oakutils.calibration package

Module contents

Module for camera calibration utilities.

Classes

CalibrationData

A class for storing calibration data.

ColorCalibrationData

A class for storing color camera calibration data.

MonoCalibrationData

A class for storing mono camera calibration data.

StereoCalibrationData

A class for storing stereo camera calibration data.

Functions

create_q_matrix

Creates a Q matrix from a stereo calibration.

get_camera_calibration

Gets the camera calibration data from any device.

get_oak1_calibration

Gets the camera calibration data from the OAK-1 device.

get_oak1_calibration_basic

Gets the camera calibration data from the OAK-1 device, without computed info.

get_oakd_calibration

Gets the camera calibration data from the OAK-D device.

get_oakd_calibration_basic

Gets the camera calibration data from the OAK-D device, without computed info.

get_oakd_calibration_primary_mono

Gets the camera calibration data from the OAK-D device, using the primary mono camera.

class oakutils.calibration.CalibrationData(rgb: ColorCalibrationData, left: MonoCalibrationData, right: MonoCalibrationData, stereo: StereoCalibrationData, l2rgb_extrinsic: np.ndarray, r2rgb_extrinsic: np.ndarray, rgb2l_extrinsic: np.ndarray, rgb2r_extrinsic: np.ndarray, T_l_rgb: np.ndarray, T_r_rgb: np.ndarray, T_rgb_l: np.ndarray, T_rgb_r: np.ndarray, primary: MonoCalibrationData | None = None)

Bases: object

An object to store calibration data for an entire OAK camera.

rgb

RGB camera calibration data.

Type:

ColorCalibrationData

left

Left mono camera calibration data.

Type:

MonoCalibrationData

right

Right mono camera calibration data.

Type:

MonoCalibrationData

stereo

Stereo camera calibration data.

Type:

StereoCalibrationData

l2rgb_extrinsic

Extrinsic matrix from the left to the RGB camera.

Type:

np.ndarray

r2rgb_extrinsic

Extrinsic matrix from the right to the RGB camera.

Type:

np.ndarray

rgb2l_extrinsic

Extrinsic matrix from the RGB to the left camera.

Type:

np.ndarray

rgb2r_extrinsic

Extrinsic matrix from the RGB to the right camera.

Type:

np.ndarray

T_l_rgb

Translation vector from the left to the RGB camera.

Type:

np.ndarray

T_r_rgb

Translation vector from the right to the RGB camera.

Type:

np.ndarray

T_rgb_l

Translation vector from the RGB to the left camera.

Type:

np.ndarray

T_rgb_r

Translation vector from the RGB to the right camera.

Type:

np.ndarray

primary

Primary camera calibration data.

Type:

Optional[MonoCalibrationData], optional

See also

ColorCalibrationData

Class to store calibration data for a color camera.

MonoCalibrationData

Class to store calibration data for a mono camera.

StereoCalibrationData

Class to store calibration data for stereo cameras.

T_l_rgb: np.ndarray
T_r_rgb: np.ndarray
T_rgb_l: np.ndarray
T_rgb_r: np.ndarray
l2rgb_extrinsic: np.ndarray
left: MonoCalibrationData
primary: MonoCalibrationData | None = None
r2rgb_extrinsic: np.ndarray
rgb: ColorCalibrationData
rgb2l_extrinsic: np.ndarray
rgb2r_extrinsic: np.ndarray
right: MonoCalibrationData
stereo: StereoCalibrationData
class oakutils.calibration.ColorCalibrationData(size: tuple[int, int], K: np.ndarray, D: np.ndarray, fx: float, fy: float, cx: float, cy: float, fov: float, fov_rad: float, P: np.ndarray | None = None, valid_region: tuple[int, int, int, int] | None = None, map_1: np.ndarray | None = None, map_2: np.ndarray | None = None, pinhole: PinholeCameraIntrinsic | None = None)

Bases: object

Class to store calibration data for a color camera.

size

Image size.

Type:

Tuple[int, int]

K

Camera matrix.

Type:

np.ndarray

D

Distortion coefficients.

Type:

np.ndarray

fx

Focal length in the x direction.

Type:

float

fy

Focal length in the y direction.

