tesseract::SampleIterator Class Reference

#include <sampleiterator.h>

Public Member Functions
	SampleIterator ()
	~SampleIterator ()
void	Clear ()
void	Init (const IndexMapBiDi *charset_map, const ShapeTable *shape_table, bool randomize, TrainingSampleSet *sample_set)
void	Begin ()
bool	AtEnd () const
const TrainingSample &	GetSample () const
TrainingSample *	MutableSample () const
int	GlobalSampleIndex () const
int	GetCompactClassID () const
int	GetSparseClassID () const
void	Next ()
int	CompactCharsetSize () const
int	SparseCharsetSize () const
const IndexMapBiDi &	charset_map () const
const ShapeTable *	shape_table () const
const TrainingSampleSet *	sample_set () const
void	MapSampleFeatures (const IntFeatureMap &feature_map)
int	UniformSamples ()
double	NormalizeSamples ()

Detailed Description

Definition at line 91 of file sampleiterator.h.

Constructor & Destructor Documentation

SampleIterator()

tesseract::SampleIterator::SampleIterator

(

)

Definition at line 29 of file sampleiterator.cpp.

    : charset_map_(nullptr)
    , shape_table_(nullptr)
    , sample_set_(nullptr)
    , randomize_(false)
    , owned_shape_table_(nullptr) {
  num_shapes_ = 0;
  Begin();
}

~SampleIterator()

tesseract::SampleIterator::~SampleIterator

(

)

Definition at line 39 of file sampleiterator.cpp.

                                {
  Clear();
}

Member Function Documentation

AtEnd()

bool tesseract::SampleIterator::AtEnd

(

)

const

Definition at line 98 of file sampleiterator.cpp.

                                 {
  return shape_index_ >= num_shapes_;
}

Begin()

void tesseract::SampleIterator::Begin

(

)

Definition at line 86 of file sampleiterator.cpp.

                           {
  shape_index_ = -1;
  shape_char_index_ = 0;
  num_shape_chars_ = 0;
  shape_font_index_ = 0;
  num_shape_fonts_ = 0;
  sample_index_ = 0;
  num_samples_ = 0;
  // Find the first indexable sample.
  Next();
}

charset_map()

const IndexMapBiDi & tesseract::SampleIterator::charset_map ( ) const

inline

Definition at line 134 of file sampleiterator.h.

                                          {
    return *charset_map_;
  }

Clear()

void tesseract::SampleIterator::Clear

(

)

Definition at line 43 of file sampleiterator.cpp.

                           {
  delete owned_shape_table_;
  owned_shape_table_ = nullptr;
}

CompactCharsetSize()

int tesseract::SampleIterator::CompactCharsetSize

(

)

const

Definition at line 195 of file sampleiterator.cpp.

                                             {
  return charset_map_ != nullptr ? charset_map_->CompactSize() : SparseCharsetSize();
}

GetCompactClassID()

int tesseract::SampleIterator::GetCompactClassID

(

)

const

Definition at line 141 of file sampleiterator.cpp.

                                            {
  return charset_map_ != nullptr ? charset_map_->SparseToCompact(shape_index_) : GetSparseClassID();
}

GetSample()

const TrainingSample & tesseract::SampleIterator::GetSample

(

)

const

Definition at line 102 of file sampleiterator.cpp.

                                                      {
  if (shape_table_ != nullptr) {
    const UnicharAndFonts *shape_entry = GetShapeEntry();
    int char_id = shape_entry->unichar_id;
    int font_id = shape_entry->font_ids[shape_font_index_];
    return *sample_set_->GetSample(font_id, char_id, sample_index_);
  } else {
    return *sample_set_->GetSample(shape_index_);
  }
}

GetSparseClassID()

int tesseract::SampleIterator::GetSparseClassID

(

)

const

Definition at line 148 of file sampleiterator.cpp.

                                           {
  return shape_table_ != nullptr ? shape_index_ : GetSample().class_id();
}

GlobalSampleIndex()

int tesseract::SampleIterator::GlobalSampleIndex

(

)

const

Definition at line 126 of file sampleiterator.cpp.

                                            {
  if (shape_table_ != nullptr) {
    const UnicharAndFonts *shape_entry = GetShapeEntry();
    int char_id = shape_entry->unichar_id;
    int font_id = shape_entry->font_ids[shape_font_index_];
    return sample_set_->GlobalSampleIndex(font_id, char_id, sample_index_);
  } else {
    return shape_index_;
  }
}

Init()

void tesseract::SampleIterator::Init	(	const IndexMapBiDi *	charset_map,
		const ShapeTable *	shape_table,
		bool	randomize,
		TrainingSampleSet *	sample_set
	)

Definition at line 49 of file sampleiterator.cpp.

