tesseract::TabFind Class Reference

#include <tabfind.h>

Inheritance diagram for tesseract::TabFind:

Collaboration 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)
virtual	~TabFind ()
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)
virtual	~AlignedBlob ()
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)
virtual	~BlobGrid ()
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)
virtual	~BBGrid ()
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, Pix *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_
int	resolution_
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_

Private Member Functions
ScrollView *	FindTabBoxes (int min_gutter_width, double tabfind_aligned_gap_fraction)
bool	TestBoxForTabs (BLOBNBOX *bbox, int min_gutter_width, double tabfind_aligned_gap_fraction)
bool	ConfirmRaggedLeft (BLOBNBOX *bbox, int min_gutter)
bool	ConfirmRaggedRight (BLOBNBOX *bbox, int min_gutter)
bool	NothingYOverlapsInBox (const TBOX &search_box, const TBOX &target_box)
void	FindAllTabVectors (int min_gutter_width)
int	FindTabVectors (int search_size_multiple, TabAlignment alignment, int min_gutter_width, TabVector_LIST *vectors, int *vertical_x, int *vertical_y)
TabVector *	FindTabVector (int search_size_multiple, int min_gutter_width, TabAlignment alignment, BLOBNBOX *bbox, int *vertical_x, int *vertical_y)
void	SetVerticalSkewAndParallelize (int vertical_x, int vertical_y)
void	SortVectors ()
void	EvaluateTabs ()
void	ComputeColumnWidths (ScrollView *tab_win, ColPartitionGrid *part_grid)
void	ApplyPartitionsToColumnWidths (ColPartitionGrid *part_grid, STATS *col_widths)
void	MakeColumnWidths (int col_widths_size, STATS *col_widths)
void	MarkVerticalText ()
int	FindMedianGutterWidth (TabVector_LIST *tab_vectors)
BLOBNBOX *	AdjacentBlob (const BLOBNBOX *bbox, bool look_left, bool ignore_images, double min_overlap_fraction, int gap_limit, int top_y, int bottom_y)
void	AddPartnerVector (BLOBNBOX *left_blob, BLOBNBOX *right_blob, TabVector *left, TabVector *right)
void	CleanupTabs ()
bool	Deskew (TabVector_LIST *hlines, BLOBNBOX_LIST *image_blobs, TO_BLOCK *block, FCOORD *deskew, FCOORD *reskew)
void	ComputeDeskewVectors (FCOORD *deskew, FCOORD *reskew)
void	ApplyTabConstraints ()

Private Attributes
ICOORD	image_origin_
TabVector_LIST	vectors_
TabVector_IT	v_it_
TabVector_LIST	dead_vectors_
ICOORDELT_LIST	column_widths_
WidthCallback *	width_cb_
GenericVector< BLOBNBOX * >	left_tab_boxes_
GenericVector< BLOBNBOX * >	right_tab_boxes_

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.

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
	)

~TabFind()

tesseract::TabFind::~TabFind ( )

virtual

Member Function Documentation

AddPartnerVector()

void tesseract::TabFind::AddPartnerVector ( BLOBNBOX *  left_blob, BLOBNBOX *  right_blob, TabVector *  left, TabVector *  right  )

private

AdjacentBlob()

BLOBNBOX * tesseract::TabFind::AdjacentBlob ( const BLOBNBOX *  bbox, bool  look_left, bool  ignore_images, double  min_overlap_fraction, int  gap_limit, int  top_y, int  bottom_y  )

private

ApplyPartitionsToColumnWidths()

void tesseract::TabFind::ApplyPartitionsToColumnWidths ( ColPartitionGrid *  part_grid, STATS *  col_widths  )

private

ApplyTabConstraints()

void tesseract::TabFind::ApplyTabConstraints ( )

private

Compute and apply constraints to the end positions of TabVectors so that where possible partners end at the same y coordinate.

CleanupTabs()

void tesseract::TabFind::CleanupTabs ( )

private

Remove separators and unused tabs from the main vectors_ list to the dead_vectors_ list.

CommonWidth()

bool tesseract::TabFind::CommonWidth

(

int

width

)

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

ComputeColumnWidths()

void tesseract::TabFind::ComputeColumnWidths ( ScrollView *  tab_win, ColPartitionGrid *  part_grid  )

private

ComputeDeskewVectors()

void tesseract::TabFind::ComputeDeskewVectors ( FCOORD *  deskew, FCOORD *  reskew  )

private

ConfirmRaggedLeft()

bool tesseract::TabFind::ConfirmRaggedLeft ( BLOBNBOX *  bbox, int  min_gutter  )

private

ConfirmRaggedRight()

bool tesseract::TabFind::ConfirmRaggedRight ( BLOBNBOX *  bbox, int  min_gutter  )

private

dead_vectors()

TabVector_LIST* tesseract::TabFind::dead_vectors ( )

inlineprotected

Deskew()

bool tesseract::TabFind::Deskew ( TabVector_LIST *  hlines, BLOBNBOX_LIST *  image_blobs, TO_BLOCK *  block, FCOORD *  deskew, FCOORD *  reskew  )

private

Deskew the tab vectors and blobs, computing the rotation and resetting the storked vertical_skew_. The deskew inverse is returned in reskew. Returns false if the detected skew angle is impossible.

