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 | /*
This file implements the different edge detection algorithms
presented in the article "A review of classic edge detectors".
All functions implemented here receive only a pointer to the
input image (grayscale) and the different parameters of each
method. All them also return a pointer to the output image.
Copyright (c) 2011-2013, Haldo Sponton <haldos@fing.edu.uy>
Copyright (c) 2011-2013, Juan Cardelino <juanc@fing.edu.uy>
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 3 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, see <http://www.gnu.org/licenses/>.
*/
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include "classic_edge_detectors.h"
// Useful macros
#define MAX(x, y) (((x) > (y)) ? (x) : (y))
#define MIN(x, y) (((x) < (y)) ? (x) : (y))
#define THRESHOLD(x, th) (((x) > (th)) ? (255) : (0))
// Fatal error, print a message to standard-error output and exit.
static void error(char * msg) {
fprintf(stderr,"Error: %s\n",msg);
exit(EXIT_FAILURE);
}
// Memory allocation, print an error and exit if fail.
static void * xmalloc(size_t size) {
void * p;
if( size == 0 ) error("xmalloc: zero size");
p = malloc(size);
if( p == NULL ) error("xmalloc: out of memory");
return p;
}
// Computes a Gaussian kernel of size n x n and standard deviation sigma.
static float *gaussian_kernel(int n, float sigma) {
// Memory allocation for the kernel
float *kernel = xmalloc(n*n*sizeof(float));
// Gaussian kernel: e^{ -(x^2+y^2) / (2sigma^2) }
float sum = 0;
for(int i=0; i<n; i++) {
for(int j=0; j<n; j++) {
int x = i - n/2;
int y = j - n/2;
kernel[i+j*n] = exp( -(x*x + y*y) / (2.0*sigma*sigma) );
sum += kernel[i+j*n];
}
}
// Kernel normalization
for(int i=0; i<n*n; i++) {
kernel[i] = kernel[i] / sum;
}
return kernel;
}
// Computes a Laplacian of a Gaussian (LoG) kernel
// of size nxn and standard deviation sigma.
static float *LoG_kernel(int n, float sigma) {
// Memory allocation for the kernel
float *kernel = xmalloc(n*n*sizeof(float));
// Laplacian of a Gaussian kernel:
// (x^2 + y^2 - 2sigma^2) / (sigma^4) e^{ -(x^2+y^2) / (2sigma^2) }
float sum = 0;
for(int i=0; i<n; i++) {
for(int j=0; j<n; j++) {
int x = i - n/2;
int y = j - n/2;
kernel[i+j*n] = ( (x*x + y*y - 2.*sigma*sigma)
/ (sigma*sigma*sigma*sigma) )
* exp( -(x*x + y*y) / (2.0*sigma*sigma) );
sum += kernel[i+j*n];
}
}
// Normalization
for(int i=0; i<n*n; i++) {
kernel[i] /= sum;
}
return kernel;
}
// 2D convolution of an input image (size w x h) with a square (n x n) kernel.
//
// Two padding methods are available:
// zero-padding (padding_method=0)
// image boundary reflection (padding_method=1)
//
// returns a pointer to the resulting (w+n-1) x (h+n-1) image.
static float *conv2d(float *input, int w, int h,
float *kernel, int n, int padding_method) {
// Boundary handling:
// aux: extension of the input image to a larger size,
// enough to apply the kernel to each output pixel
int wx = w + 2*(n-1);
int hx = h + 2*(n-1);
float *aux = xmalloc(wx*hx*sizeof(float));
for(int i=0; i<wx; i++) {
for(int j=0; j<hx; j++) {
if( i >= n-1 && j >= n-1 && i < w+n-1 && j < h+n-1 ) {
aux[i+j*wx] = input[ i-n+1 + (j-n+1)*w ]; // image at center
} else {
if(padding_method == 0) {
aux[i+j*wx] = 0.0; // zero-padding
} else {
// reflection boundary padding
int reflex_x = i<n-1 ? 2*n-3-i : i<w+n-1 ? i : 2*w+2*n-3-i;
int reflex_y = j<n-1 ? 2*n-3-j : j<h+n-1 ? j : 2*h+2*n-3-j;
aux[i+j*wx] = input[ reflex_x-n+1 + (reflex_y-n+1)*w ];
}
}
}
}
// get memory for output
int wo = w+n-1;
int ho = h+n-1;
float *out = xmalloc(wo*ho*sizeof(float));
// compute convolution
for(int i=0; i<wo; i++) {
for(int j=0; j<ho; j++) {
float sum = 0.0;
for(int k=0; k<n; k++) {
for(int l=0; l<n; l++) {
sum += kernel[k+l*n] * aux[i+k + (j+l)*wx];
}
}
out[i+j*wo] = sum;
}
}
// free memory
free(aux);
return out;
}
// Roberts edge detector
// inputs:
// float *input - pointer to input image
// int w, int h - width and height of input image
// float threshold - threshold of edge detection
// int padding_method - padding method for convolution
// output:
// float * - pointer to output image
float *edges_roberts(float *im, int w, int h,
float threshold, int padding_method) {
// Define operators
// 3x3 operators are used (although it is unnecessary) in order to
// simplify the convolution application, and also to unify the three
// detectors of the first family.
