2893 lines
109 KiB
Text
2893 lines
109 KiB
Text
/*
|
|
This is a Optical-Character-Recognition program
|
|
Copyright (C) 2000-2009 Joerg Schulenburg
|
|
|
|
This program is free software; you can redistribute it and/or
|
|
modify it under the terms of the GNU General Public License
|
|
as published by the Free Software Foundation; either version 2
|
|
of the License, or (at your option) any later version.
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
GNU General Public License for more details.
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
along with this program; if not, write to the Free Software
|
|
Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
|
|
|
|
see README for EMAIL-address
|
|
|
|
sometimes I have written comments in german language, sorry for that
|
|
|
|
- look for ??? for preliminary code
|
|
- space: avX=22 11-13 (empirical estimated)
|
|
avX=16 5-7
|
|
avX= 7 5-6
|
|
|
|
ToDo: - add filter (r/s mismatch) g300c1
|
|
- better get_line2 function (problems on high resolution)
|
|
- write parallelizable code!
|
|
- learnmode (optimize filter)
|
|
- use ispell for final control or if unsure
|
|
- better line scanning (if not even)
|
|
- step 5: same chars differ? => expert mode
|
|
- chars dx>dy and above 50% hor-crossing > 4 is char-group ?
|
|
- detect color of chars and background
|
|
- better word space calculation (look at the examples)
|
|
(distance: left-left, middle-middle, left-right, thickness of e *0.75)
|
|
|
|
GLOBAL DATA (mostly structures)
|
|
- pix : image - one byte per pixel bits0-2=working
|
|
- lines : rows of the text (points to pix)
|
|
- box : list of bounding box for character
|
|
- obj : objects (lines, splines, etc. building a character)
|
|
*/
|
|
|
|
|
|
#include <stdlib.h>
|
|
#include <stdio.h>
|
|
#include <assert.h>
|
|
#include <string.h>
|
|
#include <ctype.h>
|
|
#include "config.h"
|
|
#ifdef HAVE_WCHAR_H
|
|
#include <wchar.h>
|
|
#endif
|
|
|
|
#include "amiga.h"
|
|
#include "list.h"
|
|
#include "pgm2asc.h"
|
|
// #include "pcx.h" /* needed for writebmp (removed later) */
|
|
/* ocr1 is the test-engine - remember: this is development version */
|
|
#include "ocr1.h"
|
|
/* first engine */
|
|
#include "ocr0.h"
|
|
#include "otsu.h"
|
|
#include "barcode.h"
|
|
#include "progress.h"
|
|
|
|
#include "gocr.h"
|
|
|
|
/* wew: will be exceeded by capitals at 1200dpi */
|
|
#define MaxBox (100*200) // largest possible letter (buffersize)
|
|
#define MAX(a,b) ((a) >= (b) ? (a) : (b))
|
|
|
|
/* if the system does not know about wchar.h, define functions here */
|
|
#ifndef HAVE_WCHAR_H
|
|
/* typedef unsigned wchar_t; */
|
|
/* Find the first occurrence of WC in WCS. */
|
|
wchar_t *wcschr (wchar_t *wcs, wchar_t wc) {
|
|
int i; for(i=0;wcs[i];i++) if (wcs[i]==wc) return wcs+i; return NULL;
|
|
}
|
|
wchar_t *wcscpy (wchar_t *dest, const wchar_t *src) {
|
|
int i; for(i=0;src[i];i++) dest[i]=src[i]; dest[i]=0; return dest;
|
|
}
|
|
size_t wcslen (const wchar_t *s){
|
|
size_t i; for(i=0;s[i];i++); return i;
|
|
}
|
|
#endif
|
|
#ifndef HAVE_WCSDUP
|
|
wchar_t * wcsdup (const wchar_t *WS) { /* its a gnu extension */
|
|
wchar_t *copy;
|
|
copy = (wchar_t *) malloc((wcslen(WS)+1)*sizeof(wchar_t));
|
|
if (!copy)return NULL;
|
|
wcscpy(copy, WS);
|
|
return copy;
|
|
}
|
|
#endif
|
|
|
|
// ------------------------ feature extraction -----------------
|
|
// -------------------------------------------------------------
|
|
// detect maximas in of line overlaps (return in %) and line coordinates
|
|
// this is for future use
|
|
#define HOR 1 // horizontal
|
|
#define VER 2 // vertical
|
|
#define RIS 3 // rising=steigend
|
|
#define FAL 4 // falling=fallend
|
|
|
|
/* exchange two variables */
|
|
static void swap(int *a, int *b) {
|
|
int c = *a;
|
|
*a = *b;
|
|
*b = c;
|
|
}
|
|
|
|
// calculate the overlapping of the line (0-1) with black points
|
|
// by recursive bisection
|
|
// line: y=dy/dx*x+b, implicit form: d=F(x,y)=dy*x-dx*y+b*dx=0
|
|
// incremental y(i+1)=m*(x(i)+1)+b, F(x+1,y+1)=f(F(x,y))
|
|
// ret & 1 => inverse pixel!
|
|
// d=2*F(x,y) integer numbers
|
|
int get_line(int x0, int y0, int x1, int y1, pix *p, int cs, int ret){
|
|
int dx,dy,incrE,incrNE,d,x,y,r0,r1,ty,tx,
|
|
*px,*py,*pdx,*pdy,*ptx,*pty,*px1;
|
|
dx=abs(x1-x0); tx=((x1>x0)?1:-1); // tx=x-spiegelung (new)
|
|
dy=abs(y1-y0); ty=((y1>y0)?1:-1); // ty=y-spiegelung (new)
|
|
// rotate coordinate system if dy>dx
|
|
/*bbg: can be faster if instead of pointers we use the variables and swaps? */
|
|
/*js: Do not know, I am happy that the current code is working and is small */
|
|
if(dx>dy){ pdx=&dx;pdy=&dy;px=&x;py=&y;ptx=&tx;pty=&ty;px1=&x1; }
|
|
else { pdx=&dy;pdy=&dx;px=&y;py=&x;ptx=&ty;pty=&tx;px1=&y1; }
|
|
if( *ptx<0 ){ swap(&x0,&x1);swap(&y0,&y1);tx=-tx;ty=-ty; }
|
|
d=((*pdy)<<1)-(*pdx); incrE=(*pdy)<<1; incrNE=((*pdy)-(*pdx))<<1;
|
|
x=x0; y=y0; r0=r1=0; /* dd=tolerance (store max drift) */
|
|
while( (*px)<=(*px1) ){
|
|
if( ((getpixel(p,x,y)<cs)?1:0)^(ret&1) ) r0++; else r1++;
|
|
(*px)++; if( d<=0 ){ d+=incrE; } else { d+=incrNE; (*py)+=(*pty); }
|
|
}
|
|
return (r0*(ret&~1))/(r0+r1); // ret==100 => percentage %
|
|
}
|
|
|
|
// this function should detect whether a direct connection between points
|
|
// exists or not, not finally implemented
|
|
// ret & 1 => inverse pixel!
|
|
// d=2*F(x,y) integer numbers, ideal line: ,I pixel: I@
|
|
// ..@ @@@ .@. ...,@2@. +1..+3 floodfill around line ???
|
|
// ..@ .@@ .@. ...,.@@@ +2..+4 <= that's not implemented yet
|
|
// ..@ ..@ .@. ...,.@@@ +2..+4
|
|
// @.@ @.. .@. ...,@@@. +1..+3
|
|
// @.@ @@. .@. ...I@@@. 0..+3
|
|
// @@@ @@@ .@. ..@1@@.. 0..+2
|
|
// 90% 0% 100% 90% r1-r2
|
|
// I am not satisfied with it
|
|
int get_line2(int x0, int y0, int x1, int y1, pix *p, int cs, int ret){
|
|
int dx,dy,incrE,incrNE,d,x,y,r0,r1,ty,tx,q,ddy,rx,ry,
|
|
*px,*py,*pdx,*pdy,*ptx,*pty,*px1;
|
|
dx=abs(x1-x0); tx=((x1>x0)?1:-1); // tx=x-spiegelung (new)
|
|
dy=abs(y1-y0); ty=((y1>y0)?1:-1); // ty=y-spiegelung (new)
|
|
// rotate coordinate system if dy>dx
|
|
if(dx>dy){ pdx=&dx;pdy=&dy;px=&x;py=&y;ptx=&tx;pty=&ty;px1=&x1;rx=1;ry=0; }
|
|
else { pdx=&dy;pdy=&dx;px=&y;py=&x;ptx=&ty;pty=&tx;px1=&y1;rx=0;ry=1; }
|
|
if( *ptx<0 ){ swap(&x0,&x1);swap(&y0,&y1);tx=-tx;ty=-ty; }
|
|
d=((*pdy)<<1)-(*pdx); incrE=(*pdy)<<1; incrNE=((*pdy)-(*pdx))<<1;
|
|
x=x0; y=y0; r0=r1=0; ddy=3; // tolerance = bit 1 + bit 0 = left+right
|
|
// int t=(*pdx)/16,tl,tr; // tolerance, left-,right delimiter
|
|
while( (*px)<=(*px1) ){ // not finaly implemented
|
|
q=((getpixel(p,x,y)<cs)?1:0)^(ret&1);
|
|
if ( !q ){ // tolerance one pixel perpenticular to the line
|
|
// what about 2 or more pixels tolerance???
|
|
ddy&=(~1)|(((getpixel(p,x+ry,y+rx)<cs)?1:0)^(ret&1));
|
|
ddy&=(~2)|(((getpixel(p,x-ry,y-rx)<cs)?1:0)^(ret&1))*2;
|
|
} else ddy=3;
|
|
if( ddy ) r0++; else r1++;
|
|
(*px)++; if( d<=0 ){ d+=incrE; } else { d+=incrNE; (*py)+=(*pty); }
|
|
}
|
|
return (r0*(ret&~1))/(r0+r1); // ret==100 => percentage %
|
|
}
|
|
|
|
/* Look for dots in the rectangular region x0 <= x <= x1 and y0 <= y
|
|
<= y1 in pixmap p. The two low order bits in mask indicate the color
|
|
of dots to look for: If mask==1 then look for black dots (where a
|
|
pixel value less than cs is considered black). If mask==2 then look
|
|
for white dots. If mask==3 then look for both black and white dots.
|
|
If the dots are found, the corresponding bits are set in the returned
|
|
value. Heavily used by the engine ocr0*.cc */
|
|
char get_bw(int x0, int x1, int y0, int y1, pix * p, int cs, int mask) {
|
|
char rc = 0; // later with error < 2% (1 dot)
|
|
int x, y;
|
|
|
|
if (x0 < 0) x0 = 0;
|
|
if (x1 >= p->x) x1 = p->x - 1;
|
|
if (y0 < 0) y0 = 0;
|
|
if (y1 >= p->y) y1 = p->y - 1;
|
|
|
|
for ( y = y0; y <= y1; y++)
|
|
for ( x = x0; x <= x1; x++) {
|
|
rc |= ((getpixel(p, x, y) < cs) ? 1 : 2); // break if rc==3
|
|
if ((rc & mask) == mask)
|
|
return mask; // break loop
|
|
}
|
|
return (rc & mask);
|
|
}
|
|
|
|
/* more general Mar2000 (x0,x1,y0,y1 instead of x0,y0,x1,y1! (history))
|
|
* look for black crossings throw a line from x0,y0 to x1,y1 and count them
|
|
* follow line and count crossings ([white]-black-transitions)
|
|
* ex: horizontal num_cross of 'm' would return 3
|
|
*
|
|
* fail for: .a... a-to-b counts no transitions, but there is
|
|
* ...#.
|
|
* ..#..
|
|
* .#..b
|
|
*/
|
|
int num_cross(int x0, int x1, int y0, int y1, pix *p, int cs) {
|
|
int rc = 0, col = 0, k, x, y, i, d; // rc=crossings col=0=white
|
|
int dx = x1 - x0, dy = y1 - y0;
|
|
|
|
d = MAX(abs(dx), abs(dy));
|
|
for (i = 0, x = x0, y = y0; i <= d; i++) {
|
|
if (d) {
|
|
x = x0 + i * dx / d;
|
|
y = y0 + i * dy / d;
|
|
}
|
|
k = ((getpixel(p, x, y) < cs) ? 1 : 0); // 0=white 1=black
|
|
if (col == 0 && k == 1) // found a white-black transition
|
|
rc++;
|
|
col = k; // last color
|
|
}
|
|
return rc;
|
|
}
|
|
|
|
/* check if test matches pattern
|
|
* possible pattern: "a-zA-Z0-9+--\\" (x-y dont work for c>127)
|
|
* return: 0 means dont fit, 1 means found
|
|
* ToDo: wchar_t cc + matching UTF-8 pattern for nonASCII
|
|
*/
|
|
int my_strchr( char *pattern, wchar_t cc ) {
|
|
char *s1;
|
|
if (pattern==(char *)NULL) return 0;
|
|
|
|
/* if (!(cc&0x80)) s1=strchr(pattern,(char)cc); else */
|
|
switch (cc) {
|
|
case '-': /* used as a special character */
|
|
s1=strstr(pattern,"--"); /* search string -- in pattern */
|
|
if (s1) return 1; break;
|
|
default:
|
|
s1=strstr(pattern,decode(cc, UTF8)); /* search string cc in pattern */
|
|
if (s1) return 1; /* cc simply matches */
|
|
/* single char not found, now check the ranges */
|
|
s1=pattern;
|
|
while (s1) {
|
|
s1=strchr(s1+1,'-'); /* look for next '-' */
|
|
if ((!s1) || (!s1[0]) || (!s1[1])) return 0; /* nothing found or end */
|
|
if (*(s1-1)=='-' || *(s1+1)=='-') continue; /* skip -- pattern */
|
|
if (*(s1-1)<=cc && *(s1+1)>=cc) return 1; /* within range */
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
/* set alternate chars and its weight, called from the engine
|
|
if a char is recognized to (weight) percent
|
|
can be used for filtering (only numbers etc)
|
|
often usefull if Il1 are looking very similar
|
|
should this function stay in box.c ???
|
|
weight is between 0 and 100 in percent, 100 means absolutely sure
|
|
- not final, not time critical (js)
|
|
- replace it by a string-function setaobj(*b,"string",weight)
|
|
and let call setac the setas function
|
|
*/
|
|
|
|
int setas(struct box *b, char *as, int weight){
|
|
int i,j;
|
|
if (b->num_ac > NumAlt || b->num_ac<0) {
|
|
fprintf(stderr,"\nDBG: There is something wrong with setas()!");
|
|
b->num_ac=0;
|
|
}
|
|
if (as==NULL) {
|
|
fprintf(stderr,"\nDBG: setas(NULL) makes no sense!"); return 0; }
|
|
if (as[0]==0) {
|
|
fprintf(stderr,"\nDBG: setas(\"\") makes no sense!"
|
|
" x= %d %d", b->x0, b->y0);
|
|
// out_x(b);
|
|
return 0;
|
|
}
|
|
|
|
/* char filter (ex: only numbers) ToDo: cfilter as UTF-8 */
|
|
if (JOB->cfg.cfilter) {
|
|
/* do not accept chars which are not in the cfilter string */
|
|
if ( as[0]>0 && as[1]==0 )
|
|
if ( !my_strchr(JOB->cfg.cfilter,as[0]) ) return 0;
|
|
}
|
|
#if 0 /* obsolete, done in setac */
|
|
/* not sure that this is the right place, but where else? */
|
|
if ( as[0]>0 && as[1]==0 )
|
|
if (b->modifier != SPACE && b->modifier != 0) {
|
|
wchar_t newac;
|
|
newac = compose(as[0], b->modifier);
|
|
as = (char *)decode(newac, UTF8); /* was (const char *) */
|
|
if (newac == as[0]) { /* nothing composed */
|
|
fprintf(stderr, "\nDBG setas compose was useless %d %d",b->x0,b->y0);
|
|
// out_x(b);
|
|
}
|
|
}
|
|
#endif
|
|
|
|
/* only the first run gets the full weight */
|
|
weight=(100-JOB->tmp.n_run)*weight/100;
|
|
|
|
/* remove same entries from table */
|
|
for (i=0;i<b->num_ac;i++)
|
|
if (b->tas[i])
|
|
if (strcmp(as,b->tas[i])==0) break;
|
|
if (b->num_ac>0 && i<b->num_ac){
|
|
if (weight<=b->wac[i]) return 0; /* if found + less weight ignore it */
|
|
/* to insert the new weigth on the right place, we remove it first */
|
|
if (b->tas[i]) free(b->tas[i]);
|
|
for (j=i;j<b->num_ac-1;j++){ /* shift lower entries */
|
|
b->tac[j]=b->tac[j+1]; /* copy the char */
|
|
b->tas[j]=b->tas[j+1]; /* copy the pointer to the string */
|
|
b->wac[j]=b->wac[j+1]; /* copy the weight */
|
|
}
|
|
b->num_ac--; /* shrink table */
|
|
}
|
|
/* sorting and add it to the table */
|
|
for (i=0;i<b->num_ac;i++) if (weight>b->wac[i]) break;
|
|
if (b->num_ac<NumAlt-1) b->num_ac++; /* enlarge table */
|
|
for (j=b->num_ac-1;j>i;j--){ /* shift lower entries */
|
|
b->tac[j]=b->tac[j-1]; /* copy the char */
|
|
b->tas[j]=b->tas[j-1]; /* copy the pointer to the string */
|
|
b->wac[j]=b->wac[j-1]; /* copy the weight */
|
|
}
|
|
if (i<b->num_ac) { /* insert new entry */
|
|
b->tac[i]=0; /* insert the char=0 ... */
|
|
b->tas[i]=(char *)malloc(strlen(as)+1); /* ... string */
|
|
if (b->tas[i]) memcpy(b->tas[i],as,strlen(as)+1);
|
|
b->wac[i]=weight; /* ... and its weight */
|
|
}
|
|
if (i==0) b->c=b->tac[0]; /* char or 0 for string */
|
|
return 0;
|
|
}
|
|
|
|
/* ToDo: this function will be replaced by a call of setas() later */
|
|
int setac(struct box *b, wchar_t ac, int weight){
|
|
int i,j;
|
|
if ((!b) || b->num_ac > NumAlt || b->num_ac<0) {
|
|
fprintf(stderr,"\nDBG: This is a bad call to setac()!");
|
|
if(b && (JOB->cfg.verbose & 6)) out_x(b);
|
|
b->num_ac=0;
|
|
}
|
|
if (ac==0 || ac==UNKNOWN) {
|
|
fprintf(stderr,"\nDBG: setac(0) makes no sense!");
|
|
return 0;
|
|
}
|
|
/* char filter (ex: only numbers) ToDo: cfilter as UTF-8 */
|
|
if (JOB->cfg.cfilter) {
|
|
/* do not accept chars which are not in the cfilter string */
|
|
/* if ( ac>255 || !strchr(JOB->cfg.cfilter,(char)ac) ) return 0; */
|
|
if ( !my_strchr(JOB->cfg.cfilter,ac) ) return 0;
|
|
}
|
|
/* not sure that this is the right place, but where else? */
|
|
if (b->modifier != SPACE && b->modifier != 0) {
|
|
wchar_t newac;
|
|
newac = compose(ac, b->modifier);
|
|
if (newac == ac) { /* nothing composed */
|
|
if(JOB->cfg.verbose & 7)
|
|
fprintf(stderr, "\nDBG setac(%s): compose was useless @ %d %d",
|
|
decode(ac,ASCII), b->x0, b->y0);
|
|
/* if(JOB->cfg.verbose & 6) out_x(b); */
|
|
}
|
|
ac = newac;
|
|
}
|
|
|
|
/* only the first run gets the full weight */
|
|
weight=(100-JOB->tmp.n_run)*weight/100;
|
|
|
|
/* remove same entries from table */
|
|
for (i=0;i<b->num_ac;i++) if (ac==b->tac[i]) break;
|
|
if (b->num_ac>0 && i<b->num_ac){
|
|
if (weight<=b->wac[i]) return 0;
|
|
if (b->tas[i]) free(b->tas[i]);
|
|
for (j=i;j<b->num_ac-1;j++){ /* shift lower entries */
|
|
b->tac[j]=b->tac[j+1]; /* copy the char */
|
|
b->tas[j]=b->tas[j+1]; /* copy the pointer to the string */
|
|
b->wac[j]=b->wac[j+1]; /* copy the weight */
|
|
}
|
|
b->num_ac--; /* shrink table */
|
|
}
|
|
/* sorting it to the table */
|
|
for (i=0;i<b->num_ac;i++) if (weight>b->wac[i]) break;
|
|
if (b->num_ac<NumAlt-1) b->num_ac++; /* enlarge table */
|
|
for (j=b->num_ac-1;j>i;j--){ /* shift lower entries */
|
|
b->tac[j]=b->tac[j-1]; /* copy the char */
|
|
b->tas[j]=b->tas[j-1]; /* copy the pointer to the string */
|
|
b->wac[j]=b->wac[j-1]; /* copy the weight */
|
|
}
|
|
if (i<b->num_ac) { /* insert new entry */
|
|
b->tac[i]=ac; /* insert the char ... */
|
|
b->tas[j]=NULL; /* ... no string (?) */
|
|
b->wac[i]=weight; /* ... and its weight */
|
|
}
|
|
if (i==0) b->c=ac; /* store best result to b->c (will be obsolete) */
|
|
|
|
return 0;
|
|
}
|
|
|
|
/* test if ac in wac-table
|
|
usefull for contextcorrection and box-splitting
|
|
return 0 if not found
|
|
return wac if found (wac>0)
|
|
*/
|
|
int testac(struct box *b, wchar_t ac){
|
|
int i;
|
|
if (b->num_ac > NumAlt || b->num_ac<0) {
|
|
fprintf(stderr,"\n#DEBUG: There is something wrong with testac()!");
|
|
b->num_ac=0;
|
|
}
|
|
/* search entries in table */
|
|
for (i=0;i<b->num_ac;i++) if (ac==b->tac[i]) return b->wac[i];
|
|
return 0;
|
|
}
|
|
|
|
|
|
/* look for edges: follow a line from x0,y0 to x1,y1, record the
|
|
* location of each transition, and return their number.
