tesseract::BoxWord Class Reference

#include <boxword.h>

Public Member Functions
	BoxWord ()
	BoxWord (const BoxWord &src)
	~BoxWord ()
BoxWord &	operator= (const BoxWord &src)
void	CopyFrom (const BoxWord &src)
void	SetScriptPositions (const UNICHARSET &unicharset, bool small_caps, TWERD *tessword, WERD_CHOICE *best_choice)
void	ClipToOriginalWord (const BLOCK *block, WERD *original_word)
void	MergeBoxes (int start, int end)
void	InsertBox (int index, const TBOX &box)
void	DeleteBox (int index)
void	DeleteAllBoxes ()
void	ProcessMatchedBlobs (const TWERD &other, TessCallback1< int > *cb) const
const TBOX &	bounding_box () const
const int	length () const
const TBOX &	BlobBox (int index) const
ScriptPos	BlobPosition (int index) const

Static Public Member Functions
static BoxWord *	CopyFromNormalized (const DENORM *denorm, TWERD *tessword)

Detailed Description

Definition at line 47 of file boxword.h.

Constructor & Destructor Documentation

tesseract::BoxWord::BoxWord

(

)

Definition at line 39 of file boxword.cpp.

                 : length_(0) {
}

tesseract::BoxWord::BoxWord ( const BoxWord &  src )

explicit

Definition at line 42 of file boxword.cpp.

                                   {
  CopyFrom(src);
}

tesseract::BoxWord::~BoxWord

(

)

Definition at line 46 of file boxword.cpp.

                  {
}

Member Function Documentation

const TBOX& tesseract::BoxWord::BlobBox ( int  index ) const

inline

Definition at line 102 of file boxword.h.

                                       {
    return boxes_[index];
  }

ScriptPos tesseract::BoxWord::BlobPosition ( int  index ) const

inline

Definition at line 105 of file boxword.h.

                                          {
    if (index < 0 || index >= script_pos_.size())
      return SP_NORMAL;
    return script_pos_[index];
  }

const TBOX& tesseract::BoxWord::bounding_box ( ) const

inline

Definition at line 96 of file boxword.h.

                                   {
    return bbox_;
  }

void tesseract::BoxWord::ClipToOriginalWord	(	const BLOCK *	block,
		WERD *	original_word
	)

Definition at line 138 of file boxword.cpp.

                                                                        {
  for (int i = 0; i < length_; ++i) {
    TBOX box = boxes_[i];
    // Expand by a single pixel, as the poly approximation error is 1 pixel.
    box = TBOX(box.left() - 1, box.bottom() - 1,
               box.right() + 1, box.top() + 1);
    // Now find the original box that matches.
    TBOX original_box;
    C_BLOB_IT b_it(original_word->cblob_list());
    for (b_it.mark_cycle_pt(); !b_it.cycled_list(); b_it.forward()) {
      TBOX blob_box = b_it.data()->bounding_box();
      if (block != NULL)
        blob_box.rotate(block->re_rotation());
      if (blob_box.major_overlap(box)) {
        original_box += blob_box;
      }
    }
    if (!original_box.null_box()) {
      if (NearlyEqual<int>(original_box.left(), box.left(), kBoxClipTolerance))
        box.set_left(original_box.left());
      if (NearlyEqual<int>(original_box.right(), box.right(),
                           kBoxClipTolerance))
        box.set_right(original_box.right());
      if (NearlyEqual<int>(original_box.top(), box.top(), kBoxClipTolerance))
        box.set_top(original_box.top());
      if (NearlyEqual<int>(original_box.bottom(), box.bottom(),
                           kBoxClipTolerance))
        box.set_bottom(original_box.bottom());
    }
    original_box = original_word->bounding_box();
    if (block != NULL)
      original_box.rotate(block->re_rotation());
    boxes_[i] = box.intersection(original_box);
  }
  ComputeBoundingBox();
}

void tesseract::BoxWord::CopyFrom

(

const BoxWord &

src

)

Definition at line 54 of file boxword.cpp.

                                         {
  bbox_ = src.bbox_;
  length_ = src.length_;
  boxes_.clear();
  boxes_.reserve(length_);
  for (int i = 0; i < length_; ++i)
    boxes_.push_back(src.boxes_[i]);
}

BoxWord * tesseract::BoxWord::CopyFromNormalized ( const DENORM *  denorm, TWERD *  tessword  )

static

Definition at line 67 of file boxword.cpp.