Type:

float

cx

Principal point in the x direction.

Type:

float

cy

Principal point in the y direction.

Type:

float

fov

Field of view.

Type:

float

fov_rad

Field of view in radians.

Type:

float

P

Projection matrix.

Type:

Optional[np.ndarray], optional

valid_region

Valid region of the calibration generated by cv2.getOptimalNewCameraMatrix.

Type:

Optional[Tuple[int, int, int, int]], optional

map_1

Map 1 for undistortion generated by cv2.initUndistortRectifyMap.

Type:

Optional[np.ndarray], optional

map_2

Map 2 for undistortion generated by cv2.initUndistortRectifyMap.

Type:

Optional[np.ndarray], optional

pinhole

o3d pinhole camera intrinsic.

Type:

Optional[PinholeCameraIntrinsic], optional

D: np.ndarray
K: np.ndarray
P: np.ndarray | None = None
cx: float
cy: float
fov: float
fov_rad: float
fx: float
fy: float
map_1: np.ndarray | None = None
map_2: np.ndarray | None = None
pinhole: PinholeCameraIntrinsic | None = None
size: tuple[int, int]
valid_region: tuple[int, int, int, int] | None = None
class oakutils.calibration.MonoCalibrationData(size: tuple[int, int], K: np.ndarray, D: np.ndarray, fx: float, fy: float, cx: float, cy: float, fov: float, fov_rad: float, R: np.ndarray, T: np.ndarray, H: np.ndarray, valid_region: tuple[int, int, int, int] | None = None, map_1: np.ndarray | None = None, map_2: np.ndarray | None = None, pinhole: PinholeCameraIntrinsic | None = None)

Bases: object

Class to store calibration data for a mono camera.

size

Image size.

Type:

Tuple[int, int]

K

Camera matrix.

Type:

np.ndarray

D

Distortion coefficients.

Type:

np.ndarray

fx

Focal length in x.

Type:

float

fy

Focal length in y.

Type:

float

cx

Principal point in x.

Type:

float

cy

Principal point in y.

Type:

float

fov

Field of view.

Type:

float

fov_rad

Field of view in radians.

Type:

float

R

Rotation matrix.

Type:

np.ndarray

T

Translation matrix.

Type:

np.ndarray

H

Homography matrix.

Type:

np.ndarray

valid_region

Valid region of the calibration generated by cv2.stereoRectify.

Type:

Optional[Tuple[int, int, int, int]], optional

map_1

Map 1 for undistortion generated by cv2.initUndistortRectifyMap.

Type:

Optional[np.ndarray], optional

map_2

Map 2 for undistortion generated by cv2.initUndistortRectifyMap.

Type:

Optional[np.ndarray], optional

pinhole

o3d pinhole camera intrinsic.

Type:

Optional[PinholeCameraIntrinsic], optional

D: np.ndarray
H: np.ndarray
K: np.ndarray
R: np.ndarray
T: np.ndarray
cx: float
cy: float
fov: float
fov_rad: float
fx: float
fy: float
map_1: np.ndarray | None = None
map_2: np.ndarray | None = None
pinhole: PinholeCameraIntrinsic | None = None
size: tuple[int, int]
valid_region: tuple[int, int, int, int] | None = None
class oakutils.calibration.StereoCalibrationData(left: MonoCalibrationData, right: MonoCalibrationData, R1: np.ndarray, R2: np.ndarray, T1: np.ndarray, T2: np.ndarray, H_left: np.ndarray, H_right: np.ndarray, l2r_extrinsic: np.ndarray, r2l_extrinsic: np.ndarray, Q_left: np.ndarray, Q_right: np.ndarray, baseline: float, primary: MonoCalibrationData | None = None, Q_primary: np.ndarray | None = None, Q_cv2: np.ndarray | None = None, R1_cv2: np.ndarray | None = None, R2_cv2: np.ndarray | None = None, P1: np.ndarray | None = None, P2: np.ndarray | None = None, valid_region_primary: tuple[int, int, int, int] | None = None, pinhole_primary: PinholeCameraIntrinsic | None = None)

Bases: object

Class to store calibration data for stereo cameras.

left

Left camera calibration data.

