#include "makerow.h"
#include "blkocc.h"
#include "blobbox.h"
#include "ccstruct.h"
#include "detlinefit.h"
#include "drawtord.h"
#include "oldbasel.h"
#include "sortflts.h"
#include "statistc.h"
#include "textord.h"
#include "tordmain.h"
#include "tovars.h"
#include "tprintf.h"
#include "underlin.h"
#include <algorithm>
#include <cmath>
#include <vector>

Namespaces
namespace	tesseract

Macros
#define	MAX_HEIGHT_MODES 12

Functions
row_y_order
Sort function to sort rows in y from page top.
row_spacing_order
Qsort style function to compare 2 TO_ROWS based on their spacing value.
make_single_row
Arrange the blobs into a single row... well actually, if there is only a single blob, it makes 2 rows, in case the top-level blob is a container of the real blobs to recognize.
float	tesseract::make_single_row (ICOORD page_tr, bool allow_sub_blobs, TO_BLOCK block, TO_BLOCK_LIST blocks)

make_rows
Arrange the blobs into rows.
float	tesseract::make_rows (ICOORD page_tr, TO_BLOCK_LIST *port_blocks)

make_initial_textrows
Arrange the good blobs into rows of text.
void	tesseract::make_initial_textrows (ICOORD page_tr, TO_BLOCK *block, FCOORD rotation, bool testing_on)

fit_lms_line
Fit an LMS line to a row.
void	tesseract::fit_lms_line (TO_ROW *row)

find_best_dropout_row
Delete this row if it has a neighbour with better dropout characteristics. true is returned if the row should be deleted.
bool	tesseract::find_best_dropout_row (TO_ROW row, int32_t distance, float dist_limit, int32_t line_index, TO_ROW_IT row_it, bool testing_on)

deskew_block_coords
Compute the bounding box of all the blobs in the block if they were deskewed without actually doing it.
TBOX	tesseract::deskew_block_coords (TO_BLOCK *block, float gradient)

compute_line_occupation
Compute the pixel projection back on the y axis given the global skew. Also compute the 1st derivative.
void	tesseract::compute_line_occupation (TO_BLOCK block, float gradient, int32_t min_y, int32_t max_y, int32_t occupation, int32_t *deltas)

void	tesseract::compute_occupation_threshold (int32_t low_window, int32_t high_window, int32_t line_count, int32_t occupation, int32_t thresholds)

compute_dropout_distances
Compute the distance from each coordinate to the nearest dropout.
void	tesseract::compute_dropout_distances (int32_t occupation, int32_t thresholds, int32_t line_count)

expand_rows
Expand each row to the least of its allowed size and touching its neighbours. If the expansion would entirely swallow a neighbouring row then do so.
void	tesseract::expand_rows (ICOORD page_tr, TO_BLOCK *block, float gradient, FCOORD rotation, int32_t block_edge, bool testing_on)

void	tesseract::adjust_row_limits (TO_BLOCK *block)

compute_row_stats
Compute the linespacing and offset.
void	tesseract::compute_row_stats (TO_BLOCK *block, bool testing_on)

fill_heights
Fill the given heights with heights of the blobs that are legal candidates for estimating xheight.
void	tesseract::fill_heights (TO_ROW row, float gradient, int min_height, int max_height, STATS heights, STATS *floating_heights)

compute_xheight_from_modes
Given a STATS object heights, looks for two most frequently occurring heights that look like xheight and xheight + ascrise. If found, sets the values of xheight and ascrise accordingly, otherwise sets xheight to any most frequently occurring height and sets *ascrise to 0. Returns the number of times xheight occurred in heights. For each mode that is considered for being an xheight the count of floating blobs (stored in floating_heights) is subtracted from the total count of the blobs of this height. This is done because blobs that sit far above the baseline could represent valid ascenders, but it is highly unlikely that such a character's height will be an xheight (e.g. -, ', =, ^, ‘, ", ’, etc) If cap_only, then force finding of only the top mode.
int	tesseract::compute_xheight_from_modes (STATS heights, STATS floating_heights, bool cap_only, int min_height, int max_height, float xheight, float ascrise)

compute_row_descdrop
Estimates the descdrop of this row. This function looks for "significant" descenders of lowercase letters (those that could not just be the small descenders of upper case letters like Q,J). The function also takes into account how many potential ascenders this row might contain. If the number of potential ascenders along with descenders is close to the expected fraction of the total number of blobs in the row, the function returns the descender height, returns 0 otherwise.
int32_t	tesseract::compute_row_descdrop (TO_ROW row, float gradient, int xheight_blob_count, STATS asc_heights)

compute_height_modes
Find the top maxmodes values in the input array and put their indices in the output in the order in which they occurred.
int32_t	tesseract::compute_height_modes (STATS heights, int32_t min_height, int32_t max_height, int32_t modes, int32_t maxmodes)

correct_row_xheight
Adjust the xheight etc of this row if not within reasonable limits of the average for the block.
void	tesseract::correct_row_xheight (TO_ROW *row, float xheight, float ascrise, float descdrop)

separate_underlines
Test wide objects for being potential underlines. If they are then put them in a separate list in the block.
void	tesseract::separate_underlines (TO_BLOCK *block, float gradient, FCOORD rotation, bool testing_on)

pre_associate_blobs
Associate overlapping blobs and fake chop wide blobs.
void	tesseract::pre_associate_blobs (ICOORD page_tr, TO_BLOCK *block, FCOORD rotation, bool testing_on)

