tesseract::FPAnalyzer Class Reference

Public Member Functions
	FPAnalyzer (ICOORD page_tr, TO_BLOCK_LIST *port_blocks)
	~FPAnalyzer ()=default
void	Pass1Analyze ()
void	EstimatePitch (bool pass1)
bool	maybe_fixed_pitch ()
void	MergeFragments ()
void	FinalizeLargeChars ()
bool	Pass2Analyze ()
void	OutputEstimations ()
void	DebugOutputResult ()
size_t	num_rows ()
unsigned	max_iteration ()

Detailed Description

Definition at line 951 of file cjkpitch.cpp.

Constructor & Destructor Documentation

FPAnalyzer()

tesseract::FPAnalyzer::FPAnalyzer	(	ICOORD	page_tr,
		TO_BLOCK_LIST *	port_blocks
	)

Definition at line 1031 of file cjkpitch.cpp.

    : page_tr_(page_tr)
    , num_tall_rows_(0)
    , num_bad_rows_(0)
    , num_empty_rows_(0)
    , max_chars_per_row_(0) {
  TO_BLOCK_IT block_it(port_blocks);
 
  for (block_it.mark_cycle_pt(); !block_it.cycled_list(); block_it.forward()) {
    TO_BLOCK *block = block_it.data();
    if (!block->get_rows()->empty()) {
      ASSERT_HOST(block->xheight > 0);
      find_repeated_chars(block, false);
    }
  }
 
  for (block_it.mark_cycle_pt(); !block_it.cycled_list(); block_it.forward()) {
    TO_ROW_IT row_it = block_it.data()->get_rows();
    for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward()) {
      FPRow row;
      row.Init(row_it.data());
      rows_.push_back(row);
      size_t num_chars = rows_.back().num_chars();
      if (num_chars <= 1) {
        num_empty_rows_++;
      }
      if (num_chars > max_chars_per_row_) {
        max_chars_per_row_ = num_chars;
      }
    }
  }
}

~FPAnalyzer()

tesseract::FPAnalyzer::~FPAnalyzer ( )

default

Member Function Documentation

DebugOutputResult()

void tesseract::FPAnalyzer::DebugOutputResult ( )

inline

Definition at line 1003 of file cjkpitch.cpp.

                           {
    tprintf("FPAnalyzer: final result\n");
    for (size_t i = 0; i < rows_.size(); i++) {
      rows_[i].DebugOutputResult(i);
    }
  }

EstimatePitch()

void tesseract::FPAnalyzer::EstimatePitch

(

bool

pass1

)

Definition at line 1064 of file cjkpitch.cpp.

                                         {
  LocalCorrelation pitch_height_stats;
 
  num_tall_rows_ = 0;
  num_bad_rows_ = 0;
  pitch_height_stats.Clear();
  for (auto &row : rows_) {
    row.EstimatePitch(pass1);
    if (row.good_pitches()) {
      pitch_height_stats.Add(row.height() + row.gap(), row.pitch(), row.good_pitches());
      if (row.height_pitch_ratio() > 1.1) {
        num_tall_rows_++;
      }
    } else {
      num_bad_rows_++;
    }
  }
 
  pitch_height_stats.Finish();
  for (auto &row : rows_) {
    if (row.good_pitches() >= 5) {
      // We have enough evidences. Just use the pitch estimation
      // from this row.
      row.set_estimated_pitch(row.pitch());
    } else if (row.num_chars() > 1) {
      float estimated_pitch = pitch_height_stats.EstimateYFor(row.height() + row.gap(), 0.1f);
      // CJK characters are more likely to be fragmented than poorly
      // chopped. So trust the page-level estimation of character
      // pitch only if it's larger than row-level estimation or
      // row-level estimation is too large (2x bigger than row height).
      if (estimated_pitch > row.pitch() || row.pitch() > row.height() * 2.0) {
        row.set_estimated_pitch(estimated_pitch);
      } else {
        row.set_estimated_pitch(row.pitch());
      }
    }
  }
}

FinalizeLargeChars()

void tesseract::FPAnalyzer::FinalizeLargeChars ( )

inline

Definition at line 980 of file cjkpitch.cpp.

                            {
    for (auto &row : rows_) {
      row.FinalizeLargeChars();
    }
  }

max_iteration()

unsigned tesseract::FPAnalyzer::max_iteration ( )

inline

Definition at line 1015 of file cjkpitch.cpp.

                           {
    // We're fixing at least one character per iteration. So basically
    // we shouldn't require more than max_chars_per_row_ iterations.
    return max_chars_per_row_ + 100;
  }

maybe_fixed_pitch()

bool tesseract::FPAnalyzer::maybe_fixed_pitch ( )

inline

Definition at line 967 of file cjkpitch.cpp.

                           {
    if (rows_.empty() || rows_.size() <= num_bad_rows_ + num_tall_rows_ + 1) {
      return false;
    }
    return true;
  }

MergeFragments()

void tesseract::FPAnalyzer::MergeFragments ( )

inline

Definition at line 974 of file cjkpitch.cpp.

                        {
    for (auto &row : rows_) {
      row.MergeFragments();
    }
  }

num_rows()

size_t tesseract::FPAnalyzer::num_rows ( )

inline

Definition at line 1010 of file cjkpitch.cpp.

                    {
    return rows_.size();
  }

OutputEstimations()

void tesseract::FPAnalyzer::OutputEstimations ( )

inline

Definition at line 996 of file cjkpitch.cpp.

                           {
    for (auto &row : rows_) {
      row.OutputEstimations();
    }
    // Don't we need page-level estimation of gaps/spaces?
  }

Pass1Analyze()

void tesseract::FPAnalyzer::Pass1Analyze ( )

inline

Definition at line 956 of file cjkpitch.cpp.

                      {
    for (auto &row : rows_) {
      row.Pass1Analyze();
    }
  }

Pass2Analyze()

bool tesseract::FPAnalyzer::Pass2Analyze ( )

inline

Definition at line 986 of file cjkpitch.cpp.

                      {
    bool changed = false;
    for (auto &row : rows_) {
      if (row.Pass2Analyze()) {
        changed = true;
      }
    }
    return changed;
  }

---

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

• /media/home/debian/src/github/tesseract-ocr/tesseract/src/textord/cjkpitch.cpp