Detailed Description

Declares a class derived from "CObservation" that encapsules a 3D range scan measurement (e.g.

from a time of flight range camera). This kind of observations can carry one or more of these data fields:

3D point cloud (as float's).
2D range image (as a matrix): Each entry in the matrix "rangeImage(ROW,COLUMN)" contains a distance or a depth (in meters), depending on range_is_depth.
2D intensity (grayscale or RGB) image (as a mrpt::utils::CImage): For SwissRanger cameras, a logarithmic A-law compression is used to convert the original 16bit intensity to a more standard 8bit graylevel.
2D confidence image (as a mrpt::utils::CImage): For each pixel, a 0x00 and a 0xFF mean the lowest and highest confidence levels, respectively.

The coordinates of the 3D point cloud are in meters with respect to the depth camera origin of coordinates (in SwissRanger, the front face of the camera: a small offset ~1cm in front of the physical focal point), with the +X axis pointing forward, +Y pointing left-hand and +Z pointing up. By convention, a 3D point with its coordinates set to (0,0,0), will be considered as invalid. The field CObservation3DRangeScan::relativePoseIntensityWRTDepth describes the change of coordinates from the depth camera to the intensity (RGB or grayscale) camera. In a SwissRanger camera both cameras coincide, so this pose is just a rotation (0,0,0,-90deg,0,-90deg). But in Microsoft Kinect there is also an offset, as shown in this figure:

In any case, check the field relativePoseIntensityWRTDepth, or the method doDepthAndIntensityCamerasCoincide() to determine if both frames of reference coincide, since even for Kinect cameras both can coincide if the images have been rectified.

The 2D images and matrices are stored as common images, with an up->down rows order and left->right, as usual. Optionally, the intensity and confidence channels can be set to delayed-load images for off-rawlog storage so it saves memory by having loaded in memory just the needed images. See the methods load() and unload(). Due to the intensive storage requirements of this kind of observations, this observation is the only one in MRPT for which it's recommended to always call "load()" and "unload()" before and after using the observation, ONLY when the observation was read from a rawlog dataset, in order to make sure that all the externally stored data fields are loaded and ready in memory.

Classes that grab observations of this type are:

There are two sets of calibration parameters (see mrpt::vision::checkerBoardStereoCalibration() or the ready-to-use GUI program kinect-calibrate):

cameraParams: Projection parameters of the depth camera.
cameraParamsIntensity: Projection parameters of the intensity (gray-level or RGB) camera.

In some cameras, like SwissRanger, both are the same. It is possible in Kinect to rectify the range images such both cameras seem to coincide and then both sets of camera parameters will be identical.

Range data can be interpreted in two different ways depending on the 3D camera (this field is already set to the correct setting when grabbing observations from an mrpt::hwdrivers sensor):

range_is_depth=true -> Kinect-like ranges: entries of rangeImage are distances along the +X axis
range_is_depth=false -> Ranges in rangeImage are actual distances in 3D.

The "intensity" channel may come from different channels in sesnsors as Kinect. Look at field intensityImageChannel to find out if the image was grabbed from the visible (RGB) or IR channels.

3D point clouds can be generated at any moment after grabbing with CObservation3DRangeScan::project3DPointsFromDepthImage() and CObservation3DRangeScan::project3DPointsFromDepthImageInto(), provided the correct calibration parameters.

Note: Starting at serialization version 2 (MRPT 0.9.1+), the confidence channel is stored as an image instead of a matrix to optimize memory and disk space.; Starting at serialization version 3 (MRPT 0.9.1+), the 3D point cloud and the rangeImage can both be stored externally to save rawlog space.; Starting at serialization version 5 (MRPT 0.9.5+), the new field range_is_depth; Starting at serialization version 6 (MRPT 0.9.5+), the new field intensityImageChannel

See also: mrpt::hwdrivers::CSwissRanger3DCamera, mrpt::hwdrivers::CKinect, CObservation

Definition at line 98 of file obs/CObservation3DRangeScan.h.

#include <mrpt/obs/CObservation3DRangeScan.h>

Inheritance diagram for mrpt::obs::CObservation3DRangeScan:

[legend]

Classes
struct	TCached3DProjTables
	Look-up-table struct for project3DPointsFromDepthImageInto() More...

Public Types
enum	TIntensityChannelID { CH_VISIBLE = 0, CH_IR = 1 }
	Enum type for intensityImageChannel. More...

