mirror/tesseract
1
0
Fork 0
mirror of https://github.com/tesseract-ocr/tesseract.git synced 2024-11-30 23:49:05 +08:00
Code Issues Actions 10 Packages Projects Releases Wiki Activity
312ce37de7
tesseract/ccstruct/ratngs.h
Ray Smith 84920b92b3 Font and classifier output structure cleanup. 
Font recognition was poor, due to forcing a 1st and 2nd choice at
a character level, when the total score for the correct font is often
correct at the word level, so allowed the propagation of a full set
of fonts and scores to the word recognizer, which can now decide word
level fonts using the scores instead of simple votes.

Change precipitated a cleanup of output data structures for classifier
results, eliminating ScoredClass and INT_RESULT_STRUCT, with a few
extra elements going in UnicharRating, and using that wherever possible.
That added the extra complexity of 1-rating due to a flip between 0 is
good and 0 is bad for the internal classifier scores before they are
converted to rating and certainty.
2015-05-12 17:24:34 -07:00

643 lines
22 KiB
C++
Raw Blame History

/**********************************************************************
* File: ratngs.h (Formerly ratings.h)
* Description: Definition of the WERD_CHOICE and BLOB_CHOICE classes.
* Author: Ray Smith
* Created: Thu Apr 23 11:40:38 BST 1992
*
* (C) Copyright 1992, Hewlett-Packard Ltd.
** Licensed under the Apache License, Version 2.0 (the "License");
** you may not use this file except in compliance with the License.
** You may obtain a copy of the License at
** http://www.apache.org/licenses/LICENSE-2.0
** Unless required by applicable law or agreed to in writing, software
** distributed under the License is distributed on an "AS IS" BASIS,
** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
** See the License for the specific language governing permissions and
** limitations under the License.
*
**********************************************************************/
#ifndef RATNGS_H
#define RATNGS_H
#include <assert.h>
#include "clst.h"
#include "elst.h"
#include "fontinfo.h"
#include "genericvector.h"
#include "matrix.h"
#include "unichar.h"
#include "unicharset.h"
#include "werd.h"
class MATRIX;
struct TBLOB;
struct TWERD;
// Enum to describe the source of a BLOB_CHOICE to make it possible to determine
// whether a blob has been classified by inspecting the BLOB_CHOICEs.
enum BlobChoiceClassifier {
BCC_STATIC_CLASSIFIER, // From the char_norm classifier.
BCC_ADAPTED_CLASSIFIER, // From the adaptive classifier.
BCC_SPECKLE_CLASSIFIER, // Backup for failed classification.
BCC_AMBIG, // Generated by ambiguity detection.
BCC_FAKE, // From some other process.
};
class BLOB_CHOICE: public ELIST_LINK
{
public:
BLOB_CHOICE() {
unichar_id_ = UNICHAR_SPACE;
fontinfo_id_ = -1;
fontinfo_id2_ = -1;
rating_ = 10.0;
certainty_ = -1.0;
script_id_ = -1;
xgap_before_ = 0;
xgap_after_ = 0;
min_xheight_ = 0.0f;
max_xheight_ = 0.0f;
yshift_ = 0.0f;
classifier_ = BCC_FAKE;
}
BLOB_CHOICE(UNICHAR_ID src_unichar_id, // character id
float src_rating, // rating
float src_cert, // certainty
int script_id, // script
float min_xheight, // min xheight in image pixel units
float max_xheight, // max xheight allowed by this char
float yshift, // the larger of y shift (top or bottom)
