#include <tabfind.h>

Inheritance diagram for tesseract::TabFind:

Public Member Functions
	TabFind (int gridsize, const ICOORD &bleft, const ICOORD &tright, TabVector_LIST *vlines, int vertical_x, int vertical_y, int resolution)

	~TabFind () override

void	InsertBlobsToGrid (bool h_spread, bool v_spread, BLOBNBOX_LIST blobs, BBGrid< BLOBNBOX, BLOBNBOX_CLIST, BLOBNBOX_C_IT > grid)

bool	InsertBlob (bool h_spread, bool v_spread, BLOBNBOX blob, BBGrid< BLOBNBOX, BLOBNBOX_CLIST, BLOBNBOX_C_IT > grid)

void	SetBlockRuleEdges (TO_BLOCK *block)

void	SetBlobRuleEdges (BLOBNBOX_LIST *blobs)

int	GutterWidth (int bottom_y, int top_y, const TabVector &v, bool ignore_unmergeables, int max_gutter_width, int *required_shift)

void	GutterWidthAndNeighbourGap (int tab_x, int mean_height, int max_gutter, bool left, BLOBNBOX bbox, int gutter_width, int *neighbour_gap)

int	RightEdgeForBox (const TBOX &box, bool crossing, bool extended)

int	LeftEdgeForBox (const TBOX &box, bool crossing, bool extended)

TabVector *	RightTabForBox (const TBOX &box, bool crossing, bool extended)

TabVector *	LeftTabForBox (const TBOX &box, bool crossing, bool extended)

bool	CommonWidth (int width)

WidthCallback	WidthCB ()

const ICOORD &	image_origin () const

Public Member Functions inherited from tesseract::AlignedBlob
	AlignedBlob (int gridsize, const ICOORD &bleft, const ICOORD &tright)

	~AlignedBlob () override

ScrollView *	DisplayTabs (const char window_name, ScrollView tab_win)

TabVector *	FindVerticalAlignment (AlignedBlobParams align_params, BLOBNBOX bbox, int vertical_x, int *vertical_y)

Public Member Functions inherited from tesseract::BlobGrid
	BlobGrid (int gridsize, const ICOORD &bleft, const ICOORD &tright)

	~BlobGrid () override

void	InsertBlobList (BLOBNBOX_LIST *blobs)

Public Member Functions inherited from tesseract::BBGrid< BLOBNBOX, BLOBNBOX_CLIST, BLOBNBOX_C_IT >
	BBGrid ()

	BBGrid (int gridsize, const ICOORD &bleft, const ICOORD &tright)

	~BBGrid () override

void	Init (int gridsize, const ICOORD &bleft, const ICOORD &tright)

void	Clear ()

void	ClearGridData (void(free_method)(BLOBNBOX ))

void	InsertBBox (bool h_spread, bool v_spread, BLOBNBOX *bbox)

void	InsertPixPtBBox (int left, int bottom, Image pix, BLOBNBOX *bbox)

void	RemoveBBox (BLOBNBOX *bbox)

bool	RectangleEmpty (const TBOX &rect)

IntGrid *	CountCellElements ()

ScrollView *	MakeWindow (int x, int y, const char *window_name)

void	DisplayBoxes (ScrollView *window)

void	AssertNoDuplicates ()

virtual void	HandleClick (int x, int y)

Public Member Functions inherited from tesseract::GridBase
	GridBase ()=default

	GridBase (int gridsize, const ICOORD &bleft, const ICOORD &tright)

virtual	~GridBase ()

void	Init (int gridsize, const ICOORD &bleft, const ICOORD &tright)

int	gridsize () const

int	gridwidth () const

int	gridheight () const

const ICOORD &	bleft () const

const ICOORD &	tright () const

void	GridCoords (int x, int y, int grid_x, int grid_y) const

void	ClipGridCoords (int x, int y) const

Static Public Member Functions
static bool	DifferentSizes (int size1, int size2)

static bool	VeryDifferentSizes (int size1, int size2)

Static Public Member Functions inherited from tesseract::AlignedBlob
static bool	WithinTestRegion (int detail_level, int x, int y)

Protected Member Functions
TabVector_LIST *	vectors ()