Public Member Functions
void *	operator new (size_t size)

void *	operator new[] (size_t size)

void	operator delete (void *ptr) throw ()

void	operator delete[] (void *ptr) throw ()

void	operator delete (void memory, void ptr) throw ()

void *	operator new (size_t size, const std::nothrow_t &) throw ()

void	operator delete (void *ptr, const std::nothrow_t &) throw ()

	CObservation3DRangeScan ()
	Default constructor. More...

virtual	~CObservation3DRangeScan ()
	Destructor. More...

template<class POINTMAP >
void	project3DPointsFromDepthImageInto (POINTMAP &dest_pointcloud, const bool takeIntoAccountSensorPoseOnRobot, const mrpt::poses::CPose3D *robotPoseInTheWorld=NULL, const bool PROJ3D_USE_LUT=true)
	Project the RGB+D images into a 3D point cloud (with color if the target map supports it) and optionally at a given 3D pose. More...

void	project3DPointsFromDepthImage (const bool PROJ3D_USE_LUT=true)
	This method is equivalent to `project3DPointsFromDepthImageInto()` storing the projected 3D points (without color, in local coordinates) in this same class. More...

void	convertTo2DScan (mrpt::obs::CObservation2DRangeScan &out_scan2d, const std::string &sensorLabel, const double angle_sup=mrpt::utils::DEG2RAD(5), const double angle_inf=mrpt::utils::DEG2RAD(5), const double oversampling_ratio=1.2)
	Convert this 3D observation into an "equivalent 2D fake laser scan", with a configurable vertical FOV. More...

void	resizePoints3DVectors (const size_t nPoints)
	Use this method instead of resizing all three points3D_x, points3D_y & points3D_z to allow the usage of the internal memory pool. More...

bool	points3D_isExternallyStored () const

std::string	points3D_getExternalStorageFile () const

void	points3D_getExternalStorageFileAbsolutePath (std::string &out_path) const

std::string	points3D_getExternalStorageFileAbsolutePath () const

void	points3D_convertToExternalStorage (const std::string &fileName, const std::string &use_this_base_dir)
	Users won't normally want to call this, it's only used from internal MRPT programs. More...

void	rangeImage_setSize (const int HEIGHT, const int WIDTH)
	Similar to calling "rangeImage.setSize(H,W)" but this method provides memory pooling to speed-up the memory allocation. More...

bool	rangeImage_isExternallyStored () const

std::string	rangeImage_getExternalStorageFile () const

void	rangeImage_getExternalStorageFileAbsolutePath (std::string &out_path) const

std::string	rangeImage_getExternalStorageFileAbsolutePath () const

void	rangeImage_convertToExternalStorage (const std::string &fileName, const std::string &use_this_base_dir)
	Users won't normally want to call this, it's only used from internal MRPT programs. More...

void	rangeImage_forceResetExternalStorage ()
	Forces marking this observation as non-externally stored - it doesn't anything else apart from reseting the corresponding flag (Users won't normally want to call this, it's only used from internal MRPT programs) More...

bool	doDepthAndIntensityCamerasCoincide () const
	Return true if relativePoseIntensityWRTDepth equals the pure rotation (0,0,0,-90deg,0,-90deg) (with a small comparison epsilon) More...

void	getSensorPose (mrpt::poses::CPose3D &out_sensorPose) const
	A general method to retrieve the sensor pose on the robot. More...

void	setSensorPose (const mrpt::poses::CPose3D &newSensorPose)
	A general method to change the sensor pose on the robot. More...

virtual void	getDescriptionAsText (std::ostream &o) const
	Build a detailed, multi-line textual description of the observation contents and dump it to the output stream. More...

void	swap (CObservation3DRangeScan &o)
	Very efficient method to swap the contents of two observations. More...

void	getZoneAsObs (CObservation3DRangeScan &obs, const unsigned int &r1, const unsigned int &r2, const unsigned int &c1, const unsigned int &c2)

template<class METRICMAP >
bool	insertObservationInto (METRICMAP theMap, const mrpt::poses::CPose3D robotPose=NULL) const
	This method is equivalent to: More...

void	getSensorPose (mrpt::math::TPose3D &out_sensorPose) const
	A general method to retrieve the sensor pose on the robot. More...

void	setSensorPose (const mrpt::math::TPose3D &newSensorPose)
	A general method to change the sensor pose on the robot. More...

mrpt::utils::CObjectPtr	duplicateGetSmartPtr () const
	Returns a copy of the object, indepently of its class, as a smart pointer (the newly created object will exist as long as any copy of this smart pointer). More...

CObject *	clone () const
	Cloning interface for smart pointers. More...

Delayed-load manual control methods.
virtual void	load () const
	Makes sure all images and other fields which may be externally stored are loaded in memory. More...

virtual void	unload ()
	Unload all images, for the case they being delayed-load images stored in external files (othewise, has no effect). More...

Static Public Member Functions
static void *	operator new (size_t size, void *ptr)

static double	recoverCameraCalibrationParameters (const CObservation3DRangeScan &in_obs, mrpt::utils::TCamera &out_camParams, const double camera_offset=0.01)
	A Levenberg-Marquart-based optimizer to recover the calibration parameters of a 3D camera given a range (depth) image and the corresponding 3D point cloud. More...

