tesseract::ViterbiStateEntry Struct Reference

#include <lm_state.h>

Inheritance diagram for tesseract::ViterbiStateEntry:

Public Member Functions
	ViterbiStateEntry (ViterbiStateEntry *pe, BLOB_CHOICE *b, float c, float ol, const LMConsistencyInfo &ci, const AssociateStats &as, LanguageModelFlagsType tcf, LanguageModelDawgInfo *d, LanguageModelNgramInfo *n, const char *debug_uch)
	~ViterbiStateEntry ()
bool	Consistent () const
bool	HasAlnumChoice (const UNICHARSET &unicharset)
void	Print (const char *msg) const
Public Member Functions inherited from ELIST_LINK
	ELIST_LINK ()
	ELIST_LINK (const ELIST_LINK &)
void	operator= (const ELIST_LINK &)

Static Public Member Functions
static int	Compare (const void *e1, const void *e2)

Public Attributes
float	cost
BLOB_CHOICE *	curr_b
	Pointers to BLOB_CHOICE and parent ViterbiStateEntry (not owned by this). More...
ViterbiStateEntry *	parent_vse
ViterbiStateEntry *	competing_vse
float	ratings_sum
float	min_certainty
int	adapted
int	length
float	outline_length
LMConsistencyInfo	consistency_info
AssociateStats	associate_stats
LanguageModelFlagsType	top_choice_flags
LanguageModelDawgInfo *	dawg_info
LanguageModelNgramInfo *	ngram_info
bool	updated
STRING *	debug_str

Detailed Description

Struct for storing the information about a path in the segmentation graph explored by Viterbi search.

Definition at line 95 of file lm_state.h.

Constructor & Destructor Documentation

tesseract::ViterbiStateEntry::ViterbiStateEntry ( ViterbiStateEntry *  pe, BLOB_CHOICE *  b, float  c, float  ol, const LMConsistencyInfo &  ci, const AssociateStats &  as, LanguageModelFlagsType  tcf, LanguageModelDawgInfo *  d, LanguageModelNgramInfo *  n, const char *  debug_uch  )

inline

Definition at line 96 of file lm_state.h.

    : cost(c), curr_b(b), parent_vse(pe), competing_vse(NULL),
      ratings_sum(b->rating()),
      min_certainty(b->certainty()), adapted(b->IsAdapted()), length(1),
      outline_length(ol), consistency_info(ci), associate_stats(as),
      top_choice_flags(tcf), dawg_info(d), ngram_info(n),
      updated(true) {
    debug_str = (debug_uch == NULL) ? NULL : new STRING();
    if (pe != NULL) {
      ratings_sum += pe->ratings_sum;
      if (pe->min_certainty < min_certainty) {
        min_certainty = pe->min_certainty;
      }
      adapted += pe->adapted;
      length += pe->length;
      outline_length += pe->outline_length;
      if (debug_uch != NULL) *debug_str += *(pe->debug_str);
    }
    if (debug_str != NULL && debug_uch != NULL) *debug_str += debug_uch;
  }

tesseract::ViterbiStateEntry::~ViterbiStateEntry ( )

inline

Definition at line 123 of file lm_state.h.

                       {
    delete dawg_info;
    delete ngram_info;
    delete debug_str;
  }

Member Function Documentation

static int tesseract::ViterbiStateEntry::Compare ( const void *  e1, const void *  e2  )

inlinestatic

Comparator function for sorting ViterbiStateEntry_LISTs in non-increasing order of costs.

Definition at line 130 of file lm_state.h.

                                                     {
    const ViterbiStateEntry *ve1 =
      *reinterpret_cast<const ViterbiStateEntry * const *>(e1);
    const ViterbiStateEntry *ve2 =
      *reinterpret_cast<const ViterbiStateEntry * const *>(e2);
    return (ve1->cost < ve2->cost) ? -1 : 1;
  }

bool tesseract::ViterbiStateEntry::Consistent ( ) const

inline

Definition at line 137 of file lm_state.h.

                                 {
    if (dawg_info != NULL && consistency_info.NumInconsistentCase() == 0) {
      return true;
    }
    return consistency_info.Consistent();
  }

bool tesseract::ViterbiStateEntry::HasAlnumChoice ( const UNICHARSET &  unicharset )

inline

Returns true if this VSE has an alphanumeric character as its classifier result.

Definition at line 145 of file lm_state.h.

