Optimized vocabulary tree quantizer, templated on feature type and distance metric for maximum efficiency. More...

#include <VocabularyTree.hpp>

Inheritance diagram for aliceVision::voctree::VocabularyTree< Feature, Distance >:

Public Member Functions
	VocabularyTree (const std::string &file)
	Create from vocabulary file. More...

template<class DescriptorT >
Word	quantize (const DescriptorT &feature) const
	Quantizes a feature into a discrete word.

template<class DescriptorT >
std::vector< Word >	quantize (const std::vector< DescriptorT > &features) const
	Quantizes a set of features into visual words.

template<class DescriptorT >
SparseHistogram	quantizeToSparse (const std::vector< DescriptorT > &features) const
	Quantizes a set of features into sparse histogram of visual words.

SparseHistogram	quantizeToSparse (const void *blindDescriptors) const override
	Create a SparseHistogram from a blind vector of descriptors. More...

uint32_t	levels () const override
	Get the depth (number of levels) of the tree.

uint32_t	splits () const override
	Get the branching factor (max splits at each node) of the tree.

uint32_t	words () const override
	Get the number of words the tree contains.

void	clear () override
	Clears vocabulary, leaving an empty tree.

void	save (const std::string &file) const override
	Save vocabulary to a file. More...

void	load (const std::string &file) override
	Load vocabulary from a file.

bool	operator== (const VocabularyTree &other) const

Protected Member Functions
bool	initialized () const

void	setNodeCounts ()

Protected Attributes
std::vector< Feature >	centers_

std::vector< uint8_t >	valid_centers_

uint32_t	k_

uint32_t	levels_

uint32_t	num_words_

uint32_t	word_start_

Detailed Description

template<class Feature, template< typename, typename > class Distance = L2>
class aliceVision::voctree::VocabularyTree< Feature, Distance >

Optimized vocabulary tree quantizer, templated on feature type and distance metric for maximum efficiency.

Feature is the data type of one feature. It has no requirements except compatibility with the distance metric.

Distance is a functor that computes the distance between two Feature objects. It must have a result_type typedef specifying the type of the returned distance. For the purposes of VocabularyTree, this need not even be a metric; distances simply need to be comparable.

Constructor & Destructor Documentation

◆ VocabularyTree()

template<class Feature , template< typename, typename > class Distance>

aliceVision::voctree::VocabularyTree< Feature, Distance >::VocabularyTree ( const std::string & file )

Create from vocabulary file.

Parameters

file	vocabulary file path

Member Function Documentation

◆ quantizeToSparse()

template<class Feature , template< typename, typename > class Distance = L2>

SparseHistogram aliceVision::voctree::VocabularyTree< Feature, Distance >::quantizeToSparse ( const void * blindDescriptors ) const

inlineoverridevirtual

Create a SparseHistogram from a blind vector of descriptors.

Parameters

blindDescriptors

Returns

Implements aliceVision::voctree::IVocabularyTree.

◆ save()

template<class Feature , template< typename, typename > class Distance>

void aliceVision::voctree::VocabularyTree< Feature, Distance >::save ( const std::string & file ) const

overridevirtual

Save vocabulary to a file.

Todo:: Support serializing of non-"simple" feature classes

Todo:: Some identifying name for the distance used

Implements aliceVision::voctree::IVocabularyTree.

Member Data Documentation

◆ k_

template<class Feature , template< typename, typename > class Distance = L2>

uint32_t aliceVision::voctree::VocabularyTree< Feature, Distance >::k_

protected

Todo:: Consider bit-vector

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

/home/docs/checkouts/readthedocs.org/user_builds/alicevision/checkouts/latest/src/aliceVision/voctree/VocabularyTree.hpp

Public Member Functions

Protected Member Functions

Protected Attributes