float roberts_1[9] = {-1, 0, 0, 0, 1, 0, 0, 0, 0}; // ROBERTS
float roberts_2[9] = { 0,-1, 0, 1, 0, 0, 0, 0, 0}; // OPERATORS
for(int z=0; z<9; z++) { // normalization
roberts_1[z] /= 2.0;
roberts_2[z] /= 2.0;
}
// Convolution with operators
float *Gx = conv2d(im, w, h, roberts_1, 3, padding_method);
float *Gy = conv2d(im, w, h, roberts_2, 3, padding_method);
// Allocate memory for output image
float *im_roberts = xmalloc(w*h*sizeof(float));
// Two images are obtained (one for each operator). Then the gradient
// magnitude image is constructed using sqrt(g_x^2+g_y^2). Also
// the absolute maximum value of the constructed images is computed
float max = 0.0;
for(int i=0; i<w; i++) {
for(int j=0; j<h; j++) {
float gx = Gx[i+1 + (j+1)*(w+2)]; // Gx and Gy are (w+2) x (h+2)
float gy = Gy[i+1 + (j+1)*(w+2)]; // there is an (+1,+1) offset
im_roberts[i+j*w] = sqrt(gx*gx + gy*gy);
max = MAX(max, im_roberts[i+j*w]);
}
}
// Threshold
for(int i=0; i<w*h; i++) {
im_roberts[i] = THRESHOLD(im_roberts[i], threshold*max);
}
// Free memory
free(Gx);
free(Gy);
return im_roberts;
}
// Prewitt edge detector
// inputs:
// float *input - pointer to input image
// int w, int h - width and height of input image
// float threshold - threshold of edge detection
// int padding_method - padding method for convolution
// output:
// float * - pointer to output image
float *edges_prewitt(float *im, int w, int h,
float threshold, int padding_method) {
// Define operators
float prewitt_1[9] = {-1,-1,-1, 0, 0, 0, 1, 1, 1}; // PREWITT
float prewitt_2[9] = {-1, 0, 1,-1, 0, 1,-1, 0, 1}; // OPERATORS
for(int z=0; z<9; z++) { // normalization
prewitt_1[z] /= 6.0;
prewitt_2[z] /= 6.0;
}
// Convolution with operators
float *Gx = conv2d(im, w, h, prewitt_1, 3, padding_method);
float *Gy = conv2d(im, w, h, prewitt_2, 3, padding_method);
// get memory for output
float *im_prewitt = xmalloc(w*h*sizeof(float));
// Two images are obtained (one for each operator). Then the gradient
// magnitude image is constructed using sqrt(g_x^2+g_y^2). Also
// the absolute maximum value of the constructed images is computed
float max = 0.0;
for(int i=0; i<w; i++) {
for(int j=0; j<h; j++) {
float gx = Gx[i+1 + (j+1)*(w+2)]; // Gx and Gy are (w+2) x (h+2)
float gy = Gy[i+1 + (j+1)*(w+2)]; // there is an (+1,+1) offset
im_prewitt[i+j*w] = sqrt(gx*gx + gy*gy);
max = MAX(max, im_prewitt[i+j*w]);
}
}
// Threshold
for(int i=0; i<w*h; i++) {
im_prewitt[i] = THRESHOLD(im_prewitt[i], threshold*max);
}
// Free memory
free(Gx);
free(Gy);
return im_prewitt;
}
// Sobel edge detector
// inputs:
// float *input - pointer to input image
// int w, int h - width and height of input image
// float threshold - threshold of edge detection
// int padding_method - padding method for convolution
// output:
// float * - pointer to output image
