tesseract-doxygen/ccmain/reject.cpp File Reference

#include "mfcpch.h"
#include "tessvars.h"
#include "scanutils.h"
#include <ctype.h>
#include <string.h>
#include "memry.h"
#include "reject.h"
#include "tfacep.h"
#include "mainblk.h"
#include "charcut.h"
#include "imgs.h"
#include "scaleimg.h"
#include "control.h"
#include "docqual.h"
#include "secname.h"
#include "globals.h"
#include "callnet.h"
#include "tesseractclass.h"
#include "notdll.h"

Namespaces

namespace  tesseract

Defines

#define EXTERN

Functions

 CLISTIZEH (STRING) CLISTIZE(STRING) EXTERN int tessedit_reject_mode=0
void reject_blanks (WERD_RES *word)
void reject_I_1_L (WERD_RES *word)
void reject_poor_matches (WERD_RES *word, BLOB_CHOICE_LIST_CLIST *blob_choices)
float compute_reject_threshold (BLOB_CHOICE_LIST_CLIST *blob_choices)
void reject_edge_blobs (WERD_RES *word)
inT16 nn_match_char (IMAGE &scaled_image, float baseline_pos, BOOL8 dict_word, BOOL8 checked_dict_word, BOOL8 sensible_word, BOOL8 centre, BOOL8 good_quality_word, char tess_ch)
inT16 evaluate_net_match (char top, float top_score, char next, float next_score, char tess_ch, BOOL8 dict_word, BOOL8 checked_dict_word, BOOL8 sensible_word, BOOL8 centre, BOOL8 good_quality_word)
void dont_allow_dubious_chars (WERD_RES *word)
void reject_mostly_rejects (WERD_RES *word)

Variables

EXTERN int tessedit_ok_mode = 5
EXTERN bool tessedit_use_nn = FALSE
EXTERN bool tessedit_rejection_debug = FALSE
EXTERN bool tessedit_rejection_stats = FALSE
EXTERN bool tessedit_flip_0O = TRUE
EXTERN double tessedit_lower_flip_hyphen = 1.5
EXTERN double tessedit_upper_flip_hyphen = 1.8
EXTERN bool rej_trust_doc_dawg = FALSE
EXTERN bool rej_1Il_use_dict_word = FALSE
EXTERN bool rej_1Il_trust_permuter_type = TRUE
EXTERN bool one_ell_conflict_default = TRUE
EXTERN bool show_char_clipping = FALSE
EXTERN bool nn_debug = FALSE
EXTERN bool nn_reject_debug = FALSE
EXTERN bool nn_lax = FALSE
EXTERN bool nn_double_check_dict = FALSE
EXTERN bool nn_conf_double_check_dict = TRUE
EXTERN bool nn_conf_1Il = TRUE
EXTERN bool nn_conf_Ss = TRUE
EXTERN bool nn_conf_hyphen = TRUE
EXTERN bool nn_conf_test_good_qual = FALSE
EXTERN bool nn_conf_test_dict = TRUE
EXTERN bool nn_conf_test_sensible = TRUE
EXTERN bool nn_conf_strict_on_dodgy_chs = TRUE
EXTERN double nn_dodgy_char_threshold = 0.99
EXTERN int nn_conf_accept_level = 4
EXTERN int nn_conf_initial_i_level = 3
EXTERN bool no_unrej_dubious_chars = TRUE
EXTERN bool no_unrej_no_alphanum_wds = TRUE
EXTERN bool no_unrej_1Il = FALSE
EXTERN bool rej_use_tess_accepted = TRUE
EXTERN bool rej_use_tess_blanks = TRUE
EXTERN bool rej_use_good_perm = TRUE
EXTERN bool rej_use_sensible_wd = FALSE
EXTERN bool rej_alphas_in_number_perm = FALSE
EXTERN double rej_whole_of_mostly_reject_word_fract = 0.85
EXTERN int rej_mostly_reject_mode = 1
EXTERN double tessed_fullstop_aspect_ratio = 1.2
EXTERN int net_image_width = 40
EXTERN int net_image_height = 36
EXTERN int net_image_x_height = 22
EXTERN int tessedit_image_border = 2
EXTERN int net_bl_nodes = 20
EXTERN double nn_reject_threshold = 0.5
EXTERN double nn_reject_head_and_shoulders = 0.6
EXTERN char * ok_single_ch_non_alphanum_wds = "-?\075"
EXTERN char * ok_repeated_ch_non_alphanum_wds = "-?*\075"
EXTERN char * conflict_set_I_l_1 = "Il1[]"
EXTERN char * conflict_set_S_s = "Ss$"
EXTERN char * conflict_set_hyphen = "-_~"
EXTERN char * dubious_chars_left_of_reject = "!'+`()-./\<>;:^_,~\""
EXTERN char * dubious_chars_right_of_reject = "!'+`()-./\<>;:^_,~\""
EXTERN int min_sane_x_ht_pixels = 8

Define Documentation

#define EXTERN

Function Documentation

CLISTIZEH ( STRING   )  [pure virtual]

