File: pagesegmentation.hpp

package info (click to toggle)
gamera 1%3A3.4.2%2Bgit20160808.1725654-1
  • links: PTS, VCS
  • area: main
  • in suites: stretch
  • size: 22,312 kB
  • ctags: 24,991
  • sloc: xml: 122,324; ansic: 52,869; cpp: 50,664; python: 35,034; makefile: 118; sh: 101
file content (741 lines) | stat: -rw-r--r-- 24,596 bytes parent folder | download | duplicates (3)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
/*
 *
 * Copyright (C) 2007-2009 Stefan Ruloff, Maria Elhachimi,
 *                         Ilya Stoyanov, Robert Butz
 *               2010      Tobias Bolten
 *               2007-2011 Christoph Dalitz
 *
 * 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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
 */

#ifndef cd20070814_pagesegmentation
#define cd20070814_pagesegmentation

#include <Python.h>
#include <map>
#include <vector>
#include <iostream>
#include <algorithm>
#include <stdexcept>
#include <functional>
#include "gamera.hpp"
#include "gameramodule.hpp"
#include "gamera_limits.hpp"
#include "connected_components.hpp"
#include "plugins/listutilities.hpp"
#include "plugins/projections.hpp"
#include "plugins/segmentation.hpp"
#include "plugins/image_utilities.hpp"


namespace Gamera {

/* Function: median
 * Calculates the middle height of the CCs.
 * Used for setting defualt parameters in 
 * runlength_smearing and projection_cutting
 */
int pagesegmentation_median_height(ImageList* ccs) {
  vector<int> ccs_heights;
  ImageList::iterator i;

  if (ccs->empty()) {
    throw std::runtime_error("pagesegmentation_median_height: no CC's found in image.");
  }
  for (i = ccs->begin(); i != ccs->end(); ++i) {
    ccs_heights.push_back( (*i)->nrows() );
  }
  return median(&ccs_heights);
}


/*****************************************************************************
* Run Length Smearing
* IN:   Cx - Minimal length of white runs in the rows
*   Cy - Minimal length of white runs in the columns
*   Csm- Minimal length of white runs row-wise in the almost final image
*       
*   If you choose "-1" the algorithm will determine the
*   median character length in the image to obtain the values for Cx,Cy or 
*   Csm.
******************************************************************************/
template<class T>
ImageList* runlength_smearing(T &image, int Cx, int Cy, int Csm) {
    typedef OneBitImageView view_type;
    typedef OneBitImageData data_type;
    typedef typename T::value_type value_type;

    data_type* img1_data = new data_type(image.size(), image.origin());
    view_type* img1 = new view_type(*img1_data);
    image_copy_fill(image, *img1);
    
    data_type* img2_data = new data_type(image.size(), image.origin());
    view_type* img2 = new view_type(*img2_data);
    image_copy_fill(image, *img2);

    int Ctemp = 0;
    size_t nrows = image.nrows();
    size_t ncols = image.ncols();
    size_t x, y;
    value_type black_val = black(image);
    value_type white_val = white(image);

    // when no values given, guess them from the Cc size statistics
    if (Csm <= 0 || Cy <= 0 || Cx <= 0) {
      ImageList* ccs_temp = cc_analysis(image);
      int Median = pagesegmentation_median_height(ccs_temp);

      for (ImageList::iterator i = ccs_temp->begin(); i != ccs_temp->end(); i++) {
        delete *i;
      }
      delete ccs_temp;

      if (Csm <= 0)
        Csm = 3 * Median;
      if (Cy <= 0)
        Cy = 20 * Median;
      if (Cx <= 0)
        Cx = 20 * Median;
    }

    // horizontal smearing
    for (y = 0; y < nrows; ++y) {
      for (x = 0, Ctemp = 0; x < ncols; ++x) {
        if (is_white(image.get(Point(x, y)))) {
          Ctemp += 1;
        } else {
          if ((0 != Ctemp) && (Ctemp <= Cx)){
            for (int z = 0; z < Ctemp; z++) {
              img1->set(Point(x-z-1, y), black_val);
            }
          }
          Ctemp = 0;
        }
      }
    }