Type:

MonoCalibrationData

right

Right camera calibration data.

Type:

MonoCalibrationData

R1

Rectification transform for the left camera.

Type:

np.ndarray

R2

Rectification transform for the right camera.

Type:

np.ndarray

T1

Projection matrix for the left camera.

Type:

np.ndarray

T2

Projection matrix for the right camera.

Type:

np.ndarray

H_left

Homography matrix for the left camera.

Type:

np.ndarray

H_right

Homography matrix for the right camera.

Type:

np.ndarray

l2r_extrinsic

Extrinsic matrix from the left to the right camera.

Type:

np.ndarray

r2l_extrinsic

Extrinsic matrix from the right to the left camera.

Type:

np.ndarray

Q_left

Q matrix for the left camera.

Type:

np.ndarray

Q_right

Q matrix for the right camera.

Type:

np.ndarray

baseline

Baseline between the two cameras (in meters).

Type:

float

primary

Primary camera calibration data.

Type:

Optional[MonoCalibrationData], optional

Q_primary

Q matrix for the primary camera.

Type:

Optional[np.ndarray], optional

Q_cv2

Q matrix generated by cv2.stereoRectify.

Type:

Optional[np.ndarray], optional

R1_cv2

R1 matrix generated by cv2.stereoRectify.

Type:

Optional[np.ndarray], optional

R2_cv2

R2 matrix generated by cv2.stereoRectify.

Type:

Optional[np.ndarray], optional

P1

P1 matrix generated by cv2.stereoRectify.

Type:

Optional[np.ndarray], optional

P2

P2 matrix generated by cv2.stereoRectify.

Type:

Optional[np.ndarray], optional

valid_region_primary

Valid region of the primary camera.

Type:

Optional[Tuple[int, int, int, int]], optional

pinhole_primary

o3d pinhole camera intrinsic for the primary camera.

Type:

Optional[PinholeCameraIntrinsic], optional

See also

MonoCalibrationData

Class to store calibration data for a mono camera.

H_left: np.ndarray
H_right: np.ndarray
P1: np.ndarray | None = None
P2: np.ndarray | None = None
Q_cv2: np.ndarray | None = None
Q_left: np.ndarray
Q_primary: np.ndarray | None = None
Q_right: np.ndarray
R1: np.ndarray
R1_cv2: np.ndarray | None = None
R2: np.ndarray
R2_cv2: np.ndarray | None = None
T1: np.ndarray
T2: np.ndarray
baseline: float
l2r_extrinsic: np.ndarray
left: MonoCalibrationData
pinhole_primary: PinholeCameraIntrinsic | None = None
primary: MonoCalibrationData | None = None
r2l_extrinsic: np.ndarray
right: MonoCalibrationData
valid_region_primary: tuple[int, int, int, int] | None = None
oakutils.calibration.create_q_matrix(fx: float, fy: float, cx: float, cy: float, baseline: float) → ndarray

Use to create Q matrix for stereo depth map.

Parameters:

  • fx (float) – Focal length in x direction (in millimeters).

  • fy (float) – Focal length in y direction (in millimeters).

  • cx (float) – Principal point in x direction.

  • cy (float) – Principal point in y direction.

  • baseline (float) – Baseline distance between the left and right camera (in meters).

Returns:

Q matrix for stereo depth map.

Return type:

np.ndarray

Note

This uses the OpenCV formula for Q matrix, with an alpha value of 0. Thus, the Q matrix is: [[1, 0, 0, -cx], [0, 1, 0, -cy], [0, 0, 0, (fx + fy) / 2], [0, 0, -1 / baseline, 0]]

oakutils.calibration.get_camera_calibration(rgb_size: tuple[int, int] | None = None, mono_size: tuple[int, int] | None = None, device: dai.DeviceBase | None = None, *, basic: bool | None = None, is_primary_mono_left: bool | None = None) → CalibrationData | ColorCalibrationData

Get the calibration of your OAK camera.

This function will return a different calibration datatype depending on which model camera is being used. OAK-D devices will return a CalibrationData object. OAK-1 devices will return a ColorCalibrationData object.

Parameters:

  • device (Optional[dai.DeviceBase], optional) – DepthAI device object.