DifferentSizes()

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

static

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

DisplayTabVectors()

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

protected

Display the tab vectors found in this grid.

DontFindTabVectors()

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

protected

EvaluateTabs()

void tesseract::TabFind::EvaluateTabs ( )

private

FindAllTabVectors()

void tesseract::TabFind::FindAllTabVectors ( int  min_gutter_width )

private

FindInitialTabVectors()

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

protected

FindMedianGutterWidth()

int tesseract::TabFind::FindMedianGutterWidth ( TabVector_LIST *  tab_vectors )

private

FindTabBoxes()

ScrollView * tesseract::TabFind::FindTabBoxes ( int  min_gutter_width, double  tabfind_aligned_gap_fraction  )

private

FindTabVector()

TabVector * tesseract::TabFind::FindTabVector ( int  search_size_multiple, int  min_gutter_width, TabAlignment  alignment, BLOBNBOX *  bbox, int *  vertical_x, int *  vertical_y  )

private

FindTabVectors() [1/2]

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

FindTabVectors() [2/2]

int tesseract::TabFind::FindTabVectors ( int  search_size_multiple, TabAlignment  alignment, int  min_gutter_width, TabVector_LIST *  vectors, int *  vertical_x, int *  vertical_y  )

private

GutterWidth()

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

image_origin()

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

inline

Return the coords at which to draw the image backdrop.

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).

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.

LeftEdgeForBox()

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

As RightEdgeForBox, but finds the left Edge instead.

LeftTabForBox()

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

As RightTabForBox, but finds the left TabVector instead.

MakeColumnWidths()

void tesseract::TabFind::MakeColumnWidths ( int  col_widths_size, STATS *  col_widths  )

private

MarkVerticalText()

void tesseract::TabFind::MarkVerticalText ( )

private

NothingYOverlapsInBox()

bool tesseract::TabFind::NothingYOverlapsInBox ( const TBOX &  search_box, const TBOX &  target_box  )

private

ReflectInYAxis()

void tesseract::TabFind::ReflectInYAxis ( )

protected

Reset()

void tesseract::TabFind::Reset ( )

protected

ResetForVerticalText()

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

protected

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.

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.

RotateBlobList()

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

staticprotected

SetBlobRuleEdges()

void tesseract::TabFind::SetBlobRuleEdges

(

BLOBNBOX_LIST *

blobs

)

SetBlockRuleEdges()

void tesseract::TabFind::SetBlockRuleEdges

(

TO_BLOCK *

block

)

SetupTabSearch()

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

protected

SetVerticalSkewAndParallelize()

void tesseract::TabFind::SetVerticalSkewAndParallelize ( int  vertical_x, int  vertical_y  )

private

SortVectors()

void tesseract::TabFind::SortVectors ( )

private

TestBoxForTabs()

bool tesseract::TabFind::TestBoxForTabs ( BLOBNBOX *  bbox, int  min_gutter_width, double  tabfind_aligned_gap_fraction  )

private

TidyBlobs()

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

protected

vectors()

TabVector_LIST* tesseract::TabFind::vectors ( )

inlineprotected

Accessors

VeryDifferentSizes()

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

static

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

WidthCB()

WidthCallback* tesseract::TabFind::WidthCB ( )

inline

Return a callback for testing CommonWidth.

Member Data Documentation

column_widths_

ICOORDELT_LIST tesseract::TabFind::column_widths_

private

dead_vectors_

TabVector_LIST tesseract::TabFind::dead_vectors_

private

image_origin_

ICOORD tesseract::TabFind::image_origin_

private

left_tab_boxes_

GenericVector<BLOBNBOX*> tesseract::TabFind::left_tab_boxes_

private

resolution_

int tesseract::TabFind::resolution_

protected

right_tab_boxes_

GenericVector<BLOBNBOX*> tesseract::TabFind::right_tab_boxes_

private

v_it_

TabVector_IT tesseract::TabFind::v_it_

private

vectors_

TabVector_LIST tesseract::TabFind::vectors_

private

vertical_skew_

ICOORD tesseract::TabFind::vertical_skew_

protected

width_cb_

WidthCallback* tesseract::TabFind::width_cb_

private

Callback to test an int for being a common width.

---

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

• /home/stephane/src/tesseract/src/textord/tabfind.h
• /home/stephane/src/tesseract/src/textord/tabfind.cpp