float *edges_sobel(float *im, int w, int h,
float threshold, int padding_method) {
// Define operators
float sobel_1[9] = {-1,-2,-1, 0, 0, 0, 1, 2, 1}; // SOBEL
float sobel_2[9] = {-1, 0, 1,-2, 0, 2,-1, 0, 1}; // OPERATORS
for(int z=0; z<9; z++) { // normalization
sobel_1[z] /= 8.0;
sobel_2[z] /= 8.0;
}
// Convolution with operators
float *Gx = conv2d(im, w, h, sobel_1, 3, padding_method);
float *Gy = conv2d(im, w, h, sobel_2, 3, padding_method);
// Allocate memory for output image
float *im_sobel = xmalloc(w*h*sizeof(float));
// Two images are obtained (one for each operator). Then the gradient
// magnitude image is constructed using sqrt(g_x^2+g_y^2). Also
// the absolute maximum value of the constructed images is computed
float max = 0.0;
for(int i=0; i<w; i++) {
for(int j=0; j<h; j++) {
float gx = Gx[i+1 + (j+1)*(w+2)]; // Gx and Gy are (w+2) x (h+2)
float gy = Gy[i+1 + (j+1)*(w+2)]; // there is an (+1,+1) offset
im_sobel[i+j*w] = sqrt(gx*gx + gy*gy);
max = MAX(max, im_sobel[i+j*w]);
}
}
// Threshold
for(int i=0; i<w*h; i++) {
im_sobel[i] = THRESHOLD(im_sobel[i], threshold*max);
}
// Free memory
free(Gx);
free(Gy);
return im_sobel;
}
// Marr-Hildreth edge detector with Gaussian and Laplacian kernels
// inputs:
// float *input - pointer to input image
// int w, int h - width and height of input image
// float sigma - gaussian standard deviation
// int n - kernel size
// float tzc - threshold in zero-crossing
// int padding_method - padding method for convolution
// output:
// float * - pointer to output image
float *edges_mh(float *im, int w, int h,
float sigma, int n, float tzc, int padding_method) {
// generate Gaussian kernel
float *kernel = gaussian_kernel(n,sigma);
// smooth input image with the Gaussian kernel
float *im_smoothed = conv2d(im, w, h, kernel, n, padding_method);
// compute Laplacian of the smoothed image using a 3x3 operator
float operator[9] = {1, 1, 1, 1, -8, 1, 1, 1, 1};
float *laplacian = conv2d(im_smoothed, w+n-1, h+n-1,
operator, 3, padding_method);
// compute maximum of absolute Laplacian
float max = 0.0;
for(int i=0; i<w; i++) {
for(int j=0; j<h; j++) {
// laplacian is (w+n+1)x(h+n+1) with an offset of (n+1)/2 in x and y
float v = abs( laplacian[(i+(n+1)/2) + (j+(n+1)/2)*(w+n+1)] );
if( v > max ) max = v;
}
}
// compute laplacian zero-crossings
float *edges = xmalloc(w*h*sizeof(float));
for(int i=0; i<w; i++) {
for(int j=0; j<h; j++) {
// laplacian is (w+n+1)x(h+n+1) with an offset of (n+1)/2 in x and y
float UP_LE = laplacian[ (i-1+(n+1)/2) + (j+1+(n+1)/2)*(w+n+1) ];
float UP = laplacian[ (i +(n+1)/2) + (j+1+(n+1)/2)*(w+n+1) ];
float UP_RI = laplacian[ (i+1+(n+1)/2) + (j+1+(n+1)/2)*(w+n+1) ];
float LE = laplacian[ (i-1+(n+1)/2) + (j +(n+1)/2)*(w+n+1) ];
float RI = laplacian[ (i+1+(n+1)/2) + (j +(n+1)/2)*(w+n+1) ];
