-
Notifications
You must be signed in to change notification settings - Fork 3
/
Geometrias_3D.py
616 lines (470 loc) · 25.9 KB
/
Geometrias_3D.py
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
#!/usr/bin/python
# -*- coding: iso-8859-15 -*-
# External
import random as rnd
import numpy as np
from skimage import morphology
from skimage import transform
from skimage import filters
from collections import OrderedDict
import multiprocessing
import math
from scipy import ndimage
from skimage import measure
from joblib import Parallel, delayed
import multiprocessing
#-----------------------------------------------------------------------------#
#--------------- Functions that create the diferent volume shapes ------------#
#-----------------------------------------------------------------------------#
def fun_cilindro(tam_base):
aux = morphology.disk(tam_base[0]/2);
tam_aux = aux.shape
Salida_Base = np.zeros((tam_aux[0], tam_aux[1], tam_base[2]))
# Slice-wise drawing
for i in range(0,tam_base[2]-1):
Salida_Base[:,:,i] = morphology.disk(tam_base[0]/2)
return Salida_Base
def fun_estrella_cil_vol(tam_base):
aux = morphology.star(tam_base[0]/3)
tam_aux = aux.shape
Salida_Base = np.zeros((tam_aux[0], tam_aux[1], tam_base[2]))
# Slice-wise drawing
for i in range(0,tam_base[2]-1):
Salida_Base[:,:,i] = morphology.star(tam_base[0]/3)
return Salida_Base
def fun_estrella_rot_vol(tam_base):
aux = morphology.star(tam_base/3)
tam_aux = aux.shape
Salida_Base = np.zeros((tam_aux[0], tam_aux[0], tam_aux[0]))
# I create a star in the center of the volume, wich must be a cube
Salida_Base[:,:,tam_base/2] = morphology.star(tam_base/3)
Salida_aux = Salida_Base;
# Now I rotate it over its X axis
for i in range(0,360,1):
Salida_Base = Salida_Base + fun_rotar_vol(Salida_aux, 0, i,0)
# Saturate output...
Salida_Base[np.where(Salida_Base > 1)] = 1
return Salida_Base
#-----------------------------------------------------------------------------#
#--------------- Volume transformation ---------------------------------------#
#-----------------------------------------------------------------------------#
def fun_rotar_vol(Vol_entrada, eje, angulo, expandir):
# Rotating volume shape
tam_base = Vol_entrada.shape
# Output volume generation
if expandir == 1:
# If the volume is to be expanded due to its rotation, I check the new size
if eje == 0:
aux_rot = transform.rotate(Vol_entrada[1,:,:],angulo,1)
tam_aux = aux_rot.shape
Salida_Base = np.zeros((tam_base[0], tam_aux[0], tam_aux[1]))
if eje == 1:
aux_rot = transform.rotate(Vol_entrada[:,1,:],angulo,1)
tam_aux = aux_rot.shape
Salida_Base = np.zeros((tam_aux[0], tam_base[1], tam_aux[1]))
if eje == 2:
aux_rot = transform.rotate(Vol_entrada[:,:,1],angulo,1)
tam_aux = aux_rot.shape
Salida_Base = np.zeros((tam_aux[0], tam_aux[1], tam_base[2]))
else:
Salida_Base = np.zeros(tam_base)
# Slice-wise rotation
for i in range(0,tam_base[eje]-1):
if eje == 0:
Salida_Base[i,:,:] = transform.rotate(Vol_entrada[i,:,:],angulo,expandir, preserve_range=1)
if eje == 1:
Salida_Base[:,i,:] = transform.rotate(Vol_entrada[:,i,:],angulo,expandir, preserve_range=1)
if eje == 2:
Salida_Base[:,:,i] = transform.rotate(Vol_entrada[:,:,i],angulo,expandir, preserve_range=1)
return Salida_Base