|
|
* ex: horizontal num_cross of 'm' would return 6
|
|
* remark: this function is not used, obsolete? ToDo: remove?
|
|
*/
|
|
int follow_path(int x0, int x1, int y0, int y1, pix *p, int cs, path_t *path) {
|
|
int rc = 0, prev, x, y, i, d, color; // rc=crossings col=0=white
|
|
int dx = x1 - x0, dy = y1 - y0;
|
|
|
|
d = MAX(abs(dx), abs(dy));
|
|
prev = getpixel(p, x0, y0) < cs; // 0=white 1=black
|
|
path->start = prev;
|
|
for (i = 1, x = x0, y = y0; i <= d; i++) {
|
|
if (d) {
|
|
x = x0 + i * dx / d;
|
|
y = y0 + i * dy / d;
|
|
}
|
|
color = getpixel(p, x, y) < cs; // 0=white 1=black
|
|
if (color != prev){
|
|
if (rc>=path->max){
|
|
int n=path->max*2+10;
|
|
path->x = (int *) xrealloc(path->x, n*sizeof(int));
|
|
path->y = (int *) xrealloc(path->y, n*sizeof(int));
|
|
path->max = n;
|
|
}
|
|
path->x[rc]=x;
|
|
path->y[rc]=y;
|
|
rc++;
|
|
}
|
|
prev = color;
|
|
}
|
|
path->num=rc;
|
|
return rc;
|
|
}
|
|
|
|
/* ToDo: only used in follow_path, which is obsolete, remove? */
|
|
void *xrealloc(void *ptr, size_t size){
|
|
void *p;
|
|
p = realloc(ptr, size);
|
|
if (size>0 && (!p)){
|
|
fprintf(stderr, "insufficient memory");
|
|
exit(1);
|
|
}
|
|
return p;
|
|
}
|
|
|
|
/*
|
|
* -------------------------------------------------------------
|
|
* mark edge-points
|
|
* - first move forward until b/w-edge
|
|
* - more than 2 pixel?
|
|
* - loop around
|
|
* - if forward pixel : go up, rotate right
|
|
* - if forward no pixel : rotate left
|
|
* - stop if found first 2 pixel in same order
|
|
* go_along_the_right_wall strategy is very similar and used otherwhere
|
|
* --------------------------------------------------------------
|
|
* turmite game: inp: start-x,y, regel r_black=UP,r_white=RIght until border
|
|
* out: last-position
|
|
*
|
|
* could be used to extract more features:
|
|
* by counting stepps, dead-end streets ,xmax,ymax,ro-,ru-,lo-,lu-edges
|
|
*
|
|
* use this little animal to find features, I first was happy about it
|
|
* but now I prefer the loop() function
|
|
*/
|
|
|
|
void turmite(pix *p, int *x, int *y,
|
|
int x0, int x1, int y0, int y1, int cs, int rw, int rb) {
|
|
int r;
|
|
if (outbounds(p, x0, y0)) // out of pixmap
|
|
return;
|
|
while (*x >= x0 && *y >= y0 && *x <= x1 && *y <= y1) {
|
|
r = ((getpixel(p, *x, *y) < cs) ? rb : rw); // select rule
|
|
switch (r) {
|
|
case UP: (*y)--; break;
|
|
case DO: (*y)++; break;
|
|
case RI: (*x)++; break;
|
|
case LE: (*x)--; break;
|
|
case ST: break;
|
|
default: assert(0);
|
|
}
|
|
if( r==ST ) break; /* leave the while-loop */
|
|
}
|
|
}
|
|
|
|
/* search a way from p0 to p1 without crossing pixels of type t
|
|
* only two directions, useful to test if there is a gap 's'
|
|
* labyrinth algorithm - do you know a faster way? */
|
|
int joined(pix *p, int x0, int y0, int x1, int y1, int cs){
|
|
int t,r,x,y,dx,dy,xa,ya,xb,yb;
|
|
x=x0;y=y0;dx=1;dy=0;
|
|
if(x1>x0){xa=x0;xb=x1;} else {xb=x0;xa=x1;}
|
|
if(y1>y0){ya=y0;yb=y1;} else {yb=y0;ya=y1;}
|
|
t=((getpixel(p,x,y)<cs)?1:0);
|
|
for(;;){
|
|
if( t==((getpixel(p,x+dy,y-dx)<cs)?1:0) // right free?
|
|
&& x+dy>=xa && x+dy<=xb && y-dx>=ya && y-dx<=yb) // wall
|
|
{ r=dy;dy=-dx;dx=r;x+=dx;y+=dy; } // rotate right and step forward
|
|
else { r=dx;dx=-dy;dy=r; } // rotate left
|
|
// fprintf(stderr," path xy %d-%d %d-%d %d %d %d %d\n",xa,xb,ya,yb,x,y,dx,dy);
|
|
if( x==x1 && y==y1 ) return 1;
|
|
if( x==x0 && y==y0 && dx==1) return 0;
|
|
}
|
|
// return 0; // endless loop ?
|
|
}
|
|
|
|
/* move from x,y to direction r until pixel of color col is found
|
|
* or maximum of l steps
|
|
* return the number of steps done */
|
|
int loop(pix *p,int x,int y,int l,int cs,int col, DIRECTION r){
|
|
int i=0;
|
|
if(x>=0 && y>=0 && x<p->x && y<p->y){
|
|
switch (r) {
|
|
case UP:
|
|
for( ;i<l && y>=0;i++,y--)
|
|
if( (getpixel(p,x,y)<cs)^col )
|
|
break;
|
|
break;
|
|
case DO:
|
|
for( ;i<l && y<p->y;i++,y++)
|
|
if( (getpixel(p,x,y)<cs)^col )
|
|
break;
|
|
break;
|
|
case LE:
|
|
for( ;i<l && x>=0;i++,x--)
|
|
if( (getpixel(p,x,y)<cs)^col )
|
|
break;
|
|
break;
|
|
case RI:
|
|
for( ;i<l && x<p->x;i++,x++)
|
|
if( (getpixel(p,x,y)<cs)^col )
|
|
break;
|
|
break;
|
|
default:;
|
|
}
|
|
}
|
|
return i;
|
|
}
|
|
|
|
/* Given a point, frames a rectangle containing all points of the same
|
|
* color surrounding it, and mark these points.
|
|
* ToDo: obsolate and replaced by frame_vector
|
|
*
|
|
* looking for better algo: go horizontally and look for upper/lower non_marked_pixel/nopixel
|
|
* use lowest three bits for mark
|
|
* - recursive version removed! AmigaOS has no Stack-OVL-Event
|
|
* run around the chape using laby-robot
|
|
* bad changes can lead to endless loop!
|
|
* - this is not absolutely sure but mostly works well
|
|
* diag - 0: only pi/2 direction, 1: pi/4 directions (diagonal)
|
|
* mark - 3 bit marker, mark each valid pixel with it
|
|
*/
|
|
int frame_nn(pix *p, int x, int y,
|
|
int *x0, int *x1, int *y0, int *y1, // enlarge frame
|
|
int cs, int mark,int diag){
|
|
#if 1 /* flood-fill to detect black objects, simple and faster? */
|
|
int rc = 0, dx, col, maxstack=0; static int overflow=0;
|
|
int bmax=1024, blen=0, *buf; /* buffer as replacement for recursion stack */
|
|
|
|
/* check bounds */
|
|
if (outbounds(p, x, y)) return 0;
|
|
/* check if already marked (with mark since v0.4) */
|
|
if ((marked(p,x,y)&mark)==mark) return 0;
|
|
|
|
col = ((getpixel(p, x, y) < cs) ? 0 : 1);
|
|
buf=(int *)malloc(bmax*sizeof(int)*2);
|
|
if (!buf) { fprintf(stderr,"malloc failed (frame_nn)\n");return 0;}
|
|
buf[0]=x;
|
|
buf[1]=y;
|
|
blen=1;
|
|
|
|
g_debug(fprintf(stderr,"\nframe_nn x=%4d y=%4d",x,y);)
|
|
for ( ; blen ; ) {
|
|
/* max stack depth is complexity of the object */
|
|
if (blen>maxstack) maxstack=blen;
|
|
blen--; /* reduce the stack */
|
|
x=buf[blen*2+0];
|
|
y=buf[blen*2+1];
|
|
if (y < *y0) *y0 = y;
|
|
if (y > *y1) *y1 = y;
|
|
/* first go to leftmost pixel */
|
|
for ( ; x>0 && (col == ((getpixel(p, x-1, y) < cs) ? 0 : 1)) ; x--);
|
|
if ((marked(p,x,y)&mark)==mark) continue; /* already scanned */
|
|
for (dx=-1;dx<2;dx+=2) /* look at upper and lower line, left */
|
|
if ( diag && x<p->x && x-1>0 && y+dx >=0 && y+dx < p->y
|
|
&& col != ((getpixel(p, x , y+dx) < cs) ? 0 : 1)
|
|
&& col == ((getpixel(p, x-1, y+dx) < cs) ? 0 : 1)
|
|
&& !((marked(p,x-1,y+dx)&mark)==mark)
|
|
) {
|
|
if (blen+1>=bmax) { overflow|=1; continue; }
|
|
buf[blen*2+0]=x-1;
|
|
buf[blen*2+1]=y+dx;
|
|
blen++;
|
|
}
|
|
if (x < *x0) *x0 = x;
|
|
/* second go right, mark and get new starting points */
|
|
for ( ; x<p->x && (col == ((getpixel(p, x , y) < cs) ? 0 : 1)) ; x++) {
|
|
p->p[x + y * p->x] |= (mark & 7); rc++; /* mark pixel */
|
|
/* enlarge frame */
|
|
if (x > *x1) *x1 = x;
|
|
for (dx=-1;dx<2;dx+=2) /* look at upper and lower line */
|
|
if ( col == ((getpixel(p, x , y+dx) < cs) ? 0 : 1)
|
|
&& (
|
|
col != ((getpixel(p, x-1, y ) < cs) ? 0 : 1)
|
|
|| col != ((getpixel(p, x-1, y+dx) < cs) ? 0 : 1) )
|
|
&& !((marked(p,x,y+dx)&mark)==mark) && y+dx<p->y && y+dx>=0
|
|
) {
|
|
if (blen+1>=bmax) { overflow|=1; continue; }
|
|
buf[blen*2+0]=x;
|
|
buf[blen*2+1]=y+dx;
|
|
blen++;
|
|
}
|
|
}
|
|
for (dx=-1;dx<2;dx+=2) /* look at upper and lower line, right */
|
|
if ( diag && x<p->x && x-1>0 && y+dx >=0 && y+dx < p->y
|
|
&& col == ((getpixel(p, x-1, y ) < cs) ? 0 : 1)
|
|
&& col != ((getpixel(p, x , y ) < cs) ? 0 : 1)
|
|
&& col != ((getpixel(p, x-1, y+dx) < cs) ? 0 : 1)
|
|
&& col == ((getpixel(p, x , y+dx) < cs) ? 0 : 1)
|
|
&& !((marked(p,x,y+dx)&mark)==mark)
|
|
) {
|
|
if (blen+1>=bmax) { overflow|=1; continue; }
|
|
buf[blen*2+0]=x;
|
|
buf[blen*2+1]=y+dx;
|
|
blen++;
|
|
}
|
|
}
|
|
|
|
/* debug, ToDo: use info maxstack and pixels for image classification */
|
|
g_debug(fprintf(stderr," maxstack= %4d pixels= %6d",maxstack,rc);)
|
|
if (overflow==1){
|
|
overflow|=2;
|
|
fprintf(stderr,"# Warning: frame_nn stack oerflow\n");
|
|
}
|
|
free(buf);
|
|
#else /* old version, ToDo: improve it for tmp04/005*.pgm.gz */
|
|
int i, j, d, dx, ox, oy, od, nx, ny, rc = 0, rot = 0, x2 = x, y2 = y, ln;
|
|
|
|
static const int d0[8][2] = { { 0, -1} /* up */, {-1, -1},
|
|
{-1, 0} /* left */, {-1, 1},
|
|
{ 0, 1} /* down */, { 1, 1},
|
|
{ 1, 0} /* right */, { 1, -1}};
|
|
|
|
/* check bounds */
|
|
if (outbounds(p, x, y))
|
|
return 0;
|
|
/* check if already marked */
|
|
if ((marked(p,x,y)&mark)==mark)
|
|
return 0;
|
|
|
|
i = ((getpixel(p, x, y) < cs) ? 0 : 1);
|
|
rc = 0;
|
|
|
|
g_debug(fprintf(stderr," start frame:");)
|
|
|
|
for (ln = 0; ln < 2 && rot >= 0; ln++) { // repeat if right-loop
|
|
g_debug(fprintf(stderr," ln=%d diag=%d cs=%d x=%d y=%d - go to border\n",ln,diag,cs,x,y);)
|
|
|
|
od=d=(8+4*ln-diag)&7; // start robot looks up, right is a wall
|
|
// go to right (left) border
|
|
if (ln==1) {
|
|
x=x2; y=y2;
|
|
}
|
|
/* start on leftmost position */
|
|
for (dx = 1 - 2*ln; x + dx < p->x && x + dx >= 0 /* bounds */ &&
|
|
i == ((getpixel(p, x + dx, y) < cs) ? 0 : 1) /* color */;
|
|
x += dx);
|
|
|
|
g_debug(fprintf(stderr," ln=%d diag=%d cs=%d x=%d y=%d\n",ln,diag,cs,x,y);)
|
|
|
|
/* robot stores start-position */
|
|
ox = x; oy = y;
|
|
for (rot = 0; abs(rot) <= 64; ) { /* for sure max. 8 spirals */
|
|
/* leftmost position */
|
|
if (ln == 0 && x < x2) {
|
|
x2 = x; y2 = y;
|
|
}
|
|
|
|
g_debug(fprintf(stderr," x=%3d y=%3d d=%d i=%d p=%3d rc=%d\n",x,y,d,i,getpixel(p,x,y),rc);)
|
|
|
|
if ( abs(d0[d][1]) ) { /* mark left (right) pixels */
|
|
for (j = 0, dx = d0[d][1]; x + j >= 0 && x + j < p->x
|
|
&& i == ((getpixel(p, x + j, y) < cs) ? 0 : 1); j += dx) {
|
|
if (!((marked(p, x + j, y)&mark)==mark))
|
|
rc++;
|
|
p->p[x + j + y * p->x] |= (mark & 7);
|
|
}
|
|
}
|
|
/* look to the front of robot */
|
|
nx = x + d0[d][0];
|
|
ny = y + d0[d][1];
|
|
/* if right is a wall */
|
|
if ( outbounds(p, nx, ny) || i != ((getpixel(p,nx,ny)<cs) ? 0 : 1) ) {
|
|
/* rotate left */
|
|
d=(d+2-diag) & 7; rot-=2-diag;
|
|
}
|
|
else { /* if no wall, go, turn back and rotate left */
|
|
x=nx; y=ny; d=(d+4+2-diag) & 7; rot+=2-diag+4;
|
|
/* enlarge frame */
|
|
if (x < *x0) *x0 = x;
|
|
if (x > *x1) *x1 = x;
|
|
if (y < *y0) *y0 = y;
|
|
if (y > *y1) *y1 = y;
|
|
}
|
|
if(x==ox && y==oy && d==od) break; // round trip finished
|
|
}
|
|
}
|
|
g_debug(fprintf(stderr," rot=%d\n",rot);)
|
|
#endif
|
|
return rc;
|
|
}
|
|
|
|
/* obsolete! replaced by vectors
|
|
* mark neighbouring pixel of same color, return number
|
|
* better with neighbours of same color (more general) ???