Public Attributes
bool	hasPoints3D
	true means the field points3D contains valid data. More...

std::vector< float >	points3D_x
	If hasPoints3D=true, the X coordinates of the 3D point cloud detected by the camera. More...

std::vector< float >	points3D_y
	If hasPoints3D=true, the Y coordinates of the 3D point cloud detected by the camera. More...

std::vector< float >	points3D_z
	If hasPoints3D=true, the Z coordinates of the 3D point cloud detected by the camera. More...

bool	hasRangeImage
	true means the field rangeImage contains valid data More...

mrpt::math::CMatrix	rangeImage
	If hasRangeImage=true, a matrix of floats with the range data as captured by the camera (in meters) More...

bool	range_is_depth
	true: Kinect-like ranges: entries of rangeImage are distances along the +X axis; false: Ranges in rangeImage are actual distances in 3D. More...

bool	hasIntensityImage
	true means the field intensityImage contains valid data More...

mrpt::utils::CImage	intensityImage
	If hasIntensityImage=true, a color or gray-level intensity image of the same size than "rangeImage". More...

TIntensityChannelID	intensityImageChannel
	The source of the intensityImage; typically the visible channel. More...

bool	hasConfidenceImage
	true means the field confidenceImage contains valid data More...

mrpt::utils::CImage	confidenceImage
	If hasConfidenceImage=true, an image with the "confidence" value [range 0-255] as estimated by the capture drivers. More...

mrpt::utils::TCamera	cameraParams
	Projection parameters of the depth camera. More...

mrpt::utils::TCamera	cameraParamsIntensity
	Projection parameters of the intensity (graylevel or RGB) camera. More...

mrpt::poses::CPose3D	relativePoseIntensityWRTDepth
	Relative pose of the intensity camera wrt the depth camera (which is the coordinates origin for this observation). More...

float	maxRange
	The maximum range allowed by the device, in meters (e.g. 8.0m, 5.0m,...) More...

mrpt::poses::CPose3D	sensorPose
	The 6D pose of the sensor on the robot. More...

float	stdError
	The "sigma" error of the device in meters, used while inserting the scan in an occupancy grid. More...

Data common to any observation
mrpt::system::TTimeStamp	timestamp
	The associated time-stamp. More...

std::string	sensorLabel
	An arbitrary label that can be used to identify the sensor. More...

Static Public Attributes
static TCached3DProjTables	m_3dproj_lut
	3D point cloud projection look-up-table More...

static const mrpt::utils::TRuntimeClassId	classCObject

RTTI stuff
static const mrpt::utils::TRuntimeClassId	classCObservation

RTTI stuff
static const mrpt::utils::TRuntimeClassId	classCSerializable

Protected Member Functions
void	swap (CObservation &o)
	Swap with another observation, ONLY the data defined here in the base class CObservation. It's protected since it'll be only called from child classes that should know what else to swap appart from these common data. More...

CSerializable virtual methods
void	writeToStream (mrpt::utils::CStream &out, int *getVersion) const
	Introduces a pure virtual method responsible for writing to a CStream. More...

void	readFromStream (mrpt::utils::CStream &in, int version)
	Introduces a pure virtual method responsible for loading from a CStream This can not be used directly be users, instead use "stream >> object;" for reading it from a stream or "stream >> object_ptr;" if the class is unknown apriori. More...

Protected Attributes
bool	m_points3D_external_stored
	If set to true, m_points3D_external_file is valid. More...

std::string	m_points3D_external_file
	3D points are in CImage::IMAGES_PATH_BASE+<this_file_name> More...

bool	m_rangeImage_external_stored
	If set to true, m_rangeImage_external_file is valid. More...

std::string	m_rangeImage_external_file
	rangeImage is in CImage::IMAGES_PATH_BASE+<this_file_name> More...

RTTI stuff
typedef CObservation3DRangeScanPtr	SmartPtr

static mrpt::utils::CLASSINIT	_init_CObservation3DRangeScan

static mrpt::utils::TRuntimeClassId	classCObservation3DRangeScan

static const mrpt::utils::TRuntimeClassId *	classinfo

static const mrpt::utils::TRuntimeClassId *	_GetBaseClass ()

virtual const mrpt::utils::TRuntimeClassId *	GetRuntimeClass () const
	Returns information about the class of an object in runtime. More...

virtual mrpt::utils::CObject *	duplicate () const
	Returns a copy of the object, indepently of its class. More...

static mrpt::utils::CObject *	CreateObject ()

static CObservation3DRangeScanPtr	Create ()

Member Typedef Documentation

typedef CObservation3DRangeScanPtr mrpt::obs::CObservation3DRangeScan::SmartPtr

A typedef for the associated smart pointer

Definition at line 101 of file obs/CObservation3DRangeScan.h.