    // vertical smearing
    for (x = 0; x < ncols; ++x) {
      for (y = 0, Ctemp = 0; y < nrows; ++y) {
        if (is_white(image.get(Point(x, y)))) {
          Ctemp += 1;
        } else {
          if ((0 != Ctemp) && (Ctemp <= Cy)) {
            for (int z = 0; z < Ctemp; z++)
              img2->set(Point(x, y-z-1), black_val);
          }
          Ctemp = 0;
        }
      }
    }

    // logical AND between both images
    for(y = 0; y < nrows; ++y) {
      for(x = 0; x < ncols; ++x) {
        if ((is_black(img1->get(Point(x, y))))
            && (is_black(img2->get(Point(x, y))))){
          img1->set(Point(x, y), black_val);
        } else {
          img1->set(Point(x, y), white_val);
        }
      }
    }

    // again horizontal smearing for removal of small holes
    for (y = 0; y < nrows; ++y) {
      for (x = 0, Ctemp = 0; x < ncols; ++x) {
        if (is_white(img1->get(Point(x, y)))) {
          Ctemp += 1;
        } else {
          if ((0 != Ctemp) && (Ctemp <= Csm)){
            for (int z = 0; z < Ctemp; z++)
              img1->set(Point(x-z-1, y), black_val);
          }
          Ctemp = 0;
        }
      }
    }

    ImageList* ccs_AND = cc_analysis(*img1);
    ImageList* return_ccs = new ImageList();

    // create result Cc's 
    ImageList::iterator i;
    for (i = ccs_AND->begin(); i != ccs_AND->end(); ++i) {  
      Cc* cc = dynamic_cast<Cc*>(*i);
      int label = cc->label();
      bool containspixel = false; // some segments may not contain black pixels

      // Methods "get" and "set" operates relative to the image view
      // but the offset of the connected components is not relative
      // to the view. (here: (*i)->offset_x() and (*i)->offset_y())
      //
      // This means that these values must be adjusted for labeling
      // the image view.
      for (y = 0; y < cc->nrows(); ++y) {
        for (x = 0; x < cc->ncols(); ++x) {
          if ( is_black(image.get(Point(x+(*i)->offset_x()-image.offset_x(),
                                        y+(*i)->offset_y()-image.offset_y())))
              && is_black(cc->get(Point(x,y))) ) {
            image.set(Point(x + cc->offset_x() - image.offset_x(),
                            y + cc->offset_y() - image.offset_y()), label);
            containspixel = true;
          }
        }
      }

      // create new CC with the dimensions, offset and label from the
      // smeared image, pointing to the original image.
      if (containspixel) {
        return_ccs->push_back(new ConnectedComponent<data_type>(
                *((data_type*)image.data()),                // Data
                label,                                      // Label
                Point((*i)->offset_x(), (*i)->offset_y()),  // Point
                (*i)->dim())                                // Dim
                );
      }
    }
    
    // clean up
    for (ImageList::iterator i=ccs_AND->begin(); i!=ccs_AND->end(); i++)
      delete *i;
    delete ccs_AND;
    delete img1->data();
    delete img1;
    delete img2->data();
    delete img2;

    return return_ccs;
}


/*-------------------------------------------------------------------------
 * Functions for projection_cutting:
 * Interne_RXY_Cut(image, Tx, Ty, ccs, noise, label):recursively splits 
 * the image, sets the label and creates the CCs.
 * Start_point(image, ul, lr):search the upper_left point of the sub-image.
 * End_point(image,ul,lr):search the lower_right point of the sub-image.
 * Split_point:searchs the split point of the image
 * rxy_cut(image,Tx,Ty,noise,label):returns the ccs-list
 *-------------------------------------------------------------------------*/


/* Function: Start_Point
 * This funktion is used to search the first black pixel:
 * calculates the coordinates of the begin of the cc
 * returns the coordinates of the upper-left point of subimage
 */



template<class T>
Point proj_cut_Start_Point(T& image, Point ul, Point lr) {
    Point Start;

    for (size_t y = ul.y(); y <= lr.y(); y++) {
    for (size_t x = ul.x(); x <= lr.x(); x++) {
        if ((image.get(Point(x, y))) != 0) {
            Start.x(x);
            Start.y(y);
            goto endLoop1; // unfortunately there is no break(2) in gorgeous C++
        }
    }
    } 
    endLoop1:

    for (size_t x = ul.x(); x <= lr.x(); x++) {
    for (size_t y = ul.y(); y <= lr.y(); y++) {
        if ((image.get(Point(x, y))) != 0) {
            if (Start.x() > x)
                Start.x(x);
            goto endLoop2; // unfortunately there is no break(2) in gorgeous C++
            }
        }
    }
    endLoop2:
    return Start;
}