|
|
* parameters: (&~7)-pixmap, start-point, critical_value, mark
|
|
* recursion is removed */
|
|
int mark_nn(pix * p, int x, int y, int cs, int r) {
|
|
/* out of bounds or already marked? */
|
|
if (outbounds(p, x, y) || (marked(p, x, y)&r)==r)
|
|
return 0;
|
|
{
|
|
int x0, x1, y0, y1;
|
|
x0 = x1 = x;
|
|
y0 = y1 = y; // not used
|
|
return frame_nn(p, x, y, &x0, &x1, &y0, &y1, cs, r, JOB->tmp.n_run & 1);
|
|
// using same scheme
|
|
}
|
|
}
|
|
|
|
/* ToDo: finish to replace old frame by this new one
|
|
*
|
|
* @...........#@@@@@@@. # = marked as already scanned black pixels
|
|
* @........@@@@@@@@@@@# only left and right border
|
|
* .......#@@@@@@@@@@@@@ left side on even y
|
|
* ......@@@@@@@@#.@@@@# right side on odd y
|
|
* .....#@@@@@......#@@@ no border is marked twice
|
|
* ....@@@@@#......@@@#. works also for thinn lines
|
|
* ...#@@@@........#@@@. - outer loop is stored as first
|
|
* ..@@@@#........@@@#.. - inner loop is stored as second
|
|
* .#@@@@........#@@@@.. 1st in an extra box (think on white chars)
|
|
* @@@@#.......@@@@#.... 2nd merge in an extra step
|
|
* #@@@@@....#@@@@@.....
|
|
* @@@@@@@@@@@@@@#......
|
|
* .#@@@@@@@@@@@@.......
|
|
*
|
|
* run around the chape using laby-robot
|
|
* - used for scanning boxes, look for horizontal b/w transitions
|
|
* with unmarked black pixels and call this routine
|
|
* - stop if crossing a marked box in same direction (left=up, right=down)
|
|
* box - char box, store frame_vectors and box
|
|
* x,y - starting point
|
|
* mark - 3 bit marker, mark each valid pixel with it
|
|
* diag - 0: only pi/2 direction, 1: pi/4 directions (diagonal)
|
|
* ds - start direction, 6=right of right border, 2=left of left border
|
|
* ret - 0=ok, -1=already marked, -2=max_num_frames_exceeded
|
|
* -7=no border in direction ds
|
|
*/
|
|
#if 0
|
|
#undef g_debug
|
|
#define g_debug(x) x
|
|
#endif
|
|
/* grep keywords: scan_vectors frame_vector */
|
|
int frame_vector(struct box *box1, int x, int y,
|
|
int cs, int mark, int diag, int ds) {
|
|
int i1, i2, i2o,
|
|
new_x=1, /* flag for storing the vector x,y */
|
|
steps=1, /* steps between stored vectors, speedup for big frames */
|
|
d, /* direction */
|
|
ox, oy, /* starting point */
|
|
nx, ny, mx, my, /* used for simplification */
|
|
/* ToDo: add periphery to box (german: Umfang?) */
|
|
rc = 1, /* return code, circumference, sum vector lengths */
|
|
rot = 0, /* memory for rotation, rot=8 means one full rotation */
|
|
vol = 0; /* volume inside frame, negative for white inside black */
|
|
pix *p=box1->p;
|
|
|
|
/* translate the 8 directions to (x,y) pairs,
|
|
* if only four directions are used, only every 2nd vector is accessed,
|
|
* +1 turn left, -1 turn right
|
|
*/
|
|
static const int d0[8][2] =
|
|
{ { 0, -1}, /* up */ {-1, -1}, /* up-le */
|
|
{-1, 0}, /* left */ {-1, 1}, /* do-le */
|
|
{ 0, 1}, /* down */ { 1, 1}, /* do-ri */
|
|
{ 1, 0}, /* right */ { 1, -1} }; /* up-ri */
|
|
|
|
/* check bounds */
|
|
if (outbounds(p, x, y))
|
|
return 0;
|
|
|
|
/* pixel color we are looking for, 0=black, 1=white */
|
|
d = ds;
|
|
i1 = ((getpixel(p, x, y ) < cs) ? 0 : 1);
|
|
i2 = ((getpixel(p, x + d0[d][0], y + d0[d][1]) < cs) ? 0 : 1);
|
|
|
|
g_debug(fprintf(stderr,"\nLEV2 frame_vector @ %3d %3d d%d %2d %2d"
|
|
" %d-%d pix=%3d mark=%d cs=%d",\
|
|
x,y,ds,d0[ds][0],d0[ds][1],i1,i2,getpixel(p,x,y),mark,cs);)
|
|
|
|
if (i1==i2){
|
|
fprintf(stderr,"ERROR frame_vector: no border\n");
|
|
return -7; /* no border detected */
|
|
}
|
|
|
|
/* initialize boxframe outside this function
|
|
box1->x0=box1->x1=x;
|
|
box1->y0=box1->y1=y;
|
|
*/
|
|
|
|
/* initialize boxvector outside this function
|
|
box1->num_frames=0
|
|
num_frame_vectors[0]=0 ???
|
|
and store start value
|
|
*/
|
|
if (box1->num_frames > MaxNumFrames) return -2;
|
|
/* index to next (x,y) */
|
|
i2o=i2=( (box1->num_frames==0)?0:
|
|
box1->num_frame_vectors[ box1->num_frames ] );
|
|
#if 0 // obsolete v0.43
|
|
box1->frame_vector[i2][0]=x;
|
|
box1->frame_vector[i2][1]=y;
|
|
i2++;
|
|
box1->num_frame_vectors[ box1->num_frames ]=i2;
|
|
#endif
|
|
box1->num_frames++;
|
|
|
|
/* robot stores start-position */
|
|
ox = x; oy = y; /* look forward to white pixel */
|
|
|
|
for (;;) { /* stop if same marked pixel touched */
|
|
|
|
g_debug(fprintf(stderr,"\nLEV3: x= %3d %3d d= %d rot= %2d %3d",x,y,d,rot,i2);)
|
|
|
|
/* ToDo: store max. abs(rot) ??? for better recognition */
|
|
if (new_x) {
|
|
g_debug(fprintf(stderr,"\nLEV2: markB xy= %3d %3d ", x, y);)
|
|
p->p[x + y * p->x] |= (mark & 7); /* mark black pixel */
|
|
}
|
|
|
|
/* store a new vector or enlarge the predecessor */
|
|
if (new_x && (rc%steps)==0) { /* dont store everything on big chars */
|
|
if (i2>=MaxFrameVectors) {
|
|
box1->num_frame_vectors[ box1->num_frames-1 ]=i2;
|
|
reduce_vectors(box1,1); /* simplify loop */
|
|
i2=box1->num_frame_vectors[ box1->num_frames-1 ];
|
|
/* enlarge steps on big chars getting speedup */
|
|
steps=(box1->y1-box1->y0+box1->x1-box1->x0)/32+1;
|
|
}
|
|
/* store frame-vector */
|
|
if (i2<MaxFrameVectors) {
|
|
box1->frame_vector[i2][0]=x;
|
|
box1->frame_vector[i2][1]=y;
|
|
/* test if older vector points to the same direction */
|
|
if (i2>1) {
|
|
/* get predecessor */
|
|
nx=box1->frame_vector[i2-1][0]-box1->frame_vector[i2-2][0];
|
|
ny=box1->frame_vector[i2-1][1]-box1->frame_vector[i2-2][1];
|
|
mx=x -box1->frame_vector[i2-1][0];
|
|
my=y -box1->frame_vector[i2-1][1];
|
|
/* same direction? */
|
|
if (nx*my-ny*mx==0 && nx*mx>=0 && ny*my>=0) {
|
|
/* simplify by removing predecessor */
|
|
i2--;
|
|
box1->frame_vector[i2][0]=x;
|
|
box1->frame_vector[i2][1]=y;
|
|
} /* do not simplify */
|
|
}
|
|
i2++;
|
|
box1->num_frame_vectors[ box1->num_frames-1 ]=i2;
|
|
}
|
|
g_debug(fprintf(stderr," stored @ %3d steps= %d", i2-1, steps);)
|
|
}
|
|
new_x=0; /* work for new pixel (x,y) done */
|
|
|
|
/* check if round trip is finished */
|
|
if (x==ox && y==oy && abs(rot)>=8) break;
|
|
|
|
/* look to the front of robot (turtle or ant) */
|
|
nx = x + d0[d][0];
|
|
ny = y + d0[d][1];
|
|
|
|
/* next step, if right is a wall turn the turtle left */
|
|
if ( outbounds(p, nx, ny) || i1 != ((getpixel(p,nx,ny)<cs) ? 0 : 1) ) {
|
|
if (y==ny && nx>=0 && nx<p->x) { /* if inbound */
|
|
g_debug(fprintf(stderr,"\nLEV2: markW xy= %3d %3d ", nx, ny);)
|
|
p->p[nx + ny * p->x] |= (mark & 7); /* mark white pixel */
|
|
}
|
|
/* rotate left 90 or 45 degrees */
|
|
d=(d+2-diag) & 7; rot+=2-diag;
|
|
/* calculate volume inside frame */
|
|
switch (d+diag) {
|
|
case 2+2: vol-=x-1; break;
|
|
case 6+2: vol+=x; break;
|
|
}
|
|
}
|
|
else { /* if no wall, go forward and turn right (90 or 45 degrees) */
|
|
x=nx; y=ny;
|
|
/* turn back and rotate left */
|
|
d=(d+4+2-diag) & 7; rot+=2-diag-4;
|
|
rc++; /* counting steps, used for speedup */
|
|
|
|
/* enlarge frame */
|
|
if (x < box1->x0) box1->x0 = x;
|
|
if (x > box1->x1) box1->x1 = x;
|
|
if (y < box1->y0) box1->y0 = y;
|
|
if (y > box1->y1) box1->y1 = y;
|
|
|
|
new_x=1;
|
|
}
|
|
}
|
|
|
|
/* to distinguish inner and outer frames, store volume as +v or -v */
|
|
box1->frame_vol[ box1->num_frames-1 ] = vol;
|
|
box1->frame_per[ box1->num_frames-1 ] = rc-1;
|
|
|
|
/* dont count and store the first vector twice */
|
|
if (i2-i2o>1) {
|
|
i2--; rc--; box1->num_frame_vectors[ box1->num_frames-1 ]=i2;
|
|
}
|
|
/* output break conditions */
|
|
g_debug(fprintf(stderr,"\nLEV2 o= %3d %3d x= %3d %3d r=%d v=%d",ox,oy,x,y,rot,vol);)
|
|
/* rc=1 for a single point, rc=2 for a two pixel sized point */
|
|
g_debug(fprintf(stderr," steps= %3d vectors= %3d",rc,i2);)
|
|
/* out_x(box1); ToDo: output only the first thousend */
|
|
return rc; /* return number of bordering pixels = periphery? */
|
|
}
|
|
|
|
|
|
|
|
/* clear lowest 3 (marked) bits (they are used for marking) */
|
|
void clr_bits(pix * p, int x0, int x1, int y0, int y1) {
|
|
int x, y;
|
|
for ( y=y0; y <= y1; y++)
|
|
for ( x=x0; x <= x1; x++)
|
|
p->p[x+y*p->x] &= ~7;
|
|
}
|
|
|
|
/* look for white holes surrounded by black points
|
|
* at the moment look for white point with black in all four directions
|
|
* - store position of hole in coordinates relativ to box!
|
|
* ToDo: count only holes with vol>10% ???
|
|
* ToDo: rewrite for frame vectors (faster, no malloc)
|
|
* holes are frames rotating left hand
|
|
* obsolete, do it with vectors
|
|
*/
|
|
int num_hole(int x0, int x1, int y0, int y1, pix * p, int cs, holes_t *holes) {
|
|
int num_holes = 0, x, y, hole_size;
|
|
pix b; // temporary mini-page
|
|
int dx = x1 - x0 + 1, dy = y1 - y0 + 1;
|
|
unsigned char *buf; // 2nd copy of picture, for working
|
|
|
|
if (holes) holes->num=0;
|
|
if(dx<3 || dy<3) return 0;
|
|
b.p = buf = (unsigned char *) malloc( dx * dy );
|
|
if( !buf ){
|
|
fprintf( stderr, "\nFATAL: malloc(%d) failed, skip num_hole", dx*dy );
|
|
return 0;
|
|
}
|
|
if (copybox(p, x0, y0, dx, dy, &b, dx * dy))
|
|
{ free(b.p); return -1;}
|
|
|
|
// printf(" num_hole(");
|
|
/* --- mark white-points connected with border */
|
|
for (x = 0; x < b.x; x++) {
|
|
if (getpixel(&b, x, 0) >= cs)
|
|
mark_nn(&b, x, 0, cs, AT);
|
|
if (getpixel(&b, x, b.y - 1) >= cs)
|
|
mark_nn(&b, x, b.y - 1, cs, AT);
|
|
}
|
|
for (y = 0; y < b.y; y++) {
|
|
if (getpixel(&b, 0, y) >= cs)
|
|
mark_nn(&b, 0, y, cs, AT);
|
|
if (getpixel(&b, b.x - 1, y) >= cs)
|
|
mark_nn(&b, b.x - 1, y, cs, AT);
|
|
}
|
|
|
|
g_debug(out_b(NULL,&b,0,0,b.x,b.y,cs);)
|
|
// --- look for unmarked white points => hole
|
|
for (x = 0; x < b.x; x++)
|
|
for (y = 0; y < b.y; y++)
|
|
if (!((marked(&b, x, y)&AT)==AT)) // unmarked
|
|
if (getpixel(&b, x, y) >= cs) { // hole found
|
|
#if 0
|
|
hole_size=mark_nn(&b, x, y, cs, AT); /* old version */
|
|
if (hole_size > 1 || dx * dy <= 40)
|
|
num_holes++;
|
|
#else
|
|
{ /* new version, for future store of hole characteristics */
|
|
int x0, x1, y0, y1, i, j;
|
|
x0 = x1 = x;
|
|
y0 = y1 = y; // not used
|
|
hole_size=frame_nn(&b, x, y, &x0, &x1, &y0, &y1, cs, AT, JOB->tmp.n_run & 1);
|
|
// store hole for future use, num is initialized with 0
|
|
if (hole_size > 1 || dx * dy <= 40){
|
|
num_holes++;
|
|
if (holes) {
|
|
// sort in table
|
|
for (i=0;i<holes->num && i<MAX_HOLES;i++)
|
|
if (holes->hole[i].size < hole_size) break;
|
|
for (j=MAX_HOLES-2;j>=i;j--)
|
|
holes->hole[j+1]=holes->hole[j];
|
|
if (i<MAX_HOLES) {
|
|
// printf(" i=%d size=%d\n",i,hole_size);
|
|
holes->hole[i].size=hole_size;
|
|
holes->hole[i].x=x;
|
|
holes->hole[i].y=y;
|
|
holes->hole[i].x0=x0;
|
|
holes->hole[i].y0=y0;
|
|
holes->hole[i].x1=x1;
|
|
holes->hole[i].y1=y1;
|
|
}
|
|
holes->num++;
|
|
}
|
|
}
|
|
}
|
|
#endif
|
|
}
|
|
free(b.p);
|
|
// printf(")=%d",num_holes);
|
|
return num_holes;
|
|
}
|
|
|
|
/* count for black nonconnected objects --- used for i,auml,ouml,etc. */
|
|
/* ToDo: obsolete, replaced by vectors and box.num_boxes */
|
|
int num_obj(int x0, int x1, int y0, int y1, pix * p, int cs) {
|
|
int x, y, rc = 0; // rc=num_obj
|
|
unsigned char *buf; // 2nd copy of picture, for working
|
|
pix b;
|
|
|
|
if(x1<x0 || y1<y0) return 0;
|
|
b.p = buf = (unsigned char *) malloc( (x1-x0+1) * (y1-y0+1) );
|
|
if( !buf ){
|
|
fprintf( stderr, "\nFATAL: malloc(%d) failed, skip num_obj",(x1-x0+1)*(y1-y0+1) );
|
|
return 0;
|
|
}
|
|
if (copybox(p, x0, y0, x1 - x0 + 1, y1 - y0 + 1, &b, (x1-x0+1) * (y1-y0+1)))
|
|
{ free(b.p); return -1; }
|
|
// --- mark black-points connected with neighbours
|
|
for (x = 0; x < b.x; x++)
|
|
for (y = 0; y < b.y; y++)
|
|
if (getpixel(&b, x, y) < cs)
|
|
if (!((marked(&b, x, y)&AT)==AT)) {
|
|
rc++;
|
|
mark_nn(&b, x, y, cs, AT);
|
|
}
|
|
free(b.p);
|
|
return rc;
|
|
}
|
|
|
|
#if 0
|
|
// ----------------------------------------------------------------------
|
|
// first idea for making recognition based on probability
|
|
// - start with a list of all possible chars
|
|
// - call recognition_of_char(box *)
|
|
// - remove chars from list which could clearly excluded
|
|
// - reduce probability of chars which have wrong features
|
|
// - font types list could also build
|
|
// at the moment it is only an idea, I should put it to the todo list
|
|
//
|
|
char *list="0123456789,.\0xe4\0xf6\0xfc" // "a=228 o=246 u=252
|
|
"abcdefghijklmnopqrstuvwxyz"
|
|
"ABCDEFGHIJKLMNOPQRSTUVWXYZ";
|
|
int wert[100];
|
|
int listlen=0,numrest=0;
|
|
// initialize a new character list (for future)
|
|
void ini_list(){ int i;
|
|
for(i=0;list[i]!=0 && i<100;i++) wert[i]=0;
|
|
numrest=listlen=i; }
|
|
// exclude??? (for future) oh it was long time ago, I wrote that :/
|
|
void exclude(char *filt){ int i,j;
|
|
for(j=0;filt[j]!=0 && j<100;j++)
|
|
for(i=0;list[i]!=0 && i<100;i++)
|
|
if( filt[j]==list[i] ) { if(!wert[i])numrest--; wert[i]++; } }
|
|
// get the result after all the work (for future)
|
|
char getresult(){ int i;
|
|
if( numrest==1 )
|
|
for(i=0;list[i]!=0 && i<100;i++) if(!wert[i]) return list[i];
|
|
return '_';
|
|
}
|
|
#endif
|
|
|
|
// look at the environment of the pixel too (contrast etc.)
|
|
// detailed analysis only of diff pixels!
