This interface provide functionalities of the pinhole camera model. More...

Inheritance diagram for ILCalibPinHole:

Public Member Functions
LPVErrorCode	AddMVPoints (ILCalibPointPairs *pointPairs)

void	SetCameraProps (double cellSize, double focusLength, double viewDistance)

Public Member Functions inherited from ILCalib
LPVErrorCode	AddPoint (double imagePointX, double imagePointY, double worldPointX, double worldPointY)

LPVErrorCode	AddPoints (ILCalibPointPairs *pointPairs)

LPVErrorCode	Calibrate ()

LPVErrorCode	FixImage (ILImage image, ILImage fixedImage)

LPVErrorCode	GenPostTransformMatrix (ILCalibPointPairs *pointPairs)

void	ImageToWorld (double imagePointX, double imagePointY, double worldPointX, double worldPointY)

BOOL	IsCalibrated ()

void	SetImageSize (int w, int h)

void	SetPostTransform (double offsetX, double offsetY, double deltaAngle)

void	SetPostTransformMatrix (ILTransform *tf)

void	SetPostTransformMatrix23 (LMatrix23 *data)

void	SetPostTransformMatrix33 (LMatrix33 *data)

void	WorldToImage (double worldPointX, double worldPointY, double imagePointX, double imagePointY)

Public Member Functions inherited from ILObject
ILObject *	Copy ()

LPVErrorCode	Load (LString filename)

void	Reset ()

LPVErrorCode	Save (LString filename)

BOOL	Valid ()

Properties
LPVDistortionModel	DistortionModel `[get, set]`
	The distortion model.

Properties inherited from ILCalib
double	CalibError `[get]`
	The overall RMS re-projection error for the calibration, calculated based on the distances between the image points and the projected (using the optimized calibration matrices) world points. It's defined as: \( \sqrt{ \frac{1}{n} \sum_{i=0}^{n} distance(p^{image}_i, p^{proj}_i)^2 } \).

LPVFixImageMode	FixImageMode `[get, set]`
	The mode to fix image used in FixImage(). More...

double	FixImageOffsetX `[get, set]`
	The x-coordinate offset (in pixel unit) of the fixed image used in FixImage(), used together with FixImageOffsetY. By default, it's 0, means no offset in x-coordinate. You may change the offset to translate the fixed image, for example move the inspect region to the center of the fixed image.

double	FixImageOffsetY `[get, set]`
	The y-coordinate offset (in pixel unit) of the fixed image used in FixImage(), used together with FixImageOffsetX. By default, it's 0, means no offset in y-coordinate. You may change the offset to translate the fixed image, for example move the inspect region to the center of the fixed image.

double	FixImageScale `[get, set]`
	The scale of the fixed image used in FixImage(). By default, it's 1, means no scale. You may change the scale factor to zoom in/out the fixed image.

LPVInterpolationMethod	InterpolationMethod `[get, set]`
	The interpolation method used in fix the image for undistortion (and untilt). By default, it's LPVInterCubic.

LPVCalibModel	Model `[get]`
	The calibration model.

BOOL	UseCache `[get, set]`
	Whether to make use of cache which is used to speed-up the process of the image undistortion (and untilt), but consume more memory. By default, it's off.

Detailed Description

This interface provide functionalities of the pinhole camera model.

In the pinhole camera model, a scene view is formed by projecting 3D points into the image plane using a perspective transformation, defined as:

\[ s \; m' = K \; [R | t] \; M \]

where \( m' \) are the image points in pixels, \( M \) are the world points in physical units, \( K \) is the intrinsic camera array involves camera's principal point and focus lengths in pixel units, \( [R | t] \) is the extrinsic camera array describes the camera's relative position to the view planar, \( s \) is the scale factor for the whole system.
Intrinsic parameters are independent to the viewed scene while extrinsic parameters do, which means if we move the camera other places without changing the focusing, the intrinsic parameters stay unchanged but extrinsic parameters should be re-calibrated.
This model support radial distortion estimation, see DistortionModel for more details.

The accuracy of the calibration result is affected by many factors:

The number of point pairs and their distribution. Usually, it's better to have the point pairs distributed evenly and as many as possible throughout the image.
The chosen distortion model that better fit to the reality.
Calibrated with multi-view point pairs help estimate more accurate intrinsic parameters, then more accurate extrinsic parameters are derived. Moreover, with multi-view calibration, the system is able to eliminate the re-projection error by the thickness of the calibration board.

To use this interface, you should create a LCalibPinHole object.

Example Code

Member Function Documentation

◆ AddMVPoints()

LPVErrorCode AddMVPoints ( ILCalibPointPairs * pointPairs )

Add point pairs for multi-view calibration.

Parameters

[in] pointPairs The collection of point pairs from assistant view.

Return values

error Return error code if anything is wrong.

◆ SetCameraProps()

void SetCameraProps	(	double	cellSize,
		double	focusLength,
		double	viewDistance
	)

Set some property of the camera or the viewed scene, which is used in setting up initial value for calibration.