Implements an OctTree with adaptive leaf splitting as a post process step. More...

#include <adaptive_oc_tree.h>

Inheritance diagram for lslgeneric::AdaptiveOctTree< PointT >:

Public Member Functions
	AdaptiveOctTree ()
	dummy default constructor
	AdaptiveOctTree (pcl::PointXYZ center, double xsize, double ysize, double zsize, NDTCell< PointT > type, OctTree< PointT > _parent=NULL, unsigned int _depth=0)
	creates an oct tree node with known center and size
virtual SpatialIndex< PointT > *	clone ()
virtual void	postProcessPoints ()
void	setParameters (bool _useDornikHansen=false, bool _useFlatness=true, double _RSS_THRESHOLD=1000, double _DH_SIGNIFICANCE_LVL=0.5, double _MIN_CELL_SIZE=1, double _FLAT_FACTOR=10, double _BIG_CELL_SIZE=5, double _SMALL_CELL_SIZE=5)
virtual	~AdaptiveOctTree ()
Public Attributes
double	MIN_CELL_SIZE
Protected Member Functions
double	computeDornikHansen (NDTCell< PointT > *cell)
double	computeResidualSquare (NDTCell< PointT > *cell)
virtual void	computeTreeLeafs ()
std::vector< OctTree< PointT > * >	splitTree (OctTree< PointT > *leaf)
Protected Attributes
double	DH_SIGNIFICANCE_LVL
double	FLAT_FACTOR
std::vector< OctTree< PointT > * >	myTreeLeafs
bool	parametersSet
double	RSS_THRESHOLD
bool	useDornikHansen
bool	useFlatness

Detailed Description

template<typename PointT>
class lslgeneric::AdaptiveOctTree< PointT >

Implements an OctTree with adaptive leaf splitting as a post process step.

The OctTree is split unitl a conservative initial size. After this, a post processing step is used to determine if the gaussians in each leaf are well-fitting. At the moment one heuristic option is available - using the residual squares sum and the residual variance. A second option using the omnibus normality test will be added soon.

Definition at line 53 of file adaptive_oc_tree.h.

Constructor & Destructor Documentation

template<typename PointT >

lslgeneric::AdaptiveOctTree< PointT >::AdaptiveOctTree ( )

dummy default constructor

empty! default constructor

Definition at line 38 of file adaptive_oc_tree.hpp.

template<typename PointT >

lslgeneric::AdaptiveOctTree< PointT >::AdaptiveOctTree	(	pcl::PointXYZ	center,
		double	xsize,
		double	ysize,
		double	zsize,
		NDTCell< PointT > *	type,
		OctTree< PointT > *	_parent = `NULL`,
		unsigned int	_depth = `0`
	)

creates an oct tree node with known center and size

constructor, calls parent OctTree constructor

Definition at line 49 of file adaptive_oc_tree.hpp.

template<typename PointT >

lslgeneric::AdaptiveOctTree< PointT >::~AdaptiveOctTree ( ) [virtual]

empty destructor, all data are deallocated by parent class

Definition at line 62 of file adaptive_oc_tree.hpp.

Member Function Documentation

template<typename PointT >

SpatialIndex< PointT > * lslgeneric::AdaptiveOctTree< PointT >::clone ( ) [virtual]

creates an oct tree with the same parameters.

Note:: the points are not copied in teh returned instance

Definition at line 402 of file adaptive_oc_tree.hpp.

template<typename PointT >

double lslgeneric::AdaptiveOctTree< PointT >::computeDornikHansen ( NDTCell< PointT > * cell ) [protected]

performs the Dornik-Hansen Omnibus normality test

Definition at line 187 of file adaptive_oc_tree.hpp.

template<typename PointT >

double lslgeneric::AdaptiveOctTree< PointT >::computeResidualSquare ( NDTCell< PointT > * cell ) [protected]

fits a 3d gaussian in the cell and computes the residual squares sum

Definition at line 305 of file adaptive_oc_tree.hpp.

template<typename PointT >

void lslgeneric::AdaptiveOctTree< PointT >::computeTreeLeafs ( ) [protected, virtual]

finds all leafs of this tree and fills the vector of pointers to the leafs

Definition at line 70 of file adaptive_oc_tree.hpp.

template<typename PointT >

void lslgeneric::AdaptiveOctTree< PointT >::postProcessPoints ( ) [virtual]

go through all leafs and split the ones with high residuals

Definition at line 111 of file adaptive_oc_tree.hpp.

template<typename PointT >

void lslgeneric::AdaptiveOctTree< PointT >::setParameters	(	bool	_useDornikHansen = `false`,
		bool	_useFlatness = `true`,
		double	_RSS_THRESHOLD = `1000`,
		double	_DH_SIGNIFICANCE_LVL = `0.5`,
		double	_MIN_CELL_SIZE = `1`,
		double	_FLAT_FACTOR = `10`,
		double	_BIG_CELL_SIZE = `5`,
		double	_SMALL_CELL_SIZE = `5`
	)