float DO_LE = laplacian[ (i-1+(n+1)/2) + (j-1+(n+1)/2)*(w+n+1) ];
float DO = laplacian[ (i +(n+1)/2) + (j-1+(n+1)/2)*(w+n+1) ];
float DO_RI = laplacian[ (i+1+(n+1)/2) + (j-1+(n+1)/2)*(w+n+1) ];
if( (LE*RI < 0.0 && abs(LE-RI) > tzc*max) ||
(UP_LE*DO_RI < 0.0 && abs(UP_LE-DO_RI) > tzc*max) ||
(DO_LE*UP_RI < 0.0 && abs(DO_LE-UP_RI) > tzc*max) ||
(UP*DO < 0.0 && abs(UP-DO) > tzc*max) ) {
edges[i+j*w] = 255.0;
} else {
edges[i+j*w] = 0.0;
}
}
}
// free memory
free(kernel);
free(im_smoothed);
free(laplacian);
return edges;
}
// Marr-Hildreth edge detector with Laplacian of Gaussian kernel
// inputs:
// float *input - pointer to input image
// int w, int h - width and height of input image
// float sigma - gaussian standard deviation
// int n - kernel size
// float tzc - threshold in zero-crossing
// int padding_method - padding method for convolution
// output:
// float * - pointer to output image
float *edges_mh_log(float *im, int w, int h,
float sigma, int n, float tzc, int padding_method) {
// compute Laplacian using a LoG (Laplacian of Gaussian) kernel
float *kernel = LoG_kernel(n,sigma);
float *laplacian = conv2d(im, w, h, kernel, n, padding_method);
// compute maximum of absolute Laplacian
float max = 0.0;
for(int i=0; i<w; i++) {
for(int j=0; j<h; j++) {
// laplacian is (w+n-1)x(h+n-1) with an offset of (n-1)/2 in x and y
float v = abs( laplacian[(i+(n-1)/2) + (j+(n-1)/2)*(w+n-1)] );
if( v > max ) max = v;
}
}
// compute laplacian zero-crossings
float *edges = xmalloc(w*h*sizeof(float));
for(int i=0; i<w; i++) {
for(int j=0; j<h; j++) {
// laplacian is (w+n-1)x(h+n-1) with an offset of (n-1)/2 in x and y
float UP_LE = laplacian[ (i-1+(n-1)/2) + (j+1+(n-1)/2)*(w+n-1) ];
float UP = laplacian[ (i +(n-1)/2) + (j+1+(n-1)/2)*(w+n-1) ];
float UP_RI = laplacian[ (i+1+(n-1)/2) + (j+1+(n-1)/2)*(w+n-1) ];
float LE = laplacian[ (i-1+(n-1)/2) + (j +(n-1)/2)*(w+n-1) ];
float RI = laplacian[ (i+1+(n-1)/2) + (j +(n-1)/2)*(w+n-1) ];
float DO_LE = laplacian[ (i-1+(n-1)/2) + (j-1+(n-1)/2)*(w+n-1) ];
float DO = laplacian[ (i +(n-1)/2) + (j-1+(n-1)/2)*(w+n-1) ];
float DO_RI = laplacian[ (i+1+(n-1)/2) + (j-1+(n-1)/2)*(w+n-1) ];
if( (LE*RI < 0.0 && abs(LE-RI) > tzc*max) ||
(UP_LE*DO_RI < 0.0 && abs(UP_LE-DO_RI) > tzc*max) ||
(DO_LE*UP_RI < 0.0 && abs(DO_LE-UP_RI) > tzc*max) ||
(UP*DO < 0.0 && abs(UP-DO) > tzc*max) ) {
edges[i+j*w] = 255.0;
} else {
edges[i+j*w] = 0.0;
}
}
}
// free memory
free(kernel);
free(laplacian);
return edges;
}
// Haralick edge detector
// inputs:
// float *input - pointer to input image
// int w, int h - width and height of input image
// float rhozero - threshold
// int padding_method - padding method for convolution
// output:
// float * - pointer to output image
float *edges_haralick(float *im, int w, int h,
float rhozero, int padding_method) {
// Haralick's masks for computing k1 to k10
