DICe::Camera_System Class Reference

A class for camera calibration parameters and computing projection transformations notes: there is only one source camera per camera system, this determines how the coordinate systems are organized. All other cameras are target cameras. More...

#include <DICe_CameraSystem.h>

Public Types
enum	System_Type_3D {   UNKNOWN_SYSTEM = 0, GENERIC_SYSTEM, OPENCV, VIC3D,   DICE, MAX_SYSTEM_TYPE_3D, NO_SUCH_SYSTEM_TYPE_3D }
enum	Coordinate_Transformation_Param {   ANGLE_X = 0, ANGLE_Y, ANGLE_Z, TRANSLATION_X,   TRANSLATION_Y, TRANSLATION_Z }

Public Member Functions
	Camera_System ()
	default constructor
	Camera_System (const std::string &param_file_name)
	constructor More...
void	read_camera_system_file (const std::string &file)
	read the camera system parameters More...
void	write_camera_system_file (const std::string &file)
	write the camera system parameters to an xml file More...
void	set_system_type (const System_Type_3D system_type)
	set the system type More...
System_Type_3D	system_type () const
	get the integer value for the current system type More...
scalar_t	diff (const Camera_System &rhs) const
	diffs all the cameras that make up a camera system
bool	extrinsics_relative_camera_to_camera () const
size_t	get_camera_num_from_id (std::string &cam_id) const
	returns the number of the first camera with a matching identifier More...
Teuchos::RCP< Camera >	camera (const size_t i) const
	return a pointer to the camera with this vector index
size_t	num_cameras () const
	return the number of cameras
void	add_camera (Teuchos::RCP< Camera > camera_ptr)
	add a camera to the set
Matrix< scalar_t, 3 >	fundamental_matrix (const size_t source_cam_id=0, const size_t target_cam_id=1)
	calculate the fundamental matrix for a set of cameras
void	camera_to_camera_projection (const size_t source_id, const size_t target_id, const std::vector< scalar_t > &img_source_x, const std::vector< scalar_t > &img_source_y, std::vector< scalar_t > &img_target_x, std::vector< scalar_t > &img_target_y, const std::vector< scalar_t > &facet_params, std::vector< std::vector< scalar_t > > &img_target_dx, std::vector< std::vector< scalar_t > > &img_target_dy, const std::vector< scalar_t > &rigid_body_params=std::vector< scalar_t >())
	3 parameter projection routine from the one camera to another camera More...
void	camera_to_camera_projection (const size_t source_id, const size_t target_id, const std::vector< scalar_t > &img_source_x, const std::vector< scalar_t > &img_source_y, std::vector< scalar_t > &img_target_x, std::vector< scalar_t > &img_target_y, const std::vector< scalar_t > &facet_params, const std::vector< scalar_t > &rigid_body_params=std::vector< scalar_t >())
	same as above, only do not compute the partials
void	rot_trans_3D (const std::vector< scalar_t > &source_x, const std::vector< scalar_t > &source_y, const std::vector< scalar_t > &source_z, std::vector< scalar_t > &target_x, std::vector< scalar_t > &target_y, std::vector< scalar_t > &target_z, const std::vector< scalar_t > &rigid_body_params, std::vector< std::vector< scalar_t > > &target_dx, std::vector< std::vector< scalar_t > > &target_dy, std::vector< std::vector< scalar_t > > &target_dz)
void	rot_trans_3D (const std::vector< scalar_t > &source_x, const std::vector< scalar_t > &source_y, const std::vector< scalar_t > &source_z, std::vector< scalar_t > &target_x, std::vector< scalar_t > &target_y, std::vector< scalar_t > &target_z, const std::vector< scalar_t > &rigid_body_params)
	same as above with no partials

Static Public Member Functions
static std::string	to_string (System_Type_3D in)
static System_Type_3D	string_to_system_type_3d (std::string &in)

Private Attributes
const size_t	max_num_cameras_allowed_
	sets the number of cameras that can be in an input file
Matrix< scalar_t, 4 >	user_4x4_trans_
std::vector< scalar_t >	user_6x1_trans_
	8 parameters that define a projective transform (independent from intrinsic and extrinsic parameters)
System_Type_3D	sys_type_
	defines the camera system type (OPENCV VIC3D...)
bool	has_6_transform_
	the user defined a 6 parameter transform to modify the world coordinate system
bool	has_4x4_transform_
	the user defined a 4x transform to another coordinate system to use for the world coordinates
bool	extrinsics_relative_camera_to_camera_
std::vector< Teuchos::RCP< DICe::Camera > >	cameras_
	cameras are stored in a vector

Friends
DICE_LIB_DLL_EXPORT friend bool	operator== (const Camera_System &lhs, const Camera_System &rhs)
	comparison operator
DICE_LIB_DLL_EXPORT friend bool	operator!= (const Camera_System &lhs, const Camera_System &rhs)
	comparison operator
DICE_LIB_DLL_EXPORT friend std::ostream &	operator<< (std::ostream &os, const Camera_System &camera_system)
	overaload the ostream operator for a camera system class

Detailed Description

A class for camera calibration parameters and computing projection transformations notes: there is only one source camera per camera system, this determines how the coordinate systems are organized. All other cameras are target cameras.