  • rgb_size (Optional[Tuple[int, int]], optional) – RGB camera resolution.

  • mono_size (Optional[Tuple[int, int]], optional) – Mono camera resolution.

  • basic (Optional[bool], optional) – Whether to get basic calibration data. If True, no computed data will be included such as cv2.remap or o3d.PinholeCameraIntrinsic (even if o3d is available).

  • is_primary_mono_left (Optional[bool], optional) – Whether the primary mono camera is the left camera. This parameter is only used for cameras with depth imaging onboard.

Returns:

Object containing all the calibration data.

Return type:

CalibrationData | ColorCalibrationData

Raises:

ValueError – If the camera type is not recognized. If the RGB size is not provided for OAK-1 devices. If the RGB and mono sizes are not provided for OAK-D devices.

oakutils.calibration.get_oak1_calibration(rgb_size: tuple[int, int], device: DeviceBase | None = None) → ColorCalibrationData

Use to create the full ColorCalibrationData object.

This includes the calibration data for the RGB camera.

Note

Requires available OAK device. If device is not provided, dai.Device() will be used.

Parameters:

  • rgb_size (Tuple[int, int]) – RGB camera resolution.

  • device (Optional[dai.Device], optional) – DepthAI device object.

Returns:

Object containing all the calibration data.

Return type:

ColorCalibrationData

oakutils.calibration.get_oak1_calibration_basic(device: DeviceBase | None = None, rgb_size: tuple[int, int] = (1920, 1080)) → ColorCalibrationData

Use to get camera calibration data from OAK-1 device.

Note

Requires available OAK device. If device is not provided, dai.Device() will be used.

Parameters:

  • device (Optional[dai.Device], optional) – DepthAI device object.

  • rgb_size (Tuple[int, int], optional) – RGB camera resolution.

Returns:

Object containing all the calibration data.

Return type:

CalibrationData

oakutils.calibration.get_oakd_calibration(rgb_size: tuple[int, int], mono_size: tuple[int, int], device: DeviceBase | None = None, *, is_primary_mono_left: bool | None = None) → CalibrationData

Use to create the full CalibrationData object.

This includes the calibration data for the RGB camera, the left mono camera, the right mono camera, and the primary mono camera. As well as all OpenCV and Open3D compatible data for each camera.

Parameters:

  • rgb_size (Tuple[int, int]) – RGB camera resolution.

  • mono_size (Tuple[int, int]) – Mono camera resolution.

  • is_primary_mono_left (Optional[bool], optional) – Whether the primary mono camera is the left or right mono camera. Defaults to True.

  • device (Optional[dai.Device], optional) – DepthAI device object.

Returns:

Object containing all the calibration data.

Return type:

CalibrationData

Raises:

RuntimeError – If the mono cameras cannot be found.

oakutils.calibration.get_oakd_calibration_basic(device: DeviceBase | None = None, rgb_size: tuple[int, int] = (1920, 1080), mono_size: tuple[int, int] = (640, 400)) → CalibrationData

Use to get camera calibration data from OAK-D device.

Note

Requires available OAK device. If device is not provided, dai.Device() will be used.

Parameters:

  • device (Optional[dai.Device], optional) – DepthAI device object.

  • rgb_size (Tuple[int, int], optional) – RGB camera resolution.

  • mono_size (Tuple[int, int], optional) – Mono camera resolution.

Returns:

Object containing all the calibration data.

Return type:

CalibrationData

oakutils.calibration.get_oakd_calibration_primary_mono(device: DeviceBase | None = None, rgb_size: tuple[int, int] = (1920, 1080), mono_size: tuple[int, int] = (640, 400), *, is_primary_mono_left: bool | None = None) → CalibrationData

Use to get the calibration data for both RGB and mono cameras and primary mono camera.

Note

Requires available OAK device.

Parameters:

  • device (Optional[dai.Device], optional) – DepthAI device object.

  • rgb_size (Tuple[int, int], optional) – RGB camera resolution.

  • mono_size (Tuple[int, int], optional) – Mono camera resolution.

  • is_primary_mono_left (bool, optional) – Whether the primary mono camera is the left or right mono camera.

Returns:

Object containing all the calibration data.

Return type:

CalibrationData