org.openimaj.image.camera

Class CameraIntrinsics

java.lang.Object

org.openimaj.image.camera.CameraIntrinsics

public class CameraIntrinsics
extends Object

The intrinsic parameters of a camera (focal length in x and y; skew; principal point in x and y; 3-term radial distorion; 2-term tangential distortion).

Author:

Jonathon Hare (jsh2@ecs.soton.ac.uk)

Field Summary

Fields

Modifier and Type	Field and Description
Jama.Matrix	calibrationMatrix The camera calibration matrix
int	height The image height of the camera in pixels
double	k1 First radial distortion term
double	k2 Second radial distortion term
double	k3 Third radial distortion term
double	p1 First tangential distortion term
double	p2 Second tangential distortion term
int	width The image width of the camera in pixels

Constructor Summary

Constructors

Constructor and Description
CameraIntrinsics(Jama.Matrix m, int width, int height) Construct with the given matrix.

Method Summary

Methods

Modifier and Type	Method and Description
Point2d	applyDistortion(Point2d p) Apply the radial and tangential distortion of this camera to the given projected point (presumably computed by projecting a world point through the homography defined by the extrinsic parameters of a camera).
FImage[]	buildRectifiedUndistortionMap(int width, int height, CameraIntrinsics target, Jama.Matrix R) Build the rectified distortion map, which for every un-distorted point in the given size image contains the x and y ordinates of the corresponding rectifed distorted point.
RemapProcessor	buildRectifiedUndistortionProcessor(int width, int height, CameraIntrinsics target, Jama.Matrix R) Build a RemapProcessor capable of correcting the radial and tangential distortion of this camera.
FImage[]	buildUndistortionMap(int width, int height) Build the distortion map, which for every un-distorted point in the given size image contains the x and y ordinates of the corresponding distorted point.
FImage[]	buildUndistortionMap(int width, int height, CameraIntrinsics target) Build the distortion map, which for every un-distorted point in the given size image contains the x and y ordinates of the corresponding distorted point.
RemapProcessor	buildUndistortionProcessor() Build a RemapProcessor capable of correcting the radial and tangential distortion of this camera.
RemapProcessor	buildUndistortionProcessor(int width, int height) Build a RemapProcessor capable of correcting the radial and tangential distortion of this camera.
RemapProcessor	buildUndistortionProcessor(int width, int height, CameraIntrinsics target) Build a RemapProcessor capable of correcting the radial and tangential distortion of this camera.
double	getFocalLengthX() Get the focal length in the x direction (in pixels)
double	getFocalLengthY() Get the focal length in the y direction (in pixels)
double	getPrincipalPointX() Get the x-ordinate of the principal point (in pixels).
double	getPrincipalPointY() Get the y-ordinate of the principal point (in pixels).
CameraIntrinsics	getScaledIntrinsics(int newWidth, int newHeight) Compute a scaled set of intrinsic parameters based on the given image size.
double	getSkewFactor() Get the skew factor.
void	setFocalLengthX(double fy) Set the focal length in the x direction (in pixels)
void	setFocalLengthY(double fy) Set the focal length in the y direction (in pixels)
void	setPrincipalPointX(double cx) Set the x-ordinate of the principal point (in pixels).
void	setPrincipalPointY(double cy) Set the y-ordinate of the principal point (in pixels).
void	setSkewFactor(double skew) Set the skew factor.
String	toString()
<I extends Image<?,I> & SinglebandImageProcessor.Processable<Float,FImage,I>> I	undistort(I image) Undistort the given image by removing the radial and tangential distortions of this camera.

Methods inherited from class java.lang.Object

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

Field Detail

calibrationMatrix

public Jama.Matrix calibrationMatrix

The camera calibration matrix

k1

public double k1

First radial distortion term

k2

public double k2

Second radial distortion term

k3

public double k3

Third radial distortion term

p1

public double p1

First tangential distortion term

p2

public double p2

Second tangential distortion term

width

public int width

The image width of the camera in pixels

height

public int height

The image height of the camera in pixels

Constructor Detail

CameraIntrinsics

public CameraIntrinsics(Jama.Matrix m,
                int width,
                int height)

Construct with the given matrix.

Parameters:

m - the matrix

width - the image width of the camera in pixels

height - the image height of the camera in pixels

Method Detail

getFocalLengthX

public double getFocalLengthX()

Get the focal length in the x direction (in pixels)

Returns:

fx

getFocalLengthY

public double getFocalLengthY()

Get the focal length in the y direction (in pixels)

Returns:

fy

getPrincipalPointX

public double getPrincipalPointX()

Get the x-ordinate of the principal point (in pixels).

Returns:

cx

getPrincipalPointY

public double getPrincipalPointY()

Get the y-ordinate of the principal point (in pixels).

Returns:

cy

setFocalLengthX

public void setFocalLengthX(double fy)

Set the focal length in the x direction (in pixels)

Parameters:

fy - the focal length in the x direction

setFocalLengthY

public void setFocalLengthY(double fy)

Set the focal length in the y direction (in pixels)

Parameters:

fy - the focal length in the y direction

setPrincipalPointX

public void setPrincipalPointX(double cx)

Set the x-ordinate of the principal point (in pixels).

Parameters:

cx - the y-ordinate of the principal point

setPrincipalPointY

public void setPrincipalPointY(double cy)

Set the y-ordinate of the principal point (in pixels).