/* Function: End_Point
 * This funktion is used to search the last black pixel:the lower-right point
 * of subimage calculates the coordinates of the end of the CC.
 */
template<class T>
Point proj_cut_End_Point(T& image, Point ul, Point lr) {
    Point End;
    size_t x, y;

    for (y = lr.y(); y+1 >= ul.y()+1; y--) {
        for (x = lr.x(); x+1 >= ul.x()+1; x--) {
            if ((image.get(Point(x, y))) != 0) {
                End.x(x);
                End.y(y);
                goto endLoop1;
            }
        }
    }
    endLoop1:
    
    for (x = lr.x(); x+1 > ul.x()+1; x--) {
        for (y = lr.y(); y+1 > ul.y()+1; y--) {
            if ((image.get(Point(x,y))) != 0){
                if (End.x()<x)
                    End.x(x);
                goto endLoop2;
            }
        }
    }
    endLoop2:

    return End;
}

/* Function: Split_Point
 * calculates the coordinates of the split_point.
 * The split point is determined
 * by finding the largest possible gaps in the X and Y projection of the image.
 */
template<class T>
IntVector * proj_cut_Split_Point(T& image, Point ul, Point lr, int Tx, int Ty, int noise, int gap_treatment, char direction ) {
    IntVector * SplitPoints = new IntVector(); //empty IntVector
    size_t size;
    lr.x()-ul.x()>lr.y()-ul.y()?size=lr.x()-ul.x():size=lr.y()-ul.y();

    // probably no need for such big mem-alloc, but necessary in certain situations
    int* SplitPoints_Min = new int[size];
    int* SplitPoints_Max = new int[size]; 
    int gap_width = 0; // width of the gap
    int gap_counter = 0; //number of gaps

    if (direction == 'x'){
        // Correct Points for Rect() with offset
        Point a( ul.x() + image.offset_x(), ul.y() + image.offset_y() );
        Point b( lr.x() + image.offset_x(), lr.y() + image.offset_y() );
        IntVector *proj_x = projection_rows(image, Rect(a, b));
        SplitPoints->push_back(ul.y()); // starting point
        
        for (size_t i = 1; i < proj_x->size(); i++) {
            if ((*proj_x)[i] <= noise) {
                gap_width++;
                if (Ty <= gap_width) {// min-gap <= act-gap?
                SplitPoints_Min[gap_counter] = (i + ul.y() - gap_width+1);
                SplitPoints_Max[gap_counter] = (i + ul.y()); // finally set to last point of gap
                }
            } 
            else {
                if (Ty <= gap_width)
                    gap_counter++;
                gap_width = 0;
            }
        }
    delete proj_x;
    }
    else{ // y-direction
        // Correct Points for Rect() with offset
        Point a( ul.x() + image.offset_x(), ul.y() + image.offset_y() );
        Point b( lr.x() + image.offset_x(), lr.y() + image.offset_y() );
        IntVector *proj_y = projection_cols(image, Rect(a, b));
        SplitPoints->push_back(ul.x()); // starting point
        
        for (size_t i = 1; i < proj_y->size(); i++) {
            if ((*proj_y)[i] <= noise) {
                gap_width++;
                if (Tx <= gap_width) {// min-gap <= act-gap?
                SplitPoints_Min[gap_counter] = (i + ul.x() - gap_width+1);
                SplitPoints_Max[gap_counter] = (i + ul.x()); // finally set to last point of gap
                }
            } 
            else {
                if (Tx <= gap_width) 
                    gap_counter++;
                gap_width = 0;
            }
        }
        delete proj_y;
    }
    
    for (int i=0; i<gap_counter; i++){
        if (0==gap_treatment){ // cut exactly in the middle of the gap -> no unlabeled noise pixels
            int mid = (SplitPoints_Min[i] + SplitPoints_Max[i]) / 2;
            SplitPoints_Min[i] = mid;
            SplitPoints_Max[i] = mid;
        }
        SplitPoints->push_back(SplitPoints_Min[i]);
        SplitPoints->push_back(SplitPoints_Max[i]);
    }   
    direction=='x'? SplitPoints->push_back(lr.y()): SplitPoints->push_back(lr.x()); // ending point
    
    delete [] SplitPoints_Min;
    delete [] SplitPoints_Max;

    return SplitPoints;
}



/* Function: Interne_RXY_Cut
 * This function recursively splits the image in horizontal or 
 * vertical direction.
 * The original image will have all of its pixels "labeled" with a number
 * representing each connected component.
 */
template<class T>
void projection_cutting_intern(T& image, Point ul, Point lr, ImageList* ccs, 
        int Tx, int Ty, int noise, int gap_treatment, char direction, int& label) {
    