Constructor & Destructor Documentation

mrpt::obs::CObservation3DRangeScan::CObservation3DRangeScan ( )

Default constructor.

virtual mrpt::obs::CObservation3DRangeScan::~CObservation3DRangeScan ( )

virtual

Destructor.

Member Function Documentation

static const mrpt::utils::TRuntimeClassId* mrpt::obs::CObservation3DRangeScan::_GetBaseClass ( )

staticprotected

CObject* mrpt::utils::CObject::clone ( ) const

inlineinherited

Cloning interface for smart pointers.

Definition at line 139 of file CObject.h.

void mrpt::obs::CObservation3DRangeScan::convertTo2DScan	(	mrpt::obs::CObservation2DRangeScan &	out_scan2d,
		const std::string &	sensorLabel,
		const double	angle_sup = `mrpt::utils::DEG2RAD(5)`,
		const double	angle_inf = `mrpt::utils::DEG2RAD(5)`,
		const double	oversampling_ratio = `1.2`
	)

Convert this 3D observation into an "equivalent 2D fake laser scan", with a configurable vertical FOV.

The result is a 2D laser scan with more "rays" (N) than columns has the 3D observation (W), exactly: N = W * oversampling_ratio. This oversampling is required since laser scans sample the space at evenly-separated angles, while a range camera follows a tangent-like distribution. By oversampling we make sure we don't leave "gaps" unseen by the virtual "2D laser".

All obstacles within a frustum are considered and the minimum distance is kept in each direction. The horizontal FOV of the frustum is automatically computed from the intrinsic parameters, but the vertical FOV must be provided by the user, and can be set to be assymetric which may be useful depending on the zone of interest where to look for obstacles.

All spatial transformations are riguorosly taken into account in this class, using the depth camera intrinsic calibration parameters.

The timestamp of the new object is copied from the 3D object. Obviously, a requisite for calling this method is the 3D observation having range data, i.e. hasRangeImage must be true. It's not needed to have RGB data nor the raw 3D point clouds for this method to work.

Parameters

[out]	out_scan2d	The resulting 2D equivalent scan.
[in]	sensorLabel	The sensor label that will have the newly created observation.
[in]	angle_sup	(Default=5deg) The upper half-FOV angle (in radians)
[in]	angle_sup	(Default=5deg) The lower half-FOV angle (in radians)
[in]	oversampling_ratio	(Default=1.2=120%) How many more laser scans rays to create (read above).

See also: The example in http://www.mrpt.org/Example_Kinect_To_2D_laser_scan

static CObservation3DRangeScanPtr mrpt::obs::CObservation3DRangeScan::Create ( )

static

static mrpt::utils::CObject* mrpt::obs::CObservation3DRangeScan::CreateObject ( )

static

bool mrpt::obs::CObservation3DRangeScan::doDepthAndIntensityCamerasCoincide ( ) const

Return true if relativePoseIntensityWRTDepth equals the pure rotation (0,0,0,-90deg,0,-90deg) (with a small comparison epsilon)

See also: relativePoseIntensityWRTDepth

Referenced by mrpt::obs::detail::project3DPointsFromDepthImageInto().

virtual mrpt::utils::CObject* mrpt::obs::CObservation3DRangeScan::duplicate ( ) const

virtual

Returns a copy of the object, indepently of its class.

Implements mrpt::utils::CObject.

mrpt::utils::CObjectPtr mrpt::utils::CObject::duplicateGetSmartPtr ( ) const

inlineinherited

Returns a copy of the object, indepently of its class, as a smart pointer (the newly created object will exist as long as any copy of this smart pointer).

Definition at line 136 of file CObject.h.

virtual void mrpt::obs::CObservation3DRangeScan::getDescriptionAsText ( std::ostream & o ) const

virtual

Build a detailed, multi-line textual description of the observation contents and dump it to the output stream.

Note: If overried by derived classes, call base CObservation::getDescriptionAsText() first to show common information.; This is the text that appears in RawLogViewer when selecting an object in the dataset

Reimplemented from mrpt::obs::CObservation.

virtual const mrpt::utils::TRuntimeClassId* mrpt::obs::CObservation3DRangeScan::GetRuntimeClass ( ) const

virtual

Returns information about the class of an object in runtime.

Reimplemented from mrpt::obs::CObservation.

void mrpt::obs::CObservation::getSensorPose ( mrpt::math::TPose3D & out_sensorPose ) const

inherited

A general method to retrieve the sensor pose on the robot.

Note that most sensors will return a full (6D) CPose3D, but see the derived classes for more details or special cases.

See also: setSensorPose

void mrpt::obs::CObservation3DRangeScan::getSensorPose ( mrpt::poses::CPose3D & out_sensorPose ) const

inlinevirtual

A general method to retrieve the sensor pose on the robot.