                                                    {
    if (curr_b == NULL) return false;
    UNICHAR_ID unichar_id =  curr_b->unichar_id();
    if (unicharset.get_isalpha(unichar_id) ||
        unicharset.get_isdigit(unichar_id))
      return true;
    return false;
  }

void tesseract::ViterbiStateEntry::Print

(

const char *

msg

)

const

Definition at line 27 of file lm_state.cpp.

                                                   {
  tprintf("%s ViterbiStateEntry", msg);
  if (updated) tprintf("(NEW)");
  if (this->debug_str != NULL) {
    tprintf(" str=%s", this->debug_str->string());
  }
  tprintf(" with ratings_sum=%.4f length=%d cost=%.6f",
          this->ratings_sum, this->length, this->cost);
  if (this->top_choice_flags) {
    tprintf(" top_choice_flags=0x%x", this->top_choice_flags);
  }
  if (!this->Consistent()) {
    tprintf(" inconsistent=(punc %d case %d chartype %d script %d font %d)",
            this->consistency_info.NumInconsistentPunc(),
            this->consistency_info.NumInconsistentCase(),
            this->consistency_info.NumInconsistentChartype(),
            this->consistency_info.inconsistent_script,
            this->consistency_info.inconsistent_font);
  }
  if (this->dawg_info) tprintf(" permuter=%d", this->dawg_info->permuter);
  if (this->ngram_info) {
    tprintf(" ngram_cl_cost=%g context=%s ngram pruned=%d",
            this->ngram_info->ngram_and_classifier_cost,
            this->ngram_info->context.string(),
            this->ngram_info->pruned);
  }
  if (this->associate_stats.shape_cost > 0.0f) {
    tprintf(" shape_cost=%g", this->associate_stats.shape_cost);
  }
  tprintf(" %s",
          XHeightConsistencyEnumName[this->consistency_info.xht_decision]);

  tprintf("\n");
}

Member Data Documentation

int tesseract::ViterbiStateEntry::adapted

Definition at line 170 of file lm_state.h.

AssociateStats tesseract::ViterbiStateEntry::associate_stats

Definition at line 174 of file lm_state.h.

ViterbiStateEntry* tesseract::ViterbiStateEntry::competing_vse

Pointer to a case-competing ViterbiStateEntry in the same list that represents a path ending in the same letter of the opposite case.

Definition at line 164 of file lm_state.h.

LMConsistencyInfo tesseract::ViterbiStateEntry::consistency_info

Definition at line 173 of file lm_state.h.

float tesseract::ViterbiStateEntry::cost

The cost is an adjusted ratings sum, that is adjusted by all the language model components that use Viterbi search.

Definition at line 157 of file lm_state.h.

BLOB_CHOICE* tesseract::ViterbiStateEntry::curr_b

Pointers to BLOB_CHOICE and parent ViterbiStateEntry (not owned by this).

Definition at line 160 of file lm_state.h.

LanguageModelDawgInfo* tesseract::ViterbiStateEntry::dawg_info

Extra information maintained by Dawg laguage model component (owned by ViterbiStateEntry).

Definition at line 182 of file lm_state.h.

STRING* tesseract::ViterbiStateEntry::debug_str

UTF8 string representing the path corresponding to this vse. Populated only in when language_model_debug_level > 0.

Definition at line 191 of file lm_state.h.

int tesseract::ViterbiStateEntry::length

Definition at line 171 of file lm_state.h.

float tesseract::ViterbiStateEntry::min_certainty

Definition at line 169 of file lm_state.h.

LanguageModelNgramInfo* tesseract::ViterbiStateEntry::ngram_info

Extra information maintained by Ngram laguage model component (owned by ViterbiStateEntry).

Definition at line 186 of file lm_state.h.

float tesseract::ViterbiStateEntry::outline_length

Definition at line 172 of file lm_state.h.

ViterbiStateEntry* tesseract::ViterbiStateEntry::parent_vse

Definition at line 161 of file lm_state.h.

float tesseract::ViterbiStateEntry::ratings_sum

Various information about the characters on the path represented by this ViterbiStateEntry.

Definition at line 168 of file lm_state.h.

LanguageModelFlagsType tesseract::ViterbiStateEntry::top_choice_flags

Flags for marking the entry as a top choice path with the smallest rating or lower/upper case letters).

Definition at line 178 of file lm_state.h.

bool tesseract::ViterbiStateEntry::updated

Definition at line 188 of file lm_state.h.

---

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

• wordrec/lm_state.h
• wordrec/lm_state.cpp