                                                      {
  BoxWord* boxword = new BoxWord();
  // Count the blobs.
  boxword->length_ = 0;
  for (TBLOB* tblob = tessword->blobs; tblob != NULL; tblob = tblob->next)
    ++boxword->length_;
  // Allocate memory.
  boxword->boxes_.reserve(boxword->length_);

  for (TBLOB* tblob = tessword->blobs; tblob != NULL; tblob = tblob->next) {
    TBOX blob_box;
    for (TESSLINE* outline = tblob->outlines; outline != NULL;
         outline = outline->next) {
      EDGEPT* edgept = outline->loop;
      // Iterate over the edges.
      do {
        if (!edgept->IsHidden() || !edgept->prev->IsHidden()) {
          ICOORD pos(edgept->pos.x, edgept->pos.y);
          if (denorm != NULL) {
            TPOINT denormed;
            denorm->DenormTransform(edgept->pos, &denormed);
            pos.set_x(denormed.x);
            pos.set_y(denormed.y);
          }
          TBOX pt_box(pos, pos);
          blob_box += pt_box;
        }
        edgept = edgept->next;
      } while (edgept != outline->loop);
    }
    boxword->boxes_.push_back(blob_box);
  }
  boxword->ComputeBoundingBox();
  return boxword;
}

void tesseract::BoxWord::DeleteAllBoxes

(

)

Definition at line 213 of file boxword.cpp.

                             {
  length_ = 0;
  boxes_.clear();
  bbox_ = TBOX();
}

void tesseract::BoxWord::DeleteBox

(

int

index

)

Definition at line 205 of file boxword.cpp.

                                 {
  ASSERT_HOST(0 <= index && index < length_);
  boxes_.remove(index);
  --length_;
  ComputeBoundingBox();
}

void tesseract::BoxWord::InsertBox	(	int	index,
		const TBOX &	box
	)

Definition at line 194 of file boxword.cpp.

                                                  {
  if (index < length_)
    boxes_.insert(box, index);
  else
    boxes_.push_back(box);
  length_ = boxes_.size();
  ComputeBoundingBox();
}

const int tesseract::BoxWord::length ( ) const

inline

Definition at line 99 of file boxword.h.

                           {
    return length_;
  }

void tesseract::BoxWord::MergeBoxes	(	int	start,
		int	end
	)

Definition at line 177 of file boxword.cpp.

                                           {
  start = ClipToRange(start, 0, length_);
  end = ClipToRange(end, 0, length_);
  if (end <= start + 1)
    return;
  for (int i = start + 1; i < end; ++i) {
    boxes_[start] += boxes_[i];
  }
  int shrinkage = end - 1 - start;
  length_ -= shrinkage;
  for (int i = start + 1; i < length_; ++i)
    boxes_[i] = boxes_[i + shrinkage];
  boxes_.truncate(length_);
}

BoxWord & tesseract::BoxWord::operator=

(

const BoxWord &

src

)

Definition at line 49 of file boxword.cpp.

                                              {
  CopyFrom(src);
  return *this;
}

void tesseract::BoxWord::ProcessMatchedBlobs	(	const TWERD &	other,
		TessCallback1< int > *	cb
	)		const

Definition at line 229 of file boxword.cpp.

                                                                {
  TBLOB* blob = other.blobs;
  for (int i = 0; i < length_ && blob != NULL; ++i, blob = blob->next) {
    TBOX blob_box = blob->bounding_box();
    if (blob_box == boxes_[i])
      cb->Run(i);
  }
  delete cb;
}

void tesseract::BoxWord::SetScriptPositions	(	const UNICHARSET &	unicharset,
		bool	small_caps,
		TWERD *	tessword,
		WERD_CHOICE *	best_choice
	)

Definition at line 108 of file boxword.cpp.

                                                                            {
  // Allocate memory.
  script_pos_.init_to_size(length_, SP_NORMAL);

  int blob_index = 0;
  for (TBLOB* tblob = tessword->blobs; tblob != NULL; tblob = tblob->next,
       ++blob_index) {
    int class_id = best_choice->unichar_id(blob_index);
    TBOX blob_box = tblob->bounding_box();
    int top = blob_box.top();
    int bottom = blob_box.bottom();
    int min_bottom, max_bottom, min_top, max_top;
    unicharset.get_top_bottom(class_id, &min_bottom, &max_bottom,
                              &min_top, &max_top);
    if (bottom <= kMaxDropCapBottom) {
      script_pos_[blob_index] = SP_DROPCAP;
    } else if (!small_caps) {
      if (top + kMinSubscriptOffset < min_top) {
        script_pos_[blob_index] = SP_SUBSCRIPT;
      } else if (bottom - kMinSuperscriptOffset > max_bottom) {
        script_pos_[blob_index] = SP_SUPERSCRIPT;
      }
    }
  }
}

---

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

• /home/abuild/rpmbuild/BUILD/tesseract-ocr-3.02.02/ccstruct/boxword.h
• /home/abuild/rpmbuild/BUILD/tesseract-ocr-3.02.02/ccstruct/boxword.cpp