TabVector_LIST *	dead_vectors ()

bool	FindTabVectors (TabVector_LIST hlines, BLOBNBOX_LIST image_blobs, TO_BLOCK block, int min_gutter_width, double tabfind_aligned_gap_fraction, ColPartitionGrid part_grid, FCOORD deskew, FCOORD reskew)

void	DontFindTabVectors (BLOBNBOX_LIST image_blobs, TO_BLOCK block, FCOORD deskew, FCOORD reskew)

void	TidyBlobs (TO_BLOCK *block)

void	SetupTabSearch (int x, int y, int min_key, int max_key)

ScrollView *	DisplayTabVectors (ScrollView *tab_win)

ScrollView *	FindInitialTabVectors (BLOBNBOX_LIST image_blobs, int min_gutter_width, double tabfind_aligned_gap_fraction, TO_BLOCK block)

void	ResetForVerticalText (const FCOORD &rotate, const FCOORD &rerotate, TabVector_LIST horizontal_lines, int min_gutter_width)

void	Reset ()

void	ReflectInYAxis ()

Static Protected Member Functions
static void	RotateBlobList (const FCOORD &rotation, BLOBNBOX_LIST *blobs)

Protected Attributes
ICOORD	vertical_skew_
	Estimate of true vertical in this image. More...

int	resolution_
	Of source image in pixels per inch. More...

Protected Attributes inherited from tesseract::BBGrid< BLOBNBOX, BLOBNBOX_CLIST, BLOBNBOX_C_IT >
BLOBNBOX_CLIST *	grid_

Protected Attributes inherited from tesseract::GridBase
int	gridsize_

int	gridwidth_

int	gridheight_

int	gridbuckets_

ICOORD	bleft_

ICOORD	tright_

Detailed Description

The TabFind class contains code to find tab-stops and maintain the vectors_ list of tab vectors. Also provides an interface to find neighbouring blobs in the grid of BLOBNBOXes that is used by multiple subclasses. Searching is a complex operation because of the need to enforce rule/separator lines, and tabstop boundaries, (when available), so as the holder of the list of TabVectors this class provides the functions.

Definition at line 52 of file tabfind.h.

Constructor & Destructor Documentation

◆ TabFind()

tesseract::TabFind::TabFind	(	int	gridsize,
		const ICOORD &	bleft,
		const ICOORD &	tright,
		TabVector_LIST *	vlines,
		int	vertical_x,
		int	vertical_y,
		int	resolution
	)

Definition at line 65 of file tabfind.cpp.

    : AlignedBlob(gridsize, bleft, tright)
    , resolution_(resolution)
    , image_origin_(0, tright.y() - 1)
    , v_it_(&vectors_) {
  width_cb_ = nullptr;
  v_it_.add_list_after(vlines);
  SetVerticalSkewAndParallelize(vertical_x, vertical_y);
  using namespace std::placeholders; // for _1
  width_cb_ = std::bind(&TabFind::CommonWidth, this, _1);
}

◆ ~TabFind()

tesseract::TabFind::~TabFind ( )

overridedefault

Member Function Documentation

◆ CommonWidth()

bool tesseract::TabFind::CommonWidth ( int width )

Return true if the given width is close to one of the common widths in column_widths_.

Definition at line 393 of file tabfind.cpp.

                                   {
  width /= kColumnWidthFactor;
  ICOORDELT_IT it(&column_widths_);
  for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
    ICOORDELT *w = it.data();
    if (w->x() - 1 <= width && width <= w->y() + 1) {
      return true;
    }
  }
  return false;
}

◆ dead_vectors()

TabVector_LIST * tesseract::TabFind::dead_vectors ( )

inlineprotected

Definition at line 170 of file tabfind.h.

                                 {
    return &dead_vectors_;
  }

◆ DifferentSizes()

bool tesseract::TabFind::DifferentSizes	(	int	size1,
		int	size2
	)

static

Return true if the sizes are more than a factor of 2 different.

Definition at line 407 of file tabfind.cpp.

                                                 {
  return size1 > size2 * 2 || size2 > size1 * 2;
}

◆ DisplayTabVectors()

ScrollView * tesseract::TabFind::DisplayTabVectors ( ScrollView * tab_win )

protected

Display the tab vectors found in this grid.

Definition at line 495 of file tabfind.cpp.

                                                          {
  // For every vector, display it.
  TabVector_IT it(&vectors_);
  for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
    TabVector *vector = it.data();
    vector->Display(tab_win);
  }
  tab_win->Update();
  return tab_win;
}

◆ DontFindTabVectors()

void tesseract::TabFind::DontFindTabVectors	(	BLOBNBOX_LIST *	image_blobs,
		TO_BLOCK *	block,
		FCOORD *	deskew,
		FCOORD *	reskew
	)

protected

Definition at line 449 of file tabfind.cpp.