Note that most sensors will return a full (6D) CPose3D, but see the derived classes for more details or special cases.

See also: setSensorPose

Implements mrpt::obs::CObservation.

Definition at line 293 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation3DRangeScan::getZoneAsObs	(	CObservation3DRangeScan &	obs,
		const unsigned int &	r1,
		const unsigned int &	r2,
		const unsigned int &	c1,
		const unsigned int &	c2
	)

template<class METRICMAP >

bool mrpt::obs::CObservation::insertObservationInto	(	METRICMAP *	theMap,
		const mrpt::poses::CPose3D *	robotPose = `NULL`
	)		const

inlineinherited

This method is equivalent to:

map->insertObservation(this, robotPose)

Parameters

theMap	The map where this observation is to be inserted: the map will be updated.
robotPose	The pose of the robot base for this observation, relative to the target metric map. Set to NULL (default) to use (0,0,0deg)

Returns: Returns true if the map has been updated, or false if this observations has nothing to do with a metric map (for example, a sound observation).

See also: CMetricMap, CMetricMap::insertObservation

Definition at line 91 of file obs/CObservation.h.

virtual void mrpt::obs::CObservation3DRangeScan::load ( ) const

virtual

Makes sure all images and other fields which may be externally stored are loaded in memory.

Note that for all CImages, calling load() is not required since the images will be automatically loaded upon first access, so load() shouldn't be needed to be called in normal cases by the user. If all the data were alredy loaded or this object has no externally stored data fields, calling this method has no effects.

See also: unload

Reimplemented from mrpt::obs::CObservation.

Referenced by mrpt::maps::COctoMapBase< OCTREE, OCTREE_NODE >::internal_build_PointCloud_for_observation().

void mrpt::obs::CObservation3DRangeScan::operator delete	(	void *	ptr	)
throw	(
	)

inline

Definition at line 101 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation3DRangeScan::operator delete	(	void *	memory,
		void *	ptr
	)
throw	(
	)

inline

Definition at line 101 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation3DRangeScan::operator delete	(	void *	ptr,
		const std::nothrow_t &
	)
throw	(
	)

inline

Definition at line 101 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation3DRangeScan::operator delete[]	(	void *	ptr	)
throw	(
	)

inline

Definition at line 101 of file obs/CObservation3DRangeScan.h.

static void* mrpt::obs::CObservation3DRangeScan::operator new	(	size_t	size,
		void *	ptr
	)

inlinestatic

Definition at line 101 of file obs/CObservation3DRangeScan.h.

void* mrpt::obs::CObservation3DRangeScan::operator new	(	size_t	size,
		const std::nothrow_t &
	)
throw	(
	)

inline

Definition at line 101 of file obs/CObservation3DRangeScan.h.

void* mrpt::obs::CObservation3DRangeScan::operator new ( size_t size )

inline

Definition at line 101 of file obs/CObservation3DRangeScan.h.

void* mrpt::obs::CObservation3DRangeScan::operator new[] ( size_t size )

inline

Definition at line 101 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation3DRangeScan::points3D_convertToExternalStorage	(	const std::string &	fileName,
		const std::string &	use_this_base_dir
	)

Users won't normally want to call this, it's only used from internal MRPT programs.

std::string mrpt::obs::CObservation3DRangeScan::points3D_getExternalStorageFile ( ) const

inline

Definition at line 228 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation3DRangeScan::points3D_getExternalStorageFileAbsolutePath ( std::string & out_path ) const

std::string mrpt::obs::CObservation3DRangeScan::points3D_getExternalStorageFileAbsolutePath ( ) const

inline

Definition at line 230 of file obs/CObservation3DRangeScan.h.

bool mrpt::obs::CObservation3DRangeScan::points3D_isExternallyStored ( ) const

inline

Definition at line 227 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation3DRangeScan::project3DPointsFromDepthImage ( const bool PROJ3D_USE_LUT = true )

inline

This method is equivalent to project3DPointsFromDepthImageInto() storing the projected 3D points (without color, in local coordinates) in this same class.

For new code it's recommended to use instead project3DPointsFromDepthImageInto() which is much more versatile.

Definition at line 178 of file obs/CObservation3DRangeScan.h.

References mrpt::obs::detail::project3DPointsFromDepthImageInto().

template<class POINTMAP >

void mrpt::obs::CObservation3DRangeScan::project3DPointsFromDepthImageInto	(	POINTMAP &	dest_pointcloud,
		const bool	takeIntoAccountSensorPoseOnRobot,
		const mrpt::poses::CPose3D *	robotPoseInTheWorld = `NULL`,
		const bool	PROJ3D_USE_LUT = `true`
	)

inline

Project the RGB+D images into a 3D point cloud (with color if the target map supports it) and optionally at a given 3D pose.