BlobChoiceClassifier c); // adapted match or other
BLOB_CHOICE(const BLOB_CHOICE &other);
~BLOB_CHOICE() {}
UNICHAR_ID unichar_id() const {
return unichar_id_;
}
float rating() const {
return rating_;
}
float certainty() const {
return certainty_;
}
inT16 fontinfo_id() const {
return fontinfo_id_;
}
inT16 fontinfo_id2() const {
return fontinfo_id2_;
}
const GenericVector<tesseract::ScoredFont>& fonts() const {
return fonts_;
}
void set_fonts(const GenericVector<tesseract::ScoredFont>& fonts) {
fonts_ = fonts;
int score1 = 0, score2 = 0;
fontinfo_id_ = -1;
fontinfo_id2_ = -1;
for (int f = 0; f < fonts_.size(); ++f) {
if (fonts_[f].score > score1) {
score2 = score1;
fontinfo_id2_ = fontinfo_id_;
score1 = fonts_[f].score;
fontinfo_id_ = fonts_[f].fontinfo_id;
} else if (fonts_[f].score > score2) {
score2 = fonts_[f].score;
fontinfo_id2_ = fonts_[f].fontinfo_id;
}
}
}
int script_id() const {
return script_id_;
}
const MATRIX_COORD& matrix_cell() {
return matrix_cell_;
}
inT16 xgap_before() const {
return xgap_before_;
}
inT16 xgap_after() const {
return xgap_after_;
}
float min_xheight() const {
return min_xheight_;
}
float max_xheight() const {
return max_xheight_;
}
float yshift() const {
return yshift_;
}
BlobChoiceClassifier classifier() const {
return classifier_;
}
bool IsAdapted() const {
return classifier_ == BCC_ADAPTED_CLASSIFIER;
}
bool IsClassified() const {
return classifier_ == BCC_STATIC_CLASSIFIER ||
classifier_ == BCC_ADAPTED_CLASSIFIER ||
classifier_ == BCC_SPECKLE_CLASSIFIER;
}
void set_unichar_id(UNICHAR_ID newunichar_id) {
unichar_id_ = newunichar_id;
}
void set_rating(float newrat) {
rating_ = newrat;
}
void set_certainty(float newrat) {
certainty_ = newrat;
}
void set_script(int newscript_id) {
script_id_ = newscript_id;
}
void set_matrix_cell(int col, int row) {
matrix_cell_.col = col;
matrix_cell_.row = row;
}
void set_xgap_before(inT16 gap) {
xgap_before_ = gap;
}
void set_xgap_after(inT16 gap) {
xgap_after_ = gap;
}
void set_classifier(BlobChoiceClassifier classifier) {
classifier_ = classifier;
}
static BLOB_CHOICE* deep_copy(const BLOB_CHOICE* src) {
BLOB_CHOICE* choice = new BLOB_CHOICE;
*choice = *src;
return choice;
}
// Returns true if *this and other agree on the baseline and x-height
// to within some tolerance based on a given estimate of the x-height.
bool PosAndSizeAgree(const BLOB_CHOICE& other, float x_height,
bool debug) const;
void print(const UNICHARSET *unicharset) const {
tprintf("r%.2f c%.2f x[%g,%g]: %d %s",
rating_, certainty_,
min_xheight_, max_xheight_, unichar_id_,
(unicharset == NULL) ? "" :
unicharset->debug_str(unichar_id_).string());
}
void print_full() const {
print(NULL);
tprintf(" script=%d, font1=%d, font2=%d, yshift=%g, classifier=%d\n",
script_id_, fontinfo_id_, fontinfo_id2_, yshift_, classifier_);
}
// Sort function for sorting BLOB_CHOICEs in increasing order of rating.
static int SortByRating(const void *p1, const void *p2) {
const BLOB_CHOICE *bc1 =
*reinterpret_cast<const BLOB_CHOICE * const *>(p1);
const BLOB_CHOICE *bc2 =
*reinterpret_cast<const BLOB_CHOICE * const *>(p2);
return (bc1->rating_ < bc2->rating_) ? -1 : 1;
}
private:
UNICHAR_ID unichar_id_; // unichar id
// Fonts and scores. Allowed to be empty.
