vpRBVisualOdometryUtils Class Reference

#include <vpRBVisualOdometryUtils.h>

Static Public Member Functions
static std::pair< std::vector< unsigned int >, std::vector< unsigned int > >	computeIndicesObjectAndEnvironment (const vpMatrix &keypoints, const vpRBFeatureTrackerInput &frame, double minMaskConfidence, double minEdgeDistObject, double minEdgeDistEnv)
static void	levenbergMarquardtKeypoints2D (const vpMatrix &points3d, const vpMatrix &observations, const vpLevenbergMarquardtParameters &parameters, vpHomogeneousMatrix &cTw)
static void	levenbergMarquardtKeypoints3D (const vpMatrix &points3d, const vpMatrix &observations, const vpLevenbergMarquardtParameters &parameters, vpHomogeneousMatrix &cTw)

Detailed Description

Tutorials & Examples

Tutorials
If you want to have an in-depth presentation of the Render-Based Tracker (RBT), you may have a look at:

• Tutorial: Tracking with the Render-Based Tracker

Definition at line 63 of file vpRBVisualOdometryUtils.h.

Member Function Documentation

computeIndicesObjectAndEnvironment()

BEGIN_VISP_NAMESPACE std::pair< std::vector< unsigned int >, std::vector< unsigned int > > vpRBVisualOdometryUtils::computeIndicesObjectAndEnvironment ( const vpMatrix & keypoints, const vpRBFeatureTrackerInput & frame, double minMaskConfidence, double minEdgeDistObject, double minEdgeDistEnv )

static

Compute the indices of the keypoints that belong to either the object (the renderered depth is greater than 0) or the environment Keypoints that are too close to the silhouette edges are discarded. There are two thresholds, one for object candidates and another for environment candidates Points can also be filtered with the probability mask, if it is available and minMaskConfidence is greater than 0.

Parameters

keypoints	the keypoints that should be split between object and environment
frame	Input frame data, containing the renders
minMaskConfidence	The minimum mask confidence value to consider a point has belonging to the object. If 0, it is unused
minEdgeDistObject	Minimum euclidean distance between an object keypoint and the object silhouette's points to consider the keypoint as valid
minEdgeDistEnv	Minimum euclidean distance between an environment keypoint and the object silhouette's points to consider the keypoint as a valid one

Returns

std::pair<std::vector<unsigned int>, std::vector<unsigned int>>

Definition at line 46 of file vpRBVisualOdometryUtils.cpp.

References vpRBRenderData::depth, vpImage< Type >::getHeight(), vpArray2D< Type >::getRows(), vpImage< Type >::getWidth(), vpRBFeatureTrackerInput::hasMask(), vpRBFeatureTrackerInput::mask, vpRBFeatureTrackerInput::renders, vpRBFeatureTrackerInput::silhouettePoints, and vpMath::sqr().

levenbergMarquardtKeypoints2D()

void vpRBVisualOdometryUtils::levenbergMarquardtKeypoints2D ( const vpMatrix & points3d, const vpMatrix & observations, const vpLevenbergMarquardtParameters & parameters, vpHomogeneousMatrix & cTw )

static

Definition at line 113 of file vpRBVisualOdometryUtils.cpp.

References vpException::dimensionError, vpExponentialMap::direct(), vpMatrix::eye(), vpLevenbergMarquardtParameters::gain, vpArray2D< Type >::getCols(), vpArray2D< Type >::getRows(), vpHomogeneousMatrix::inverse(), vpLevenbergMarquardtParameters::maxNumIters, vpRobust::MEstimator(), vpLevenbergMarquardtParameters::minImprovementFactor, vpLevenbergMarquardtParameters::muInit, vpLevenbergMarquardtParameters::muIterFactor, vpMatrix::mult2Matrices(), vpMatrix::t(), and vpRobust::TUKEY.

levenbergMarquardtKeypoints3D()

void vpRBVisualOdometryUtils::levenbergMarquardtKeypoints3D ( const vpMatrix & points3d, const vpMatrix & observations, const vpLevenbergMarquardtParameters & parameters, vpHomogeneousMatrix & cTw )

static

Definition at line 186 of file vpRBVisualOdometryUtils.cpp.

References vpException::dimensionError, vpExponentialMap::direct(), vpMatrix::eye(), vpLevenbergMarquardtParameters::gain, vpArray2D< Type >::getCols(), vpArray2D< Type >::getRows(), vpHomogeneousMatrix::inverse(), vpLevenbergMarquardtParameters::maxNumIters, vpRobust::MEstimator(), vpLevenbergMarquardtParameters::minImprovementFactor, vpLevenbergMarquardtParameters::muInit, vpLevenbergMarquardtParameters::muIterFactor, vpMatrix::mult2Matrices(), vpMatrix::t(), and vpRobust::TUKEY.