Support class for segmentation algorithms. More...

#include <SegmenterUtils.h>

Public Types
enum	Mode { BEST, GPU, CPU }
Public Member Functions
	SegmenterUtils (Mode mode=BEST)
	Constructor.
	~SegmenterUtils ()
	Destructor.
core::Img8u	createColorImage (core::Img32s &labelImage)
	Creates a color image (e.g. for pointcloud coloring) from a given segmentation label image.
core::Img8u	createROIMask (core::DataSegment< float, 4 > &xyzh, core::Img32f &depthImage, float xMin, float xMax, float yMin, float yMax, float zMin=-10000, float zMax=10000)
	Creates the mask image for segmentation (including 3D ROI).
core::Img8u	createMask (core::Img32f &depthImage)
	Creates the mask image for segmentation.
core::Img32s	stabelizeSegmentation (core::Img32s &labelImage)
	Minimizes the label ID changes from frame to frame. The overlaps between the current and the previous label image are calculated and relabeled for the result.
math::DynMatrix< bool >	calculateAdjacencyMatrix (core::DataSegment< float, 4 > &xyzh, core::Img32s &labelImage, core::Img8u &maskImage, int radius, float euclideanDistance, int numSurfaces)
	Calculates the adjacency between segments. Use edgePointAssignmentAndAdjacencyMatrix(...) if edge point assignment is needed as well.
void	edgePointAssignment (core::DataSegment< float, 4 > &xyzh, core::Img32s &labelImage, core::Img8u &maskImage, int radius, float euclideanDistance, int numSurfaces)
	Assigns the edge points to the surfaces. Use edgePointAssignmentAndAdjacencyMatrix(...) if adjacency matrix is needed as well.
math::DynMatrix< bool >	edgePointAssignmentAndAdjacencyMatrix (core::DataSegment< float, 4 > &xyzh, core::Img32s &labelImage, core::Img8u &maskImage, int radius, float euclideanDistance, int numSurfaces)
	Calculates the adjacency between segments and assigns the edge points to the surfaces.
Private Member Functions
void	createColorImageCL (core::Img32s &labelImage, core::Img8u &colorImage)
void	createColorImageCPU (core::Img32s &labelImage, core::Img8u &colorImage)
std::vector< int >	calculateLabelReassignment (int countCur, int countLast, core::Channel32s &labelImageC, core::Channel32s &lastLabelImageC, utils::Size size)
math::DynMatrix< bool >	edgePointAssignmentAndAdjacencyMatrixCL (core::DataSegment< float, 4 > &xyzh, core::Img32s &labelImage, core::Img8u &maskImage, int radius, float euclideanDistance, int numSurfaces, bool pointAssignment)
math::DynMatrix< bool >	edgePointAssignmentAndAdjacencyMatrixCPU (core::DataSegment< float, 4 > &xyzh, core::Img32s &labelImage, core::Img8u &maskImage, int radius, float euclideanDistance, int numSurfaces, bool pointAssignment)
Static Private Member Functions
static float	dist3 (const Vec &a, const Vec &b)
Private Attributes
Data *	m_data
	internal data type

Detailed Description

Support class for segmentation algorithms.

This class provides supporting methods for segmentation algorithms.

Member Enumeration Documentation

enum icl::geom::SegmenterUtils::Mode

Enumerator:

BEST
GPU
CPU

Constructor & Destructor Documentation

icl::geom::SegmenterUtils::SegmenterUtils ( Mode mode = BEST )

Constructor.

Constructs an object of this class.

Parameters:

mode	the selected mode: CPU, GPU or BEST (uses GPU if available)

icl::geom::SegmenterUtils::~SegmenterUtils ( )

Destructor.

Member Function Documentation

math::DynMatrix<bool> icl::geom::SegmenterUtils::calculateAdjacencyMatrix	(	core::DataSegment< float, 4 > &	xyzh,
		core::Img32s &	labelImage,
		core::Img8u &	maskImage,
		int	radius,
		float	euclideanDistance,
		int	numSurfaces
	)

Calculates the adjacency between segments. Use edgePointAssignmentAndAdjacencyMatrix(...) if edge point assignment is needed as well.

Parameters:

xyzh	the input pointcloud (point position)
labelImage	the input label image
maskImage	the input mask image
radius	in pixel (distance of surfaces/segments around separating edge)
euclideanDistance	the maximum euclidean distance between adjacent surfaces/segments
numSurfaces	the number of surfaces/segments in the label image

Returns:: the adjacency matrix.

std::vector<int> icl::geom::SegmenterUtils::calculateLabelReassignment	(	int	countCur,
		int	countLast,
		core::Channel32s &	labelImageC,
		core::Channel32s &	lastLabelImageC,
		utils::Size	size
	)		`[private]`

core::Img8u icl::geom::SegmenterUtils::createColorImage ( core::Img32s & labelImage )

Creates a color image (e.g. for pointcloud coloring) from a given segmentation label image.