                                                                         {
  Clear();
  charset_map_ = charset_map;
  shape_table_ = shape_table;
  sample_set_ = sample_set;
  randomize_ = randomize;
  if (shape_table_ == nullptr && charset_map_ != nullptr) {
    // The caller wishes to iterate by class. The easiest way to do this
    // is to create a dummy shape_table_ that we will own.
    int num_fonts = sample_set_->NumFonts();
    owned_shape_table_ = new ShapeTable(sample_set_->unicharset());
    int charsetsize = sample_set_->unicharset().size();
    for (int c = 0; c < charsetsize; ++c) {
      // We always add a shape for each character to keep the index in sync
      // with the unichar_id.
      int shape_id = owned_shape_table_->AddShape(c, 0);
      for (int f = 1; f < num_fonts; ++f) {
        if (sample_set_->NumClassSamples(f, c, true) > 0) {
          owned_shape_table_->AddToShape(shape_id, c, f);
        }
      }
    }
    shape_table_ = owned_shape_table_;
  }
  if (shape_table_ != nullptr) {
    num_shapes_ = shape_table_->NumShapes();
  } else {
    num_shapes_ = randomize ? sample_set_->num_samples() : sample_set_->num_raw_samples();
  }
  Begin();
}

MapSampleFeatures()

void tesseract::SampleIterator::MapSampleFeatures

(

const IntFeatureMap &

feature_map

)

Definition at line 219 of file sampleiterator.cpp.

                                                                       {
  for (Begin(); !AtEnd(); Next()) {
    TrainingSample *sample = MutableSample();
    MapFeatures(*sample, feature_map);
  }
}

MutableSample()

TrainingSample * tesseract::SampleIterator::MutableSample

(

)

const

Definition at line 113 of file sampleiterator.cpp.

                                                    {
  if (shape_table_ != nullptr) {
    const UnicharAndFonts *shape_entry = GetShapeEntry();
    int char_id = shape_entry->unichar_id;
    int font_id = shape_entry->font_ids[shape_font_index_];
    return sample_set_->MutableSample(font_id, char_id, sample_index_);
  } else {
    return sample_set_->mutable_sample(shape_index_);
  }
}

Next()

void tesseract::SampleIterator::Next

(

)

Definition at line 154 of file sampleiterator.cpp.

                          {
  if (shape_table_ != nullptr) {
    // Next sample in this class/font combination.
    ++sample_index_;
    if (sample_index_ < num_samples_) {
      return;
    }
    // Next font in this class in this shape.
    sample_index_ = 0;
    do {
      ++shape_font_index_;
      if (shape_font_index_ >= num_shape_fonts_) {
        // Next unichar in this shape.
        shape_font_index_ = 0;
        ++shape_char_index_;
        if (shape_char_index_ >= num_shape_chars_) {
          // Find the next shape that is mapped in the charset_map_.
          shape_char_index_ = 0;
          do {
            ++shape_index_;
          } while (shape_index_ < num_shapes_ && charset_map_ != nullptr &&
                   charset_map_->SparseToCompact(shape_index_) < 0);
          if (shape_index_ >= num_shapes_) {
            return; // The end.
          }
          num_shape_chars_ = shape_table_->GetShape(shape_index_).size();
        }
      }
      const UnicharAndFonts *shape_entry = GetShapeEntry();
      num_shape_fonts_ = shape_entry->font_ids.size();
      int char_id = shape_entry->unichar_id;
      int font_id = shape_entry->font_ids[shape_font_index_];
      num_samples_ = sample_set_->NumClassSamples(font_id, char_id, randomize_);
    } while (num_samples_ == 0);
  } else {
    // We are just iterating over the samples.
    ++shape_index_;
  }
}

NormalizeSamples()

double tesseract::SampleIterator::NormalizeSamples

(

)

Definition at line 241 of file sampleiterator.cpp.

                                        {
  double total_weight = 0.0;
  int sample_count = 0;
  for (Begin(); !AtEnd(); Next()) {
    const TrainingSample &sample = GetSample();
    total_weight += sample.weight();
    ++sample_count;
  }
  // Normalize samples.
  double min_assigned_sample_weight = 1.0;
  if (total_weight > 0.0) {
    for (Begin(); !AtEnd(); Next()) {
      TrainingSample *sample = MutableSample();
      double weight = sample->weight() / total_weight;
      if (weight < min_assigned_sample_weight) {
        min_assigned_sample_weight = weight;
      }
      sample->set_weight(weight);
    }
  }
  return min_assigned_sample_weight;
}

sample_set()

const TrainingSampleSet * tesseract::SampleIterator::sample_set ( ) const

inline

Definition at line 141 of file sampleiterator.h.

                                              {
    return sample_set_;
  }

shape_table()

const ShapeTable * tesseract::SampleIterator::shape_table ( ) const

inline

Definition at line 137 of file sampleiterator.h.

                                        {
    return shape_table_;
  }

SparseCharsetSize()

int tesseract::SampleIterator::SparseCharsetSize

(

)

const

Definition at line 200 of file sampleiterator.cpp.

                                            {
  return charset_map_ != nullptr
             ? charset_map_->SparseSize()
             : (shape_table_ != nullptr ? shape_table_->NumShapes() : sample_set_->charsetsize());
}

UniformSamples()

int tesseract::SampleIterator::UniformSamples

(

)

Definition at line 228 of file sampleiterator.cpp.

                                   {
  int num_good_samples = 0;
  for (Begin(); !AtEnd(); Next()) {
    TrainingSample *sample = MutableSample();
    sample->set_weight(1.0);
    ++num_good_samples;
  }
  NormalizeSamples();
  return num_good_samples;
}

---

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

• /media/home/debian/src/github/tesseract-ocr/tesseract/src/training/common/sampleiterator.h
• /media/home/debian/src/github/tesseract-ocr/tesseract/src/training/common/sampleiterator.cpp