"Rejection algorithm"

float compute_reject_threshold ( BLOB_CHOICE_LIST_CLIST *  blob_choices  ) 
void dont_allow_dubious_chars ( WERD_RES word  ) 
inT16 evaluate_net_match ( char  top,
float  top_score,
char  next,
float  next_score,
char  tess_ch,
BOOL8  dict_word,
BOOL8  checked_dict_word,
BOOL8  sensible_word,
BOOL8  centre,
BOOL8  good_quality_word 
)
inT16 nn_match_char ( IMAGE scaled_image,
float  baseline_pos,
BOOL8  dict_word,
BOOL8  checked_dict_word,
BOOL8  sensible_word,
BOOL8  centre,
BOOL8  good_quality_word,
char  tess_ch 
)
void reject_blanks ( WERD_RES word  ) 
void reject_edge_blobs ( WERD_RES word  ) 
void reject_I_1_L ( WERD_RES word  ) 
void reject_mostly_rejects ( WERD_RES word  ) 
void reject_poor_matches ( WERD_RES word,
BLOB_CHOICE_LIST_CLIST *  blob_choices 
)

Variable Documentation

EXTERN char* conflict_set_hyphen = "-_~"

"hyphen conflict set"

EXTERN char* conflict_set_I_l_1 = "Il1[]"

"Il1 conflict set"

EXTERN char* conflict_set_S_s = "Ss$"

"Ss conflict set"

EXTERN char* dubious_chars_left_of_reject = "!'+`()-./\<>;:^_,~\""

"Unreliable chars"

EXTERN char* dubious_chars_right_of_reject = "!'+`()-./\<>;:^_,~\""

"Unreliable chars"

EXTERN int min_sane_x_ht_pixels = 8

"Reject any x-ht lt or eq than this"

EXTERN int net_bl_nodes = 20

"Number of baseline nodes"

EXTERN int net_image_height = 36

"NN input image height"

EXTERN int net_image_width = 40

"NN input image width"

EXTERN int net_image_x_height = 22

"NN input image x_height"

EXTERN bool nn_conf_1Il = TRUE

"NN use 1Il conflicts"

EXTERN int nn_conf_accept_level = 4

"NN accept dodgy 1Il matches? "

"Double check for confusions"

"NN hyphen conflicts"

EXTERN int nn_conf_initial_i_level = 3

"NN accept initial Ii match level "

EXTERN bool nn_conf_Ss = TRUE

"NN use Ss conflicts"

"Require stronger NN match"

"NN dodgy 1Il cross check"

EXTERN bool nn_conf_test_good_qual = FALSE

"NN dodgy 1Il cross check"

"NN dodgy 1Il cross check"

EXTERN bool nn_debug = FALSE

"NN DEBUGGING?"

EXTERN double nn_dodgy_char_threshold = 0.99

"min accept score"

EXTERN bool nn_double_check_dict = FALSE

"Double check"

EXTERN bool nn_lax = FALSE

"Use 2nd rate matches"

EXTERN bool nn_reject_debug = FALSE

"NN DEBUG each char?"

EXTERN double nn_reject_head_and_shoulders = 0.6

"top scores sep factor"

EXTERN double nn_reject_threshold = 0.5

"NN min accept score"

EXTERN bool no_unrej_1Il = FALSE

"Stop unrej of 1Ilchars?"

"Dubious chars next to reject?"

"Stop unrej of non A/N wds?"

EXTERN char* ok_repeated_ch_non_alphanum_wds = "-?*\075"

"Allow NN to unrej"

EXTERN char* ok_single_ch_non_alphanum_wds = "-?\075"

"Allow NN to unrej"

"one_ell_conflict default"

"Dont double check"

EXTERN bool rej_1Il_use_dict_word = FALSE

"Use dictword test"

"Extend permuter check"

EXTERN int rej_mostly_reject_mode = 1

"0-never, 1-afterNN, 2-after new xht"

EXTERN bool rej_trust_doc_dawg = FALSE

"Use DOC dawg in 11l conf. detector"

"Individual rejection control"

EXTERN bool rej_use_sensible_wd = FALSE

"Extend permuter check"

"Individual rejection control"

"Individual rejection control"

"if >this fract"

EXTERN bool show_char_clipping = FALSE

"Show clip image window?"

EXTERN double tessed_fullstop_aspect_ratio = 1.2

"if >this fract then reject"

"Contextual 0O O0 flips"

EXTERN int tessedit_image_border = 2

"Rej blbs near image edge limit"

EXTERN double tessedit_lower_flip_hyphen = 1.5

"Aspect ratio dot/hyphen test"

EXTERN int tessedit_ok_mode = 5

"Acceptance decision algorithm"

"Adaption debug"

"Show NN stats"

EXTERN double tessedit_upper_flip_hyphen = 1.8

"Aspect ratio dot/hyphen test"

EXTERN bool tessedit_use_nn = FALSE

""

Generated on Sun Jul 18 17:10:49 2010 for Tesseract by  doxygen 1.6.3