|
|
//
|
|
// 100% * "distance", 0 is ideal fit
|
|
// = similarity of two chars for recognition of garbled (verstuemmelter) chars
|
|
// weight of pixels with only one same neighbour set to 0
|
|
// look at contours too! v0.2.4: B==H
|
|
// changed for v0.41, Mar06
|
|
int distance( pix *p1, struct box *box1,
|
|
pix *p2, struct box *box2, int cs){
|
|
int rc=0,x,y,v1,v2,i1,i2,rgood=0,rbad=0,x1,y1,x2,y2,dx,dy,dx1,dy1,dx2,dy2;
|
|
x1=box1->x0;y1=box1->y0;x2=box2->x0;y2=box2->y0;
|
|
dx1=box1->x1-box1->x0+1; dx2=box2->x1-box2->x0+1; dx=((dx1>dx2)?dx1:dx2);
|
|
dy1=box1->y1-box1->y0+1; dy2=box2->y1-box2->y0+1; dy=((dy1>dy2)?dy1:dy2);
|
|
if(abs(dx1-dx2)>1+dx/16 || abs(dy1-dy2)>1+dy/16) return 100;
|
|
// compare relations to baseline and upper line
|
|
if(2*box1->y1>box1->m3+box1->m4 && 2*box2->y1<box2->m3+box2->m4) rbad+=128;
|
|
if(2*box1->y0>box1->m1+box1->m2 && 2*box2->y0<box2->m1+box2->m2) rbad+=128;
|
|
// compare pixels
|
|
for( y=0;y<dy;y++ )
|
|
for( x=0;x<dx;x++ ) { // try global shift too ???
|
|
v1 =((getpixel(p1,x1+x ,y1+y )<cs)?1:0); i1=8; // better gray?
|
|
v2 =((getpixel(p2,x2+x ,y2+y )<cs)?1:0); i2=8; // better gray?
|
|
if(v1==v2) { rgood+=8; continue; } // all things are right!
|
|
// what about different pixel???
|
|
// test overlap of 8 surounding pixels ??? bad if two nb. are bad
|
|
v1=-1;
|
|
for(i1=-1;i1<2;i1++)
|
|
for(i2=-1;i2<2;i2++)if(i1!=0 || i2!=0){
|
|
if( ((getpixel(p1,x1+x+i1*(1+dx/32),y1+y+i2*(1+dy/32))<cs)?1:0)
|
|
!=((getpixel(p2,x2+x+i1*(1+dx/32),y2+y+i2*(1+dy/32))<cs)?1:0) ) v1++;
|
|
}
|
|
if (v1>0) rbad+=16*v1;
|
|
else rbad++;
|
|
}
|
|
if(rgood+rbad) rc= (100*rbad+(rgood+rbad-1))/(rgood+rbad); else rc=99;
|
|
if(rc<10 && JOB->cfg.verbose & 7){
|
|
fprintf(stderr,"\n# distance rc=%d good=%d bad=%d",rc,rgood,rbad);
|
|
// out_x(box1);out_x(box2);
|
|
}
|
|
return rc;
|
|
}
|
|
|
|
|
|
|
|
// ============================= call OCR engine ================== ;)
|
|
// nrun=0 from outside, nrun=1 from inside (allows modifications, oobsolete)
|
|
wchar_t whatletter(struct box *box1, int cs, int nrun){
|
|
wchar_t bc=UNKNOWN; // best letter
|
|
wchar_t um=SPACE; // umlaut? '" => modifier
|
|
pix *p=box1->p; // whole image
|
|
int x,y,dots,xa,ya,x0,x1,y0,y1,dx,dy,i;
|
|
pix b; // box
|
|
struct box bbuf=*box1; // restore after modifikation!
|
|
|
|
if (box1->num_ac>0 && box1->wac[0]>=JOB->cfg.certainty && bc==UNKNOWN) {
|
|
bc=box1->tac[0];
|
|
}
|
|
// if (bc!=UNKNOWN) return bc;
|
|
// if whatletter() called again, only unknown chars are processed
|
|
// bad for splitting!
|
|
|
|
// store box data, which can be modified for modified chars in 2nd run
|
|
bbuf.x0=box1->x0; bbuf.y0=box1->y0;
|
|
bbuf.x1=box1->x1; bbuf.y1=box1->y1;
|
|
|
|
xa=box1->x; ya=box1->y;
|
|
x0=box1->x0; y0=box1->y0;
|
|
x1=box1->x1; y1=box1->y1;
|
|
// int vol=(y1-y0+1)*(x1-x0+1); // volume
|
|
// crossed l-m , divided chars
|
|
while( get_bw(x0,x1,y0,y0,p,cs,1)!=1 && y0+1<y1) y0++;
|
|
while( get_bw(x0,x1,y1,y1,p,cs,1)!=1 && y0+1<y1) y1--;
|
|
dx=x1-x0+1;
|
|
dy=y1-y0+1; // size
|
|
|
|
// better to proof the white frame too!!! ????
|
|
// --- test for german umlaut and points above, not robust enough???
|
|
// if three chars are connected i-dots (ari) sometimes were not detected
|
|
// - therefore after division a test could be useful
|
|
// modify y0 only in second run!?
|
|
// we need it here to have the right copybox
|
|
if (um==SPACE && dy>5 && box1->num_boxes>1)
|
|
testumlaut(box1,cs,2,&um); /* set box1->modifier + new y0 */
|
|
|
|
dots=box1->dots;
|
|
y0 =box1->y0; // dots==2 => y0 below double dots
|
|
dy =y1-y0+1;
|
|
|
|
// move upper and lower border (for divided letters)
|
|
while( get_bw(x0,x1,y0,y0,p,cs,1)==0 && y0+1<y1) y0++;
|
|
while( get_bw(x0,x1,y1,y1,p,cs,1)==0 && y0+1<y1) y1--;
|
|
while( get_bw(x0,x0,y0,y1,p,cs,1)==0 && x0+1<x1) x0++;
|
|
while( get_bw(x1,x1,y0,y1,p,cs,1)==0 && x0+1<x1) x1--;
|
|
dx=x1-x0+1;
|
|
dy=y1-y0+1; // size
|
|
box1->x0=x0; box1->y0=y0; // set reduced frame
|
|
box1->x1=x1; box1->y1=y1;
|
|
|
|
// set good startpoint (probably bad from division)?
|
|
if( xa<x0 || xa>x1 || ya<y0 || ya>y1
|
|
|| getpixel(p,xa,ya)>=cs /* || 2*ya<y0+y1 */ || dots>0 ){
|
|
// subfunction? also called after division of two glued chars?
|
|
for(y=y1;y>=y0;y--) // low to high (not i-dot)
|
|
for(x=(x0+x1)/2,i=0;x>=x0 && x<=x1;i++,x+=((2*i&2)-1)*i) /* is that ok? */
|
|
if (getpixel(p,x,y)<cs && (getpixel(p,x+1,y)<cs
|
|
|| getpixel(p,x,y+1)<cs)){ xa=x;ya=y;y=-1;break; }
|
|
/* should box1->x,y be set? */
|
|
}
|
|
|
|
// ----- create char-only-box -------------------------------------
|
|
// ToDo: this will be obsolete if vectors are used only
|
|
if(dx<1 || dy<1) return bc; /* should not happen */
|
|
b.p = (unsigned char *) malloc( dx * dy );
|
|
if (!b.p) fprintf(stderr,"Warning: malloc failed L%d\n",__LINE__);
|
|
if( copybox(p,x0,y0,dx,dy,&b,dx*dy) )
|
|
{ free(b.p); return bc; }
|
|
// clr_bits(&b,0,b.x-1,0,b.y-1);
|
|
// ------ use diagonal too (only 2nd run?)
|
|
/* following code failes on ! and ? obsolete if vectors are used
|
|
ToDo:
|
|
- mark pixels neighoured to pixels outside and remove them from &b
|
|
v0.40
|
|
will be replaced by list of edge vectors
|
|
- mark accents, dots and remove them from &b
|
|
*/
|
|
#if 1 /* becomes obsolate by vector code */
|
|
if (y0>0) // mark upper overlap
|
|
for ( x=x0; x<=x1; x++) {
|
|
if (getpixel(p,x,y0-1)<cs
|
|
&& getpixel(p,x,y0 )<cs && (marked(&b,x-x0,0)&1)!=1)
|
|
mark_nn(&b,x-x0,0,cs,1);
|
|
}
|
|
if (x0>0) // mark left overlap
|
|
for ( y=y0; y<=y1; y++) {
|
|
if (getpixel(p,x0-1,y)<cs
|
|
&& getpixel(p,x0 ,y)<cs && (marked(&b,0,y-y0 )&1)!=1)
|
|
mark_nn(&b,0,y-y0,cs,1);
|
|
}
|
|
if (x1<p->x-1) // mark right overlap
|
|
for ( y=y0; y<=y1; y++) {
|
|
if (getpixel(p,x1+1,y)<cs
|
|
&& getpixel(p,x1 ,y)<cs && (marked(&b,x1-x0,y-y0)&1)!=1)
|
|
mark_nn(&b,x1-x0,y-y0,cs,1);
|
|
}
|
|
mark_nn(&b,xa-x0,ya-y0,cs,2); // not glued chars
|
|
for(x=0;x<b.x;x++)
|
|
for(y=0;y<b.y;y++){
|
|
if ( (marked(&b,x,y )&3)==1 && getpixel(&b,x,y )<cs )
|
|
b.p[x+y*b.x] = 255&~7; /* reset pixel */
|
|
}
|
|
#endif
|
|
|
|
// if (bc == UNKNOWN) // cause split to fail
|
|
bc=ocr0(box1,&b,cs);
|
|
|
|
/* ToDo: try to change pixels near cs?? or melt? */
|
|
if (box1->num_ac>0 && box1->wac[0]>=JOB->cfg.certainty && bc==UNKNOWN) {
|
|
bc=box1->tac[0];
|
|
}
|
|
|
|
if (um!=0 && um!=SPACE && bc<127) { /* ToDo: is that obsolete now? */
|
|
wchar_t newbc;
|
|
newbc = compose(bc, um );
|
|
if (newbc == bc) { /* nothing composed */
|
|
if(JOB->cfg.verbose & 7)
|
|
fprintf(stderr, "\nDBG whatletter: compose(%s) was useless (%d,%d)",
|
|
decode(bc,ASCII), box1->x0, box1->y0);
|
|
// if(JOB->cfg.verbose & 6) out_x(box1);
|
|
}
|
|
bc = newbc;
|
|
}
|
|
// restore modified boxes
|
|
box1->x0=bbuf.x0; box1->y0=bbuf.y0;
|
|
box1->x1=bbuf.x1; box1->y1=bbuf.y1;
|
|
// if (box1->c==UNKNOWN) out_b(box1,&b,0,0,dx,dy,cs); // test
|
|
|
|
free(b.p);
|
|
return bc;
|
|
}
|
|
|
|
/*
|
|
** creates a list of boxes/frames around objects detected
|
|
** on the pixmap p for further work
|
|
** returns number of boxes created.
|
|
** - by the way: get average X, Y (avX=sumX/numC,..)
|
|
*/
|
|
int scan_boxes( pix *p ){
|
|
int x, y, nx, cs, rc, ds;
|
|
struct box *box3;
|
|
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr,"# scanning boxes");
|
|
|
|
cs = JOB->cfg.cs;
|
|
JOB->res.sumX = JOB->res.sumY = JOB->res.numC = 0;
|
|
|
|
/* clear the lowest bits of each pixel, later used as "scanned"-marker */
|
|
clr_bits( p, 0, p->x - 1, 0, p->y - 1);
|
|
|
|
for (y=0; y < p->y; y++)
|
|
for (x=0; x < p->x; x++)
|
|
for (ds=2; ds<7; ds+=4) { // NO - dust of size 1 is not removed !!!
|
|
nx=x+((ds==2)?-1:+1);
|
|
if (nx<0 || nx>=p->x) continue; /* out of image, ex: recframe */
|
|
if ( getpixel(p, x,y)>=cs || getpixel(p,nx,y)< cs) // b/w transition?
|
|
continue;
|
|
if ((marked(p, x,y) & 1)&&(marked(p, nx, y) & 1))
|
|
continue;
|
|
/* check (and mark) only horizontal b/w transitions */
|
|
// --- insert new box in list
|
|
box3 = (struct box *)malloc_box(NULL);
|
|
box3->x0=box3->x1=box3->x=x;
|
|
box3->y0=box3->y1=box3->y=y;
|
|
box3->num_frames=0;
|
|
box3->dots=0;
|
|
box3->num_boxes=1;
|
|
box3->num_subboxes=0;
|
|
box3->modifier='\0';
|
|
box3->num=JOB->res.numC;
|
|
box3->line=0; // not used here
|
|
box3->m1=0; box3->m2=0; box3->m3=0; box3->m4=0;
|
|
box3->p=p;
|
|
box3->num_ac=0; // for future use
|
|
|
|
/* frame, vectorize and mark only odd/even horizontal b/w transitions
|
|
* args: box, x,y, cs, mark, diag={0,1}, ds={2,6}
|
|
* ds - start direction, 6=right of right border, 2=left of left border
|
|
* ret - 0=ok, -1=already marked, -2=max_num_frames_exceeded
|
|
* -7=no border in direction ds
|
|
* ToDo: count errors and print out for debugging
|
|
*/
|
|
rc=frame_vector(box3, x, y, cs, 1, 1, ds);
|
|
g_debug(fprintf(stderr,"\n# ... scan xy= %3d %3d rc= %2d", x, y, rc);)
|
|
if (rc<0) { free_box(box3); continue; }
|
|
if (box3->num_frames && !box3->num_frame_vectors[0])
|
|
fprintf(stderr,"\nERROR scan_boxes: no vector in frame (%d,%d)",x,y);
|
|
|
|
JOB->res.numC++;
|
|
JOB->res.sumX += box3->x1 - box3->x0 + 1;
|
|
JOB->res.sumY += box3->y1 - box3->y0 + 1;
|
|
|
|
box3->c=(((box3->y1-box3->y0+1)
|
|
*(box3->x1-box3->x0+1)>=MaxBox)? PICTURE : UNKNOWN);
|
|
list_app(&(JOB->res.boxlist), box3); // append to list
|
|
// ToDo: debug
|
|
// if (JOB->cfg.verbose && box3->y0==29) out_x(box3);
|
|
}
|
|
if(JOB->res.numC){
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr," nC= %3d avD= %2d %2d\n",JOB->res.numC,
|
|
(JOB->res.sumX+JOB->res.numC/2)/JOB->res.numC,
|
|
(JOB->res.sumY+JOB->res.numC/2)/JOB->res.numC);
|
|
}
|
|
return JOB->res.numC;
|
|
}
|
|
|
|
/* compare ints for sorting. Return -1, 0, or 1 according to
|
|
whether *vr < *vs, vr == *vs, or *vr > *vs */
|
|
int
|
|
intcompare (const void *vr, const void *vs)
|
|
{
|
|
int *r=(int *)vr;
|
|
int *s=(int *)vs;
|
|
|
|
if (*r < *s) return -1;
|
|
if (*r > *s) return 1;
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* measure_pitch - detect monospaced font and measure the pitch
|
|
* measure overall pitch for difficult lines,
|
|
* after that measure pitch per line
|
|
* dists arrays are limited to 1024 elements to reduce
|
|
* cpu usage for qsort on images with extreme high number of objects
|
|
* insert space if dist>=pitch in list_insert_spaces()
|
|
* ToDo: ???
|
|
* - min/max distance-matrix a-a,a-b,a-c,a-d ... etc; td,rd > ie,el,es
|
|
* - OR measuring distance as min. pixel distance instead of box distance
|
|
* especially useful for italic font!
|
|
*/
|
|
void measure_pitch( job_t *job ){
|
|
int numdists=0, spc=0, /* number of stored distances */
|
|
pitch_p=2, pdist, pdists[1024], /* proportional distances */
|
|
pitch_m=6, mdist, mdists[1024], /* monospaced distances */
|
|
monospaced=0, l1;
|
|
struct box *box2, *prev=NULL;
|
|
|
|
if(job->cfg.verbose){ fprintf(stderr,"# check for word pitch"); }
|
|
for (l1=0; l1<job->res.lines.num; l1++)
|
|
{ /* 0 means all lines */
|
|
if(job->cfg.verbose){ fprintf(stderr,"\n# line %2d",l1); }
|
|
numdists = 0; /* clear distance lists */
|
|
for_each_data(&(job->res.boxlist)) {
|
|
box2 = (struct box *)list_get_current(&(job->res.boxlist));
|
|
if (l1>0 && box2->line!=l1) continue; /* ignore other lines */
|
|
/* ignore dots and pictures (min. font is 4x6) */
|
|
if (box2->y1 - box2->y0 + 1 < 4 || box2->c==PICTURE) { prev=NULL; }
|
|
if (!prev) { prev=box2; continue; } /* we need a predecessor */
|
|
/* use center distance for monospaced fonts */
|
|
mdist = ((box2->x0 + box2->x1) - (prev->x0 + prev->x1) + 1)/2;
|
|
/* use gap for proportional fonts */
|
|
pdist = box2->x0 - prev->x1 + 1;
|
|
/* ToDo: better take 3 instead of 2 neighbours?, smallest font 4x6 */
|
|
/* fonts are expected to be 6 to 60 pixels high, which is about
|
|
4 to 50 pixels wide. We allow some extra margin. */
|
|
if (3 < mdist && mdist < 150) { /* better mdist < 3*Xaverage ? */
|
|
/* two options for overflow: 1) ignore, 2) store randomly */
|
|
if (numdists<1024) { /* we do ignore here */
|
|
mdists[numdists] = mdist;
|
|
pdists[numdists] = pdist;
|
|
numdists++;
|
|
}
|
|
}
|
|
prev = box2;
|
|
} end_for_each(&(job->res.boxlist));
|
|
|
|
if(job->cfg.verbose){ fprintf(stderr," num_gaps= %2d",numdists); }
|
|
if( numdists<8 ){
|
|
if (job->cfg.verbose && l1==0) /* only for all lines */
|
|
fprintf(stderr," (WARNING num_gaps<8)");
|
|
}
|
|
if (numdists>0) {
|
|
int i,diff,ni_min,max,best_m,best_p,ni; double v;
|
|
/* aware: takes long time for big data sets */
|
|
/* dilute? (german: ausduennen?) */
|
|
qsort (mdists, numdists, sizeof (int), intcompare);
|
|
qsort (pdists, numdists, sizeof (int), intcompare);
|
|
/* the new method, div0? */
|
|
v = (mdists[numdists*7/10]-mdists[numdists/5])
|
|
/(double)mdists[numdists/5];
|
|
/* measurements showed v=.09 for Courier and .44 for Times-Roman */
|
|
if (l1==0) monospaced = (v < .22);
|
|
best_m= numdists/5;
|
|
best_p=4*numdists/5;
|
|
/* try to find better pitch for monospaced font (ok for prop) */
|
|
for (i=numdists/5+1;i<numdists;i++) {
|
|
if (2*mdists[i]>=3*mdists[best_m]) { best_m=i-1; break; }
|
|
}
|
|
/* try to find better pitch for proportional font */
|
|
// the largest diff could be the best, if diff is always 1,
|
|
// take the diff with the lowest weight
|
|
for (ni=ni_min=1024,max=0,i=numdists/2+1;i<numdists-numdists/16;i++) {
|
|
diff=pdists[i]-pdists[i-1];
|
|
if (diff>max) {
|
|
max=diff; best_p=i-1;
|
|
if ((job->cfg.verbose&(32+16))==48)
|
|
fprintf(stderr," best_p=%d maxdiff=%d\n# ...", pdists[best_p], max);
|
|
if (max>3 && 3*pdists[i]>=4*pdists[i-1]) { break; }
|
|
}
|
|
if (diff) {
|
|
if (ni<ni_min) {
|
|
// do not try to divide one word per line
|
|
ni_min=ni; if (max<=1 && numdists>16) best_p=i-1;
|
|
if ((job->cfg.verbose&(32+16))==48)
|
|
fprintf(stderr," best_p=%d ni_min=%d\n# ...", pdists[best_p], ni_min);
|
|
}
|
|
ni=1;
|
|
} else ni++;
|
|
}
|
|
if (numdists<16 && max<=1 && ni_min>1) best_p=numdists-1; // one word
|
|
#if 1 /* debugging */
|
|
if ((job->cfg.verbose&(32+16))==48) {
|
|
fprintf(stderr,"\n# ...");
|
|
for (i=0;i<numdists;i++) fprintf(stderr," %2d",mdists[i]);
|
|
fprintf(stderr," <- mdist[%d]\n# ...",l1);
|
|
for (i=0;i<numdists;i++) fprintf(stderr," %2d",pdists[i]);
|
|
fprintf(stderr," <- pdist[%d]\n# ...",l1);
|
|
fprintf(stderr," maxdiff=%d min_samediffs=%d\n# ...",max,ni_min);
|
|
}
|
|
#endif
|
|
/* we measure spaces in two different ways (mono, prop) */
|
|
/* prop: gap between boxes, mono: distance of middle */
|
|
if (best_p<numdists-1) pitch_p = ((pdists[best_p]+pdists[best_p+1])/2+1);
|
|
else pitch_p = (pdists[best_p]+1 );
|
|
pitch_m = (mdists[best_m]*4/3);
|
|
if (numdists)
|
|
if ( pdists[numdists-1]*2 <= pdists[0]*3
|
|
|| pdists[numdists-1] <= pdists[0]+3) {
|
|
/* line is just a single word */
|
|
pitch_p = pdists[numdists-1]+10;
|
|
}
|
|
if (l1>0 && job->cfg.spc==0) {
|
|
job->res.lines.pitch[l1]=(monospaced?pitch_m:pitch_p);
|
|
job->res.lines.mono[l1]=monospaced;
|
|
}
|
|
if (job->cfg.verbose) {
|
|
fprintf(stderr,"\n# ..."