                                                 {
  InsertBlobsToGrid(false, false, image_blobs, this);
  InsertBlobsToGrid(true, false, &block->blobs, this);
  deskew->set_x(1.0f);
  deskew->set_y(0.0f);
  reskew->set_x(1.0f);
  reskew->set_y(0.0f);
}

◆ FindInitialTabVectors()

ScrollView * tesseract::TabFind::FindInitialTabVectors	(	BLOBNBOX_LIST *	image_blobs,
		int	min_gutter_width,
		double	tabfind_aligned_gap_fraction,
		TO_BLOCK *	block
	)

protected

Definition at line 512 of file tabfind.cpp.

                                                                                                 {
#ifndef GRAPHICS_DISABLED
  if (textord_tabfind_show_initialtabs) {
    ScrollView *line_win = MakeWindow(0, 0, "VerticalLines");
    line_win = DisplayTabVectors(line_win);
  }
#endif
  // Prepare the grid.
  if (image_blobs != nullptr) {
    InsertBlobsToGrid(true, false, image_blobs, this);
  }
  InsertBlobsToGrid(true, false, &block->blobs, this);
  ScrollView *initial_win = FindTabBoxes(min_gutter_width, tabfind_aligned_gap_fraction);
  FindAllTabVectors(min_gutter_width);
 
  TabVector::MergeSimilarTabVectors(vertical_skew_, &vectors_, this);
  SortVectors();
  EvaluateTabs();
#ifndef GRAPHICS_DISABLED
  if (textord_tabfind_show_initialtabs && initial_win != nullptr) {
    initial_win = DisplayTabVectors(initial_win);
  }
#endif
  MarkVerticalText();
  return initial_win;
}

◆ FindTabVectors()

bool tesseract::TabFind::FindTabVectors	(	TabVector_LIST *	hlines,
		BLOBNBOX_LIST *	image_blobs,
		TO_BLOCK *	block,
		int	min_gutter_width,
		double	tabfind_aligned_gap_fraction,
		ColPartitionGrid *	part_grid,
		FCOORD *	deskew,
		FCOORD *	reskew
	)

protected

Top-level function to find TabVectors in an input page block. Returns false if the detected skew angle is impossible. Applies the detected skew angle to deskew the tabs, blobs and part_grid. tabfind_aligned_gap_fraction should be the value of parameter textord_tabfind_aligned_gap_fraction

Definition at line 422 of file tabfind.cpp.

                                                                                          {
  ScrollView *tab_win =
      FindInitialTabVectors(image_blobs, min_gutter_width, tabfind_aligned_gap_fraction, block);
  ComputeColumnWidths(tab_win, part_grid);
  TabVector::MergeSimilarTabVectors(vertical_skew_, &vectors_, this);
  SortVectors();
  CleanupTabs();
  if (!Deskew(hlines, image_blobs, block, deskew, reskew)) {
    return false; // Skew angle is too large.
  }
  part_grid->Deskew(*deskew);
  ApplyTabConstraints();
#ifndef GRAPHICS_DISABLED
  if (textord_tabfind_show_finaltabs) {
    tab_win = MakeWindow(640, 50, "FinalTabs");
    DisplayBoxes(tab_win);
    DisplayTabs("FinalTabs", tab_win);
    tab_win = DisplayTabVectors(tab_win);
  }
#endif // !GRAPHICS_DISABLED
  return true;
}

◆ GutterWidth()

int tesseract::TabFind::GutterWidth	(	int	bottom_y,
		int	top_y,
		const TabVector &	v,
		bool	ignore_unmergeables,
		int	max_gutter_width,
		int *	required_shift
	)

Definition at line 156 of file tabfind.cpp.