The 3D point coordinates are computed from the depth image (rangeImage) and the depth camera camera parameters (cameraParams). There exist two set of formulas for projecting the i'th point, depending on the value of "range_is_depth". In all formulas below, "rangeImage" is the matrix of ranges and the pixel coordinates are (r,c).

1) [range_is_depth=true] With "range equals depth" or "Kinect-like depth mode": the range values are in fact distances along the "+X" axis, not real 3D ranges (this is the way Kinect reports ranges):

x(i) = rangeImage(r,c)
y(i) = (r_cx - c) * x(i) / r_fx
z(i) = (r_cy - r) * x(i) / r_fy

2) [range_is_depth=false] With "normal ranges": range means distance in 3D. This must be set when processing data from the SwissRange 3D camera, among others.

Ky = (r_cx - c)/r_fx
Kz = (r_cy - r)/r_fy
x(i) = rangeImage(r,c) / sqrt( 1 + Ky^2 + Kz^2 )
y(i) = Ky * x(i)
z(i) = Kz * x(i)

The color of each point is determined by projecting the 3D local point into the RGB image using cameraParamsIntensity.

By default the local coordinates of points are directly stored into the local map, but if indicated so in takeIntoAccountSensorPoseOnRobot the points are transformed with sensorPose. Furthermore, if provided, those coordinates are transformed with robotPoseInTheWorld

Parameters

[in] PROJ3D_USE_LUT (Only when range_is_depth=true) Whether to use a Look-up-table (LUT) to speed up the conversion. It's thread safe in all situations except when you call this method from different threads and with different camera parameter matrices. In all other cases, it's a good idea to left it enabled.

Template Parameters

POINTMAP Supported maps are all those covered by mrpt::utils::PointCloudAdapter (mrpt::maps::CPointsMap and derived, mrpt::opengl::CPointCloudColoured, PCL point clouds,...)

Note: In MRPT < 0.9.5, this method always assumes that ranges were in Kinect-like format.

Definition at line 166 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation3DRangeScan::rangeImage_convertToExternalStorage	(	const std::string &	fileName,
		const std::string &	use_this_base_dir
	)

Users won't normally want to call this, it's only used from internal MRPT programs.

void mrpt::obs::CObservation3DRangeScan::rangeImage_forceResetExternalStorage ( )

inline

Forces marking this observation as non-externally stored - it doesn't anything else apart from reseting the corresponding flag (Users won't normally want to call this, it's only used from internal MRPT programs)

Definition at line 255 of file obs/CObservation3DRangeScan.h.

std::string mrpt::obs::CObservation3DRangeScan::rangeImage_getExternalStorageFile ( ) const

inline

Definition at line 246 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation3DRangeScan::rangeImage_getExternalStorageFileAbsolutePath ( std::string & out_path ) const

std::string mrpt::obs::CObservation3DRangeScan::rangeImage_getExternalStorageFileAbsolutePath ( ) const

inline

Definition at line 248 of file obs/CObservation3DRangeScan.h.

bool mrpt::obs::CObservation3DRangeScan::rangeImage_isExternallyStored ( ) const

inline

Definition at line 245 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation3DRangeScan::rangeImage_setSize	(	const int	HEIGHT,
		const int	WIDTH
	)

Similar to calling "rangeImage.setSize(H,W)" but this method provides memory pooling to speed-up the memory allocation.

void mrpt::obs::CObservation3DRangeScan::readFromStream	(	mrpt::utils::CStream &	in,
		int	version
	)

protectedvirtual

Introduces a pure virtual method responsible for loading from a CStream This can not be used directly be users, instead use "stream >> object;" for reading it from a stream or "stream >> object_ptr;" if the class is unknown apriori.

Parameters

in	The input binary stream where the object data must read from.
version	The version of the object stored in the stream: use this version number in your code to know how to read the incoming data.

Exceptions

std::exception On any error, see CStream::ReadBuffer

See also: CStream

Implements mrpt::utils::CSerializable.

static double mrpt::obs::CObservation3DRangeScan::recoverCameraCalibrationParameters	(	const CObservation3DRangeScan &	in_obs,
		mrpt::utils::TCamera &	out_camParams,
		const double	camera_offset = `0.01`
	)

static

A Levenberg-Marquart-based optimizer to recover the calibration parameters of a 3D camera given a range (depth) image and the corresponding 3D point cloud.

Parameters

camera_offset The offset (in meters) in the +X direction of the point cloud. It's 1cm for SwissRanger SR4000.

Returns: The final average reprojection error per pixel (typ <0.05 px)

void mrpt::obs::CObservation3DRangeScan::resizePoints3DVectors ( const size_t nPoints )

Use this method instead of resizing all three points3D_x, points3D_y & points3D_z to allow the usage of the internal memory pool.

