tesseract  5.0.0-alpha-619-ge9db
blkocc.h File Reference
#include "params.h"
#include "elst.h"

Go to the source code of this file.

Classes

class  REGION_OCC
 
class  BAND
 

Macros

#define RANGE_IN_BAND(band_max, band_min, range_max, range_min)   (((range_min) >= (band_min)) && ((range_max) < (band_max)))
 
#define RANGE_OVERLAPS_BAND(band_max, band_min, range_max, range_min)   (((range_max) >= (band_min)) && ((range_min) < (band_max)))
 
#define MAX_NUM_BANDS   5
 
#define UNDEFINED_BAND   99
 
#define NO_LOWER_LIMIT   -9999
 
#define NO_UPPER_LIMIT   9999
 
#define DOT_BAND   0
 
#define END_OF_WERD_CODE   255
 

Functions

bool test_underline (bool testing_on, C_BLOB *blob, int16_t baseline, int16_t xheight)
 

Variables

bool blockocc_show_result = false
 
int blockocc_desc_height = 0
 
int blockocc_asc_height = 255
 
int blockocc_band_count = 4
 
double textord_underline_threshold = 0.9
 

Macro Definition Documentation

◆ DOT_BAND

#define DOT_BAND   0

Definition at line 220 of file blkocc.h.

◆ END_OF_WERD_CODE

#define END_OF_WERD_CODE   255

Definition at line 224 of file blkocc.h.

◆ MAX_NUM_BANDS

#define MAX_NUM_BANDS   5

Definition at line 215 of file blkocc.h.

◆ NO_LOWER_LIMIT

#define NO_LOWER_LIMIT   -9999

Definition at line 217 of file blkocc.h.

◆ NO_UPPER_LIMIT

#define NO_UPPER_LIMIT   9999

Definition at line 218 of file blkocc.h.

◆ RANGE_IN_BAND

#define RANGE_IN_BAND (   band_max,
  band_min,
  range_max,
  range_min 
)    (((range_min) >= (band_min)) && ((range_max) < (band_max)))

Definition at line 61 of file blkocc.h.

◆ RANGE_OVERLAPS_BAND

#define RANGE_OVERLAPS_BAND (   band_max,
  band_min,
  range_max,
  range_min 
)    (((range_max) >= (band_min)) && ((range_min) < (band_max)))

Definition at line 79 of file blkocc.h.

◆ UNDEFINED_BAND

#define UNDEFINED_BAND   99

Definition at line 216 of file blkocc.h.

Function Documentation

◆ test_underline()

bool test_underline ( bool  testing_on,
C_BLOB blob,
int16_t  baseline,
int16_t  xheight 
)

test_underline

Check to see if the blob is an underline. Return true if it is.

Parameters
testing_ondrawing blob
blobblob to test
baselinecoords of baseline
xheightheight of line

Definition at line 48 of file blkocc.cpp.

53  {
54  int16_t occ;
55  int16_t blob_width; //width of blob
56  TBOX blob_box; //bounding box
57  int32_t desc_occ;
58  int32_t x_occ;
59  int32_t asc_occ;
60  STATS projection;
61 
62  blob_box = blob->bounding_box ();
63  blob_width = blob->bounding_box ().width ();
64  projection.set_range (blob_box.bottom (), blob_box.top () + 1);
65  if (testing_on) {
66  // blob->plot(to_win,GOLDENROD,GOLDENROD);
67  // line_color_index(to_win,GOLDENROD);
68  // move2d(to_win,blob_box.left(),baseline);
69  // draw2d(to_win,blob_box.right(),baseline);
70  // move2d(to_win,blob_box.left(),baseline+xheight);
71  // draw2d(to_win,blob_box.right(),baseline+xheight);
72  tprintf
73  ("Testing underline on blob at (%d,%d)->(%d,%d), base=%d\nOccs:",
74  blob->bounding_box ().left (), blob->bounding_box ().bottom (),
75  blob->bounding_box ().right (), blob->bounding_box ().top (),
76  baseline);
77  }
78  horizontal_cblob_projection(blob, &projection);
79  desc_occ = 0;
80  for (occ = blob_box.bottom (); occ < baseline; occ++)
81  if (occ <= blob_box.top () && projection.pile_count (occ) > desc_occ)
82  //max in region
83  desc_occ = projection.pile_count (occ);
84  x_occ = 0;
85  for (occ = baseline; occ <= baseline + xheight; occ++)
86  if (occ >= blob_box.bottom () && occ <= blob_box.top ()
87  && projection.pile_count (occ) > x_occ)
88  //max in region
89  x_occ = projection.pile_count (occ);
90  asc_occ = 0;
91  for (occ = baseline + xheight + 1; occ <= blob_box.top (); occ++)
92  if (occ >= blob_box.bottom () && projection.pile_count (occ) > asc_occ)
93  asc_occ = projection.pile_count (occ);
94  if (testing_on) {
95  tprintf ("%d %d %d\n", desc_occ, x_occ, asc_occ);
96  }
97  if (desc_occ == 0 && x_occ == 0 && asc_occ == 0) {
98  tprintf ("Bottom=%d, top=%d, base=%d, x=%d\n",
99  blob_box.bottom (), blob_box.top (), baseline, xheight);
100  projection.print();
101  }
102  if (desc_occ > x_occ + x_occ
103  && desc_occ > blob_width * textord_underline_threshold)
104  return true; //real underline
105  return asc_occ > x_occ + x_occ &&
106  asc_occ > blob_width * textord_underline_threshold; //overline
107  //neither
108 }

Variable Documentation

◆ blockocc_asc_height

int blockocc_asc_height = 255

"Ascender height after normalisation"

◆ blockocc_band_count

int blockocc_band_count = 4

"Number of bands used"

◆ blockocc_desc_height

int blockocc_desc_height = 0

"Descender height after normalisation"

◆ blockocc_show_result

bool blockocc_show_result = false

"Show intermediate results"

◆ textord_underline_threshold

double textord_underline_threshold = 0.9

"Fraction of width occupied"

Definition at line 33 of file blkocc.cpp.

C_BLOB::bounding_box
TBOX bounding_box() const
Definition: stepblob.cpp:247
baseline
Definition: mfoutline.h:62
TBOX::top
int16_t top() const
Definition: rect.h:57
STATS::pile_count
int32_t pile_count(int32_t value) const
Definition: statistc.h:75
textord_underline_threshold
double textord_underline_threshold
Definition: blkocc.cpp:33
TBOX::width
int16_t width() const
Definition: rect.h:114
TBOX::bottom
int16_t bottom() const
Definition: rect.h:64
STATS
Definition: statistc.h:30
TBOX::left
int16_t left() const
Definition: rect.h:71
TBOX::right
int16_t right() const
Definition: rect.h:78
tprintf
DLLSYM void tprintf(const char *format,...)
Definition: tprintf.cpp:34
STATS::set_range
bool set_range(int32_t min_bucket_value, int32_t max_bucket_value_plus_1)
Definition: statistc.cpp:53
STATS::print
void print() const
Definition: statistc.cpp:509
TBOX
Definition: rect.h:33