// New masks calculated by 2-d fitting using LS, with the function
// f(x,y) = k1 + k2*x + k3*y + k4*x^2 + k5*xy
// + k6*y^2 + k7*x^3 + k8*x^2y + k9*xy^2 + k10*y^3
float mask[10][25] = { { 425, 275, 225, 275, 425,
275, 125, 75, 125, 275,
225, 75, 25, 75, 225,
275, 125, 75, 125, 275,
425, 275, 225, 275, 425},
{ -2260, -620, 0, 620, 2260,
-1660, -320, 0, 320, 1660,
-1460, -220, 0, 220, 1460,
-1660, -320, 0, 320, 1660,
-2260, -620, 0, 620, 2260},
{ 2260, 1660, 1460, 1660, 2260,
620, 320, 220, 320, 620,
0, 0, 0, 0, 0,
-620, -320, -220, -320, -620,
-2260, -1660,-1460,-1660,-2260},
{ 1130, 620, 450, 620, 1130,
830, 320, 150, 320, 830,
730, 220, 50, 220, 730,
830, 320, 150, 320, 830,
1130, 620, 450, 620, 1130},
{ -400, -200, 0, 200, 400,
-200, -100, 0, 100, 200,
0, 0, 0, 0, 0,
200, 100, 0, -100, -200,
400, 200, 0, -200, -400},
{ 1130, 830, 730, 830, 1130,
620, 320, 220, 320, 620,
450, 150, 50, 150, 450,
620, 320, 220, 320, 620,
1130, 830, 730, 830, 1130},
{ -8260, -2180, 0, 2180, 8260,
-6220, -1160, 0, 1160, 6220,
-5540, -820, 0, 820, 5540,
-6220, -1160, 0, 1160, 6220,
-8260, -2180, 0, 2180, 8260},
{ 5640, 3600, 2920, 3600, 5640,
1800, 780, 440, 780, 1800,
0, 0, 0, 0, 0,
-1800, -780, -440, -780,-1800,
-5640, -3600,-2920,-3600,-5640},
{ -5640, -1800, 0, 1800, 5640,
-3600, -780, 0, 780, 3600,
-2920, -440, 0, 440, 2920,
-3600, -780, 0, 780, 3600,
-5640, -1800, 0, 1800, 5640},
{ 8260, 6220, 5540, 6220, 8260,
2180, 1160, 820, 1160, 2180,
0, 0, 0, 0, 0,
-2180, -1160, -820,-1160,-2180,
-8260, -6220, 5540,-6220,-8260 } };
// apply the masks operators, this will lead to coefficients k1 to k10
float *aux[10];
for(int i=0; i<10; i++) {
aux[i] = conv2d(im, w, h, mask[i], 5, padding_method);
}
// compute Haralick edges
float *edges = xmalloc(w*h*sizeof(float)); // get memory for output
for(int i=0; i<w; i++) {
for(int j=0; j<h; j++) {
// aux is (w+4)x(h+4) and there is an offset of (2,2)
// k1 is not used
float k2 = aux[1][i+2 + (j+2)*(w+4)];
float k3 = aux[2][i+2 + (j+2)*(w+4)];
float k4 = aux[3][i+2 + (j+2)*(w+4)];
float k5 = aux[4][i+2 + (j+2)*(w+4)];
float k6 = aux[5][i+2 + (j+2)*(w+4)];
float k7 = aux[6][i+2 + (j+2)*(w+4)];
float k8 = aux[7][i+2 + (j+2)*(w+4)];
float k9 = aux[8][i+2 + (j+2)*(w+4)];
float k10 = aux[9][i+2 + (j+2)*(w+4)];
float sintheta = - k2 / sqrt(k2*k2 + k3*k3);
float costheta = - k3 / sqrt(k2*k2 + k3*k3);
float C2 = k4 * sintheta * sintheta
+ k5 * sintheta * costheta
+ k6 * costheta * costheta;
float C3 = k7 * sintheta * sintheta * sintheta
+ k8 * sintheta * sintheta * costheta
+ k9 * sintheta * costheta * costheta
+ k10 * costheta * costheta * costheta;
if( fabs( C2 / (3.0 * C3) ) <= rhozero && C3 < 0.0 ) {
edges[i+j*w] = 255.0;
} else {
edges[i+j*w] = 0.0;
}
}
}
// free memory
for(int i=0; i<10; i++) {
free(aux[i]);
}
return edges;
}
|