Class libCalibGeom::Calibration

Main interface to specify the calibration structure and run optimization.

The class is responsible for holding the calibration data and initializing/optimizing all parameters.

The typical workflow consists of: adding the appropriate number of empty Camera, Pose and Target objects. collecting Detection objects which reference specific cameras, targets and poses. calling initialize() calling optimize()

Constructors and destructors

Calibration()

Calibration()

Returns:

void

Public member functions

addCamera()

size_taddCamera(const Camera & c)

Adds a camera to the calibration setup.

Parameters:

c	The Camera object to add.

Returns:

size_t

The index of the added camera.

addRobot()

size_taddRobot(const Robot & r = Robot())

Adds a robot to the calibration setup.

Parameters:

r	The Robot object to add.

Returns:

size_t

The index of the added robot.

addLidar()

size_taddLidar(const Lidar & l = Lidar())

Adds a lidar to the calibration setup.

Parameters:

l	The Lidar object to add.

Returns:

size_t

The index of the added lidar.

addPose()

size_taddPose(const Pose & p = Pose())

Adds a pose to the calibration setup.

Parameters:

p	The Pose object to add.

Returns:

size_t

The index of the added pose.

addTarget()

size_taddTarget(const Target & t = Target())

Adds a target to the calibration setup.

Parameters:

t	The Target object to add.

Returns:

size_t

The index of the added target.

addDistanceConstraint()

size_taddDistanceConstraint(const DistanceConstraint & dc = DistanceConstraint())

Adds a distance constraint to the calibration setup.

Parameters:

dc	The DistanceConstraint object to add.

Returns:

size_t

The index of the added constraint.

checkConsistency()

boolcheckConsistency(std::string &errorString) const

Checks if the calibration setup is consistent.

Perform various checks on the current calibration data to ensure that all entities are referenced correctly, such that initialization and calibration will be possible.

Parameters:

errorString

Populated with an error message if the check fails.

Returns:

bool	True if the setup is consistent, false otherwise.

reset()

voidreset()

Resets the calibration setup and deletes all owned Camera, Pose, Target, Detection and DistanceConstrain objects.

Returns:

void

initialize()

InitializationResultinitialize(const std::vector< ImageDetection > & imageDetections,const std::vector< PointCloudDetection > & pointCloudDetections,const std::vector< PoseRecord > & robotPoses,const InitializationSettings & settings = InitializationSettings())

Initialize the intrinsic and extrinsic parameters.

Initializes parameters for all devices in the current calibration setup, based on the provided detections.

Parameters:

imageDetections	Detection data from images.
pointCloudDetections	Detection data from point clouds (lidars).
robotPoses	Robot pose records.
settings	Initialization settings.

Returns:

InitializationResult

Result of the initialization.

initialize()

InitializationResultinitialize(const std::vector< ImageDetection > & imageDetections,const InitializationSettings & settings = InitializationSettings())

Initialize camera parameters (convenience overload).

Parameters:

imageDetections	Detection data from images.
settings	Initialization settings.

Returns:

InitializationResult

Result of the initialization.

optimize()

OptimizationResultoptimize(const std::vector< ImageDetection > & imageDetections,const std::vector< PointCloudDetection > & pointCloudDetections,const std::vector< PoseRecord > & robotPoses,const OptimizationSettings & settings = OptimizationSettings(),std::function< bool(const int, const double)> iterationCallback = nullptr)

Run bundle adjustment optimization.

Run bundle block adjustment on the provided detections. Requires that the Calibration is initialized first.

Parameters:

imageDetections	Detection data from images.
pointCloudDetections	Detection data from point clouds.
robotPoses	Robot pose records.
settings	Optimization settings.
iterationCallback	Optional callback function called at each iteration.

Returns:

OptimizationResult

Result of the optimization.

optimize()

OptimizationResultoptimize(const std::vector< ImageDetection > & imageDetections,const OptimizationSettings & settings = OptimizationSettings(),std::function< bool(const int, const double)> iterationCallback = nullptr)

Run bundle adjustment optimization (convenience overload).

Parameters:

imageDetections	Detection data from images.
settings	Optimization settings.
iterationCallback	Optional callback function called at each iteration.

Returns:

OptimizationResult

Result of the optimization.

estimateCovariance()

CovarianceResultestimateCovariance(const std::vector< ImageDetection > &detections) const

Estimate variances/covariances of the model parameters.

Parameters:

detections

Image detections used for covariance estimation.

Returns:

CovarianceResult

The estimated covariance result.

estimateProjectionError()

boolestimateProjectionError(const CovarianceResult &cov, std::vector< double > &rms, std::vector< double > &max, std::vector< std::vector< std::pair< libCalibGeom::Point2, libCalibGeom::PixelCovariance >>> &pce) const

Estimate projection errors.

Propagates unprojected 3d points through the calibration covariance of each camera individually. Returns 2D pixel covariances for the projected points evenly sampled on the image planes.

Parameters:

cov	The covariance result used for estimation.
rms	RMS error for each camera.
max	Max error for each camera.
pce	Vector of point/covariance pairs for visualization.

Returns:

bool	True if successful.

estimateTriangulationError()

boolestimateTriangulationError(const CovarianceResult &cov, double &rms, double &max, std::vector< std::pair< Point3, PointCovariance >> &pc, std::vector< std::array< Point3, 100 >> &errors, const double planeOffset=0.0, const double featureNoise=0.0) const

Estimate triangulation errors.

Propagates 3d points through the calibration covariance structure. Samples points evenly on a plane in the volume occupied by calibration object points. The plane is aligned to the cameras' mean optical axis and mean x-axis.

Parameters:

cov	The covariance result used for estimation.
rms	RMS error.
max	Max error.
pc	Point clouds with covariances.
errors	Error vectors.
planeOffset	[m] Offsets the sample plane along the plane normal.
featureNoise	[px] Specifies the standard deviation of error that should be assumed on the detection of points in camera images.

Returns:

bool	True if successful.

computeNextBestPose()

PosecomputeNextBestPose(const std::vector< ImageDetection > &detections) const

Compute optimal next target pose.

Calculates the next best pose for the calibration target given the current calibration and detections.

Parameters:

detections

Current image detections.

Returns:

Pose	The recommended next pose.

Public data members

cameras

std::vector< Camera >cameras

No documentation provided.

robots

std::vector< Robot >robots

No documentation provided.

lidars

std::vector< Lidar >lidars

No documentation provided.

poses

std::vector< Pose >poses

Roto-translation transforms moving from the calibration target rig frame to the camera rig frame at a specific configuration.

targets

std::vector< Target >targets

Optimized target geometries.

distanceConstraints

std::vector< DistanceConstraint >distanceConstraints

Soft constraints for absolute distances between specified features.

isInitialized

boolisInitialized

No documentation provided.

isOptimized

boolisOptimized

No documentation provided.