GenericVector<tesseract::ScoredFont> fonts_;
inT16 fontinfo_id_; // char font information
inT16 fontinfo_id2_; // 2nd choice font information
// Rating is the classifier distance weighted by the length of the outline
// in the blob. In terms of probability, classifier distance is -klog p such
// that the resulting distance is in the range [0, 1] and then
// rating = w (-k log p) where w is the weight for the length of the outline.
// Sums of ratings may be compared meaningfully for words of different
// segmentation.
float rating_; // size related
// Certainty is a number in [-20, 0] indicating the classifier certainty
// of the choice. In terms of probability, certainty = 20 (k log p) where
// k is defined as above to normalize -klog p to the range [0, 1].
float certainty_; // absolute
int script_id_;
// Holds the position of this choice in the ratings matrix.
// Used to location position in the matrix during path backtracking.
MATRIX_COORD matrix_cell_;
inT16 xgap_before_;
inT16 xgap_after_;
// X-height range (in image pixels) that this classification supports.
float min_xheight_;
float max_xheight_;
// yshift_ - The vertical distance (in image pixels) the character is
// shifted (up or down) from an acceptable y position.
float yshift_;
BlobChoiceClassifier classifier_; // What generated *this.
};
// Make BLOB_CHOICE listable.
ELISTIZEH(BLOB_CHOICE)
// Return the BLOB_CHOICE in bc_list matching a given unichar_id,
// or NULL if there is no match.
BLOB_CHOICE *FindMatchingChoice(UNICHAR_ID char_id, BLOB_CHOICE_LIST *bc_list);
// Permuter codes used in WERD_CHOICEs.
enum PermuterType {
NO_PERM, // 0
PUNC_PERM, // 1
TOP_CHOICE_PERM, // 2
LOWER_CASE_PERM, // 3
UPPER_CASE_PERM, // 4
NGRAM_PERM, // 5
NUMBER_PERM, // 6
USER_PATTERN_PERM, // 7
SYSTEM_DAWG_PERM, // 8
DOC_DAWG_PERM, // 9
USER_DAWG_PERM, // 10
FREQ_DAWG_PERM, // 11
COMPOUND_PERM, // 12
NUM_PERMUTER_TYPES
};
namespace tesseract {
// ScriptPos tells whether a character is subscript, superscript or normal.
enum ScriptPos {
SP_NORMAL,
SP_SUBSCRIPT,
SP_SUPERSCRIPT,
SP_DROPCAP
};
const char *ScriptPosToString(tesseract::ScriptPos script_pos);
} // namespace tesseract.
class WERD_CHOICE : public ELIST_LINK {
public:
static const float kBadRating;
static const char *permuter_name(uinT8 permuter);
WERD_CHOICE(const UNICHARSET *unicharset)
: unicharset_(unicharset) { this->init(8); }
WERD_CHOICE(const UNICHARSET *unicharset, int reserved)
: unicharset_(unicharset) { this->init(reserved); }
WERD_CHOICE(const char *src_string,
const char *src_lengths,
float src_rating,
float src_certainty,
uinT8 src_permuter,
const UNICHARSET &unicharset)
: unicharset_(&unicharset) {
this->init(src_string, src_lengths, src_rating,
src_certainty, src_permuter);
}
WERD_CHOICE(const char *src_string, const UNICHARSET &unicharset);
WERD_CHOICE(const WERD_CHOICE &word) : unicharset_(word.unicharset_) {
this->init(word.length());
this->operator=(word);
}
~WERD_CHOICE();
const UNICHARSET *unicharset() const {
return unicharset_;
}
inline int length() const {
return length_;
}
float adjust_factor() const {
return adjust_factor_;
}
void set_adjust_factor(float factor) {
adjust_factor_ = factor;
}
inline const UNICHAR_ID *unichar_ids() const {
return unichar_ids_;
}
inline const UNICHAR_ID unichar_id(int index) const {
assert(index < length_);
return unichar_ids_[index];
}
inline int state(int index) const {
return state_[index];
}
tesseract::ScriptPos BlobPosition(int index) const {
if (index < 0 || index >= length_)
return tesseract::SP_NORMAL;
return script_pos_[index];
}
inline float rating() const {
return rating_;
}
inline float certainty() const {
return certainty_;
}
inline float certainty(int index) const {
return certainties_[index];
}
inline float min_x_height() const {
return min_x_height_;
}
inline float max_x_height() const {
return max_x_height_;
}
inline void set_x_heights(float min_height, float max_height) {
min_x_height_ = min_height;
max_x_height_ = max_height;
}
inline uinT8 permuter() const {
return permuter_;
}
const char *permuter_name() const;
// Returns the BLOB_CHOICE_LIST corresponding to the given index in the word,
// taken from the appropriate cell in the ratings MATRIX.