|
|
" mono: v=%f (v<0.22) line=%d numdists=%d\n# ...",
|
|
v, l1, numdists);
|
|
fprintf(stderr," mono: min=%3d max=%3d pitch=%3d @ %2d%%\n# ...",
|
|
mdists[0],mdists[numdists-1],pitch_m,best_m*100/numdists);
|
|
fprintf(stderr," prop: min=%3d max=%3d pitch=%3d @ %2d%%\n# ...",
|
|
pdists[0],pdists[numdists-1],pitch_p,best_p*100/numdists);
|
|
fprintf(stderr," result: distance >= %d considered space\n# ...",
|
|
job->res.lines.pitch[l1]);
|
|
}
|
|
} /* if (not) enough spaces */
|
|
if (l1==0) { /* set default spaces to each line */
|
|
int l2;
|
|
spc = job->cfg.spc;
|
|
if (spc==0) /* set only if not set by option */
|
|
spc = ((monospaced)?pitch_m:pitch_p);
|
|
for (l2=0; l2<job->res.lines.num; l2++ )
|
|
job->res.lines.pitch[l2]=spc;
|
|
}
|
|
} /* each line */
|
|
if (job->cfg.spc==0)
|
|
job->cfg.spc = spc;
|
|
if (job->cfg.verbose)
|
|
fprintf(stderr," overall space width is %d %s\n",
|
|
spc, ((monospaced)?"monospaced":"proportional"));
|
|
|
|
|
|
}
|
|
|
|
/* ---- count subboxes (white holes within black area) --------
|
|
* new: count boxes lying inside another box (usually holes, ex: "aeobdg")
|
|
* needed for glue_boxes, dont glue textboxes, tables and other complex
|
|
* objects
|
|
* ToDo: count only frames of invers spin? do we need sorted list here? -> no
|
|
*/
|
|
int count_subboxes( pix *pp ){
|
|
int ii=0, num_mini=0, num_same=0, cnt=0;
|
|
struct box *box2,*box4;
|
|
progress_counter_t *pc = NULL;
|
|
if (JOB->cfg.verbose) { fprintf(stderr,"# count subboxes\n# ..."); }
|
|
|
|
pc = open_progress(JOB->res.boxlist.n,"count_subboxes");
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
box2->num_subboxes=0;
|
|
progress(cnt++,pc);
|
|
if ( (box2->x1 - box2->x0)<2
|
|
|| (box2->y1 - box2->y0)<2) continue; /* speedup for dotted bg */
|
|
// holes inside box2 char, aoebdqg, 0.41
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box4=(struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if (box4->y0 > box2->y1) break; // faster, but boxes need to be sorted
|
|
// ToDo: better use binary tree (above/below x) to find near boxes?
|
|
if (box4==box2) continue;
|
|
if( box4->x0==box2->x0 && box4->x1==box2->x1
|
|
&& box4->y0==box2->y0 && box4->y1==box2->y1)
|
|
num_same++; /* erroneous!? */
|
|
if ( box4->x0 >= box2->x0 && box4->x1 <= box2->x1
|
|
&& box4->y0 >= box2->y0 && box4->y1 <= box2->y1
|
|
&& box4->num_subboxes==0 ) /* box4 inside box2? */
|
|
{
|
|
box2->num_subboxes++; ii++;
|
|
if ((box4->x1 - box4->x0 + 1)
|
|
*(box4->y1 - box4->y0 + 1)<17) num_mini++;
|
|
}
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
#if 0
|
|
if (cnt < 1000 && JOB->cfg.verbose)
|
|
fprintf(stderr," %4d box %4d %4d %+3d %+3d subboxes %4d\n# ...",
|
|
cnt, box2->x0, box2->y0, box2->x1-box2->x0,
|
|
box2->y1-box2->y0, box2->num_subboxes);
|
|
#endif
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
close_progress(pc);
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr," %3d subboxes counted (mini=%d, same=%d) nC= %d\n",
|
|
ii, num_mini, num_same/2 /* counted twice */, cnt);
|
|
return 0;
|
|
}
|
|
|
|
/* ---- glue holes tochars( before step1 ) v0.42 -----------------------
|
|
glue boxes lying inside another box (usually holes, ex: "aeobdg46890")
|
|
Dont add dust to a char!
|
|
lines are not detected yet
|
|
*/
|
|
int glue_holes_inside_chars( pix *pp ){
|
|
int ii, cs, x0, y0, x1, y1, cnt=0,
|
|
glued_same=0, glued_holes=0;
|
|
struct box *box2, *box4;
|
|
progress_counter_t *pc = NULL;
|
|
cs=JOB->cfg.cs;
|
|
{
|
|
count_subboxes( pp ); /* move to pgm2asc() later */
|
|
|
|
pc = open_progress(JOB->res.boxlist.n,"glue_holes_inside_chars");
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr,"# glue holes to chars nC= %d\n# ...",JOB->res.numC);
|
|
ii=0;
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
// get the smaller box which may be extended by bigger boxes around it
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
x0 = box2->x0; x1 = box2->x1;
|
|
y0 = box2->y0; y1 = box2->y1;
|
|
|
|
progress(cnt++,pc);
|
|
|
|
// would it better than moving vectors to build a sub-box-tree?
|
|
|
|
// do not remove chars inside pictures (car plates on photos)
|
|
if( box2->c == PICTURE || box2->num_subboxes > 7) continue;
|
|
|
|
// holes inside char, aoebdqg, 0.41
|
|
// dont merge boxes which have subboxes by itself!
|
|
// search boxes inside box2
|
|
// if (x1-x0+1>2 || y1-y0+1>2) /* skip tiny boxes, bad for 4x6 */
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box4=(struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if(box4!=box2 && box4->c != PICTURE )
|
|
{
|
|
// ToDo: dont glue, if size differs by big factors (>16?)
|
|
if ( ( box4->x0==x0 && box4->x1==x1
|
|
&& box4->y0==y0 && box4->y1==y1 ) /* do not happen !? */
|
|
|| ( box4->x0>=x0 && box4->x1<=x1
|
|
&& box4->y0>=y0 && box4->y1<=y1
|
|
&& box4->num_subboxes==0 ) ) /* no or very small subboxes? */
|
|
{ // fkt melt(box2,box4)
|
|
// same box, if very small but hollow char (4x5 o)
|
|
if( box4->x0==x0 && box4->x1==x1
|
|
&& box4->y0==y0 && box4->y1==y1) glued_same++; else glued_holes++;
|
|
// fprintf(stderr,"\n# DEBUG merge:");
|
|
// out_x(box2); // small
|
|
// out_x(box4); // big
|
|
if ((JOB->cfg.verbose & 7)==7) // LEV3
|
|
fprintf(stderr," glue hole (%4d %4d %+3d %+3d %+4d)"
|
|
" (%4d %4d %+3d %+3d %+4d) %d\n# ...",
|
|
x0, y0, x1-x0+1, y1-y0+1, box2->frame_vol[0],
|
|
box4->x0, box4->y0,
|
|
box4->x1-box4->x0+1, box4->y1-box4->y0+1,
|
|
box4->frame_vol[0], glued_same);
|
|
if ((box4->x1-box4->x0+1)< 8*(x1-x0+1)
|
|
|| (box4->y1-box4->y0+1)<12*(y1-y0+1)) // skip dust
|
|
merge_boxes( box2, box4 ); // add box4 to box2
|
|
// out_x(box2);
|
|
x0 = box2->x0; x1 = box2->x1;
|
|
y0 = box2->y0; y1 = box2->y1;
|
|
JOB->res.numC--; // dont count fragments as chars
|
|
ii++; // count removed
|
|
list_del(&(JOB->res.boxlist), box4); // remove box4
|
|
free_box(box4);
|
|
// now search another hole inside box2
|
|
}
|
|
}
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr," glued: %3d holes, %3d same, nC= %d\n",
|
|
glued_holes, glued_same, JOB->res.numC);
|
|
close_progress(pc);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
|
|
/* ---- glue broken chars ( before step1 ??? ) -----------------------
|
|
use this carefully, do not destroy previous detection ~fi, broken K=k' g
|
|
glue if boxes are near or diagonally connected
|
|
other strategy: mark boxes for deleting and delete in extra loop at end
|
|
faster: check only next two following boxes because list is sorted!
|
|
ToDo: store m4 of upper line to m4_of_prev_line, and check that "-points are below
|
|
done: glue boxes lying inside another box (usually holes, ex: "aeobdg")
|
|
Dont add dust to a char!
|
|
lines should be detected already (Test it for m1-m4 unknown)
|
|
ToDo: divide in glue_idots, glue_thin_chars etc. and optimize it
|
|
*/
|
|
int glue_broken_chars( pix *pp ){
|
|
int ii, y, cs, x0, y0, x1, y1, cnt=0,
|
|
num_frags=0, glued_frags=0, glued_hor=0;
|
|
struct box *box2, *box4;
|
|
progress_counter_t *pc = NULL;
|
|
cs=JOB->cfg.cs;
|
|
{
|
|
count_subboxes( pp ); /* move to pgm2asc() later */
|
|
|
|
pc = open_progress(JOB->res.boxlist.n,"glue_broken_chars");
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr,"# glue broken chars nC= %d\n# ...",JOB->res.numC);
|
|
ii=0;
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
// get the box which may be extended by boxes around it
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
x0 = box2->x0; x1 = box2->x1;
|
|
y0 = box2->y0; y1 = box2->y1;
|
|
|
|
progress(cnt++,pc);
|
|
|
|
// vertical broken (g965T umlauts etc.)
|
|
// not: f,
|
|
|
|
// would it better than moving vectors to build a sub-box-tree?
|
|
|
|
// do not remove chars inside pictures (car plates on photos)
|
|
if( box2->c == PICTURE || box2->num_subboxes > 7) continue;
|
|
|
|
/* continue loop if box is below or above line */
|
|
if( box2->m4>0 && y0>box2->m4 ) continue; /* dust outside ? */
|
|
if( box2->m1>0 && y0<box2->m1-(box2->m3-box2->m2) ) continue;
|
|
/* ToDo:
|
|
* - check that y0 is greater as m3 of the char/line above
|
|
*/
|
|
|
|
// check small boxes (box2) whether they belong
|
|
// to near same size or bigger boxes (box4)
|
|
if( 2*(y1-y0) < box2->m4 - box2->m1 // care for dots etc.
|
|
&& ( 2*y1<=(box2->m3+box2->m2) // upper fragments
|
|
|| 2*y0>=(box2->m3+box2->m2)) ) { // lower fragments
|
|
struct box *box5=NULL, *box6=NULL; // nearest and next nearest box
|
|
box4=NULL;
|
|
num_frags++; /* count for debugging */
|
|
// get the [2nd] next x-nearest box in the same line
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box4=(struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if (box4 == box2 || box4->c == PICTURE) continue;
|
|
/* 0.42 speed up for backround pixel pattern, box4 to small */
|
|
if ( box4->x1 - box4->x0 + 1 < x1-x0+1
|
|
&& box4->y1 - box4->y0 + 1 < y1-y0+1 ) continue;
|
|
// have in mind that line number may be wrong for dust
|
|
if (box4->line>=0 && box2->line>=0 && box4->line==box2->line)
|
|
{
|
|
if (!box5) box5=box4;
|
|
if ( abs(box4->x0 + box4->x1 - 2*box2->x0)
|
|
<abs(box5->x0 + box5->x1 - 2*box2->x0))
|
|
{ box6=box5; box5=box4; }
|
|
}
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
box4=box5; // next nearest box within the same line
|
|
if (box4) {
|
|
#if 0 /* set this to 1 for debugging of melting bugs */
|
|
if (JOB->cfg.verbose & 7) {
|
|
fprintf(stderr,"\n# next two boxes are candidates for melting ");
|
|
out_x(box2);
|
|
out_x(box4); }
|
|
#endif
|
|
if( /* umlaut "a "o "u, ij; box2 is the small dot, box4 the body */
|
|
( y1 <= box2->m2
|
|
&& box4->y1 >= box2->m2 // dont melt dots together
|
|
&& 2* y1 < box4->y1 + box4->y0 // box2 above box4
|
|
&& box4->x1+JOB->res.avX/2>=x0
|
|
&& box4->x0-JOB->res.avX/2<=x1
|
|
&& (y1 < box4->y0 || x0 < box4->x1) // dont melt "d'"
|
|
&& 3* ( y1 - box4->y0)
|
|
<= 2* (box4->y1 - box4->y0) // too far away? dust!
|
|
&& 8* ( x1 - x0 + 1)
|
|
>= (box4->x1 - box4->x0 + 1) // dot must have minimum size
|
|
&& 10* ( y1 - y0 + 1)
|
|
>= (box4->y1 - box4->y0 + 1) // dot must have minimum size
|
|
) || ( 0 && /* broken T */
|
|
3*(box2->x1 - box2->x0) > 2*JOB->res.avX
|
|
&& 4*box4->x0>3*box2->x0+box2->x1
|
|
&& 4*box4->x1<box2->x0+3*box2->x1
|
|
)
|
|
|| /* !?; box2 is the dot, box4 the body */
|
|
( 2*box4->x1>=x0+x1 /* test if box4 is around box2 */
|
|
&& 2*box4->x0<=2*x1 /* +x0+1 Jan00 */
|
|
&& ( x1-x0 <= box4->x1-box4->x0+2 )
|
|
&& 2*y0>=box2->m2+box2->m3
|
|
&& 4*y1>=box2->m2+3*box2->m3
|
|
&& 4*(y1-y0)<box2->m4-box2->m1
|
|
&& (8*box4->y1 < box4->m2+7*box4->m3
|
|
|| box4->m4-box4->m1<16) /* Jan00 */
|
|
)
|
|
|| /* =;: box2 is the upper box, box4 the lower box */
|
|
( 2*box4->x1>=x0+x1 /* test if box4 is around box2 */
|
|
&& 2*box4->x0<=2*x1 /* +x0+1 */
|
|
&& ( x1-x0 <= box4->x1-box4->x0+4 )
|
|
&& ( 4*x0 <= 3*box4->x1+box4->x0 )
|
|
&& (( box2->m2 && box4->m2
|
|
&& y1< box2->m3
|
|
&& 2*box4->y1 > box4->m3+box4->m2 // can be bigger than m3
|
|
&& 4*box4->y0 >= 3*box4->m2+box4->m3
|
|
&& 2*box2->y0 < box2->m3+box2->m2
|
|
)
|
|
|| ( (!box2->m2) || (!box4->m2) )
|
|
)
|
|
)
|
|
)
|
|
{ // fkt melt(box2,box4)
|
|
if (JOB->cfg.verbose & 7)
|
|
fprintf(stderr," glue objects (%3d %3d %+3d %+3d)"
|
|
" (%3d %3d %+3d %+3d)\n# ...",
|
|
x0, y0, x1-x0+1, y1-y0+1, box4->x0, box4->y0,
|
|
box4->x1-box4->x0+1, box4->y1-box4->y0+1);
|
|
// fprintf(stderr,"\n# DEBUG merge:"); // d=7x34 @ (109,51) ???