                                                                    {
  bool right_to_left = v.IsLeftTab();
  int bottom_x = v.XAtY(bottom_y);
  int top_x = v.XAtY(top_y);
  int start_x = right_to_left ? std::max(top_x, bottom_x) : std::min(top_x, bottom_x);
  BlobGridSearch sidesearch(this);
  sidesearch.StartSideSearch(start_x, bottom_y, top_y);
  int min_gap = max_gutter_width;
  *required_shift = 0;
  BLOBNBOX *blob = nullptr;
  while ((blob = sidesearch.NextSideSearch(right_to_left)) != nullptr) {
    const TBOX &box = blob->bounding_box();
    if (box.bottom() >= top_y || box.top() <= bottom_y) {
      continue; // Doesn't overlap enough.
    }
    if (box.height() >= gridsize() * 2 && box.height() > box.width() * kLineFragmentAspectRatio) {
      // Skip likely separator line residue.
      continue;
    }
    if (ignore_unmergeables && BLOBNBOX::UnMergeableType(blob->region_type())) {
      continue; // Skip non-text if required.
    }
    int mid_y = (box.bottom() + box.top()) / 2;
    // We use the x at the mid-y so that the required_shift guarantees
    // to clear all the blobs on the tab-stop. If we use the min/max
    // of x at top/bottom of the blob, then exactness would be required,
    // which is not a good thing.
    int tab_x = v.XAtY(mid_y);
    int gap;
    if (right_to_left) {
      gap = tab_x - box.right();
      if (gap < 0 && box.left() - tab_x < *required_shift) {
        *required_shift = box.left() - tab_x;
      }
    } else {
      gap = box.left() - tab_x;
      if (gap < 0 && box.right() - tab_x > *required_shift) {
        *required_shift = box.right() - tab_x;
      }
    }
    if (gap > 0 && gap < min_gap) {
      min_gap = gap;
    }
  }
  // Result may be negative, in which case,  this is a really bad tabstop.
  return min_gap - abs(*required_shift);
}

◆ GutterWidthAndNeighbourGap()

void tesseract::TabFind::GutterWidthAndNeighbourGap	(	int	tab_x,
		int	mean_height,
		int	max_gutter,
		bool	left,
		BLOBNBOX *	bbox,
		int *	gutter_width,
		int *	neighbour_gap
	)

Find the gutter width and distance to inner neighbour for the given blob.

Definition at line 206 of file tabfind.cpp.

                                                                                                {
  const TBOX &box = bbox->bounding_box();
  // The gutter and internal sides of the box.
  int gutter_x = left ? box.left() : box.right();
  int internal_x = left ? box.right() : box.left();
  // On ragged edges, the gutter side of the box is away from the tabstop.
  int tab_gap = left ? gutter_x - tab_x : tab_x - gutter_x;
  *gutter_width = max_gutter;
  // If the box is away from the tabstop, we need to increase
  // the allowed gutter width.
  if (tab_gap > 0) {
    *gutter_width += tab_gap;
  }
  bool debug = WithinTestRegion(2, box.left(), box.bottom());
  if (debug) {
    tprintf("Looking in gutter\n");
  }
  // Find the nearest blob on the outside of the column.
  BLOBNBOX *gutter_bbox = AdjacentBlob(bbox, left, bbox->flow() == BTFT_TEXT_ON_IMAGE, 0.0,
                                       *gutter_width, box.top(), box.bottom());
  if (gutter_bbox != nullptr) {
    const TBOX &gutter_box = gutter_bbox->bounding_box();
    *gutter_width = left ? tab_x - gutter_box.right() : gutter_box.left() - tab_x;
  }
  if (*gutter_width >= max_gutter) {
    // If there is no box because a tab was in the way, get the tab coord.
    TBOX gutter_box(box);
    if (left) {
      gutter_box.set_left(tab_x - max_gutter - 1);
      gutter_box.set_right(tab_x - max_gutter);
      int tab_gutter = RightEdgeForBox(gutter_box, true, false);
      if (tab_gutter < tab_x - 1) {
        *gutter_width = tab_x - tab_gutter;
      }
    } else {
      gutter_box.set_left(tab_x + max_gutter);
      gutter_box.set_right(tab_x + max_gutter + 1);
      int tab_gutter = LeftEdgeForBox(gutter_box, true, false);
      if (tab_gutter > tab_x + 1) {
        *gutter_width = tab_gutter - tab_x;
      }
    }
  }
  if (*gutter_width > max_gutter) {
    *gutter_width = max_gutter;
  }
  // Now look for a neighbour on the inside.
  if (debug) {
    tprintf("Looking for neighbour\n");
  }
  BLOBNBOX *neighbour = AdjacentBlob(bbox, !left, bbox->flow() == BTFT_TEXT_ON_IMAGE, 0.0,
                                     *gutter_width, box.top(), box.bottom());
  int neighbour_edge = left ? RightEdgeForBox(box, true, false) : LeftEdgeForBox(box, true, false);
  if (neighbour != nullptr) {
    const TBOX &n_box = neighbour->bounding_box();
    if (debug) {
      tprintf("Found neighbour:");
      n_box.print();
    }
    if (left && n_box.left() < neighbour_edge) {
      neighbour_edge = n_box.left();
    } else if (!left && n_box.right() > neighbour_edge) {
      neighbour_edge = n_box.right();
    }
  }
  *neighbour_gap = left ? neighbour_edge - internal_x : internal_x - neighbour_edge;
}