// Borrowed pointer, so do not delete.
BLOB_CHOICE_LIST* blob_choices(int index, MATRIX* ratings) const;
// Returns the MATRIX_COORD corresponding to the location in the ratings
// MATRIX for the given index into the word.
MATRIX_COORD MatrixCoord(int index) const;
inline void set_unichar_id(UNICHAR_ID unichar_id, int index) {
assert(index < length_);
unichar_ids_[index] = unichar_id;
}
bool dangerous_ambig_found() const {
return dangerous_ambig_found_;
}
void set_dangerous_ambig_found_(bool value) {
dangerous_ambig_found_ = value;
}
inline void set_rating(float new_val) {
rating_ = new_val;
}
inline void set_certainty(float new_val) {
certainty_ = new_val;
}
inline void set_permuter(uinT8 perm) {
permuter_ = perm;
}
// Note: this function should only be used if all the fields
// are populated manually with set_* functions (rather than
// (copy)constructors and append_* functions).
inline void set_length(int len) {
ASSERT_HOST(reserved_ >= len);
length_ = len;
}
/// Make more space in unichar_id_ and fragment_lengths_ arrays.
inline void double_the_size() {
if (reserved_ > 0) {
unichar_ids_ = GenericVector<UNICHAR_ID>::double_the_size_memcpy(
reserved_, unichar_ids_);
script_pos_ = GenericVector<tesseract::ScriptPos>::double_the_size_memcpy(
reserved_, script_pos_);
state_ = GenericVector<int>::double_the_size_memcpy(
reserved_, state_);
certainties_ = GenericVector<float>::double_the_size_memcpy(
reserved_, certainties_);
reserved_ *= 2;
} else {
unichar_ids_ = new UNICHAR_ID[1];
script_pos_ = new tesseract::ScriptPos[1];
state_ = new int[1];
certainties_ = new float[1];
reserved_ = 1;
}
}
/// Initializes WERD_CHOICE - reserves length slots in unichar_ids_ and
/// fragment_length_ arrays. Sets other values to default (blank) values.
inline void init(int reserved) {
reserved_ = reserved;
if (reserved > 0) {
unichar_ids_ = new UNICHAR_ID[reserved];
script_pos_ = new tesseract::ScriptPos[reserved];
state_ = new int[reserved];
certainties_ = new float[reserved];
} else {
unichar_ids_ = NULL;
script_pos_ = NULL;
state_ = NULL;
certainties_ = NULL;
}
length_ = 0;
adjust_factor_ = 1.0f;
rating_ = 0.0;
certainty_ = MAX_FLOAT32;
min_x_height_ = 0.0f;
max_x_height_ = MAX_FLOAT32;
permuter_ = NO_PERM;
unichars_in_script_order_ = false; // Tesseract is strict left-to-right.
dangerous_ambig_found_ = false;
}
/// Helper function to build a WERD_CHOICE from the given string,
/// fragment lengths, rating, certainty and permuter.
/// The function assumes that src_string is not NULL.
