tesseract  4.0.0-1-g2a2b
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 230 of file blkocc.h.

◆ END_OF_WERD_CODE

#define END_OF_WERD_CODE   255

Definition at line 234 of file blkocc.h.

◆ MAX_NUM_BANDS

#define MAX_NUM_BANDS   5

Definition at line 225 of file blkocc.h.

◆ NO_LOWER_LIMIT

#define NO_LOWER_LIMIT   -9999

Definition at line 227 of file blkocc.h.

◆ NO_UPPER_LIMIT

#define NO_UPPER_LIMIT   9999

Definition at line 228 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 68 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 87 of file blkocc.h.

◆ UNDEFINED_BAND

#define UNDEFINED_BAND   99

Definition at line 226 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.

Definition at line 53 of file blkocc.cpp.

58  {
59  int16_t occ;
60  int16_t blob_width; //width of blob
61  TBOX blob_box; //bounding box
62  int32_t desc_occ;
63  int32_t x_occ;
64  int32_t asc_occ;
65  STATS projection;
66 
67  blob_box = blob->bounding_box ();
68  blob_width = blob->bounding_box ().width ();
69  projection.set_range (blob_box.bottom (), blob_box.top () + 1);
70  if (testing_on) {
71  // blob->plot(to_win,GOLDENROD,GOLDENROD);
72  // line_color_index(to_win,GOLDENROD);
73  // move2d(to_win,blob_box.left(),baseline);
74  // draw2d(to_win,blob_box.right(),baseline);
75  // move2d(to_win,blob_box.left(),baseline+xheight);
76  // draw2d(to_win,blob_box.right(),baseline+xheight);
77  tprintf
78  ("Testing underline on blob at (%d,%d)->(%d,%d), base=%d\nOccs:",
79  blob->bounding_box ().left (), blob->bounding_box ().bottom (),
80  blob->bounding_box ().right (), blob->bounding_box ().top (),
81  baseline);
82  }
83  horizontal_cblob_projection(blob, &projection);
84  desc_occ = 0;
85  for (occ = blob_box.bottom (); occ < baseline; occ++)
86  if (occ <= blob_box.top () && projection.pile_count (occ) > desc_occ)
87  //max in region
88  desc_occ = projection.pile_count (occ);
89  x_occ = 0;
90  for (occ = baseline; occ <= baseline + xheight; occ++)
91  if (occ >= blob_box.bottom () && occ <= blob_box.top ()
92  && projection.pile_count (occ) > x_occ)
93  //max in region
94  x_occ = projection.pile_count (occ);
95  asc_occ = 0;
96  for (occ = baseline + xheight + 1; occ <= blob_box.top (); occ++)
97  if (occ >= blob_box.bottom () && projection.pile_count (occ) > asc_occ)
98  asc_occ = projection.pile_count (occ);
99  if (testing_on) {
100  tprintf ("%d %d %d\n", desc_occ, x_occ, asc_occ);
101  }
102  if (desc_occ == 0 && x_occ == 0 && asc_occ == 0) {
103  tprintf ("Bottom=%d, top=%d, base=%d, x=%d\n",
104  blob_box.bottom (), blob_box.top (), baseline, xheight);
105  projection.print();
106  }
107  if (desc_occ > x_occ + x_occ
108  && desc_occ > blob_width * textord_underline_threshold)
109  return true; //real underline
110  return asc_occ > x_occ + x_occ &&
111  asc_occ > blob_width * textord_underline_threshold; //overline
112  //neither
113 }
int32_t pile_count(int32_t value) const
Definition: statistc.h:78
double textord_underline_threshold
Definition: blkocc.cpp:38
Definition: rect.h:34
Definition: statistc.h:33
int16_t width() const
Definition: rect.h:115
int16_t left() const
Definition: rect.h:72
int16_t top() const
Definition: rect.h:58
bool set_range(int32_t min_bucket_value, int32_t max_bucket_value_plus_1)
Definition: statistc.cpp:63
DLLSYM void tprintf(const char *format,...)
Definition: tprintf.cpp:37
TBOX bounding_box() const
Definition: stepblob.cpp:255
void print() const
Definition: statistc.cpp:533
int16_t right() const
Definition: rect.h:79
int16_t bottom() const
Definition: rect.h:65

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 38 of file blkocc.cpp.