◆ image_origin()

const ICOORD & tesseract::TabFind::image_origin ( ) const

inline

Return the coords at which to draw the image backdrop.

Definition at line 159 of file tabfind.h.

                                     {
    return image_origin_;
  }

◆ InsertBlob()

bool tesseract::TabFind::InsertBlob	(	bool	h_spread,
		bool	v_spread,
		BLOBNBOX *	blob,
		BBGrid< BLOBNBOX, BLOBNBOX_CLIST, BLOBNBOX_C_IT > *	grid
	)

Insert a single blob into the given grid (not necessarily this). If h_spread, then all cells covered horizontally by the box are used, otherwise, just the bottom-left. Similarly for v_spread. A side effect is that the left and right rule edges of the blob are set according to the tab vectors in this (not grid).

Definition at line 113 of file tabfind.cpp.

                                                                                {
  TBOX box = blob->bounding_box();
  blob->set_left_rule(LeftEdgeForBox(box, false, false));
  blob->set_right_rule(RightEdgeForBox(box, false, false));
  blob->set_left_crossing_rule(LeftEdgeForBox(box, true, false));
  blob->set_right_crossing_rule(RightEdgeForBox(box, true, false));
  if (blob->joined_to_prev()) {
    return false;
  }
  grid->InsertBBox(h_spread, v_spread, blob);
  return true;
}

◆ InsertBlobsToGrid()

void tesseract::TabFind::InsertBlobsToGrid	(	bool	h_spread,
		bool	v_spread,
		BLOBNBOX_LIST *	blobs,
		BBGrid< BLOBNBOX, BLOBNBOX_CLIST, BLOBNBOX_C_IT > *	grid
	)

Insert a list of blobs into the given grid (not necessarily this). See InsertBlob for the other arguments. It would seem to make more sense to swap this and grid, but this way around allows grid to not be derived from TabFind, eg a ColPartitionGrid, while the grid that provides the tab stops(this) has to be derived from TabFind.

Definition at line 89 of file tabfind.cpp.

                                                                                       {
  BLOBNBOX_IT blob_it(blobs);
  int b_count = 0;
  int reject_count = 0;
  for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
    BLOBNBOX *blob = blob_it.data();
    //    if (InsertBlob(true, true, blob, grid)) {
    if (InsertBlob(h_spread, v_spread, blob, grid)) {
      ++b_count;
    } else {
      ++reject_count;
    }
  }
  if (textord_debug_tabfind) {
    tprintf("Inserted %d blobs into grid, %d rejected.\n", b_count, reject_count);
  }
}

◆ LeftEdgeForBox()

int tesseract::TabFind::LeftEdgeForBox	(	const TBOX &	box,
		bool	crossing,
		bool	extended
	)

As RightEdgeForBox, but finds the left Edge instead.

Definition at line 284 of file tabfind.cpp.

                                                                         {
  TabVector *v = LeftTabForBox(box, crossing, extended);
  return v == nullptr ? bleft_.x() : v->XAtY((box.top() + box.bottom()) / 2);
}

◆ LeftTabForBox()

TabVector * tesseract::TabFind::LeftTabForBox	(	const TBOX &	box,
		bool	crossing,
		bool	extended
	)

As RightTabForBox, but finds the left TabVector instead.

Definition at line 347 of file tabfind.cpp.

