opentl::modalities::DataFusion Class Reference

Static data fusion class. It inherits from the modalities:Modality interface, so it can be flexibly combined with the features. ASSUMPTIONS:

• the children output levels of one DataFusion class instance must be of the same level E.g. if you want to fuse the output of two pixel level children and one feature level child you have to cascade two DataFusion classes. DF1 instance fuses pixel level to feature level and DF2 instance fuses the output of DF1 with the feature level child.

More...

Inherits opentl::modalities::Modality.

List of all members.

Public Types
Public Member Functions
virtual modalities::Modality *	clone () const
	Deep cloning.
	DataFusion (const DataFusion &c)
	Copy Constructor.
	DataFusion (opentl::modelprojection::Warp *warp, int camIdx)
virtual void	init ()
	Initialize class (clean parameter status).
virtual int	matchFeatLevel (const TargetPtrVector &targets, T_MEAS_FEATPtrVector &outputMeas, std::size_t partitionIdx)
	Matching on feature level match projected model features with detected image features.
virtual int	matchObjLevel (const TargetPtrVector &targets, T_MEAS_OBJPtrVector &outputMeas, std::size_t partitionIdx)
	Matching on object level compute a local, maximum-likelihood estimate of pose (evtl. state), using either pixel- or feature-level likelihood functions.
virtual int	matchPixLevel (const TargetPtrVector &targets, T_MEAS_PIXPtrVector &outputMeas, std::size_t partitionIdx)
	Here "MATCH" = "FUSION"!
virtual int	preProcess (const opentl::core::cvdata::Image &image)
	Model independent pre processing.
virtual int	sampleModelFeatures (const TargetPtrVector &targets)
	Sample model features for all children.
virtual int	updateModelFeatures (const TargetPtrVector &targets)
	Update online features for all children.
int	upgradeObjectLevel (const TargetPtrVector &targets, T_MEAS_OBJPtrVector &outputMeas, std::size_t partitionIdx)
	Static data fusion: Upgrade from multiple levels to object level Method: Gauss-Newton optimization (LSE), using Jacobians (H).
virtual	~DataFusion ()
	Destructor.

---

Detailed Description

Static data fusion class. It inherits from the modalities:Modality interface, so it can be flexibly combined with the features. ASSUMPTIONS:

• the children output levels of one DataFusion class instance must be of the same level E.g. if you want to fuse the output of two pixel level children and one feature level child you have to cascade two DataFusion classes. DF1 instance fuses pixel level to feature level and DF2 instance fuses the output of DF1 with the feature level child.

---

Member Enumeration Documentation

enum opentl::modalities::DataFusion::OfflineParameters

Common flags to all modalities: they specify which field will be filled in the output measurement, inside the matchXLevel() function. Default values = all true.

Enumerator:

OFFLINE_COUNT

Reimplemented from opentl::modalities::Modality.

enum opentl::modalities::DataFusion::OnlineParameters

Enumerator:

fusionAlgo	Data fusion modality (scheme for static fusion, input/output levels for z).
fusionAlgoObjectUpgrade_Epsilon	Exit condition: minimum error for one delta step.
fusionAlgoObjectUpgrade_CovRVector	vector containing the covariance output values TODO: use math::Vector
fusionAlgoObjectUpgrade_EnableMultiThreadingPerChildFeature	create a thread per feature matchxxxLevel() function call
fusionAlgoObjectUpgrade_MaxIter	Maximum iterations performed for Gauss-Newton.
fusionAlgoObjectUpgrade_usePrior	Whether to use the prediction (prior) Gaussian to regularize Gauss-Newton optimization.
fusionAlgoObjectUpgrade_GaussNewton_weights	For w.a. we need a weight for each modality and camera.
fusionAlgoObjectUpgrade_GaussNewton_enableDampingFactor	flag to enable damping factor in case of a singular matrix
fusionAlgoObjectUpgrade_GaussNewton_useRinv	Whether to use the invert covariance in the Gauss-Newton algorithm NOTE: it makes sense only if the measurements have a variable covariance.
fusionAlgoObjectUpgrade_GaussNewton_dampingFactorDeterminantMinThreshold	min. threshold of Hessian determinant value for enabling damping
fusionAlgoObjectUpgrade_GaussNewton_dampingFactorMax	maximum damping factor
ONLINE_COUNT

Reimplemented from opentl::modalities::Modality.

enum opentl::modalities::DataFusion::T_FUSION_ALGO

Possible static fusion schemes.