Parameters:

labelImage the input label image

Returns:: the output color image.

void icl::geom::SegmenterUtils::createColorImageCL	(	core::Img32s &	labelImage,
		core::Img8u &	colorImage
	)		`[private]`

void icl::geom::SegmenterUtils::createColorImageCPU	(	core::Img32s &	labelImage,
		core::Img8u &	colorImage
	)		`[private]`

core::Img8u icl::geom::SegmenterUtils::createMask ( core::Img32f & depthImage )

Creates the mask image for segmentation.

Parameters:

depthImage the input depth image

Returns:: the output mask image.

core::Img8u icl::geom::SegmenterUtils::createROIMask	(	core::DataSegment< float, 4 > &	xyzh,
		core::Img32f &	depthImage,
		float	xMin,
		float	xMax,
		float	yMin,
		float	yMax,
		float	zMin = `-10000`,
		float	zMax = `10000`
	)

Creates the mask image for segmentation (including 3D ROI).

Parameters:

xyzh	the input pointcloud (point position)
depthImage	the input depth image parameter for ROI (in mm for world coordinates) parameter for ROI (in mm for world coordinates) parameter for ROI (in mm for world coordinates) parameter for ROI (in mm for world coordinates) parameter for ROI (in mm for world coordinates) parameter for ROI (in mm for world coordinates)

Returns:: the output mask image.

static float icl::geom::SegmenterUtils::dist3	(	const Vec &	a,
		const Vec &	b
	)		`[inline, static, private]`

void icl::geom::SegmenterUtils::edgePointAssignment	(	core::DataSegment< float, 4 > &	xyzh,
		core::Img32s &	labelImage,
		core::Img8u &	maskImage,
		int	radius,
		float	euclideanDistance,
		int	numSurfaces
	)

Assigns the edge points to the surfaces. Use edgePointAssignmentAndAdjacencyMatrix(...) if adjacency matrix is needed as well.

Parameters:

xyzh	the input pointcloud (point position)
labelImage	the input label image (changed by the method)
maskImage	the input mask image (changed by the method)
radius	in pixel (distance of surfaces/segments around separating edge)
euclideanDistance	the maximum euclidean distance between adjacent surfaces/segments
numSurfaces	the number of surfaces/segments in the label image

math::DynMatrix<bool> icl::geom::SegmenterUtils::edgePointAssignmentAndAdjacencyMatrix	(	core::DataSegment< float, 4 > &	xyzh,
		core::Img32s &	labelImage,
		core::Img8u &	maskImage,
		int	radius,
		float	euclideanDistance,
		int	numSurfaces
	)

Calculates the adjacency between segments and assigns the edge points to the surfaces.

Parameters:

xyzh	the input pointcloud (point position)
labelImage	the input label image (changed by the method)
maskImage	the input mask image (changed by the method)
radius	in pixel (distance of surfaces/segments around separating edge)
euclideanDistance	the maximum euclidean distance between adjacent surfaces/segments
numSurfaces	the number of surfaces/segments in the label image

Returns:: the adjacency matrix.

math::DynMatrix<bool> icl::geom::SegmenterUtils::edgePointAssignmentAndAdjacencyMatrixCL	(	core::DataSegment< float, 4 > &	xyzh,
		core::Img32s &	labelImage,
		core::Img8u &	maskImage,
		int	radius,
		float	euclideanDistance,
		int	numSurfaces,
		bool	pointAssignment
	)		`[private]`

math::DynMatrix<bool> icl::geom::SegmenterUtils::edgePointAssignmentAndAdjacencyMatrixCPU	(	core::DataSegment< float, 4 > &	xyzh,
		core::Img32s &	labelImage,
		core::Img8u &	maskImage,
		int	radius,
		float	euclideanDistance,
		int	numSurfaces,
		bool	pointAssignment
	)		`[private]`

core::Img32s icl::geom::SegmenterUtils::stabelizeSegmentation ( core::Img32s & labelImage )

Minimizes the label ID changes from frame to frame. The overlaps between the current and the previous label image are calculated and relabeled for the result.

Parameters:

labelImage the input label image

Returns:: the stabelized output label image.

Member Data Documentation

Data* icl::geom::SegmenterUtils::m_data [private]

internal data type

internal data pointer

The documentation for this class was generated from the following file:

/home/jenkins/workspace/icl-manual-trunk/icl-manual/ICLGeom/src/ICLGeom/SegmenterUtils.h

Public Types

Public Member Functions

Private Member Functions

Static Private Member Functions

Private Attributes

Detailed Description

Member Enumeration Documentation

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation