PointCloudHierarchy::PointCloudHierarchyManager Class Reference

Class to manage a hierarchy of point clouds with different resolutions The hierarchy is organized with coarser levels having fewer points (parents) and finer levels having more points (children). More...

#include <PointCloudHierarchy.h>

Public Member Functions
	PointCloudHierarchyManager (int min_points=100, int max_points=1000)
	Constructor.
template<int spacedim>
int	generateHierarchy (const std::vector< Point< spacedim > > &input_points, const std::vector< double > &input_density, const std::string &output_dir)
	Generate hierarchy of point clouds from input points.
void	setMaxPoints (int max_points)
	Set the maximum number of points for level 1.
void	setMinPoints (int min_points)
	Set the minimum number of points for coarsest level.
int	getNumLevels () const
	Get the number of levels in the last generated hierarchy.
int	getPointCount (int level) const
	Get the number of points at a specific level.
int	getPointsForLevel (int base_points, int level) const
	Calculate number of points for a given level.
const std::vector< std::vector< size_t > > &	getParentIndices (int level) const
	Get the parent indices for points at level L-1.
const std::vector< std::vector< size_t > > &	getChildIndices (int level) const
	Get the child indices for points at level L.

Private Member Functions
void	ensureDirectoryExists (const std::string &path) const
	Ensure directory exists, create if it doesn't.
template<int spacedim>
std::tuple< std::vector< std::array< double, spacedim > >, std::vector< double >, std::vector< int > >	kmeansClustering (const std::vector< std::array< double, spacedim > > &points, const std::vector< double > &densities, size_t k)
	Performs parallel k-means clustering on a set of points with parent-child tracking.

Private Attributes
int	min_points_
int	max_points_
int	num_levels_
std::vector< int >	level_point_counts_
std::vector< std::vector< std::vector< size_t > > >	parent_indices_
std::vector< std::vector< std::vector< size_t > > >	child_indices_

Detailed Description

Class to manage a hierarchy of point clouds with different resolutions The hierarchy is organized with coarser levels having fewer points (parents) and finer levels having more points (children).

Definition at line 23 of file PointCloudHierarchy.h.

Constructor & Destructor Documentation

PointCloudHierarchyManager()

PointCloudHierarchy::PointCloudHierarchyManager::PointCloudHierarchyManager	(	int	min_points = 100,
		int	max_points = 1000
	)

Constructor.

Parameters

min_points	Minimum number of points for the coarsest level (parent level)
max_points	Maximum number of points for level 1 point cloud

Definition at line 18 of file PointCloudHierarchy.cc.

Member Function Documentation

generateHierarchy()

template<int spacedim>

template int PointCloudHierarchy::PointCloudHierarchyManager::generateHierarchy< 3 >	(	const std::vector< Point< spacedim > > &	input_points,
		const std::vector< double > &	input_density,
		const std::string &	output_dir
	)

Generate hierarchy of point clouds from input points.

Parameters

input_points	Vector of input points (finest level)
input_density	Vector of input density values (optional, uniform density if empty)
output_dir	Directory to save the point cloud hierarchy

Returns

Number of levels generated

Exceptions

std::runtime_error

if processing fails

Definition at line 127 of file PointCloudHierarchy.cc.

Here is the call graph for this function:

Here is the caller graph for this function:

setMaxPoints()

void PointCloudHierarchy::PointCloudHierarchyManager::setMaxPoints

(

int

max_points

)

Set the maximum number of points for level 1.

Definition at line 24 of file PointCloudHierarchy.cc.

setMinPoints()

void PointCloudHierarchy::PointCloudHierarchyManager::setMinPoints

(

int

min_points

)

Set the minimum number of points for coarsest level.

Definition at line 28 of file PointCloudHierarchy.cc.

getNumLevels()

int PointCloudHierarchy::PointCloudHierarchyManager::getNumLevels

(

)

const

Get the number of levels in the last generated hierarchy.

Definition at line 32 of file PointCloudHierarchy.cc.

getPointCount()

int PointCloudHierarchy::PointCloudHierarchyManager::getPointCount

(

int

level

)

const

Get the number of points at a specific level.

Definition at line 36 of file PointCloudHierarchy.cc.

getPointsForLevel()

int PointCloudHierarchy::PointCloudHierarchyManager::getPointsForLevel	(	int	base_points,
		int	level
	)		const

Calculate number of points for a given level.

Definition at line 63 of file PointCloudHierarchy.cc.

Here is the caller graph for this function:

getParentIndices()

const std::vector< std::vector< size_t > > & PointCloudHierarchy::PointCloudHierarchyManager::getParentIndices

(

int

level

)

const

Get the parent indices for points at level L-1.

Parameters

level

The level of the parents (must be > 0)

Returns

For each point at level L-1, returns index of its parent at level L

Definition at line 43 of file PointCloudHierarchy.cc.

getChildIndices()

const std::vector< std::vector< size_t > > & PointCloudHierarchy::PointCloudHierarchyManager::getChildIndices

(

int

level

)

const

Get the child indices for points at level L.

Parameters

level

The level of points whose children we want (must be < num_levels-1)

Returns

For each point at level L, returns indices of its children at level L-1

Definition at line 50 of file PointCloudHierarchy.cc.

ensureDirectoryExists()

void PointCloudHierarchy::PointCloudHierarchyManager::ensureDirectoryExists ( const std::string &  path ) const

private

Ensure directory exists, create if it doesn't.

Definition at line 57 of file PointCloudHierarchy.cc.

Here is the caller graph for this function:

kmeansClustering()

template<int spacedim>

template std::tuple< std::vector< std::array< double, 3 > >, std::vector< double >, std::vector< int > > PointCloudHierarchy::PointCloudHierarchyManager::kmeansClustering< 3 > ( const std::vector< std::array< double, spacedim > > &  points, const std::vector< double > &  densities, size_t  k  )

private

Performs parallel k-means clustering on a set of points with parent-child tracking.

Parameters

points	Input points from finer level
densities	Input densities
k	Number of clusters (points at coarser level)

Returns

Tuple of:

• cluster centers (parent points)
• aggregated densities for parents
• assignments (mapping of each child to its parent cluster)

Definition at line 80 of file PointCloudHierarchy.cc.

Here is the call graph for this function:

Member Data Documentation

min_points_

int PointCloudHierarchy::PointCloudHierarchyManager::min_points_

private

Definition at line 86 of file PointCloudHierarchy.h.

max_points_

int PointCloudHierarchy::PointCloudHierarchyManager::max_points_

private

Definition at line 87 of file PointCloudHierarchy.h.

num_levels_

int PointCloudHierarchy::PointCloudHierarchyManager::num_levels_

private

Definition at line 88 of file PointCloudHierarchy.h.

level_point_counts_

std::vector<int> PointCloudHierarchy::PointCloudHierarchyManager::level_point_counts_

private

Definition at line 89 of file PointCloudHierarchy.h.

parent_indices_

std::vector<std::vector<std::vector<size_t> > > PointCloudHierarchy::PointCloudHierarchyManager::parent_indices_

private

Definition at line 93 of file PointCloudHierarchy.h.

child_indices_

std::vector<std::vector<std::vector<size_t> > > PointCloudHierarchy::PointCloudHierarchyManager::child_indices_

private

Definition at line 97 of file PointCloudHierarchy.h.

---

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

• include/SemiDiscreteOT/core/PointCloudHierarchy.h
• src/core/PointCloudHierarchy.cc