    Point Start = proj_cut_Start_Point(image, ul, lr);
    Point End = proj_cut_End_Point(image, ul, lr);
    IntVector * SplitPoints = proj_cut_Split_Point(image, Start, End, Tx, Ty, noise, gap_treatment, direction);
    IntVector::iterator It;
    
    ul.x(Start.x());
    ul.y(Start.y());
    lr.x(End.x());
    lr.y(End.y());
    
        

    if (!(direction=='y' && SplitPoints->size() == 2)){ // ending condition, SplitPoints==2 => only Start- and Endpoint no gaps
        if (direction=='x'){
            direction = 'y';
            for(It = SplitPoints->begin(); It != SplitPoints->end(); It++){
                Point begin, end; // note the lowercase of end, which is not End
                begin.x(Start.x());
                begin.y(*It);
                It++;
                end.x(End.x());
                end.y(*It);
                projection_cutting_intern(image, begin, end, ccs, Tx, Ty, noise, gap_treatment, direction, label);
            }
        }
        else { // direction==y
            direction = 'x';
            for(It = SplitPoints->begin(); It != SplitPoints->end(); ++It){
                Point begin, end; // note the lowercase of end, which is not End
                begin.x(*It);
                begin.y(Start.y());
                It++;
                end.x(*It);
                end.y(End.y());
                projection_cutting_intern(image, begin, end, ccs, Tx, Ty, noise, gap_treatment, direction, label);
            }
        }
    } else {
        label++;
        for (size_t y = ul.y(); y <= lr.y(); y++) {
            for (size_t x = ul.x(); x <= lr.x(); x++) {
                if((image.get(Point(x, y))) != 0){
                    image.set(Point(x, y), label);
                }
            }
        }

        Point cc(Start.x() + image.offset_x(), Start.y() + image.offset_y());
        ccs->push_back(
                new ConnectedComponent<typename T::data_type>(
                    *((typename T::data_type*)image.data()),
                    OneBitPixel(label),
                    cc,
                    Dim((End.x() - Start.x() + 1), (End.y() - Start.y() + 1))
                )
            );
    }
    delete SplitPoints;
}

/*
 * Function: rxy_cut 
 * Returns a list of ccs found in the image.
 */
template<class T>
ImageList* projection_cutting(T& image, int Tx, int Ty, int noise, int gap_treatment) {
    int Label = 1;
    char direction = 'x';

    if (noise < 0) {
        noise = 0;
    }

    // set default values
    if (Tx < 1 || Ty < 1) {
        ImageList* ccs_temp = cc_analysis(image);
        int Median = pagesegmentation_median_height(ccs_temp);
        for (ImageList::iterator i = ccs_temp->begin(); 
                i != ccs_temp->end(); i++) {
            delete *i;
        }
        delete ccs_temp;
        if (Tx < 1) {
          Tx = Median * 7;
        }
        if (Ty < 1) {
          if (Median > 1) Ty = Median / 2;
          else Ty = 1;
        }
    }
    // set minimal gap_width
    /*if (Tx <= 2){
        if (gap_treatment)
            Tx=2;
        else
            Tx=3;
    }
    if (Ty <= 2){
        if (gap_treatment)
            Ty=2;
        else
            Ty=3;
    }*/