|
|
// out_x(box2);
|
|
// out_x(box4);
|
|
merge_boxes( box2, box4 ); // add box4 to box2
|
|
x0 = box2->x0; x1 = box2->x1;
|
|
y0 = box2->y0; y1 = box2->y1;
|
|
// if (JOB->cfg.verbose & 4) out_x(box2);
|
|
// JOB->res.numC--; // dont count fragments as chars
|
|
ii++; glued_frags++; // remove
|
|
// output_list(JOB);
|
|
list_del(&(JOB->res.boxlist), box4); /* ret&1: error-message ??? */
|
|
// output_list(JOB);
|
|
free_box(box4);
|
|
}
|
|
}
|
|
}
|
|
// continue;
|
|
|
|
// horizontally broken w' K'
|
|
if( 2*y1 < (box2->m3+box2->m2) )
|
|
if( 2*(y1-y0) < (box2->m3+box2->m2) ) // fragment
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box4=(struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if(box4!=box2 && box4->c != PICTURE )
|
|
{
|
|
if( box4->line>=0 && box4->line==box2->line
|
|
&& box4->x1>=x0-1 && box4->x1<x0 // do not glue 6-
|
|
&& box4->x0+3*box4->x1<4*x0)
|
|
if( get_bw(x0 ,x0 ,y1,y1 ,pp,cs,1) == 1)
|
|
if( get_bw(x0-2,x0-1,y1,y1+2,pp,cs,1) == 1)
|
|
{ // fkt melt(box2,box4)
|
|
put(pp,x0,y1+1,~(128+64),0);
|
|
merge_boxes( box2, box4 );
|
|
x0 = box2->x0; x1 = box2->x1;
|
|
y0 = box2->y0; y1 = box2->y1;
|
|
JOB->res.numC--; ii++; // remove
|
|
glued_hor++;
|
|
list_del(&(JOB->res.boxlist), box4);
|
|
free_box(box4);
|
|
}
|
|
}
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
|
|
// horizontally broken n h (h=l_) v0.2.5 Jun00
|
|
if( abs(box2->m2-y0)<=(y1-y0)/8 )
|
|
if( abs(box2->m3-y1)<=(y1-y0)/8 )
|
|
if( num_cross(x0, x1,(y0+ y1)/2,(y0+ y1)/2,pp,cs) == 1)
|
|
if( num_cross(x0, x1,(y0+3*y1)/4,(y0+3*y1)/4,pp,cs) == 1)
|
|
if( get_bw((3*x0+x1)/4,(3*x0+x1)/4,(3*y0+y1)/4,y1,pp,cs,1) == 0)
|
|
if( get_bw(x0,(3*x0+x1)/4,(3*y0+y1)/4,(y0+3*y1)/4,pp,cs,1) == 0)
|
|
if( get_bw(x0, x0, y0,(3*y0+y1)/4,pp,cs,1) == 1)
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box4=(struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if(box4!=box2 && box4->c != PICTURE )
|
|
{
|
|
if( box4->line>=0 && box4->line==box2->line
|
|
&& box4->x1>x0-3 && box4->x1-2<x0
|
|
&& abs(box4->y1-box2->m3)<2)
|
|
{ // fkt melt(box2,box4)
|
|
y=loop(pp,x0,y0,y1-y0,cs,0,DO);if(2*y>y1-y0) continue;
|
|
put(pp,x0-1,y0+y ,~(128+64),0);
|
|
put(pp,x0-1,y0+y+1,~(128+64),0);
|
|
merge_boxes( box2, box4 ); // add box4 to box2
|
|
x0 = box2->x0; x1 = box2->x1;
|
|
y0 = box2->y0; y1 = box2->y1;
|
|
JOB->res.numC--; ii++; // remove
|
|
glued_hor++;
|
|
list_del(&(JOB->res.boxlist), box4);
|
|
free_box(box4);
|
|
}
|
|
}
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr," glued: %3d fragments (found %3d), %3d rest, nC= %d\n",
|
|
glued_frags, num_frags, glued_hor, JOB->res.numC);
|
|
close_progress(pc);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
** this is a simple way to improve results on noisy images:
|
|
** - find similar chars (build cluster of same chars)
|
|
** - analyze clusters (could be used for generating unknown font-base)
|
|
** - the quality of the result depends mainly on the distance function
|
|
*/
|
|
// ---- analyse boxes, compare chars, compress picture ------------
|
|
// ToDo: - error-correction only on large chars!
|
|
int find_same_chars( pix *pp){
|
|
int i,k,d,cs,dist,n1,dx; struct box *box2,*box3,*box4,*box5;
|
|
pix p=(*pp);
|
|
cs=JOB->cfg.cs;
|
|
{
|
|
if(JOB->cfg.verbose)fprintf(stderr,"# packing");
|
|
i = list_total(&(JOB->res.boxlist));
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box4 = box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
dist=1000; // 100% maximum
|
|
dx = box2->x1 - box2->x0 + 1;
|
|
|
|
if(JOB->cfg.verbose)fprintf(stderr,"\r# packing %5d",i);
|
|
if( dx>3 )
|
|
for(box3=(struct box *)list_next(&(JOB->res.boxlist),box2);box3;
|
|
box3=(struct box *)list_next(&(JOB->res.boxlist),box3)) {
|
|
if(box2->num!=box3->num){
|
|
int d=distance(&p,box2,&p,box3,cs);
|
|
if ( d<dist ) { dist=d; box4=box3; } // best fit
|
|
if ( d<5 ){ // good limit = 5% ???
|
|
i--;n1=box3->num; // set all num==box2.num to box2.num
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box5=(struct box *)(struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if(box5!=box2)
|
|
if( box5->num==n1 ) box5->num=box2->num;
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
// out_x2(box2,box5);
|
|
// fprintf(stderr," dist=%d\n",d);
|
|
}
|
|
}
|
|
}
|
|
// nearest dist to box2 has box4
|
|
// out_b2(box2,box4);
|
|
// fprintf(stderr," dist=%d\n",dist);
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
k=0;
|
|
if(JOB->cfg.verbose)fprintf(stderr," %d different chars",i);
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
struct box *box3,*box4;
|
|
int j,dist;
|
|
box2=(struct box *)list_get_current(&(JOB->res.boxlist));
|
|
for(box3=(struct box *)list_get_header(&(JOB->res.boxlist));
|
|
box3!=box2 && box3!=NULL;
|
|
box3=(struct box *)list_next(&(JOB->res.boxlist), box3))
|
|
if(box3->num==box2->num)break;
|
|
if(box3!=box2 && box3!=NULL)continue;
|
|
i++;
|
|
// count number of same chars
|
|
dist=0;box4=box2;
|
|
|
|
for(box3=box2,j=0;box3;
|
|
box3=(struct box *)list_next(&(JOB->res.boxlist), box3)) {
|
|
if(box3->num==box2->num){
|
|
j++;
|
|
d=distance(&p,box2,&p,box3,cs);
|
|
if ( d>dist ) { dist=d; box4=box3; } // worst fit
|
|
}
|
|
}
|
|
if(JOB->cfg.verbose&8){
|
|
out_x2(box2,box4);
|
|
fprintf(stderr," no %d char %4d %5d times maxdist=%d\n",i,box2->num,j,dist);
|
|
}
|
|
// calculate mean-char (error-correction)
|
|
// ToDo: calculate maxdist in group
|
|
k+=j;
|
|
// if(j>1)
|
|
// out_b(box1,NULL,0,0,0,0,cs);
|
|
if(JOB->cfg.verbose&8)
|
|
fprintf(stderr," no %d char %4d %5d times sum=%d\n",i,box2->num,j,k);
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if(JOB->cfg.verbose)fprintf(stderr," ok\n");
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
** call the first engine for all boxes and set box->c=result;
|
|
**
|
|
*/
|
|
int char_recognition( pix *pp, int mo){
|
|
int i,ii,ni,cs,x0,y0,x1,y1;
|
|
struct box *box2;
|
|
progress_counter_t *pc;
|
|
wchar_t cc;
|
|
cs=JOB->cfg.cs;
|
|
// ---- analyse boxes, find chars ---------------------------------
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr,"# char recognition");
|
|
i=ii=ni=0;
|
|
for_each_data(&(JOB->res.boxlist)) { /* count boxes */
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
/* wew: isn't this just JOB->res.numC? */
|
|
/* js: The program is very complex. I am not sure anymore
|
|
wether numC is the number of boxes or the number of valid
|
|
characters.
|
|
Because its not time consuming I count the boxes here. */
|
|
if (box2->c==UNKNOWN) i++;
|
|
if (box2->c==PICTURE) ii++;
|
|
ni++;
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if(JOB->cfg.verbose)
|
|
fprintf(stderr," unknown= %d picts= %d boxes= %d\n# ",i,ii,ni);
|
|
if (!ni) return 0;
|
|
i=ii=0;
|
|
pc = open_progress(ni,"char_recognition");
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
x0=box2->x0;x1=box2->x1;
|
|
y0=box2->y0;y1=box2->y1; // box
|
|
cc=box2->c;
|
|
if (cc==PICTURE) continue;
|
|
|
|
if ((mo&256)==0) { /* this case should be default (main engine) */
|
|
if(cc==UNKNOWN || box2->num_ac==0 || box2->wac[0]<JOB->cfg.certainty)
|
|
cc=whatletter(box2,cs ,0);
|
|
}
|
|
|
|
if(mo&2)
|
|
if(cc==UNKNOWN || box2->num_ac==0 || box2->wac[0]<JOB->cfg.certainty)
|
|
cc=ocr_db(box2);
|
|
|
|
|
|
// box2->c=cc; bad idea (May03 removed)
|
|
// set(box2,cc,95); ToDo: is that better?
|
|
|
|
if(cc==UNKNOWN)
|
|
i++;
|
|
ii++;
|
|
|
|
if(JOB->cfg.verbose&8) {
|
|
fprintf(stderr,"\n# code= %04lx %c",(long)cc,(char)((cc<255)?cc:'_'));
|
|
out_b(box2,pp,x0,y0,x1-x0+1,y1-y0+1,cs);
|
|
}
|
|
progress(ii,pc); /* ii = 0..ni */
|
|
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
close_progress(pc);
|
|
if(JOB->cfg.verbose)fprintf(stderr," %d of %d chars unidentified\n",i,ii);
|
|
return 0;
|
|
}
|
|
|
|
|
|
/*
|
|
** compare unknown with known chars,
|
|
** very similar to the find_similar_char_function but here only to
|
|
** improve the result
|
|
*/
|
|
int compare_unknown_with_known_chars(pix * pp, int mo) {
|
|
int i, cs = JOB->cfg.cs, dist, d, ad, wac, ni, ii;
|
|
struct box *box2, *box3, *box4;
|
|
progress_counter_t *pc=NULL;
|
|
wchar_t bc;
|
|
i = ii = 0; // ---- -------------------------------
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr, "# try to compare unknown with known chars !(mode&8)");
|
|
if (!(mo & 8))
|
|
{
|
|
ii=ni=0;
|
|
for_each_data(&(JOB->res.boxlist)) { ni++; } end_for_each(&(JOB->res.boxlist));
|
|
pc = open_progress(ni,"compare_chars");
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist)); ii++;
|
|
if (box2->c == UNKNOWN || (box2->num_ac>0 && box2->wac[0]<97))
|
|
if (box2->y1 - box2->y0 > 4 && box2->x1 - box2->x0 > 1) { // no dots!
|
|
box4 = (struct box *)list_get_header(&(JOB->res.boxlist));;
|
|
dist = 1000; /* 100% maximum */
|
|
bc = UNKNOWN; /* best fit char */
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box3 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
wac=((box3->num_ac>0)?box3->wac[0]:100);
|
|
if (box3 == box2 || box3->c == UNKNOWN
|
|
|| wac<JOB->cfg.certainty) continue;
|
|
if (box2->y1 - box2->y0 < 5 || box2->x1 - box2->x0 < 3) continue;
|
|
d = distance(pp, box2, pp, box3, cs);
|
|
if (d < dist) {
|
|
dist = d; bc = box3->c; box4 = box3;
|
|
}
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if (dist < 10) {
|
|
/* sureness can be maximal of box3 */
|
|
if (box4->num_ac>0) ad = box4->wac[0];
|
|
else ad = 97;
|
|
ad-=dist; if(ad<1) ad=1;
|
|
/* ToDo: ad should depend on ad of bestfit */
|
|
setac(box2,(wchar_t)bc,ad);
|
|
i++;
|
|
} // limit as option???
|
|
// => better max distance('e','e') ???
|
|
if (dist < 50 && (JOB->cfg.verbose & 7)) { // only for debugging
|
|
fprintf(stderr,"\n# L%02d best fit was %04x=%c dist=%3d%% i=%d",
|
|
box2->line, (int)bc, (char)((bc<128)?bc:'_'), dist, i);
|
|
if(box4->num_ac>0)fprintf(stderr," w= %3d%%",box4->wac[0]);
|
|
if ((JOB->cfg.verbose & 4) && dist < 10)
|
|
out_x2(box2, box4);
|
|
}
|
|
progress(ii,pc);
|
|
}
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
close_progress(pc);
|
|
}
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr, " - found %d (nC=%d)\n", i, ii);
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
// ---- divide overlapping chars which !strchr("_,.:;",c);
|
|
// block-splitting (two ore three glued chars)
|
|
// division if dots>0 does not work properly! ???
|
|
//
|
|
// what about glued "be"?
|
|
// what about recursive division?
|
|
// ToDo: mark divided boxes to give the engine a chance to
|
|
// handle wrong divisions
|
|
*/
|
|
int try_to_divide_boxes( pix *pp, int mo){
|
|
struct box *box2, boxa, boxb;
|
|
int cs=JOB->cfg.cs, ad=100,
|
|
a2[8], ar, // certainty of each part, ar = product of all certainties
|
|
cbest; // best certainty, skip search of certainty<cbest-1 for speed
|
|
wchar_t ci[8], // split max. 8 chars
|
|
s1[]={ UNKNOWN, '_', '.', ',', '\'', '!', ';', '?', ':', '-',
|
|
'=', '(', ')', '/', '\\', '\0' }; // not accepted chars, \0-terminated!
|
|
int x0, x1, y0, y1,
|
|
xi[8+1]; // cutting positions
|
|
int i, ii, n1, dy, dx;
|
|
// pix p=(*pp); // remove!
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr,"# try to divide unknown chars !(mode&16)");
|
|
if(!(mo&16)) // put this to the caller
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
// don't try to split simple structures (ex: 400x30 square)
|
|
if ((!box2->num_frames)
|
|
|| box2->num_frame_vectors[ box2->num_frames-1 ]<9) continue;
|
|
if((box2->c==UNKNOWN || (box2->num_ac && box2->wac[0]<JOB->cfg.certainty))
|
|
&& box2->x1-box2->x0>5 && box2->y1-box2->y0>4){
|
|
x0=box2->x0; x1=box2->x1;
|
|
y0=box2->y0; y1=box2->y1;
|
|
ad=100;
|
|
cbest=0;
|
|
|
|
/* get minimum vertical lines */
|
|
n1 = num_cross(x0,x1,( y1+y0)/2,( y1+y0)/2,pp,cs);
|
|
ii = num_cross(x0,x1,(3*y1+y0)/4,(3*y1+y0)/4,pp,cs); if (ii<n1) n1=ii;
|
|
if (box2->m2 && box2->m3 > box2->m2+2)
|
|
for (i=box2->m2+1;i<=box2->m3-1;i++) {
|
|
if (loop(pp,x0+1,i,x1-x0,cs,1,RI) > (x1-x0-2)) continue; // ll
|
|
ii = num_cross(x0,x1,i,i,pp,cs); if (ii<n1) n1=ii;
|
|
} if (n1<2) continue; // seems to make no sense to divide
|
|
if (n1<4) ad=99*ad/100; // not to strong because m2+m3 could be wrong
|
|
if (n1<3) ad=99*ad/100;
|
|
|
|
if( 2*y1 < box2->m3+box2->m4 /* baseline char ? */
|
|
&& num_cross(x0,x1,y1-1,y1-1,pp,cs)==1 // -1 for slopes
|
|
&& num_cross((x0+2*x1)/3,(x0+3*x1)/4,y0,y1,pp,cs)<3 // not exclude tz
|
|
&& num_cross((3*x0+x1)/4,(2*x0+x1)/3,y0,y1,pp,cs)<3 // not exclude zl
|
|
&& loop(pp,x0,y1-(y1-y0)/32,x1-x0,cs,0,RI)
|
|
+loop(pp,x1,y1-(y1-y0)/32,x1-x0,cs,0,LE) > (x1-x0+1)/2
|
|
) continue; /* do not try on bvdo"o etc. */
|
|
|
|
// one vertical line can not be two glued chars, lc?
|
|
if ( num_cross(x0,x1,(y1+y0)/2,(y1+y0)/2,pp,cs)<=1 ) continue;
|
|
{ // doublet = 2 letters
|
|
// char buf[4]="\0\0\0"; // 4th byte is string end == \0
|
|
// buf[0]=c1; // c1 is wchar_t! (0xbf00 to 0) failes
|
|
// buf[1]=c2;
|
|
char buf[64]=""; // end == \0
|
|
if (JOB->cfg.verbose&2){
|
|
fprintf(stderr, "\n#\n# divide box: %4d %4d %3d %3d\n",
|
|
x0, y0, x1-x0+1, y1-y0+1);
|
|
if (JOB->cfg.verbose&4) out_x(box2);
|
|
}
|
|
// it would be better if testing is only if most right and left char
|
|
// is has no horizontal gap (below m2) ex: be
|
|
i=0; // num splittet chars
|
|
xi[0]=x0; xi[1]=x0+3; xi[2]=x1;
|
|
for ( ; ; xi[i+1]++) { // x[i] .. x[i+1], slower? but better v0.42
|
|
/* break if x is to near to the right border */
|
|
if (xi[i+1]>x1-3) { if (i==0) break; i--; xi[i+2]=x1; continue; }
|
|
// ToDo: skip if not a local dy-min for speedup
|
|
{ int ymin=y1, ymax=y0, bow=0, // min max at cutting point
|
|
max0=y0, max1=y0, // max y on left and right side
|
|
min0=y1, min1=y1; // min y on left and right side
|
|
for (dy=0,ii=0;ii<box2->num_frame_vectors[ 0 ];ii++) {
|
|
int pre=ii-1, next=(ii+1)%box2->num_frame_vectors[ 0 ];
|
|
if (pre<0) pre=box2->num_frame_vectors[ 0 ]-1;
|
|
// check if vector is inside box to cut
|
|
if ( box2->frame_vector[ii ][0]<=xi[i ]) continue;
|
|
if ( box2->frame_vector[ii ][0]> xi[i+2]) continue;
|
|
// 2nd derivation of y(x)
|
|
if (abs(box2->frame_vector[ii ][0]-xi[i+1])<2) {
|
|
dy= 2*box2->frame_vector[ii ][1]
|
|
-box2->frame_vector[next][1]
|
|
-box2->frame_vector[pre ][1];
|
|
dx= box2->frame_vector[next][0]
|
|
-box2->frame_vector[pre ][0];
|
|
// rotate 180 degree if dx<0
|
|
if (((dx>0)?dy:-dy)<-abs(dx)/2) { bow=1; }
|
|
}
|
|
// its not the best if we think on glued fi fo etc.