Referenced by mrpt::utils::PointCloudAdapter< mrpt::obs::CObservation3DRangeScan >::resize().

void mrpt::obs::CObservation::setSensorPose ( const mrpt::math::TPose3D & newSensorPose )

inherited

A general method to change the sensor pose on the robot.

Note that most sensors will use the full (6D) CPose3D, but see the derived classes for more details or special cases.

See also: getSensorPose

void mrpt::obs::CObservation3DRangeScan::setSensorPose ( const mrpt::poses::CPose3D & newSensorPose )

inlinevirtual

A general method to change the sensor pose on the robot.

Note that most sensors will use the full (6D) CPose3D, but see the derived classes for more details or special cases.

See also: getSensorPose

Implements mrpt::obs::CObservation.

Definition at line 295 of file obs/CObservation3DRangeScan.h.

void mrpt::obs::CObservation::swap ( CObservation & o )

protectedinherited

Swap with another observation, ONLY the data defined here in the base class CObservation. It's protected since it'll be only called from child classes that should know what else to swap appart from these common data.

void mrpt::obs::CObservation3DRangeScan::swap ( CObservation3DRangeScan & o )

Very efficient method to swap the contents of two observations.

virtual void mrpt::obs::CObservation3DRangeScan::unload ( )

virtual

Unload all images, for the case they being delayed-load images stored in external files (othewise, has no effect).

See also: load

Reimplemented from mrpt::obs::CObservation.

void mrpt::obs::CObservation3DRangeScan::writeToStream	(	mrpt::utils::CStream &	out,
		int *	getVersion
	)		const

protectedvirtual

Introduces a pure virtual method responsible for writing to a CStream.

This can not be used directly be users, instead use "stream << object;" for writing it to a stream.

Parameters

out	The output binary stream where object must be dumped.
getVersion	If NULL, the object must be dumped. If not, only the version of the object dump must be returned in this pointer. This enables the versioning of objects dumping and backward compatibility with previously stored data.

Exceptions

std::exception On any error, see CStream::WriteBuffer

See also: CStream

Implements mrpt::utils::CSerializable.

Member Data Documentation

mrpt::utils::CLASSINIT mrpt::obs::CObservation3DRangeScan::_init_CObservation3DRangeScan

staticprotected

Definition at line 101 of file obs/CObservation3DRangeScan.h.

mrpt::utils::TCamera mrpt::obs::CObservation3DRangeScan::cameraParams

Projection parameters of the depth camera.

Definition at line 272 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::obs::detail::project3DPointsFromDepthImageInto().

mrpt::utils::TCamera mrpt::obs::CObservation3DRangeScan::cameraParamsIntensity

Projection parameters of the intensity (graylevel or RGB) camera.

Definition at line 273 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::obs::detail::project3DPointsFromDepthImageInto().

const mrpt::utils::TRuntimeClassId mrpt::utils::CObject::classCObject

staticinherited

Definition at line 124 of file CObject.h.

const mrpt::utils::TRuntimeClassId mrpt::obs::CObservation::classCObservation

staticinherited

Definition at line 53 of file obs/CObservation.h.

mrpt::utils::TRuntimeClassId mrpt::obs::CObservation3DRangeScan::classCObservation3DRangeScan

static

Definition at line 101 of file obs/CObservation3DRangeScan.h.

const mrpt::utils::TRuntimeClassId mrpt::utils::CSerializable::classCSerializable

staticinherited

Definition at line 38 of file CSerializable.h.

const mrpt::utils::TRuntimeClassId* mrpt::obs::CObservation3DRangeScan::classinfo

static

Definition at line 101 of file obs/CObservation3DRangeScan.h.

mrpt::utils::CImage mrpt::obs::CObservation3DRangeScan::confidenceImage

If hasConfidenceImage=true, an image with the "confidence" value [range 0-255] as estimated by the capture drivers.

Definition at line 270 of file obs/CObservation3DRangeScan.h.

bool mrpt::obs::CObservation3DRangeScan::hasConfidenceImage

true means the field confidenceImage contains valid data

Definition at line 269 of file obs/CObservation3DRangeScan.h.

bool mrpt::obs::CObservation3DRangeScan::hasIntensityImage

true means the field intensityImage contains valid data

Definition at line 265 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::obs::detail::project3DPointsFromDepthImageInto().

bool mrpt::obs::CObservation3DRangeScan::hasPoints3D

true means the field points3D contains valid data.

Definition at line 218 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::maps::COctoMapBase< OCTREE, OCTREE_NODE >::internal_build_PointCloud_for_observation(), and mrpt::utils::PointCloudAdapter< mrpt::obs::CObservation3DRangeScan >::resize().

bool mrpt::obs::CObservation3DRangeScan::hasRangeImage

true means the field rangeImage contains valid data

Definition at line 238 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::obs::detail::project3DPointsFromDepthImageInto().

mrpt::utils::CImage mrpt::obs::CObservation3DRangeScan::intensityImage

If hasIntensityImage=true, a color or gray-level intensity image of the same size than "rangeImage".