    ImageList* ccs = new ImageList();
    Point ul, lr;

    ul.x(0);
    ul.y(0);
    lr.x(image.ncols() - 1);
    lr.y(image.nrows() - 1);
    projection_cutting_intern(image, ul, lr, ccs, Tx, Ty, noise, gap_treatment, direction, Label);
    
    return ccs;
}




/*
sub_cc_analysis
@param cclist The list of CCs inside the image

@return A tuple with two values
    1. the image with the new labels from the new CCs
    2. a list of ImageLists
        a list-entry is a cc_analysis of a cclist from the argument
*/
template<class T>
PyObject* sub_cc_analysis(T& image, ImageVector &cclist) {
    unsigned int pos;
    int label = 2; // one is reserved for unlabeled pixels
    OneBitImageData *ret_image;
    OneBitImageView *ret_view;
    OneBitImageData *temp_image;
    OneBitImageView *temp_view;
    Cc* cc;
    ImageVector::iterator iv;
    ImageList::iterator il;
    typename T::value_type Black = black(image);

    ret_image = new OneBitImageData(image.dim(), image.origin());
    ret_view = new OneBitImageView(*ret_image, image.origin(), image.dim());
    temp_image = new OneBitImageData(image.dim(), image.origin());
    temp_view = new OneBitImageView(*temp_image, image.origin(), image.dim());

    // Generate a list to store the CCs of all lines
    PyObject *return_cclist = PyList_New(cclist.size());
    
    for (iv = cclist.begin(), pos = 0; iv != cclist.end(); iv++, pos++) {
        cc = static_cast<Cc*>(iv->first);

        // copy the needed CC from the original image(image) 
        // to the temporary image temp_view
        for (size_t y = 0; y < cc->nrows(); y++) {
          for (size_t x = 0; x < cc->ncols(); x++) {
            if (!is_white(cc->get(Point(x, y)))) {
              temp_view->set(Point(x+cc->offset_x()-temp_view->offset_x(), y+cc->offset_y()-temp_view->offset_y()), Black);
            }
          }
        }
        
        // generate a temp image for the cc_analysis,
        // it's simply a copy of one cclist entry
        OneBitImageView *cc_temp = new OneBitImageView(*temp_image, cc->origin(), cc->dim() );

        // Cc_analysis of one list entry
        ImageList* ccs_orig = cc_analysis(*cc_temp);

        ImageList* return_ccs = new ImageList();
        il = ccs_orig->begin();
        while (il != ccs_orig->end()) {
            cc = static_cast<Cc*>(*il);
           
            return_ccs->push_back(
                    new ConnectedComponent<typename T::data_type>(
                        *((typename T::data_type*)ret_view->data()),
                        OneBitPixel(label),
                        cc->origin(),
                        cc->dim()
                    )
                );
    
            // Copy CC over to return image
            for (size_t y = 0; y < cc->nrows(); y++) {
              for (size_t x = 0; x < cc->ncols(); x++) {
                if (!is_white(cc->get(Point(x, y)))) {
                  ret_view->set(Point(x+cc->offset_x()-ret_view->offset_x(), y+cc->offset_y()-ret_view->offset_y()), label);
                }
              }
            }

            // delete the temporary used CCs from the cc_analysis
            delete *il;

            il++;
            label++; // we use consecutive labels in return image
        }
        // remove copy of Cc in temporary image and clean up
        fill_white(*cc_temp);
        delete ccs_orig;
        delete cc_temp;

        // Set the Imagelist into the PyList
        // ImageList must be converted to be a valid datatype for the PyList
        PyList_SetItem(return_cclist, pos, ImageList_to_python(return_ccs));
        delete return_ccs;
    }
    // delete temporary image
    delete temp_view;
    delete temp_image;

    // Finaly create the return type, a tuple with a image 
    // and a list of ImageLists
    PyObject *return_values = PyTuple_New(2);
    PyTuple_SetItem(return_values, 0, create_ImageObject(ret_view));
    PyTuple_SetItem(return_values, 1, return_cclist);

    return return_values;
}


//
// evaluation of segmentation
//

// for distinguishing Ccs from Gccs and Sccs
class CcLabel {
public:
  char image; // 'G' or 'S'
  int  cclabel;
  CcLabel(char i, int c) {image = i; cclabel = c;}
  friend int operator<(const CcLabel& c1, const CcLabel& c2) { 
    if (c1.image == c2.image) return (c1.cclabel < c2.cclabel);
    else return c1.image < c2.image;
  }
};

// the plugin function
template<class T, class U>
IntVector* segmentation_error(T &Gseg, U &Sseg) {