|
|
if (( box2->frame_vector[pre ][0]<=xi[i+1]
|
|
&& box2->frame_vector[next][0]>=xi[i+1])
|
|
|| ( box2->frame_vector[pre ][0]>=xi[i+1]
|
|
&& box2->frame_vector[next][0]<=xi[i+1])) {
|
|
if ( box2->frame_vector[ii ][1]>ymax)
|
|
ymax= box2->frame_vector[ii ][1];
|
|
if ( box2->frame_vector[ii ][1]<ymin)
|
|
ymin= box2->frame_vector[ii ][1];
|
|
}
|
|
// min and max of left and right side
|
|
if ( box2->frame_vector[ii ][1]>max0
|
|
&& box2->frame_vector[ii ][0]<=xi[i+1])
|
|
max0=box2->frame_vector[ii ][1];
|
|
if ( box2->frame_vector[ii ][1]>max1
|
|
&& box2->frame_vector[ii ][0]> xi[i+1])
|
|
max1=box2->frame_vector[ii ][1];
|
|
if ( box2->frame_vector[ii ][1]<min0
|
|
&& box2->frame_vector[ii ][0]<=xi[i+1])
|
|
min0=box2->frame_vector[ii ][1];
|
|
if ( box2->frame_vector[ii ][1]<min1
|
|
&& box2->frame_vector[ii ][0]> xi[i+1])
|
|
min1=box2->frame_vector[ii ][1];
|
|
}
|
|
if(JOB->cfg.verbose&2)
|
|
fprintf(stderr,"\n# test if to split at x%d= %2d %2d %2d"
|
|
" bow,(max-min)[i,0,1] %d %3d %3d %3d"
|
|
, i, xi[i]-x0, xi[i+1]-x0, xi[i+2]-x0, bow, ymax-ymin, max0-min0, max1-min1);
|
|
/* skip if no local minimum at xi[i+1] or if its not thin enough */
|
|
if (bow==0 || 4*(ymax-ymin)>2*(y1-y0)) continue;
|
|
// cuttet parts should have about the same height (max-min)
|
|
// we dont want to cut an 'n' in three parts!
|
|
if (2*(max0-min0+1)<(y1-y0+1)) continue; // left height
|
|
if (2*(max1-min1+1)<(y1-y0+1)) continue; // right height
|
|
// ToDo: thickness on xi[i+1]?
|
|
}
|
|
// try to split successive right box if left box is recognised,
|
|
// else shift the splitting point further to the right border
|
|
// removing ->dots if dot only above one char !!! ??? not implemented
|
|
if(JOB->cfg.verbose&2)
|
|
fprintf(stderr,"\n# try to split, newbox[%d].x= %2d ... %2d "
|
|
"dy= %d ", i, xi[i]-x0, xi[i+1]-x0, dy);
|
|
boxa=*box2; // copy contents, ToDo: reset ac-list (in cut_box?)
|
|
boxa.x=xi[i]; boxa.y=y0; // obsolete? mark pixel, overlap?
|
|
boxa.x0=xi[i];boxa.x1=xi[i+1]; // new horizontal box range
|
|
cut_box(&boxa); boxa.num_ac=0;
|
|
// out_x(&boxa);
|
|
// get wchar + certainty
|
|
ci[i]=whatletter(&boxa,cs,0); a2[i]=testac(&boxa,ci[i]);
|
|
if(JOB->cfg.verbose&2)
|
|
fprintf(stderr,"\n# certainty %d limit= %d cbest= %d ",
|
|
a2[i], JOB->cfg.certainty, cbest);
|
|
if (a2[i]<JOB->cfg.certainty || a2[i]<cbest-1
|
|
|| wcschr(s1,ci[i]) ) { continue; } // dont split here
|
|
|
|
for (ar=ad,ii=0;ii<=i;ii++) {
|
|
ar=a2[ii]*ar/100; } // multiply all probabilities
|
|
if (ar<98*JOB->cfg.certainty/100 || ar<cbest) {
|
|
continue; } // dont go deeper, no longer string
|
|
|
|
i++; if (i==8) break; // maximum splits
|
|
if (i==4) break; // at the moment its to slow to go further
|
|
if (i+1<8) xi[i+1]=x1; // right border of next box
|
|
if (i+2<8) xi[i+2]=x1;
|
|
|
|
if(JOB->cfg.verbose&2)
|
|
fprintf(stderr,"\n try end split [%d]=%d [%d]=%d ",
|
|
i, xi[i]-x0, i+1, xi[i+1]-x0);
|
|
boxb=*box2; // try rest if it has to be split again
|
|
boxb.x=xi[i]+1; boxb.y=y0;
|
|
boxb.x0=xi[i]+1;boxb.x1=xi[i+1];
|
|
cut_box(&boxb); boxb.num_ac=0;
|
|
ci[i]=whatletter(&boxb,cs,0); a2[i]=testac(&boxb,ci[i]);
|
|
if (a2[i]<JOB->cfg.certainty || a2[i]<cbest-1
|
|
|| wcschr(s1,ci[i]) ) { xi[i+1]=xi[i]+2; continue; } // split rest
|
|
// now we have everything splittet
|
|
|
|
if(JOB->cfg.verbose&2) {
|
|
fprintf(stderr,"\n split at/to: ");
|
|
for (ii=0;ii<=i;ii++)
|
|
fprintf(stderr," %2d %s (%3d)", xi[ii+1]-x0,
|
|
decode(ci[ii],ASCII), a2[ii]);
|
|
fprintf(stderr,"\n");
|
|
}
|
|
// boxa..c changed!!! dots should be modified!!!
|
|
// Question: cut it into boxes v0.40 or set a string v0.41?
|
|
// new way of building a string v0.41 (can call setas multiple)
|
|
// usefull if compare unknown with known strings (except barcode?)
|
|
// ToDo: also create alternate variants? ex: I <-> l
|
|
for (buf[0]=0,ar=ad,ii=0;ii<=i;ii++) {
|
|
ar=a2[ii]*ar/100; // multiply all probabilities
|
|
if (i>0 && ci[ii]=='n' && ci[ii-1]=='r') ar--; // m == rn
|
|
strncat(buf,decode(ci[ii],JOB->cfg.out_format),20);
|
|
}
|
|
|
|
if (ar>cbest) cbest=ar; // best (highest) certainty found
|
|
// reduce, but not if we cross certainty border
|
|
if (99*ar/100 > JOB->cfg.certainty) ar=99*ar/100;
|
|
if (JOB->cfg.verbose&2)
|
|
fprintf(stderr,"\n split result= %s (%3d) ",buf, ar);
|
|
setas(box2,buf,ar); // char *, does it disturb further splitting?
|
|
buf[0]=0;
|
|
i--; xi[i+2]=x1;
|
|
}
|
|
}
|
|
}
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if (JOB->cfg.verbose) fprintf(stderr,", numC %d\n",JOB->res.numC);
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
// ---- divide vertical glued boxes (ex: g above T);
|
|
*/
|
|
int divide_vert_glued_boxes( pix *pp, int mo){
|
|
struct box *box2,*box3,*box4;
|
|
int y0,y1,y,dy,flag_found,dx;
|
|
if(JOB->cfg.verbose)fprintf(stderr,"# divide vertical glued boxes");
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if (box2->c != UNKNOWN) continue; /* dont try on pictures */
|
|
y0=box2->y0; y1=box2->y1; dy=y1-y0+1;
|
|
dx=4*(JOB->res.avX+box2->x1-box2->x0+1); // we want to be sure to look at 4ex distance
|
|
if ( dy>2*JOB->res.avY && dy<6*JOB->res.avY && box2->m1
|
|
&& y0<=box2->m2+2 && y0>=box2->m1-2
|
|
&& y1>=box2->m4+JOB->res.avY-2)
|
|
{ // test if lower end fits one of the other lines?
|
|
box4=box2; flag_found=0;
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box4 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if (box4->c != UNKNOWN) continue; /* dont try on pictures */
|
|
if (box4->x1<box2->x0-dx || box4->x0>box2->x1+dx) continue; // ignore far boxes
|
|
if (box4->line==box2->line ) flag_found|=1; // near char on same line
|
|
if (box4->line==box2->line+1) flag_found|=2; // near char on next line
|
|
if (flag_found==3) break; // we have two vertical glued chars
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if (flag_found!=3) continue; // do not divide big chars or special symbols
|
|
y=box2->m4; // lower end of the next line
|
|
if(JOB->cfg.verbose&2){
|
|
fprintf(stderr,"\n# divide box below y=%4d",y-y0);
|
|
if(JOB->cfg.verbose&6)out_x(box2);
|
|
}
|
|
// --- insert box3 before box2
|
|
box3= (struct box *) malloc_box(box2);
|
|
box3->y1=y;
|
|
box2->y0=y+1; box2->line++; // m1..m4 should be corrected!
|
|
if (box4->line == box2->line){
|
|
box2->m1=box4->m1; box2->m2=box4->m2;
|
|
box2->m3=box4->m3; box2->m4=box4->m4;
|
|
}
|
|
box3->num=JOB->res.numC;
|
|
if (list_ins(&(JOB->res.boxlist), box2, box3)) {
|
|
fprintf(stderr,"ERROR list_ins\n"); };
|
|
JOB->res.numC++;
|
|
}
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if(JOB->cfg.verbose)fprintf(stderr,", numC %d\n",JOB->res.numC);
|
|
return 0;
|
|
}
|
|
|
|
|
|
/*
|
|
on some systems isupper(>255) cause a segmentation fault SIGSEGV
|
|
therefore this function
|
|
ToDo: should be replaced (?) by wctype if available on every system
|
|
*/
|
|
int wisupper(wchar_t cc){ return ((cc<128)?isupper(cc):0); }
|
|
int wislower(wchar_t cc){ return ((cc<128)?islower(cc):0); }
|
|
int wisalpha(wchar_t cc){ return ((cc<128)?isalpha(cc):0); }
|
|
int wisdigit(wchar_t cc){ return ((cc<128)?isdigit(cc):0); }
|
|
int wisspace(wchar_t cc){ return ((cc<128)?isspace(cc):0); }
|
|
|
|
/* set box2->c to cc if cc is in the ac-list of box2, return 1 on success */
|
|
int setc(struct box *box2, wchar_t cc){
|
|
int ret=0, w1, w2;
|
|
w1=((box2->num_ac) ? box2->wac[0] : 0); // weight of replaced char
|
|
w2=testac(box2,cc);
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr, "\n# change %s (%d) to %s (%d to %d) at (%d,%d)",
|
|
decode(box2->c,ASCII), w1, decode(cc,ASCII), w2, (100+w2+1)/2,
|
|
box2->x0, box2->y0);
|
|
if (w2) { if (box2->c!=cc) { ret=1; setac(box2,cc,(100+w2+1)/2); } }
|
|
// if(JOB->cfg.verbose & 4) out_x(box2);
|
|
// ToDo: modify per setac (shift ac)
|
|
return ret;
|
|
}
|
|
|
|
|
|
/* ---- proof difficult chars Il1 by context view ----
|
|
context: separator, number, vowel, nonvowel, upper case ????
|
|
could be also used to find unknown chars if the environment (nonumbers)
|
|
can be found in other places!
|
|
ToDo:
|
|
- box->tac[] as set of possible chars, ac set by engine, example:
|
|
ac="l/" (not "Il|/\" because serifs detected and slant>0)
|
|
correction only to one of the ac-set (alternative chars)!
|
|
- should be language-settable; Unicode compatible
|
|
- box2->ad and wac should be changed? (not proper yet)
|
|
* ------------- */
|
|
int context_correction( job_t *job ) {
|
|
// const static char
|
|
char *l_vowel="aeiouy";
|
|
// *l_Vowel="AEIOU",chars if the environment (nonumbers)
|
|
char *l_nonvo = "bcdfghjklmnpqrstvwxz";
|
|
struct box *box4, *box3, *box2, *prev, *next;
|
|
// pix *pp = &(job->src.p);
|
|
int nc=0, ns=0; // num corrections
|
|
|
|
if (job->cfg.verbose)
|
|
fprintf(stderr, "# context correction Il1 0O");
|
|
|
|
for_each_data(&(job->res.boxlist)) {
|
|
box2 = (struct box *)list_get_current(&(job->res.boxlist));
|
|
if (box2->c > 0xFF) continue; // temporary UNICODE fix
|
|
prev = (struct box *)list_get_cur_prev(&(job->res.boxlist));
|
|
next = (struct box *)list_get_cur_next(&(job->res.boxlist));
|
|
if( (prev) && (prev->c > 0xFF)) continue; // temporary UNICODE fix 2
|
|
if( (next) && (next->c > 0xFF)) continue; // temporary UNICODE fix 3
|
|
if (box2->num_ac<2) continue; // no alternatives
|
|
if (box2->wac[0]==100 && box2->wac[1]<100) continue;
|
|
if (box2->num_ac && box2->tas[0]) continue; // buggy space_remove 0.42
|
|
|
|
/* check for Il1| which are general difficult to distinguish */
|
|
/* bbg: not very good. Should add some tests to check if is preceded by '.',
|
|
spelling, etc */
|
|
/* ToDo: only correct if not 100% sure (wac[i]<100)
|
|
and new char is in wat[] */
|
|
if (strchr("Il1|", box2->c) && next && prev) {
|
|
// if( strchr(" \n",prev->c) // SPC
|
|
// && strchr(" \n",next->c) ) box2->c='I'; else // bad idea! I have ...
|
|
if (wisalpha(next->c) && next->c!='i' &&
|
|
( prev->c == '\n' ||
|
|
( prev->c == ' ' &&
|
|
( box4=(struct box *)list_prev(&(job->res.boxlist), prev)) &&
|
|
box4->c == '.' ) ) ) { nc+=setc(box2,(wchar_t)'I'); }
|
|
else if (box2->c!='1' && strchr(l_nonvo,next->c) &&
|
|
strchr("\" \n",prev->c)) /* lnt => Int, but 1st */
|
|
/* do not change he'll to he'Il! */
|
|
{ nc+=setc(box2,(wchar_t)'I'); } // set box2->c to 'I' if 'I' is in the ac-list
|
|
else if (strchr(l_vowel,next->c)) /* unusual? Ii Ie Ia Iy Iu */
|
|
/* && strchr("KkBbFfgGpP",prev->c)) */ /* kle Kla Kli */
|
|
{ nc+=setc(box2,(wchar_t)'l'); }
|
|
else if (wisupper(next->c)
|
|
&& !strchr("O0I123456789",next->c)
|
|
&& !strchr("O0I123456789",prev->c)) /* avoid lO => IO (10) */
|
|
{ nc+=setc(box2,(wchar_t)'I'); }
|
|
else if (wislower(prev->c))
|
|
{ nc+=setc(box2,(wchar_t)'l'); }
|
|
else if (wisdigit(prev->c) || wisdigit(next->c)
|
|
|| (next->c=='O' && !wisalpha(prev->c))) /* lO => 10 */
|
|
{ nc+=setc(box2,(wchar_t)'1'); }
|
|
}
|
|
|
|
/* check for O0 */
|
|
else if (strchr("O0", box2->c) && next && prev) {
|
|
if (wisspace(prev->c) && wisalpha(next->c)) /* initial letter */
|
|
{ nc+=setc(box2,(wchar_t)'O'); }
|
|
else if (wisalpha(prev->c) && wisalpha(next->c)
|
|
&& wisupper(next->c)) /* word in upper case */
|
|
{ nc+=setc(box2,(wchar_t)'O'); }
|
|
else if (wisdigit(prev->c) || wisdigit(next->c))
|
|
{ nc+=setc(box2,(wchar_t)'0'); }
|
|
}
|
|
|
|
/* check for 5S */
|
|
else if (strchr("5S", box2->c) && next && prev) {
|
|
if (wisspace(prev->c) && wisalpha(next->c)) /* initial letter */
|
|
{ nc+=setc(box2,(wchar_t)'S'); }
|
|
else if (wisalpha(prev->c) && wisalpha(next->c)
|
|
&& wisupper(next->c)) /* word in upper case */
|
|
{ nc+=setc(box2,(wchar_t)'S'); }
|
|
else if (wisdigit(prev->c) || wisdigit(next->c))
|
|
{ nc+=setc(box2,(wchar_t)'5'); }
|
|
}
|
|
|
|
/* was a space not found? xXx => x Xx ??? */
|
|
if (wisupper(box2->c) && next && prev) {
|
|
if (wislower(prev->c) && wislower(next->c)
|
|
&& 2 * (box2->x0 - prev->x1) > 3 * (next->x0 - box2->x1)) {
|
|
struct box *box3 = malloc_box((struct box *) NULL);
|
|
box3->x0 = prev->x1 + 2;
|
|
box3->x1 = box2->x0 - 2;
|
|
box3->y0 = box2->y0;
|
|
box3->y1 = box2->y1;
|
|
box3->x = box2->x0 - 1;
|
|
box3->y = box2->y0;
|
|
box3->dots = 0;
|
|
box3->num_boxes = 0;
|
|
box3->num_subboxes = 0;
|
|
box3->c = ' ';
|
|
box3->modifier = 0;
|
|
setac(box3,' ',99); /* ToDo: weight depends from distance */
|
|
box3->num = -1;
|
|
box3->line = prev->line;
|
|
box3->m1 = box3->m2 = box3->m3 = box3->m4 = 0;
|
|
box3->p = &(job->src.p);
|
|
list_ins(&(job->res.boxlist), box2, box3);
|
|
}
|
|
}
|
|
|
|
/* a space before punctuation? but not " ./file" */
|
|
if ( prev && next)
|
|
if (prev->c == ' ' && strchr(" \n" , next->c)
|
|
&& strchr(".,;:!?)", box2->c))
|
|
if (prev->x1 - prev->x0 < 2 * job->res.avX) { // carefully on tables
|
|
box3 = prev;
|
|
if ( !list_del(&(job->res.boxlist), box3) ) free_box(box3);
|
|
prev = (struct box *)list_get_cur_prev(&(job->res.boxlist));
|
|
ns++;
|
|
}
|
|
|
|
/* \'\' to \" */
|
|
if ( prev )
|
|
if ( (prev->c == '`' || prev->c == '\'')
|
|
&& (box2->c == '`' || box2->c == '\'') )
|
|
if (prev->x1 - box2->x0 < job->res.avX) { // carefully on tables
|
|
box2->c='\"';
|
|
box3 = prev;
|
|
list_del(&(job->res.boxlist), box3);
|
|
free_box(box3);
|
|
}
|
|
} end_for_each(&(job->res.boxlist));
|
|
if (job->cfg.verbose)
|
|
fprintf(stderr, " num_corrected= %d removed_spaces= %d\n", nc, ns);
|
|
return 0;
|
|
}
|
|
|
|
|
|
/* ---- insert spaces ----
|
|
* depends strongly from the outcome of measure_pitch()
|
|
* ------------------------ */
|
|
int list_insert_spaces( pix *pp, job_t *job ) {
|
|
int i=0, j1, j2, i1, maxline=-1, dy=0; char cc;
|
|
struct box *box2, *box3=NULL, *box4=NULL;
|
|
|
|
// measure mean line height
|
|
for(i1=1;i1<job->res.lines.num;i1++) {
|
|
dy+=job->res.lines.m4[i1]-job->res.lines.m1[i1]+1;
|
|
} if (job->res.lines.num>1) dy/=(job->res.lines.num-1);
|
|
i=0; j2=0;
|
|
for(i1=1;i1<job->res.lines.num;i1++) {
|
|
j1=job->res.lines.m4[i1]-job->res.lines.m1[i1]+1;
|
|
if (j1>dy*120/100 || j1<dy*80/100) continue; // only most frequently
|
|
j2+=j1; i++;
|
|
} if (i>0 && j2/i>7) dy=j2/i;
|
|
if( job->cfg.verbose&1 )
|
|
fprintf(stderr,"# insert space between words (dy=%d) ...",dy);
|
|
if (!dy) dy=(job->res.avY)*110/100+1;
|
|
|
|
i=0;
|
|
for_each_data(&(job->res.boxlist)) {
|
|
box2 =(struct box *)list_get_current(&(job->res.boxlist));
|
|
cc=0;
|
|
if (box2->line>maxline) { // lines and chars must be sorted!