                                                                               {
  if (v_it_.empty()) {
    return nullptr;
  }
  int top_y = box.top();
  int bottom_y = box.bottom();
  int mid_y = (top_y + bottom_y) / 2;
  int left = crossing ? (box.left() + box.right()) / 2 : box.left();
  int min_key, max_key;
  SetupTabSearch(left, mid_y, &min_key, &max_key);
  // Position the iterator at the last TabVector with sort_key <= max_key.
  while (!v_it_.at_last() && v_it_.data()->sort_key() <= max_key) {
    v_it_.forward();
  }
  while (!v_it_.at_first() && v_it_.data()->sort_key() > max_key) {
    v_it_.backward();
  }
  // Find the rightmost tab vector that overlaps and has XAtY(mid_y) <= left.
  TabVector *best_v = nullptr;
  int best_x = -1;
  int key_limit = -1;
  do {
    TabVector *v = v_it_.data();
    int x = v->XAtY(mid_y);
    if (x <= left && (v->VOverlap(top_y, bottom_y) > 0 ||
                      (extended && v->ExtendedOverlap(top_y, bottom_y) > 0))) {
      if (best_v == nullptr || x > best_x) {
        best_v = v;
        best_x = x;
        // We can guarantee that no better vector can be found if the
        // sort key is less than that of the best by max_key - min_key.
        key_limit = v->sort_key() - (max_key - min_key);
      }
    }
    // Break when the search is done to avoid wrapping the iterator and
    // thereby potentially slowing the next search.
    if (v_it_.at_first() || (best_v != nullptr && v->sort_key() < key_limit)) {
      break; // Prevent restarting list for next call.
    }
    v_it_.backward();
  } while (!v_it_.at_last());
  return best_v;
}

◆ ReflectInYAxis()

void tesseract::TabFind::ReflectInYAxis ( )

protected

Definition at line 1380 of file tabfind.cpp.

                             {
  TabVector_LIST temp_list;
  TabVector_IT temp_it(&temp_list);
  v_it_.move_to_first();
  // The TabVector list only contains vertical lines, but they need to be
  // reflected and the list needs to be reversed, so they are still in
  // sort_key order.
  while (!v_it_.empty()) {
    TabVector *v = v_it_.extract();
    v_it_.forward();
    v->ReflectInYAxis();
    temp_it.add_before_then_move(v);
  }
  v_it_.add_list_after(&temp_list);
  v_it_.move_to_first();
  // Reset this grid with reflected bounding boxes.
  TBOX grid_box(bleft(), tright());
  int tmp = grid_box.left();
  grid_box.set_left(-grid_box.right());
  grid_box.set_right(-tmp);
  Init(gridsize(), grid_box.botleft(), grid_box.topright());
}

◆ Reset()

void tesseract::TabFind::Reset ( )

protected

Definition at line 1368 of file tabfind.cpp.

                    {
  v_it_.move_to_first();
  for (v_it_.mark_cycle_pt(); !v_it_.cycled_list(); v_it_.forward()) {
    if (!v_it_.data()->IsSeparator()) {
      delete v_it_.extract();
    }
  }
  Clear();
}

◆ ResetForVerticalText()

void tesseract::TabFind::ResetForVerticalText	(	const FCOORD &	rotate,
		const FCOORD &	rerotate,
		TabVector_LIST *	horizontal_lines,
		int *	min_gutter_width
	)

protected

Definition at line 1323 of file tabfind.cpp.

                                                                                            {
  // Rotate the horizontal and vertical vectors and swap them over.
  // Only the separators are kept and rotated; other tabs are used
  // to estimate the gutter width then thrown away.
  TabVector_LIST ex_verticals;
  TabVector_IT ex_v_it(&ex_verticals);
  TabVector_LIST vlines;
  TabVector_IT v_it(&vlines);
  while (!v_it_.empty()) {
    TabVector *v = v_it_.extract();
    if (v->IsSeparator()) {
      v->Rotate(rotate);
      ex_v_it.add_after_then_move(v);
    } else {
      v_it.add_after_then_move(v);
    }
    v_it_.forward();
  }
 
  // Adjust the min gutter width for better tabbox selection
  // in 2nd call to FindInitialTabVectors().
  int median_gutter = FindMedianGutterWidth(&vlines);
  if (median_gutter > *min_gutter_width) {
    *min_gutter_width = median_gutter;
  }
 
  TabVector_IT h_it(horizontal_lines);
  for (h_it.mark_cycle_pt(); !h_it.cycled_list(); h_it.forward()) {
    TabVector *h = h_it.data();
    h->Rotate(rotate);
  }
  v_it_.add_list_after(horizontal_lines);
  v_it_.move_to_first();
  h_it.set_to_list(horizontal_lines);
  h_it.add_list_after(&ex_verticals);
 
  // Rebuild the grid to the new size.
  TBOX grid_box(bleft(), tright());
  grid_box.rotate_large(rotate);
  Init(gridsize(), grid_box.botleft(), grid_box.topright());
}

◆ RightEdgeForBox()

int tesseract::TabFind::RightEdgeForBox	(	const TBOX &	box,
		bool	crossing,
		bool	extended
	)

Return the x-coord that corresponds to the right edge for the given box. If there is a rule line to the right that vertically overlaps it, then return the x-coord of the rule line, otherwise return the right edge of the page. For details see RightTabForBox below.

