalibrator()

icl::io::IntrinsicCalibrator::~IntrinsicCalibrator

(

)

Destructor.

Member Function Documentation

calibrate()

Result icl::io::IntrinsicCalibrator::calibrate	(	const math::DynMatrix< icl64f > &	impoints,
		const math::DynMatrix< icl64f > &	worldpoints
	)

calibrates the camera

Since for each image the worldcoordinates are/look exactly the same, just pass the worldcoordinates for the first image. This will be used for all computations.

Parameters

impoints	image coordinates
worldpoints	coordinates for the first image, sorting depends on image coordinates

comp_distortion_oulu()

void icl::io::IntrinsicCalibrator::comp_distortion_oulu ( const math::DynMatrix< icl64f >  xd, const math::DynMatrix< icl64f >  k, math::DynMatrix< icl64f > &  x  )

private

Compensates for radial and tangential distortion (Model From Oulu university)

Parameters

xd	distorted, normalized image coordinates
k	distortion coefficients
x	undistorted, normalized image coordinates

comp_ext_calib()

void icl::io::IntrinsicCalibrator::comp_ext_calib ( const math::DynMatrix< icl64f > &  x_kk, const math::DynMatrix< icl64f > &  X_kk, const math::DynMatrix< icl64f > &  fc, const math::DynMatrix< icl64f > &  cc, const math::DynMatrix< icl64f > &  kc, const double  alpha_c, const double  thresh_cond, math::DynMatrix< icl64f > &  omckk, math::DynMatrix< icl64f > &  Tckk, math::DynMatrix< icl64f > &  Rckk  )

private

Computes the extrinsic parameters.

Parameters

x_kk	image coordinates
X_kk	world coordinates
fc	focal length
cc	principal point
kc	distortion coefficients
alpha_c	the skew
thresh_cond	number of iterations for optimization
omckk	rotation vector
Tckk	translation vector
Rckk	rotation matrix

compute_extrinsic_init()

void icl::io::IntrinsicCalibrator::compute_extrinsic_init ( const math::DynMatrix< icl64f > &  x_kk, const math::DynMatrix< icl64f > &  X_kk, const math::DynMatrix< icl64f > &  fc, const math::DynMatrix< icl64f > &  cc, const math::DynMatrix< icl64f > &  kc, const double &  alpha_c, math::DynMatrix< icl64f > &  omckk, math::DynMatrix< icl64f > &  Tckk, math::DynMatrix< icl64f > &  Rckk  )

private

Computes extrinsic initial parameters.

Parameters

x_kk	image coordinates
X_kk	world coordinates
fc	focal length
cc	principal point
kc	distortion coefficients
alpha_c	skew
omckk	rotation vector
Tckk	translation vector
Rckk	rotation matrix

compute_extrinsic_refine()

void icl::io::IntrinsicCalibrator::compute_extrinsic_refine ( const math::DynMatrix< icl64f > &  omc_init, const math::DynMatrix< icl64f > &  Tc_init, const math::DynMatrix< icl64f > &  x_kk, const math::DynMatrix< icl64f > &  X_kk, const math::DynMatrix< icl64f > &  fc, const math::DynMatrix< icl64f > &  cc, const math::DynMatrix< icl64f > &  kc, const double  alpha_c, const int  MaxIter, double  thresh_cond, math::DynMatrix< icl64f > &  omckk, math::DynMatrix< icl64f > &  Tckk, math::DynMatrix< icl64f > &  Rckk, math::DynMatrix< icl64f > &  JJ  )

private

optimizes precomputed extrinsic parameters

compute_homography()

void icl::io::IntrinsicCalibrator::compute_homography ( const math::DynMatrix< icl64f > &  m, const math::DynMatrix< icl64f > &  M, math::DynMatrix< icl64f > &  H  )

private

Computes the planar homography between the point coordinates on the plane (M) and the image point coordinates (m).

Parameters

m	homogeneous image coordinates
M	homogeneous world coordinates
H	Homography matrix

getResult()

Result icl::io::IntrinsicCalibrator::getResult ( ) const

inline

return the result of computation

init_intrinsic_param()

void icl::io::IntrinsicCalibrator::init_intrinsic_param ( const math::DynMatrix< icl64f > &  x, const math::DynMatrix< icl64f > &  X, math::DynMatrix< icl64f > &  fc, math::DynMatrix< icl64f > &  cc, math::DynMatrix< icl64f > &  kc, double &  alpha_c  )

private

Initialization of the intrinsic parameters.

Parameters

x	images coordinate
X	world coordinates
fc	focal length
cc	principal point coordinates
kc	distortion coefficients
alpha_c	skew coefficient

loadIntrinsics()

void icl::io::IntrinsicCalibrator::loadIntrinsics

(

const std::string &

filename

)

loads intrincs from xml-file

Parameters

filename

name of xml-file to load

mean()

void icl::io::IntrinsicCalibrator::mean ( const math::DynMatrix< icl64f > &  x_k, math::DynMatrix< icl64f > &  res  )

private

Computes the mean of each col of x_k.