  ImageList* Gccs = ccs_from_labeled_image(Gseg);
  ImageList* Sccs = ccs_from_labeled_image(Sseg);
  ImageList::iterator ccs_it;
  size_t x,y;
  int classlabel, Gclasslabel, Sclasslabel;
  CcLabel Gcclabel('G',0), Scclabel('S',0), cclabel('A',0);
  map<CcLabel,int> classoflabel; // cclabel -> classlabel
  multimap<int,CcLabel> labelsofclass; // classlabel -> cclabel
  typedef multimap<int,CcLabel>::iterator mm_iterator;
  mm_iterator mmit;
  pair<mm_iterator,mm_iterator> fromto;
  vector<CcLabel> tmplabels;
  vector<CcLabel>::iterator vit;

  // check for overlaps from Gseg
  for (ccs_it = Gccs->begin(), classlabel = 0; ccs_it != Gccs->end(); ++ccs_it, ++classlabel) {
    Gclasslabel = classlabel;
    Cc* cc = static_cast<Cc*>(*ccs_it);
    Gcclabel.cclabel = cc->label();
    classoflabel[Gcclabel] = Gclasslabel;
    labelsofclass.insert(make_pair(Gclasslabel,Gcclabel));
    for (y=0; y < cc->nrows(); y++)
      for (x=0; x < cc->ncols(); x++) {
        if (!cc->get(Point(x,y))) continue;
        Scclabel.cclabel = Sseg.get(Point(cc->ul_x() + x, cc->ul_y() + y));
        // in case of overlap:
        if (Scclabel.cclabel) {
          // check whether segment from S is new
          if (classoflabel.find(Scclabel) == classoflabel.end()) {
            classoflabel[Scclabel] = Gclasslabel;
            labelsofclass.insert(make_pair(Gclasslabel,Scclabel));
          } else {
            Sclasslabel = classoflabel[Scclabel];
            if (Sclasslabel != Gclasslabel) {
              // unite both classes, i.e. move Sclasslabel into Gclasslabel
              tmplabels.clear();
              fromto = labelsofclass.equal_range(Sclasslabel);
              for (mmit = fromto.first; mmit != fromto.second; ++mmit) {
                cclabel = mmit->second;
                classoflabel[cclabel] = Gclasslabel;
                tmplabels.push_back(cclabel);
              }
              labelsofclass.erase(Sclasslabel);
              for (vit = tmplabels.begin(); vit != tmplabels.end(); ++vit)
                labelsofclass.insert(make_pair(Gclasslabel,*vit));
            }
          }
        }
      }
  }

  // check for CCs from Sseg without overlap (false positives)
  for (ccs_it = Sccs->begin(); ccs_it != Sccs->end(); ++ccs_it) {
    Cc* cc = static_cast<Cc*>(*ccs_it);
    Scclabel.cclabel = cc->label();
    if (classoflabel.find(Scclabel) == classoflabel.end()) {
        classlabel++;
        classoflabel[Scclabel] = classlabel;
        labelsofclass.insert(make_pair(classlabel,Scclabel));
    }
  }

  // build up class population numbers and classify error types
  int n1,n2,n3,n4,n5,n6,nG,nS;
  n1 = n2 = n3 = n4 = n5 = n6 = 0;
  for (mmit = labelsofclass.begin(); mmit != labelsofclass.end(); ) {
    nG = nS = 0;
    fromto = labelsofclass.equal_range(mmit->first);
    for (mmit = fromto.first; mmit != fromto.second; ++mmit) {
      if (mmit->second.image == 'G') nG++; else nS++;
    }
    // determine error category
    if (nG == 1 && nS == 1) n1++;
    else if (nG == 1 && nS == 0) n2++;
    else if (nG == 0 && nS == 1) n3++;
    else if (nG == 1 && nS  > 1) n4++;
    else if (nG  > 1 && nS == 1) n5++;
    else if (nG  > 1 && nS  > 1) n6++;
    else printf("Plugin segment_error: empty equivalence"
                " constructed which should not happen\n");
  }

  // clean up
  for (ccs_it = Sccs->begin(); ccs_it != Sccs->end(); ++ccs_it)
    delete *ccs_it;
  delete Sccs;
  for (ccs_it = Gccs->begin(); ccs_it != Gccs->end(); ++ccs_it)
    delete *ccs_it;
  delete Gccs;

  // build return value
  IntVector* errors = new IntVector();
  errors->push_back(n1); errors->push_back(n2);
  errors->push_back(n3); errors->push_back(n4);
  errors->push_back(n5); errors->push_back(n6);
  return errors;
}

} // end of namespace Gamera

#endif