Member Enumeration Documentation

Coordinate_Transformation_Param

enum DICe::Camera_System::Coordinate_Transformation_Param

rigid body rotation parameter index

System_Type_3D

enum DICe::Camera_System::System_Type_3D

valid system types

Constructor & Destructor Documentation

Camera_System()

DICe::Camera_System::Camera_System

(

const std::string &

param_file_name

)

constructor

Parameters

param_file_name

the name of the file to parse the calibration parameters from

Member Function Documentation

camera_to_camera_projection()

void DICe::Camera_System::camera_to_camera_projection	(	const size_t	source_id,
		const size_t	target_id,
		const std::vector< scalar_t > &	img_source_x,
		const std::vector< scalar_t > &	img_source_y,
		std::vector< scalar_t > &	img_target_x,
		std::vector< scalar_t > &	img_target_y,
		const std::vector< scalar_t > &	facet_params,
		std::vector< std::vector< scalar_t > > &	img_target_dx,
		std::vector< std::vector< scalar_t > > &	img_target_dy,
		const std::vector< scalar_t > &	rigid_body_params = std::vector< scalar_t >()
	)

3 parameter projection routine from the one camera to another camera

Parameters

source_id	the number id of the source camera
target_id	the camera id of the target camera.
img_source_x	array of x image location in the source camera
img_source_y	array of y image location in the source camera
img_targe_x	array of projected x image location in the target camera
img_target_y	array of projected y image location in the target camera
facet_params	array of ZP, THETA, PHI values that govern the projection
img_target_dx	array of arrays of the partials in the x location wrt (ZP, THETA, PHI)
img_target_dy	array of arrays of the partials in the y location wrt (ZP, THETA, PHI)
rigid_body_params	6 rotation/translation parameters to describe the rigid body motion if the rigid_body_params argument is specified points in space are established by projecting them using the 3 parameter projection, they are then transformed in a rigid body manner to a second location which is then projected into the target camera. ZP, THETA and PHI are considered fixed

extrinsics_relative_camera_to_camera()

bool DICe::Camera_System::extrinsics_relative_camera_to_camera ( ) const

inline

returns true if the extrinsic parameters are camera to camera relative rather than from world to camera (if this is the case, the first camera's extrinsics are world to camera 0, the second camera's extrinsics are the camera 0 to camera 1 transformation) This is included for legacy reasons

get_camera_num_from_id()

size_t DICe::Camera_System::get_camera_num_from_id ( std::string &  cam_id ) const

inline

returns the number of the first camera with a matching identifier

Parameters

cam_id

camera identifying string

read_camera_system_file()

void DICe::Camera_System::read_camera_system_file

(

const std::string &

file

)

read the camera system parameters

Parameters

file	File name of the parameters file

rot_trans_3D()

void DICe::Camera_System::rot_trans_3D	(	const std::vector< scalar_t > &	source_x,
		const std::vector< scalar_t > &	source_y,
		const std::vector< scalar_t > &	source_z,
		std::vector< scalar_t > &	target_x,
		std::vector< scalar_t > &	target_y,
		std::vector< scalar_t > &	target_z,
		const std::vector< scalar_t > &	rigid_body_params,
		std::vector< std::vector< scalar_t > > &	target_dx,
		std::vector< std::vector< scalar_t > > &	target_dy,
		std::vector< std::vector< scalar_t > > &	target_dz
	)

rigid body 3D rotation/translation transformation with no incoming partials and outgoing partials wrt params

Parameters

source_x	array of initial x world position of the point
source_y	array of initial y world position of the point
source_z	array of initial z world position of the point
targe_x	array of x world position of the point after the transformation
target_y	array of y world position of the point after the transformation
target_z	array of z world position of the point after the transformation
rigid_body_params	6 rotation/translation parameters to describe the rigid body motion (3 angles 3 translations)
target_dx	array of arrays of the partials in the x world location wrt params
target_dy	array of arrays of the partials in the y world location wrt params
target_dy	array of arrays of the partials in the z world location wrt params

set_system_type()

void DICe::Camera_System::set_system_type ( const System_Type_3D  system_type )

inline

set the system type

Parameters

system_type

a System_Type_3D enum

system_type()

System_Type_3D DICe::Camera_System::system_type ( ) const

inline

get the integer value for the current system type

Parameters

system_type

integer system type value

write_camera_system_file()

void DICe::Camera_System::write_camera_system_file

(

const std::string &

file

)

write the camera system parameters to an xml file

Parameters

file	File name of the parameters file to write to

Member Data Documentation

extrinsics_relative_camera_to_camera_

bool DICe::Camera_System::extrinsics_relative_camera_to_camera_

private

the standard convention is to have the camera extrinsics be from world to camera when this flag is true, the extrinsics are the transformation from the left camera to the right instead

user_4x4_trans_

Matrix<scalar_t,4> DICe::Camera_System::user_4x4_trans_

private

these are for compatibility with triangulation and only support 2 camera systems 12 parameters that define a user supplied transforamation (independent from intrinsic and extrinsic parameters)