Enumerator:

NOFUSION
PIXEL_WA
PIXEL_VOTING
PIXEL_FUZZY
OBJECT_UPGRADE

enum opentl::modalities::DataFusion::VotingMode

Possible voting schemes for pixel-level.

Enumerator:

UNANIMITY
BYZANTINE
MAJORITY
MOUTN

---

Constructor & Destructor Documentation

opentl::modalities::DataFusion::DataFusion	(	opentl::modelprojection::Warp *	warp,
		int	camIdx
	)

opentl::modalities::DataFusion::DataFusion

(

const DataFusion &

c

)

Copy Constructor.

virtual opentl::modalities::DataFusion::~DataFusion

(

)

[virtual]

Destructor.

---

Member Function Documentation

virtual modalities::Modality* opentl::modalities::DataFusion::clone

(

)

const [virtual]

Deep cloning.

Implements opentl::modalities::Modality.

virtual void opentl::modalities::DataFusion::init

(

)

[virtual]

Initialize class (clean parameter status).

Reimplemented from opentl::modalities::Modality.

virtual int opentl::modalities::DataFusion::matchFeatLevel	(	const TargetPtrVector &	targets,
		T_MEAS_FEATPtrVector &	outputMeas,
		std::size_t	partitionIdx
	)			[virtual]

Matching on feature level

• match projected model features with detected image features.

Parameters:

	states	Predicted state
	outputMeas	Feature-space measurements and residuals

Reimplemented from opentl::modalities::Modality.

virtual int opentl::modalities::DataFusion::matchObjLevel	(	const TargetPtrVector &	targets,
		T_MEAS_OBJPtrVector &	outputMeas,
		std::size_t	partitionIdx
	)			[virtual]

Matching on object level

• compute a local, maximum-likelihood estimate of pose (evtl. state), using either pixel- or feature-level likelihood functions.

Parameters:

	states	Predicted state
	outputMeas	State-space measurement and residuals (expected state = prediction, observed state = ML estimate)

Reimplemented from opentl::modalities::Modality.

virtual int opentl::modalities::DataFusion::matchPixLevel	(	const TargetPtrVector &	targets,
		T_MEAS_PIXPtrVector &	outputMeas,
		std::size_t	partitionIdx
	)			[virtual]

Here "MATCH" = "FUSION"!

Reimplemented from opentl::modalities::Modality.

virtual int opentl::modalities::DataFusion::preProcess

(

const opentl::core::cvdata::Image &

image

)

[virtual]

Model independent pre processing.

virtual int opentl::modalities::DataFusion::sampleModelFeatures

(

const TargetPtrVector &

targets

)

[virtual]

Sample model features for all children.

Reimplemented from opentl::modalities::Modality.

virtual int opentl::modalities::DataFusion::updateModelFeatures

(

const TargetPtrVector &

targets

)

[virtual]

Update online features for all children.

Reimplemented from opentl::modalities::Modality.

int opentl::modalities::DataFusion::upgradeObjectLevel	(	const TargetPtrVector &	targets,
		T_MEAS_OBJPtrVector &	outputMeas,
		std::size_t	partitionIdx
	)

Static data fusion: Upgrade from multiple levels to object level Method: Gauss-Newton optimization (LSE), using Jacobians (H).

Parameters:

	targets	Targets with initial states (for the moment only 1 state-per-target is supported!)
	inMeas	Lower-level measurements (typically features) to be combined
	outMeas	ObjModel-level measurement (in pose space) NOTE: This function calls the individual match() functions for each camera/modality and afterwards computes a GN optimization in order to update each object pose. The final update poses are returned in the output measurements.