/// src_lengths argument could be NULL, in which case the unichars
/// in src_string are assumed to all be of length 1.
void init(const char *src_string, const char *src_lengths,
float src_rating, float src_certainty,
uinT8 src_permuter);
/// Set the fields in this choice to be default (bad) values.
inline void make_bad() {
length_ = 0;
rating_ = kBadRating;
certainty_ = -MAX_FLOAT32;
}
/// This function assumes that there is enough space reserved
/// in the WERD_CHOICE for adding another unichar.
/// This is an efficient alternative to append_unichar_id().
inline void append_unichar_id_space_allocated(
UNICHAR_ID unichar_id, int blob_count,
float rating, float certainty) {
assert(reserved_ > length_);
length_++;
this->set_unichar_id(unichar_id, blob_count,
rating, certainty, length_-1);
}
void append_unichar_id(UNICHAR_ID unichar_id, int blob_count,
float rating, float certainty);
inline void set_unichar_id(UNICHAR_ID unichar_id, int blob_count,
float rating, float certainty, int index) {
assert(index < length_);
unichar_ids_[index] = unichar_id;
state_[index] = blob_count;
certainties_[index] = certainty;
script_pos_[index] = tesseract::SP_NORMAL;
rating_ += rating;
if (certainty < certainty_) {
certainty_ = certainty;
}
}
// Sets the entries for the given index from the BLOB_CHOICE, assuming
// unit fragment lengths, but setting the state for this index to blob_count.
void set_blob_choice(int index, int blob_count,
const BLOB_CHOICE* blob_choice);
bool contains_unichar_id(UNICHAR_ID unichar_id) const;
void remove_unichar_ids(int index, int num);
inline void remove_last_unichar_id() { --length_; }
inline void remove_unichar_id(int index) {
this->remove_unichar_ids(index, 1);
}
bool has_rtl_unichar_id() const;
void reverse_and_mirror_unichar_ids();
// Returns the half-open interval of unichar_id indices [start, end) which
// enclose the core portion of this word -- the part after stripping
// punctuation from the left and right.
void punct_stripped(int *start_core, int *end_core) const;
// Returns the indices [start, end) containing the core of the word, stripped
// of any superscript digits on either side. (i.e., the non-footnote part
// of the word). There is no guarantee that the output range is non-empty.
void GetNonSuperscriptSpan(int *start, int *end) const;
// Return a copy of this WERD_CHOICE with the choices [start, end).
// The result is useful only for checking against a dictionary.
WERD_CHOICE shallow_copy(int start, int end) const;
void string_and_lengths(STRING *word_str, STRING *word_lengths_str) const;
const STRING debug_string() const {
STRING word_str;
for (int i = 0; i < length_; ++i) {
word_str += unicharset_->debug_str(unichar_ids_[i]);
word_str += " ";
}
return word_str;
}
// Call this to override the default (strict left to right graphemes)
// with the fact that some engine produces a "reading order" set of
// Graphemes for each word.
bool set_unichars_in_script_order(bool in_script_order) {
return unichars_in_script_order_ = in_script_order;
}
bool unichars_in_script_order() const {
return unichars_in_script_order_;
}
// Returns a UTF-8 string equivalent to the current choice
// of UNICHAR IDs.
const STRING &unichar_string() const {
this->string_and_lengths(&unichar_string_, &unichar_lengths_);
return unichar_string_;
}
// Returns the lengths, one byte each, representing the number of bytes
// required in the unichar_string for each UNICHAR_ID.
const STRING &unichar_lengths() const {
this->string_and_lengths(&unichar_string_, &unichar_lengths_);
return unichar_lengths_;
}
// Sets up the script_pos_ member using the blobs_list to get the bln
// bounding boxes, *this to get the unichars, and this->unicharset
// to get the target positions. If small_caps is true, sub/super are not
// considered, but dropcaps are.
