parseutils.h

Ocr source code. provides an Ocr engine that works in C++ language.

/* parseutils.h : utils for parsing and handling scans
 * Author: Maxie D. Schmidt (created 5/26/2006)        */

#ifndef _PARSEUTILS_H_
#define _PARSEUTILS_H_

#include <stdlib.h>

#include <vector>
using namespace std;

#include "types.h"
#include "classnotes.h"
#include "parselist.h"
#include "runutils.h"
#include "profile.h"

// lines:
box_t find_line_box(int entry_y, png_file *pfile);

// traverse char (finds boxes around characters):
xy_point n_point(xy_point pt);
xy_point ne_point(xy_point pt);
xy_point e_point(xy_point pt);
xy_point se_point(xy_point pt);
xy_point s_point(xy_point pt);
xy_point sw_point(xy_point pt);
xy_point w_point(xy_point pt);
xy_point nw_point(xy_point pt);

box_points traverse_char(xy_point entry_point, png_file *pfile);
box_points traverse_char(xy_point entry_point, direction_t d, png_file *pfile);

// find chars (returns a vector of char boxes sorted from l to r):
vector<box_t> find_chars(box_t line_box, png_file *pfile);

// fill the parse_list(assumes parse_list is empty to begin with):
bool fill_parse_list(int_interval_t line_int, png_file *pfile, 
                                                             parse_list *plist);

// gridding general utility:
box_t find_biggest_char_box_hlen(parse_list *plist);
box_t find_biggest_char_box_vlen(parse_list *plist);

// used in gridding config dialog:
grid_t grid_char(unsigned char *buf, int buf_rows, int buf_cols, 
                 int pos_offset_x, int pos_offset_y, int gsx, int gsy, 
                 int gcx, int gcy);

// uses pf image as buffer and externs for grid sizes:
grid_t grid_char(box_t char_box);

// returns _sorted_ list:
vector<char_probability> identify_character(grid_t char_grid);
void parse_chars(parse_list *plist);

// other utility:
//vector<translated_line> get_text(parse_list *plist);
plist_parsed_char point_to_char(parse_list *plist, int x, int y, 
                                bool search_alternate);

#endif