Parameters:

cy - the y-ordinate of the principal point

setSkewFactor

public void setSkewFactor(double skew)

Set the skew factor.

Parameters:

skew - the skew.

getSkewFactor

public double getSkewFactor()

Get the skew factor.

Returns:

skew

applyDistortion

public Point2d applyDistortion(Point2d p)

Apply the radial and tangential distortion of this camera to the given projected point (presumably computed by projecting a world point through the homography defined by the extrinsic parameters of a camera). The point is modified in-place.

Parameters:

p - the projected point

Returns:

the input point (with the distortion added)

getScaledIntrinsics

public CameraIntrinsics getScaledIntrinsics(int newWidth,
                                   int newHeight)

Compute a scaled set of intrinsic parameters based on the given image size.

Parameters:

newWidth - the target image size

newHeight - the target image width

Returns:

the new scaled intrinsics

toString

public String toString()

Overrides:

toString in class Object

buildUndistortionProcessor

public RemapProcessor buildUndistortionProcessor()

Build a RemapProcessor capable of correcting the radial and tangential distortion of this camera.

Note: Skew is currently assumed to be zero

Returns:

the processor for undistorting the image

buildUndistortionProcessor

public RemapProcessor buildUndistortionProcessor(int width,
                                        int height)

Build a RemapProcessor capable of correcting the radial and tangential distortion of this camera.

Note: Skew is currently assumed to be zero

Parameters:

width - the target width of images produced by the processor

height - the target height of images produced by the processor

Returns:

the processor for undistorting the image

buildUndistortionProcessor

public RemapProcessor buildUndistortionProcessor(int width,
                                        int height,
                                        CameraIntrinsics target)

Build a RemapProcessor capable of correcting the radial and tangential distortion of this camera. The distortion map is computed such that the warped image will appear as if it were viewed through the undistorted target camera, rather than this one.

Note: Skew is currently assumed to be zero

Parameters:

width - the target width of images produced by the processor

height - the target height of images produced by the processor

target - the target camera

Returns:

the processor for undistorting the image

buildRectifiedUndistortionProcessor

public RemapProcessor buildRectifiedUndistortionProcessor(int width,
                                                 int height,
                                                 CameraIntrinsics target,
                                                 Jama.Matrix R)

Build a RemapProcessor capable of correcting the radial and tangential distortion of this camera. The distortion map is computed such that the warped image will appear as if it were viewed through the undistorted target camera rather than this one, and the 3x3matrix R controls the rectification (i.e. the relative change in position of the camera in world space).

Note: Skew is currently assumed to be zero

Parameters:

width - the target width of images produced by the processor

height - the target height of images produced by the processor

target - the target camera

R - the rectification matrix

Returns:

the processor for undistorting the image

buildUndistortionMap

public FImage[] buildUndistortionMap(int width,
                            int height)

Build the distortion map, which for every un-distorted point in the given size image contains the x and y ordinates of the corresponding distorted point. The resultant map can be used with a RemapProcessor to undistort imaaes.

Note: Skew is currently assumed to be zero

Parameters:

width - the desired width; typically the same as the image width used to calibrate the camera

height - the desired height; typically the same as the image height used to calibrate the camera

Returns:

the distortion map

buildUndistortionMap

public FImage[] buildUndistortionMap(int width,
                            int height,
                            CameraIntrinsics target)

Build the distortion map, which for every un-distorted point in the given size image contains the x and y ordinates of the corresponding distorted point. The resultant map can be used with a RemapProcessor to undistort imaaes. The distortion map is computed such that the warped image will appear as if it were viewed through the undistorted target camera, rather than this one.

Note: Skew is currently assumed to be zero

Parameters:

width - the desired width; typically the same as the image width used to calibrate the camera

height - the desired height; typically the same as the image height used to calibrate the camera

target - the target camera intrinsics

Returns:

the distortion map

buildRectifiedUndistortionMap

public FImage[] buildRectifiedUndistortionMap(int width,
                                     int height,
                                     CameraIntrinsics target,
                                     Jama.Matrix R)

Build the rectified distortion map, which for every un-distorted point in the given size image contains the x and y ordinates of the corresponding rectifed distorted point. The resultant map can be used with a RemapProcessor to undistort imaaes. The distortion map is computed such that the warped image will appear as if it were viewed through the undistorted target camera rather than this one, and the 3x3matrix R controls the rectification (i.e. the relative change in position of the camera in world space).

Note: Skew is currently assumed to be zero

Parameters:

width - the desired width; typically the same as the image width used to calibrate the camera

height - the desired height; typically the same as the image height used to calibrate the camera

target - the target camera intrinsics

R - the rectification matrix

Returns:

the distortion map

undistort

public <I extends Image<?,I> & SinglebandImageProcessor.Processable<Float,FImage,I>> I undistort(I image)

Undistort the given image by removing the radial and tangential distortions of this camera. It is assumed that the input image has the same dimensions as images produced by this CameraIntrinsics.

This method is inefficient if you need to process many images as the distortion map is computed each time. For more efficient operation, use buildUndistortionProcessor() to make a reusable RemapProcessor capable of efficiently undistorting multiple images.

Parameters:

image - the image to undistort

Returns:

the undistorted image