mimas::camera_calibration::Camera Class Reference
[Camera Calibration]

This class represents a camera, and provides some functions to calibrate it. More...

#include <camera.h>

Collaboration diagram for mimas::camera_calibration::Camera:


Public Types
typedef boost::numeric::ublas::matrix< double, boost::numeric::ublas::column_major >	Matrix
	Matrix definition.
typedef boost::numeric::ublas::vector< double >	Vector
	Vector definition.
typedef boost::numeric::ublas::triangular_matrix< double, boost::numeric::ublas::upper, boost::numeric::ublas::column_major >	TriangularMatrix
	Triangular matrix definition.
typedef boost::numeric::ublas::symmetric_matrix< double, boost::numeric::ublas::upper, boost::numeric::ublas::column_major >	SymmetricMatrix
	Symmetric matrix definition.
Public Member Functions
	Camera () throw (mimasexception)
	Constructor : instanciate ONE class.
	~Camera ()
	Free the class.
void	calibrate (std::vector< Chessboard > &chess) throw (mimasexception)
	Calibrate the camera.
Matrix	getIntrinsicsInit () const
	Returns first approximation of the matrix of the intrinsic parameters.
Matrix	getIntrinsics () const
	Returns the optimized matrix of the intrinsic parameters.
Vector	getDistortion () const
	Returns the distortion.
bool	isOptimized () const
	Returns if the optimisation algorithm worked or not.
Static Public Member Functions
static void	fcn (int m, int , double x, double fvec, int *)
	Calculate the projection of the chessboard points according to the given values of the intrinsic matrix and the distortion coefficients, and return an array containing the error between the mesured point and the calculated one.
static void	findHomography (const Chessboard &chess, Matrix &homography)
	Find the homography.
static void	findHomographies (std::vector< Matrix > &homographies, std::vector< Chessboard > &chess)
	Find all the homographies.
static Matrix	initCalibration (std::vector< Matrix > &homographies, std::vector< Chessboard > &chess) throw (mimasexception)
	Find a first approximation of the intrinsic matrix.
static void	findExtrinsics (std::vector< Matrix > &extrinsics, std::vector< Matrix > &homographies, Matrix &intrinsics, int nbPictures)
	Find the extrinsic parameters.
Static Public Attributes
static Camera *	current_cam
	pointer to the current class.
Protected Member Functions
void	maxLikelyhood (Matrix &intrinsics, int nbPictures) throw (mimasexception)
	This function applies the Levenberg-Marquardt algorithm to the first approximation of the intrinsic matrix.

Detailed Description

This class represents a camera, and provides some functions to calibrate it.

The class can be used with a float or double template, depending on the precision required in the calculus.

Important : because of a limitation of a FORTRAN library used by the calibration algorithm, it's only possible to instanciate one class Camera at the same time.

Author:: Julien Faucher (faucherj@gmail.com)

Definition at line 57 of file camera.h.

Member Typedef Documentation

typedef boost::numeric::ublas::matrix< double, boost::numeric::ublas::column_major > mimas::camera_calibration::Camera::Matrix

Matrix definition.

Definition at line 61 of file camera.h.

typedef boost::numeric::ublas::vector< double > mimas::camera_calibration::Camera::Vector

Vector definition.

Definition at line 64 of file camera.h.

typedef boost::numeric::ublas::triangular_matrix< double, boost::numeric::ublas::upper, boost::numeric::ublas::column_major > mimas::camera_calibration::Camera::TriangularMatrix

Triangular matrix definition.

Definition at line 68 of file camera.h.

typedef boost::numeric::ublas::symmetric_matrix< double, boost::numeric::ublas::upper, boost::numeric::ublas::column_major > mimas::camera_calibration::Camera::SymmetricMatrix

Symmetric matrix definition.

Definition at line 72 of file camera.h.

Constructor & Destructor Documentation

mimas::camera_calibration::Camera::Camera ( ) throw (mimasexception)

Constructor : instanciate ONE class.

mimas::camera_calibration::Camera::~Camera ( )

Free the class.

Member Function Documentation

void mimas::camera_calibration::Camera::maxLikelyhood	(	Matrix &	intrinsics,
		int	nbPictures
	)			throw (mimasexception) `[protected]`

This function applies the Levenberg-Marquardt algorithm to the first approximation of the intrinsic matrix.

void mimas::camera_calibration::Camera::calibrate ( std::vector< Chessboard > & chess ) throw (mimasexception)

Calibrate the camera.

This is the function one should call to get the calibration parameters.

To calibrate the camera, one should provide four points of the chessboard in the coordinate of the chessboard (the unit is the square), and the coordinates of the same points in the picture of the chessboard taken by the camera (the unit is the pixel).

A minimum of 4 images is required to compute the calibration, but you will need much more to get an accurate result.

Parameters:

chess

a vector of Chessboard classes, to know the images of the chessboard points by the camera.

static void mimas::camera_calibration::Camera::fcn	(	int *	m,
		int *	,
		double *	x,
		double *	fvec,
		int *
	)			`[static]`

Calculate the projection of the chessboard points according to the given values of the intrinsic matrix and the distortion coefficients, and return an array containing the error between the mesured point and the calculated one.

This function is used by the MINPACK library (in FORTRAN) to minimize the error using the Levenberg-Marquardt algorithm.

Parameters:

	m	the number of functions ie 4 * nbPictures (there are 4 points per picture)
	x	an arrray containing the 5 intrinsic parameters and the 2 distortion parameters
	fvec	an output array, containing the result of each function.

static void mimas::camera_calibration::Camera::findHomography	(	const Chessboard &	chess,
		Matrix &	homography
	)			`[static]`

Find the homography.

This function finds the homography between a set of $nb$ points and $nb$ images.

Parameters:

	chess	a Chessboard class
	homography	a 3*3 matrix, must be initialised.

static void mimas::camera_calibration::Camera::findHomographies	(	std::vector< Matrix > &	homographies,
		std::vector< Chessboard > &	chess
	)			`[static]`

Find all the homographies.

Given 4 points and their images for some pictures, this function calculates all the homographies (one by picture).

static Matrix mimas::camera_calibration::Camera::initCalibration	(	std::vector< Matrix > &	homographies,
		std::vector< Chessboard > &	chess
	)			throw (mimasexception) `[static]`

Find a first approximation of the intrinsic matrix.

This function returns a first approximation of the intrinsics, in the ideal case where there is no noise in the pictures, and no distortion.

static void mimas::camera_calibration::Camera::findExtrinsics	(	std::vector< Matrix > &	extrinsics,
		std::vector< Matrix > &	homographies,
		Matrix &	intrinsics,
		int	nbPictures
	)			`[static]`