Definition at line 279 of file tabfind.cpp.

                                                                          {
  TabVector *v = RightTabForBox(box, crossing, extended);
  return v == nullptr ? tright_.x() : v->XAtY((box.top() + box.bottom()) / 2);
}

◆ RightTabForBox()

TabVector * tesseract::TabFind::RightTabForBox	(	const TBOX &	box,
		bool	crossing,
		bool	extended
	)

Return the TabVector that corresponds to the right edge for the given box. If there is a TabVector to the right that vertically overlaps it, then return it, otherwise return nullptr. Note that Right and Left refer to the position of the TabVector, not its type, ie RightTabForBox returns the nearest TabVector to the right of the box, regardless of its type. If a TabVector crosses right through the box (as opposed to grazing one edge or missing entirely), then crossing false will ignore such a line. Crossing true will return the line for BOTH left and right edges. If extended is true, then TabVectors are considered to extend to their extended_start/end_y, otherwise, just the startpt_ and endpt_. These functions make use of an internal iterator to the vectors_ list for speed when used repeatedly on neighbouring boxes. The caveat is that the iterator must be updated whenever the list is modified.

Definition at line 302 of file tabfind.cpp.

                                                                                {
  if (v_it_.empty()) {
    return nullptr;
  }
  int top_y = box.top();
  int bottom_y = box.bottom();
  int mid_y = (top_y + bottom_y) / 2;
  int right = crossing ? (box.left() + box.right()) / 2 : box.right();
  int min_key, max_key;
  SetupTabSearch(right, mid_y, &min_key, &max_key);
  // Position the iterator at the first TabVector with sort_key >= min_key.
  while (!v_it_.at_first() && v_it_.data()->sort_key() >= min_key) {
    v_it_.backward();
  }
  while (!v_it_.at_last() && v_it_.data()->sort_key() < min_key) {
    v_it_.forward();
  }
  // Find the leftmost tab vector that overlaps and has XAtY(mid_y) >= right.
  TabVector *best_v = nullptr;
  int best_x = -1;
  int key_limit = -1;
  do {
    TabVector *v = v_it_.data();
    int x = v->XAtY(mid_y);
    if (x >= right && (v->VOverlap(top_y, bottom_y) > 0 ||
                       (extended && v->ExtendedOverlap(top_y, bottom_y) > 0))) {
      if (best_v == nullptr || x < best_x) {
        best_v = v;
        best_x = x;
        // We can guarantee that no better vector can be found if the
        // sort key exceeds that of the best by max_key - min_key.
        key_limit = v->sort_key() + max_key - min_key;
      }
    }
    // Break when the search is done to avoid wrapping the iterator and
    // thereby potentially slowing the next search.
    if (v_it_.at_last() || (best_v != nullptr && v->sort_key() > key_limit)) {
      break; // Prevent restarting list for next call.
    }
    v_it_.forward();
  } while (!v_it_.at_first());
  return best_v;
}

◆ RotateBlobList()

void tesseract::TabFind::RotateBlobList	(	const FCOORD &	rotation,
		BLOBNBOX_LIST *	blobs
	)

staticprotected

Definition at line 1278 of file tabfind.cpp.

                                                                         {
  BLOBNBOX_IT it(blobs);
  for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
    it.data()->rotate_box(rotation);
  }
}

◆ SetBlobRuleEdges()

void tesseract::TabFind::SetBlobRuleEdges ( BLOBNBOX_LIST * blobs )

Definition at line 137 of file tabfind.cpp.

                                                   {
  BLOBNBOX_IT blob_it(blobs);
  for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
    BLOBNBOX *blob = blob_it.data();
    TBOX box = blob->bounding_box();
    blob->set_left_rule(LeftEdgeForBox(box, false, false));
    blob->set_right_rule(RightEdgeForBox(box, false, false));
    blob->set_left_crossing_rule(LeftEdgeForBox(box, true, false));
    blob->set_right_crossing_rule(RightEdgeForBox(box, true, false));
  }
}

◆ SetBlockRuleEdges()

void tesseract::TabFind::SetBlockRuleEdges ( TO_BLOCK * block )

Definition at line 128 of file tabfind.cpp.