// NOTE: blobs_list should be the chopped_word blobs. (Fully segemented.)
void SetScriptPositions(bool small_caps, TWERD* word);
// Sets the script_pos_ member from some source positions with a given length.
void SetScriptPositions(const tesseract::ScriptPos* positions, int length);
// Sets all the script_pos_ positions to the given position.
void SetAllScriptPositions(tesseract::ScriptPos position);
static tesseract::ScriptPos ScriptPositionOf(bool print_debug,
const UNICHARSET& unicharset,
const TBOX& blob_box,
UNICHAR_ID unichar_id);
// Returns the "dominant" script ID for the word. By "dominant", the script
// must account for at least half the characters. Otherwise, it returns 0.
// Note that for Japanese, Hiragana and Katakana are simply treated as Han.
int GetTopScriptID() const;
// Fixes the state_ for a chop at the given blob_posiiton.
void UpdateStateForSplit(int blob_position);
// Returns the sum of all the state elements, being the total number of blobs.
int TotalOfStates() const;
void print() const { this->print(""); }
void print(const char *msg) const;
// Prints the segmentation state with an introductory message.
void print_state(const char *msg) const;
// Displays the segmentation state of *this (if not the same as the last
// one displayed) and waits for a click in the window.
void DisplaySegmentation(TWERD* word);
WERD_CHOICE& operator+= ( // concatanate
const WERD_CHOICE & second);// second on first
WERD_CHOICE& operator= (const WERD_CHOICE& source);
private:
const UNICHARSET *unicharset_;
// TODO(rays) Perhaps replace the multiple arrays with an array of structs?
// unichar_ids_ is an array of classifier "results" that make up a word.
// For each unichar_ids_[i], script_pos_[i] has the sub/super/normal position
// of each unichar_id.
// state_[i] indicates the number of blobs in WERD_RES::chopped_word that
// were put together to make the classification results in the ith position
// in unichar_ids_, and certainties_[i] is the certainty of the choice that
// was used in this word.
// == Change from before ==
// Previously there was fragment_lengths_ that allowed a word to be
// artificially composed of multiple fragment results. Since the new
// segmentation search doesn't do fragments, treatment of fragments has
// been moved to a lower level, augmenting the ratings matrix with the
// combined fragments, and allowing the language-model/segmentation-search
// to deal with only the combined unichar_ids.
UNICHAR_ID *unichar_ids_; // unichar ids that represent the text of the word
tesseract::ScriptPos* script_pos_; // Normal/Sub/Superscript of each unichar.
int* state_; // Number of blobs in each unichar.
float* certainties_; // Certainty of each unichar.
int reserved_; // size of the above arrays
int length_; // word length
// Factor that was used to adjust the rating.
float adjust_factor_;
// Rating is the sum of the ratings of the individual blobs in the word.
float rating_; // size related
// certainty is the min (worst) certainty of the individual blobs in the word.
float certainty_; // absolute
// xheight computed from the result, or 0 if inconsistent.
float min_x_height_;
float max_x_height_;
uinT8 permuter_; // permuter code
// Normally, the ratings_ matrix represents the recognition results in order
// from left-to-right. However, some engines (say Cube) may return
// recognition results in the order of the script's major reading direction
// (for Arabic, that is right-to-left).
bool unichars_in_script_order_;
// True if NoDangerousAmbig found an ambiguity.
bool dangerous_ambig_found_;
// The following variables are populated and passed by reference any
// time unichar_string() or unichar_lengths() are called.
mutable STRING unichar_string_;
mutable STRING unichar_lengths_;
};
// Make WERD_CHOICE listable.
ELISTIZEH(WERD_CHOICE)
typedef GenericVector<BLOB_CHOICE_LIST *> BLOB_CHOICE_LIST_VECTOR;
// Utilities for comparing WERD_CHOICEs
bool EqualIgnoringCaseAndTerminalPunct(const WERD_CHOICE &word1,
const WERD_CHOICE &word2);
// Utilities for debug printing.
void print_ratings_list(
const char *msg, // intro message
BLOB_CHOICE_LIST *ratings, // list of results
const UNICHARSET &current_unicharset // unicharset that can be used
// for id-to-unichar conversion
);
#endif
Reference in New Issue View Git Blame Copy Permalink