boofcv.abst.geo.calibration

Class CalibrateMonoPlanar

java.lang.Object

boofcv.abst.geo.calibration.CalibrateMonoPlanar

public class CalibrateMonoPlanar
extends java.lang.Object

Performs the full processing loop for calibrating a mono camera from a planar grid. A directory is specified that the images are read in from. Calibration points are detected inside the image and feed into the Zhang99 algorithm for parameter estimation.

Internally it supports status updates for a GUI and skips over bad images. Invoke functions in the following order:

1. configure(boolean, int, boolean)
2. reset()
3. addImage(boofcv.struct.image.GrayF32)
4. process()
5. getIntrinsic()

Most 3D operations in BoofCV assume that the image coordinate system is right handed and the +Z axis is pointing out of the camera. In standard image coordinate the origin (0,0) is at the top left corner with +x going to the right and +y going down, then if it is right handed +z will be out of the image. However some times this pseudo standard is not followed and the y-axis needs to be inverted by setting isInverted to true.

Field Summary

Fields

Modifier and Type	Field and Description
protected DetectorFiducialCalibration	detector
protected java.util.List<ImageResults>	errors
protected boofcv.struct.calib.CameraPinholeRadial	foundIntrinsic
protected Zhang99ParamAll	foundZhang
protected java.util.List<CalibrationObservation>	observations
boolean	verbose
protected CalibrationPlanarGridZhang99	zhang99

Constructor Summary

Constructors

Constructor and Description
CalibrateMonoPlanar(DetectorFiducialCalibration detector) High level configuration

Method Summary

Modifier and Type	Method and Description
boolean	addImage(boofcv.struct.image.GrayF32 image) Adds a new view of the calibration target and processes it.
static java.util.List<ImageResults>	computeErrors(java.util.List<CalibrationObservation> observation, Zhang99ParamAll param, java.util.List<georegression.struct.point.Point2D_F64> grid) After the parameters have been estimated this computes the error for each calibration point in each image and summary error statistics.
void	configure(boolean assumeZeroSkew, int numRadialParam, boolean includeTangential) Specify calibration assumptions.
java.util.List<ImageResults>	getErrors()
boofcv.struct.calib.CameraPinholeRadial	getIntrinsic()
java.util.List<CalibrationObservation>	getObservations()
Zhang99ParamAll	getZhangParam()
static void	printErrors(java.util.List<ImageResults> results) Prints out error information to standard out
void	printStatistics()
boofcv.struct.calib.CameraPinholeRadial	process() After calibration points have been found this invokes the Zhang99 algorithm to estimate calibration parameters.
void	removeLatestImage() Removes the most recently added image
void	reset() Resets internal data structures.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

detector

protected DetectorFiducialCalibration detector

zhang99

protected CalibrationPlanarGridZhang99 zhang99

foundZhang

protected Zhang99ParamAll foundZhang

foundIntrinsic

protected boofcv.struct.calib.CameraPinholeRadial foundIntrinsic

observations

protected java.util.List<CalibrationObservation> observations

errors

protected java.util.List<ImageResults> errors

verbose

public boolean verbose

Constructor Detail

CalibrateMonoPlanar

public CalibrateMonoPlanar(DetectorFiducialCalibration detector)

High level configuration

Parameters:

detector - Target detection algorithm.

Method Detail

configure

public void configure(boolean assumeZeroSkew,
                      int numRadialParam,
                      boolean includeTangential)

Specify calibration assumptions.

Parameters:

assumeZeroSkew - If true then zero skew is assumed. Typically this will be true.

numRadialParam - Number of radial parameters. Typically set to 2.

includeTangential - If true it will estimate tangential distortion parameters. Try false then true

reset

public void reset()

Resets internal data structures. Must call before adding images

addImage

public boolean addImage(boofcv.struct.image.GrayF32 image)

Adds a new view of the calibration target and processes it. If processing fails then false is returned.

Parameters:

image - Image of a calibration target

Returns:

true if a target was detected in the image or not

removeLatestImage

public void removeLatestImage()

Removes the most recently added image

process

public boofcv.struct.calib.CameraPinholeRadial process()

After calibration points have been found this invokes the Zhang99 algorithm to estimate calibration parameters. Error statistics are also computed.

printStatistics

public void printStatistics()

computeErrors

public static java.util.List<ImageResults> computeErrors(java.util.List<CalibrationObservation> observation,
                                                         Zhang99ParamAll param,
                                                         java.util.List<georegression.struct.point.Point2D_F64> grid)

After the parameters have been estimated this computes the error for each calibration point in each image and summary error statistics.

Parameters:

observation - Observed control point location

param - Found calibration parameters

Returns:

List of error statistics

printErrors

public static void printErrors(java.util.List<ImageResults> results)

Prints out error information to standard out

getObservations

public java.util.List<CalibrationObservation> getObservations()

getErrors

public java.util.List<ImageResults> getErrors()

getZhangParam

public Zhang99ParamAll getZhangParam()

getIntrinsic

public boofcv.struct.calib.CameraPinholeRadial getIntrinsic()