Definition at line 266 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::obs::detail::project3DPointsFromDepthImageInto().

TIntensityChannelID mrpt::obs::CObservation3DRangeScan::intensityImageChannel

The source of the intensityImage; typically the visible channel.

See also: TIntensityChannelID

Definition at line 267 of file obs/CObservation3DRangeScan.h.

TCached3DProjTables mrpt::obs::CObservation3DRangeScan::m_3dproj_lut

static

3D point cloud projection look-up-table

See also: project3DPointsFromDepthImage

Definition at line 318 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::obs::detail::project3DPointsFromDepthImageInto().

std::string mrpt::obs::CObservation3DRangeScan::m_points3D_external_file

protected

3D points are in CImage::IMAGES_PATH_BASE+<this_file_name>

Definition at line 105 of file obs/CObservation3DRangeScan.h.

bool mrpt::obs::CObservation3DRangeScan::m_points3D_external_stored

protected

If set to true, m_points3D_external_file is valid.

Definition at line 104 of file obs/CObservation3DRangeScan.h.

std::string mrpt::obs::CObservation3DRangeScan::m_rangeImage_external_file

protected

rangeImage is in CImage::IMAGES_PATH_BASE+<this_file_name>

Definition at line 108 of file obs/CObservation3DRangeScan.h.

bool mrpt::obs::CObservation3DRangeScan::m_rangeImage_external_stored

protected

If set to true, m_rangeImage_external_file is valid.

Definition at line 107 of file obs/CObservation3DRangeScan.h.

float mrpt::obs::CObservation3DRangeScan::maxRange

The maximum range allowed by the device, in meters (e.g. 8.0m, 5.0m,...)

Definition at line 288 of file obs/CObservation3DRangeScan.h.

std::vector<float> mrpt::obs::CObservation3DRangeScan::points3D_x

If hasPoints3D=true, the X coordinates of the 3D point cloud detected by the camera.

See also: resizePoints3DVectors

Definition at line 219 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::utils::PointCloudAdapter< mrpt::obs::CObservation3DRangeScan >::getPointXYZ(), mrpt::maps::COctoMapBase< OCTREE, OCTREE_NODE >::internal_build_PointCloud_for_observation(), mrpt::utils::PointCloudAdapter< mrpt::obs::CObservation3DRangeScan >::setPointXYZ(), and mrpt::utils::PointCloudAdapter< mrpt::obs::CObservation3DRangeScan >::size().

std::vector<float> mrpt::obs::CObservation3DRangeScan::points3D_y

If hasPoints3D=true, the Y coordinates of the 3D point cloud detected by the camera.

See also: resizePoints3DVectors

Definition at line 220 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::utils::PointCloudAdapter< mrpt::obs::CObservation3DRangeScan >::getPointXYZ(), mrpt::maps::COctoMapBase< OCTREE, OCTREE_NODE >::internal_build_PointCloud_for_observation(), and mrpt::utils::PointCloudAdapter< mrpt::obs::CObservation3DRangeScan >::setPointXYZ().

std::vector<float> mrpt::obs::CObservation3DRangeScan::points3D_z

If hasPoints3D=true, the Z coordinates of the 3D point cloud detected by the camera.

See also: resizePoints3DVectors

Definition at line 221 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::utils::PointCloudAdapter< mrpt::obs::CObservation3DRangeScan >::getPointXYZ(), mrpt::maps::COctoMapBase< OCTREE, OCTREE_NODE >::internal_build_PointCloud_for_observation(), and mrpt::utils::PointCloudAdapter< mrpt::obs::CObservation3DRangeScan >::setPointXYZ().

bool mrpt::obs::CObservation3DRangeScan::range_is_depth

true: Kinect-like ranges: entries of rangeImage are distances along the +X axis; false: Ranges in rangeImage are actual distances in 3D.

Definition at line 240 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::obs::detail::project3DPointsFromDepthImageInto().

mrpt::math::CMatrix mrpt::obs::CObservation3DRangeScan::rangeImage

If hasRangeImage=true, a matrix of floats with the range data as captured by the camera (in meters)

See also: range_is_depth

Definition at line 239 of file obs/CObservation3DRangeScan.h.

Referenced by mrpt::obs::detail::project3DPointsFromDepthImageInto().

mrpt::poses::CPose3D mrpt::obs::CObservation3DRangeScan::relativePoseIntensityWRTDepth

Relative pose of the intensity camera wrt the depth camera (which is the coordinates origin for this observation).

In a SwissRanger camera, this will be (0,0,0,-90deg,0,-90deg) since both cameras coincide. In a Kinect, this will include a small lateral displacement and a rotation, according to the drawing on the top of this page.