                                               {
  SetBlobRuleEdges(&block->blobs);
  SetBlobRuleEdges(&block->small_blobs);
  SetBlobRuleEdges(&block->noise_blobs);
  SetBlobRuleEdges(&block->large_blobs);
}

◆ SetupTabSearch()

void tesseract::TabFind::SetupTabSearch	(	int	x,
		int	y,
		int *	min_key,
		int *	max_key
	)

protected

Definition at line 486 of file tabfind.cpp.

                                                                     {
  int key1 = TabVector::SortKey(vertical_skew_, x, (y + tright_.y()) / 2);
  int key2 = TabVector::SortKey(vertical_skew_, x, (y + bleft_.y()) / 2);
  *min_key = std::min(key1, key2);
  *max_key = std::max(key1, key2);
}

◆ TidyBlobs()

void tesseract::TabFind::TidyBlobs ( TO_BLOCK * block )

protected

Definition at line 462 of file tabfind.cpp.

                                       {
  BLOBNBOX_IT large_it = &block->large_blobs;
  BLOBNBOX_IT blob_it = &block->blobs;
  int b_count = 0;
  for (large_it.mark_cycle_pt(); !large_it.cycled_list(); large_it.forward()) {
    BLOBNBOX *large_blob = large_it.data();
    if (large_blob->owner() != nullptr) {
      blob_it.add_to_end(large_it.extract());
      ++b_count;
    }
  }
  if (textord_debug_tabfind) {
    tprintf("Moved %d large blobs to normal list\n", b_count);
#ifndef GRAPHICS_DISABLED
    ScrollView *rej_win = MakeWindow(500, 300, "Image blobs");
    block->plot_graded_blobs(rej_win);
    block->plot_noise_blobs(rej_win);
    rej_win->Update();
#endif // !GRAPHICS_DISABLED
  }
  block->DeleteUnownedNoise();
}

◆ vectors()

TabVector_LIST * tesseract::TabFind::vectors ( )

inlineprotected

Accessors

Definition at line 167 of file tabfind.h.

                            {
    return &vectors_;
  }

◆ VeryDifferentSizes()

bool tesseract::TabFind::VeryDifferentSizes	(	int	size1,
		int	size2
	)

static

Return true if the sizes are more than a factor of 5 different.

Definition at line 413 of file tabfind.cpp.

                                                     {
  return size1 > size2 * 5 || size2 > size1 * 5;
}

◆ WidthCB()

WidthCallback tesseract::TabFind::WidthCB ( )

inline

Return a callback for testing CommonWidth.

Definition at line 152 of file tabfind.h.

                          {
    return width_cb_;
  }

Member Data Documentation

◆ resolution_

int tesseract::TabFind::resolution_

protected

Of source image in pixels per inch.

Definition at line 346 of file tabfind.h.

◆ vertical_skew_

ICOORD tesseract::TabFind::vertical_skew_

protected

Estimate of true vertical in this image.

Definition at line 345 of file tabfind.h.

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

/media/home/debian/src/github/tesseract-ocr/tesseract/src/textord/tabfind.h
/media/home/debian/src/github/tesseract-ocr/tesseract/src/textord/tabfind.cpp

Public Member Functions

Static Public Member Functions

Protected Member Functions

Static Protected Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

◆ TabFind()

◆ ~TabFind()

Member Function Documentation

◆ CommonWidth()

◆ dead_vectors()

◆ DifferentSizes()

◆ DisplayTabVectors()

◆ DontFindTabVectors()

◆ FindInitialTabVectors()

◆ FindTabVectors()

◆ GutterWidth()

◆ GutterWidthAndNeighbourGap()

◆ image_origin()

◆ InsertBlob()

◆ InsertBlobsToGrid()

◆ LeftEdgeForBox()

◆ LeftTabForBox()

◆ ReflectInYAxis()

◆ Reset()

◆ ResetForVerticalText()

◆ RightEdgeForBox()

◆ RightTabForBox()

◆ RotateBlobList()

◆ SetBlobRuleEdges()

◆ SetBlockRuleEdges()

◆ SetupTabSearch()

◆ TidyBlobs()

◆ vectors()

◆ VeryDifferentSizes()

◆ WidthCB()

Member Data Documentation

◆ resolution_

◆ vertical_skew_