If x_k is is rowvector the mean of the rowvector is computed.

Parameters

x_k	Inputmatrix
res	resultmatrix

normalize_pixel()

void icl::io::IntrinsicCalibrator::normalize_pixel ( const math::DynMatrix< icl64f >  x_kk, const math::DynMatrix< icl64f >  fc, const math::DynMatrix< icl64f >  cc, const math::DynMatrix< icl64f >  kc, const double  alpha_c, math::DynMatrix< icl64f > &  xn  )

private

Computes the normalized coordinates xn given the pixel coordinates x_kk and the intrinsic camera parameters fc, cc and kc.

Parameters

x_kk	image coordinates
fc	focal length
cc	principal point
kc	distortion coefficients
alpha_c	skew
xn	normalized image coordinates

optimize() [1/2]

void icl::io::IntrinsicCalibrator::optimize ( const math::DynMatrix< icl64f > &  impoints, const math::DynMatrix< icl64f > &  worldPoints, double *  params  )

private

Levenberg-Marquardt sparse algorithm optimizes all parameters.

Parameters

impoints	image coordinates
worldPoints	world coordinates
params	parameters to be optimized

optimize() [2/2]

static Result icl::io::IntrinsicCalibrator::optimize ( const CalibrationData &  data )

static

computes the calibration for given CalibrationData

Parameters

data	contains detected grids for calibration

project_points2()

void icl::io::IntrinsicCalibrator::project_points2 ( const math::DynMatrix< icl64f > &  X, const math::DynMatrix< icl64f > &  om, const math::DynMatrix< icl64f > &  T, const math::DynMatrix< icl64f > &  f, const math::DynMatrix< icl64f > &  c, const math::DynMatrix< icl64f > &  k, const double  alpha, math::DynMatrix< icl64f > &  xp, math::DynMatrix< icl64f > &  dxpdom, math::DynMatrix< icl64f > &  dxpdT, math::DynMatrix< icl64f > &  dxpdf, math::DynMatrix< icl64f > &  dxpdc, math::DynMatrix< icl64f > &  dxpdk, math::DynMatrix< icl64f > &  dxpdalpha  )

private

Projects a 3D structure onto the image plane.

Parameters

X	3D world coordinates
om	rotation vector
T	translation vector
f	focal length
c	principal point
k	distortion coefficients
alpha	skew
xp	Projected pixel coordinates (2xN matrix for N points)
dxpdom	Derivative of xp with respect to om
dxpdT	Derivative of xp with respect to T
dxpdf	Derivative of xp with respect to f
dxpdc	Derivative of xp with respect to c
dxpdk	Derivative of xp with respect to k
dxpdalpha	Derivative of xp with respect to alpha

resetData()

void icl::io::IntrinsicCalibrator::resetData	(	unsigned int	boardWidth,
		unsigned int	boardHeight,
		unsigned int	boardCount,
		unsigned int	imageWidth,
		unsigned int	imageHeight
	)

resets data to compute new intrinsics with

Parameters

boardWidth	the new width of the chessboard
boardHeight	the new height of the chessboard
boardCount	the new number of board for computations
imageWidth	the new imagewidth
imageHeight	the new imageheight

rigid_motion()

void icl::io::IntrinsicCalibrator::rigid_motion ( const math::DynMatrix< icl64f > &  X, const math::DynMatrix< icl64f > &  om, const math::DynMatrix< icl64f > &  T, math::DynMatrix< icl64f > &  Y, math::DynMatrix< icl64f > &  dYdom, math::DynMatrix< icl64f > &  dYdT  )

private

Computes the rigid motion transformation Y = R*X+T.

Parameters

X	3D structure in the world coordinates
om	rotation vector
T	translation vector
Y	3D coordinates of the structure points in the camera frame
dYdom	Derivative of Y with respect to om
dYdT	Derivative of Y with respect to T

rodrigues()

void icl::io::IntrinsicCalibrator::rodrigues ( const math::DynMatrix< icl64f > &  in, math::DynMatrix< icl64f > &  out, math::DynMatrix< icl64f > &  dout  )

private

Computes the rodrigues transformation and derivative.

saveIntrinsics()

void icl::io::IntrinsicCalibrator::saveIntrinsics

(

const std::string &

filename

)

saves computed/current intrinsics to xml-file

Parameters

filename

name of xml-file for saving

Member Data Documentation

m_calres

Result icl::io::IntrinsicCalibrator::m_calres

private

struct for handling result of computation

m_data

Data* icl::io::IntrinsicCalibrator::m_data

private

---

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

• /Users/alneuman/vm/icl/ICLIO/src/ICLIO/IntrinsicCalibrator.h