fit_parallel_rows
Re-fit the rows in the block to the given gradient.
void	tesseract::fit_parallel_rows (TO_BLOCK *block, float gradient, FCOORD rotation, int32_t block_edge, bool testing_on)

fit_parallel_lms
Fit an LMS line to a row. Make the fit parallel to the given gradient and set the row accordingly.
void	tesseract::fit_parallel_lms (float gradient, TO_ROW *row)

make_baseline_spline
Fit an LMS line to a row. Make the fit parallel to the given gradient and set the row accordingly.
void	tesseract::make_baseline_spline (TO_ROW row, TO_BLOCK block)

segment_baseline
Divide the baseline up into segments which require a different quadratic fitted to them. Return true if enough blobs were far enough away to need a quadratic.
bool	tesseract::segment_baseline (TO_ROW row, TO_BLOCK block, int32_t &segments, int32_t *xstarts)

linear_spline_baseline
Divide the baseline up into segments which require a different quadratic fitted to them. Returns true if enough blobs were far enough away to need a quadratic.
double *	tesseract::linear_spline_baseline (TO_ROW row, TO_BLOCK block, int32_t &segments, int32_t xstarts[])

assign_blobs_to_rows
Make enough rows to allocate all the given blobs to one. If a block skew is given, use that, else attempt to track it.
void	tesseract::assign_blobs_to_rows (TO_BLOCK block, float gradient, int pass, bool reject_misses, bool make_new_rows, bool drawing_skew)

most_overlapping_row
Return the row which most overlaps the blob.
OVERLAP_STATE	tesseract::most_overlapping_row (TO_ROW_IT row_it, TO_ROW &best_row, float top, float bottom, float rowsize, bool testing_blob)

blob_x_order
Sort function to sort blobs in x from page left.
int	tesseract::blob_x_order (const void item1, const void item2)

mark_repeated_chars
Mark blobs marked with BTFT_LEADER in repeated sets using the repeated_set member of BLOBNBOX.
void	tesseract::mark_repeated_chars (TO_ROW *row)

Variables
bool	tesseract::textord_heavy_nr = false

bool	tesseract::textord_show_initial_rows = false

bool	tesseract::textord_show_parallel_rows = false

bool	tesseract::textord_show_expanded_rows = false

bool	tesseract::textord_show_final_rows = false

bool	tesseract::textord_show_final_blobs = false

bool	tesseract::textord_test_landscape = false

bool	tesseract::textord_parallel_baselines = true

bool	tesseract::textord_straight_baselines = false

bool	tesseract::textord_old_baselines = true

bool	tesseract::textord_old_xheight = false

bool	tesseract::textord_fix_xheight_bug = true

bool	tesseract::textord_fix_makerow_bug = true

bool	tesseract::textord_debug_xheights = false

int	tesseract::textord_test_x = -INT32_MAX

int	tesseract::textord_test_y = -INT32_MAX

int	tesseract::textord_min_blobs_in_row = 4

int	tesseract::textord_spline_minblobs = 8

int	tesseract::textord_spline_medianwin = 6

int	tesseract::textord_min_xheight = 10

double	tesseract::textord_spline_shift_fraction = 0.02

double	tesseract::textord_skew_ile = 0.5

double	tesseract::textord_skew_lag = 0.02

double	tesseract::textord_linespace_iqrlimit = 0.2

double	tesseract::textord_width_limit = 8

double	tesseract::textord_chop_width = 1.5

double	tesseract::textord_minxh = 0.25

double	tesseract::textord_min_linesize = 1.25

double	tesseract::textord_excess_blobsize = 1.3

double	tesseract::textord_occupancy_threshold = 0.4

double	tesseract::textord_underline_width = 2.0

double	tesseract::textord_min_blob_height_fraction = 0.75

double	tesseract::textord_xheight_mode_fraction = 0.4

double	tesseract::textord_ascheight_mode_fraction = 0.08

double	tesseract::textord_ascx_ratio_min = 1.25

double	tesseract::textord_ascx_ratio_max = 1.8

double	tesseract::textord_descx_ratio_min = 0.25

double	tesseract::textord_descx_ratio_max = 0.6

double	tesseract::textord_xheight_error_margin = 0.1

int	tesseract::textord_lms_line_trials = 12

bool	tesseract::textord_new_initial_xheight = true

bool	tesseract::textord_debug_blob = false

compute_page_skew
Compute the skew over a full page by averaging the gradients over all the lines. Get the error of the same row.
const double	tesseract::kNoiseSize = 0.5

const int	tesseract::kMinSize = 8

void	tesseract::compute_page_skew (TO_BLOCK_LIST *blocks, float &page_m, float &page_err)

void	tesseract::vigorous_noise_removal (TO_BLOCK *block)

void	tesseract::cleanup_rows_making (ICOORD page_tr, TO_BLOCK *block, float gradient, FCOORD rotation, int32_t block_edge, bool testing_on)

void	tesseract::delete_non_dropout_rows (TO_BLOCK *block, float gradient, FCOORD rotation, int32_t block_edge, bool testing_on)

Macro Definition Documentation

◆ MAX_HEIGHT_MODES

#define MAX_HEIGHT_MODES 12

Definition at line 98 of file makerow.cpp.

Namespaces

Macros

Functions

Variables

compute_page_skew

Macro Definition Documentation

◆ MAX_HEIGHT_MODES