|
|
if (maxline>=0) cc='\n'; // NL
|
|
maxline=box2->line;
|
|
}
|
|
if((box3 = (struct box *)list_prev(&(job->res.boxlist), box2))){
|
|
if (maxline && !box2->line && cc==0) cc=' ';
|
|
if (box2->line<=maxline && cc==0) { // lines and chars must be sorted!
|
|
int thispitch = job->res.lines.pitch[box2->line];
|
|
int thismono = job->res.lines.mono[box2->line];
|
|
int mdist = (box2->x1 + box2->x0 - (box3->x1 + box3->x0) + 1)/2;
|
|
int pdist = box2->x0 - box3->x1 + 1;
|
|
if (box2->x1 - box2->x0 < thispitch) pdist=pdist*4/3;
|
|
/* allow extra pixels around small characters .,'!: etc */
|
|
// fprintf(stderr,"#\n ... mono= %2d pitch= %2d mdist= %2d pdist= %2d",
|
|
// thismono, thispitch, mdist, pdist);
|
|
if ((thismono!=0 && mdist >= thispitch)
|
|
|| (thismono==0 && pdist >= thispitch))
|
|
cc=' '; // insert SPACE
|
|
}
|
|
}
|
|
if(cc){
|
|
box4=(struct box *)list_prev(&(job->res.boxlist), box2);
|
|
box3=(struct box *)malloc_box(NULL);
|
|
box3->x0=box2->x0-2; box3->x1=box2->x0-2;
|
|
box3->y0=box2->y0; box3->y1=box2->y1;
|
|
if(cc!='\n' && box4)
|
|
box3->x0=box4->x1+2;
|
|
if(cc=='\n' || !box4)
|
|
box3->x0=job->res.lines.x0[box2->line];
|
|
if(cc=='\n' && box4){
|
|
box3->y0=box4->y1; // better use lines.y1[box2->pre] ???
|
|
box3->y1=box2->y0;
|
|
}
|
|
box3->x =box2->x0-1; box3->y=box2->y0;
|
|
box3->dots=0; box3->c=cc;
|
|
box3->num_boxes = 0;
|
|
box3->num_subboxes = 0;
|
|
box3->modifier='\0';
|
|
box3->num=-1; box3->line=box2->line;
|
|
box3->m1=box2->m1; box3->m2=box2->m2;
|
|
box3->m3=box2->m3; box3->m4=box2->m4;
|
|
box3->p=pp;
|
|
setac(box3,cc,100); /* ToDo: weight depends from distance */
|
|
list_ins(&(job->res.boxlist),box2,box3);
|
|
if( job->cfg.verbose&1 ) {
|
|
fprintf(stderr,"\n# insert space &%d; at x= %4d %4d box= %p",
|
|
(int)cc, box3->x0, box3->y0, (void*)box3);
|
|
/* out_x(box3); */
|
|
}
|
|
i++;
|
|
}
|
|
} end_for_each(&(job->res.boxlist));
|
|
if( job->cfg.verbose&1 ) fprintf(stderr," found %d\n",i);
|
|
return 0;
|
|
}
|
|
|
|
|
|
/*
|
|
add infos where the box is positioned to the box
|
|
this is useful for better recognition
|
|
*/
|
|
int add_line_info(/* List *boxlist2 */){
|
|
// pix *pp=&JOB->src.p;
|
|
struct tlines *lines = &JOB->res.lines;
|
|
struct box *box2;
|
|
int i,xx,m1,m2,m3,m4,num_line_members=0,num_rest=0;
|
|
if( JOB->cfg.verbose&1 ) fprintf(stderr,"# add line infos to boxes ...");
|
|
for_each_data(&(JOB->res.boxlist)) {
|
|
box2 =(struct box *)list_get_current(&(JOB->res.boxlist));
|
|
for(i=1;i<JOB->res.lines.num;i++) /* line 0 is a place holder */
|
|
{
|
|
if (lines->dx) xx=lines->dy*((box2->x1+box2->x0)/2)/lines->dx; else xx=0;
|
|
m1=lines->m1[i]+xx;
|
|
m2=lines->m2[i]+xx;
|
|
m3=lines->m3[i]+xx;
|
|
m4=lines->m4[i]+xx;
|
|
// fprintf(stderr," test line %d m1=%d %d %d %d\n",i,m1,m2,m3,m4);
|
|
if (m4-m1==0) continue; /* no text line (line==0) */
|
|
#if 0
|
|
if( box2->y1+2*JOB->res.avY >= m1
|
|
&& box2->y0-2*JOB->res.avY <= m4 ) /* not to far away */
|
|
#endif
|
|
/* give also a comma behind the line a chance */
|
|
if( box2->x0 >= lines->x0[i] && box2->x1 <= lines->x1[i]+JOB->res.avX )
|
|
if( box2->m2==0 || abs(box2->y0-box2->m2) > abs(box2->y0-m2) )
|
|
{ /* found nearest line */
|
|
box2->m1=m1;
|
|
box2->m2=m2;
|
|
box2->m3=m3;
|
|
box2->m4=m4;
|
|
box2->line=i;
|
|
}
|
|
}
|
|
if( box2->y1+2 < box2->m1
|
|
|| box2->y0 < box2->m1 - (box2->m3-box2->m1)/2
|
|
|| box2->y0-2 > box2->m4
|
|
|| box2->y1 > box2->m3 + (box2->m3-box2->m1)
|
|
) /* to far away */
|
|
{ /* reset */
|
|
box2->m1=0;
|
|
box2->m2=0;
|
|
box2->m3=0;
|
|
box2->m4=0;
|
|
box2->line=0;
|
|
num_rest++;
|
|
} else num_line_members++;
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if( JOB->cfg.verbose&1 )
|
|
fprintf(stderr," done, num_line_chars=%d rest=%d\n",
|
|
num_line_members, num_rest);
|
|
return 0;
|
|
}
|
|
|
|
|
|
/*
|
|
* bring the boxes in right order
|
|
* add_line_info must be executed first!
|
|
*/
|
|
int sort_box_func (const void *a, const void *b) {
|
|
struct box *boxa, *boxb;
|
|
|
|
boxa = (struct box *)a;
|
|
boxb = (struct box *)b;
|
|
|
|
if ( ( boxb->line < boxa->line ) ||
|
|
( boxb->line == boxa->line && boxb->x0 < boxa->x0 ) )
|
|
return 1;
|
|
return -1;
|
|
}
|
|
|
|
// -------------------------------------------------------------
|
|
// ------ use this for entry from other programs
|
|
// include pnm.h pgm2asc.h
|
|
// -------------------------------------------------------------
|
|
// entry point for gocr.c or if it is used as lib
|
|
// better name is call_ocr ???
|
|
// jb: OLD COMMENT: not removed due to set_options_* ()
|
|
// args after pix *pp should be removed and new functions
|
|
// set_option_mode(int mode), set_option_spacewidth() .... etc.
|
|
// should be used instead, before calling pgm2asc(pix *pp)
|
|
// ! change if you can ! - used by X11 frontend
|
|
int pgm2asc(job_t *job)
|
|
{
|
|
pix *pp;
|
|
progress_counter_t *pc;
|
|
|
|
assert(job);
|
|
/* FIXME jb: remove pp */
|
|
pp = &(job->src.p);
|
|
|
|
if( job->cfg.verbose )
|
|
fprintf(stderr, "# db_path= %s\n", job->cfg.db_path);
|
|
|
|
pc = open_progress(100,"pgm2asc_main");
|
|
progress(0,pc); /* start progress output 0% 0% */
|
|
|
|
/* ----- count colors ------ create histogram -------
|
|
- this should be used to create a upper and lower limit for cs
|
|
- cs is the optimum gray value between cs_min and cs_max
|
|
- also inverse scans could be detected here later */
|
|
if (job->cfg.cs==0)
|
|
job->cfg.cs=otsu( pp->p,pp->y,pp->x,0,0,pp->x,pp->y, job->cfg.verbose & 1 );
|
|
/* renormalize the image and set the normalized threshold value */
|
|
job->cfg.cs=thresholding( pp->p,pp->y,pp->x,0,0,pp->x,pp->y, job->cfg.cs );
|
|
if( job->cfg.verbose )
|
|
fprintf(stderr, "# thresholding new_threshold= %d\n", job->cfg.cs);
|
|
|
|
progress(5,pc); /* progress is only estimated */
|
|
|
|
#if 0 /* dont vast memory */
|
|
/* FIXME jb: malloc */
|
|
if ( job->cfg.verbose & 32 ) {
|
|
// generate 2nd imagebuffer for debugging output
|
|
job->tmp.ppo.p = (unsigned char *)malloc(job->src.p.y * job->src.p.x);
|
|
// buffer
|
|
assert(job->tmp.ppo.p);
|
|
copybox(&job->src.p,
|
|
0, 0, job->src.p.x, job->src.p.y,
|
|
&job->tmp.ppo,
|
|
job->src.p.x * job->src.p.y);
|
|
}
|
|
#else
|
|
job->tmp.ppo=job->src.p; /* temporarely, removed later */
|
|
#endif
|
|
|
|
/* load character data base */
|
|
if ( job->cfg.mode&2 )
|
|
load_db();
|
|
|
|
/* this is first step for reorganize the PG
|
|
---- look for letters, put rectangular frames around letters
|
|
letter = connected points near color F
|
|
should be used by dust removing (faster) and line detection!
|
|
---- 0..cs = black letters, last change = Mai99 */
|
|
|
|
progress(8,pc); /* progress is only estimated */
|
|
|
|
scan_boxes( pp );
|
|
if ( !job->res.numC ){
|
|
fprintf( stderr,"# no boxes found - stopped\n" );
|
|
if(job->cfg.verbose&32) debug_img("out01",job,8);
|
|
/***** should free stuff, etc) */
|
|
return(1);
|
|
}
|
|
// if (job->cfg.verbose&32) debug_img("out00",job,4+8);
|
|
|
|
progress(10,pc); /* progress is only estimated */
|
|
// if(job->cfg.verbose&32) debug_img("out01",job,4+8);
|
|
// output_list(job); // for debugging
|
|
// ToDo: matrix printer preprocessing
|
|
|
|
remove_dust( job ); /* from the &(job->res.boxlist)! */
|
|
// if(job->cfg.verbose&32) debug_img("out02",job,4+8);
|
|
// output_list(job); // for debugging
|
|
smooth_borders( job ); /* only for big chars */
|
|
progress(12,pc); /* progress is only estimated */
|
|
// if(job->cfg.verbose&32) debug_img("out03",job,4+8);
|
|
// output_list(job); // for debugging
|
|
|
|
detect_barcode( job ); /* mark barcode */
|
|
// if(job->cfg.verbose&32) debug_img("out04",job,4+8);
|
|
// output_list(job); // for debugging
|
|
|
|
detect_pictures( job ); /* mark pictures */
|
|
// if(job->cfg.verbose&32) debug_img("out05",job,4+8);
|
|
// output_list(job); // for debugging
|
|
|
|
remove_pictures( job ); /* do this as early as possible, before layout */
|
|
// if(job->cfg.verbose&32) debug_img("out06",job,4+8);
|
|
// output_list(job); // for debugging
|
|
|
|
glue_holes_inside_chars( pp ); /* including count subboxes (holes) */
|
|
|
|
detect_rotation_angle( job );
|
|
|
|
#if 1 /* Rotate the whole picture! move boxes */
|
|
if( job->res.lines.dy!=0 ){ // move down lowest first, move up highest first
|
|
// in work! ??? (at end set dy=0) think on ppo!
|
|
}
|
|
#endif
|
|
detect_text_lines( pp, job->cfg.mode ); /* detect and mark JOB->tmp.ppo */
|
|
// if(job->cfg.verbose&32) debug_img("out07",job,4+8);
|
|
progress(20,pc); /* progress is only estimated */
|
|
|
|
add_line_info(/* &(job->res.boxlist) */);
|
|
if (job->cfg.verbose&32) debug_img("out10",job,4+8);
|
|
|
|
divide_vert_glued_boxes( pp, job->cfg.mode); /* after add_line_info, before list_sort! */
|
|
// if(job->cfg.verbose&32) debug_img("out11",job,0);
|
|
|
|
remove_melted_serifs( pp ); /* make some corrections on pixmap */
|
|
/* list_ins seems to sort in the boxes on the wrong place ??? */
|
|
// if(job->cfg.verbose&32) debug_img("out12",job,4+8);
|
|
|
|
glue_broken_chars( pp ); /* 2nd glue */
|
|
// if(job->cfg.verbose&32) debug_img("out14",job,4+8);
|
|
|
|
remove_rest_of_dust( );
|
|
// if(job->cfg.verbose&32) debug_img("out15",job,4+8);
|
|
|
|
/* better sort after dust is removed (slow for lot of pixels) */
|
|
list_sort(&(job->res.boxlist), sort_box_func);
|
|
|
|
measure_pitch( job );
|
|
|
|
if(job->cfg.mode&64) find_same_chars( pp );
|
|
progress(30,pc); /* progress is only estimated */
|
|
// if(job->cfg.verbose&32) debug_img("out16",job,4+8);
|
|
|
|
char_recognition( pp, job->cfg.mode);
|
|
progress(60,pc); /* progress is only estimated */
|
|
// if(job->cfg.verbose&32) debug_img("out17",job,4+8);
|
|
|
|
if ( adjust_text_lines( pp, job->cfg.mode ) ) { /* correct using chars */
|
|
/* may be, characters/pictures have changed line number */
|
|
list_sort(&(job->res.boxlist), sort_box_func);
|
|
// 2nd recognition call if lines are adjusted
|
|
char_recognition( pp, job->cfg.mode);
|
|
}
|
|
|
|
#define BlownUpDrawing 1 /* german: Explosionszeichnung, temporarly */
|
|
#if BlownUpDrawing == 1 /* german: Explosionszeichnung */
|
|
{ /* just for debugging */
|
|
int i,ii,ni; struct box *box2;
|
|
i=ii=ni=0;
|
|
for_each_data(&(JOB->res.boxlist)) { /* count boxes */
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if (box2->c==UNKNOWN) i++;
|
|
if (box2->c==PICTURE) ii++;
|
|
ni++;
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr,"# debug: unknown= %d picts= %d boxes= %d\n",i,ii,ni);
|
|
}
|
|
#endif
|
|
// ----------- write out20.pgm ----------- mark lines + boxes
|
|
if (job->cfg.verbose&32) debug_img("out20",job,1+4+8);
|
|
|
|
compare_unknown_with_known_chars( pp, job->cfg.mode);
|
|
progress(70,pc); /* progress is only estimated */
|
|
|
|
try_to_divide_boxes( pp, job->cfg.mode);
|
|
progress(80,pc); /* progress is only estimated */
|
|
|
|
/* --- list output ---- for debugging --- */
|
|
if (job->cfg.verbose&6) output_list(job);
|
|
|
|
/* ---- insert spaces ---- */
|
|
list_insert_spaces( pp , job );
|
|
|
|
// ---- proof difficult chars Il1 by context view ----
|
|
if (JOB->cfg.verbose)
|
|
fprintf(stderr,"# context correction if !(mode&32)\n");
|
|
if (!(job->cfg.mode&32)) context_correction( job );
|
|
|
|
store_boxtree_lines( job->cfg.mode );
|
|
progress(90,pc); /* progress is only estimated */
|
|
|
|
/* 0050002.pgm.gz ca. 109 digits, only 50 recognized (only in lines?)
|
|
* ./gocr -v 39 -m 56 -e - -m 4 -C 0-9 -f XML tmp0406/0050002.pbm.gz
|
|
* awk 'BEGIN{num=0}/1<\/box>/{num++;}END{print num}' o
|
|
* 15*0 24*1 18*2 19*3 15*4 6*5 6*6 6*7 4*8 8*9 sum=125digits counted boxes
|
|
* 9*0 19*1 14*2 15*3 11*4 6*5 5*6 6*7 4*8 8*9 sum=97digits recognized
|
|
* 1*1 1*7 not recognized (Oct04)
|
|
* 33*SPC 76*NL = 109 spaces + 36*unknown sum=241 * 16 missed
|
|
*/
|
|
#if BlownUpDrawing == 1 /* german: Explosionszeichnung */
|
|
{ /* just for debugging */
|
|
int i,ii,ni; struct box *box2; const char *testc="0123456789ABCDEFGHIJK";
|
|
i=ii=ni=0;
|
|
for_each_data(&(JOB->res.boxlist)) { /* count boxes */
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if (box2->c==UNKNOWN) i++;
|
|
if (box2->c==PICTURE) ii++;
|
|
if (box2->c>' ' && box2->c<='z') ni++;
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if(JOB->cfg.verbose)
|
|
fprintf(stderr,"# debug: (_)= %d picts= %d chars= %d",i,ii,ni);
|
|
for (i=0;i<20;i++) {
|
|
ni=0;
|
|
for_each_data(&(JOB->res.boxlist)) { /* count boxes */
|
|
box2 = (struct box *)list_get_current(&(JOB->res.boxlist));
|
|
if (box2->c==testc[i]) ni++;
|
|
} end_for_each(&(JOB->res.boxlist));
|
|
if(JOB->cfg.verbose && ni>0)
|
|
fprintf(stderr," (%c)=%d",testc[i],ni);
|
|
}
|
|
if(JOB->cfg.verbose)
|
|
fprintf(stderr,"\n");
|
|
}
|
|
#endif
|
|
|
|
// ---- frame-size-histogram
|
|
// ---- (my own defined) distance between letters
|
|
// ---- write internal picture of textsite
|
|
// ----------- write out30.pgm -----------
|
|
if( job->cfg.verbose&32 ) debug_img("out30",job,2+4);
|
|
|
|
progress(100,pc); /* progress is only estimated */
|
|
|
|
close_progress(pc);
|
|
|
|
return 0; /* what should I return? error-state? num-of-chars? */
|
|
}
|