#-----------------------------------------------------------------------------#
#--------------- Random Volume generation ------------------------------------#
#-----------------------------------------------------------------------------#
def fun_generar_vol_random(tipo_forma, tam_limite, transp_forma ):
# Base volume generation
if (tipo_forma == 1):
# Random size
tam_aux = rnd.randint(tam_limite[0],tam_limite[1])
# Shape creation
vol_forma = morphology.cube(tam_aux)
elif(tipo_forma == 2):
# Random size
tam_aux = rnd.randint(tam_limite[0],tam_limite[1])
# Shape creation
vol_forma = morphology.ball(tam_aux)
elif(tipo_forma == 3):
# Random size
tam_forma = (rnd.randint(tam_limite[0],tam_limite[1]) ,
rnd.randint(tam_limite[0],tam_limite[1]) ,
rnd.randint(tam_limite[0]*2,tam_limite[1]*2))
# Shape creation
vol_forma = fun_cilindro(tam_forma)
elif(tipo_forma == 4):
# Random size
tam_aux = rnd.randint(tam_limite[0],tam_limite[1])
# Shape creation
vol_forma = morphology.octahedron(tam_aux)
elif(tipo_forma == 5):
# Random size
tam_forma = (rnd.randint(tam_limite[0],tam_limite[1]) ,
rnd.randint(tam_limite[0],tam_limite[1]) ,
rnd.randint(tam_limite[0]*2,tam_limite[1]*2))
# Shape creation
vol_forma = fun_estrella_cil_vol(tam_forma)
elif(tipo_forma == 6):
# Random size
tam_aux = rnd.randint(tam_limite[0],tam_limite[1])
# Shape creation
vol_forma = fun_estrella_rot_vol(tam_aux)
# Shape label population
label_forma = np.multiply(vol_forma,tipo_forma)
# Transparency
vol_forma = np.multiply(vol_forma,transp_forma)
tam_forma = vol_forma.shape
return (tam_forma, vol_forma, label_forma)
def fun_poblar_vol_random(vol_poblar, cantidad_elementos, margenes, tam_limite, transparencia_limite, train_single_class):
if (train_single_class != 0):
Numero_de_canales_verdad = 3
else:
Numero_de_canales_verdad = 6+1
tam_base = vol_poblar.shape
verdad_base = np.zeros((tam_base[0], tam_base[1], tam_base[2], Numero_de_canales_verdad))
for i in range(0,cantidad_elementos):
# Random shape, according to following key:
# -- Cube
# -- Ball
# -- Cilinder
# -- Octahedro
# -- Extruded Star
# -- Revolution Star
Numero_de_formas = 6
tipo_forma = rnd.randint(1,Numero_de_formas)
# If this function is called from a network training only one shape type, then
# I assure that at least that shape appears one time.
if train_single_class != 0 and i == 0:
tipo_forma = train_single_class
# Random transparency
transp_forma = rnd.uniform(transparencia_limite[0], transparencia_limite[1])
# Shape generation
(tam_forma, vol_forma, label_forma) = fun_generar_vol_random(tipo_forma, tam_limite, transp_forma )
# Random Rotation
eje_rot = rnd.randint(0,2)
grados_rot = rnd.uniform(-90,90)
vol_forma_aux = fun_rotar_vol(vol_forma, eje_rot, grados_rot, 1)
label_forma_aux = fun_rotar_vol(label_forma, eje_rot, grados_rot, 1)
tam_forma_aux = vol_forma_aux.shape
# Chek bounds after rotations, and abort rotation in case it violates any bound
if ( (tam_base[0] - margenes[1] - (tam_forma_aux[0]/2)) - (margenes[0]+(tam_forma_aux[0]/2)) ) >= 1 and ( (tam_base[1] - margenes[3] - (tam_forma_aux[1]/2)) - (margenes[2]+(tam_forma_aux[1]/2)) ) >= 1 and ( (tam_base[2] - margenes[5] - (tam_forma_aux[2]/2)) - (margenes[4]+(tam_forma_aux[2]/2)) ) >= 1:
vol_forma = vol_forma_aux
label_forma = label_forma_aux
tam_forma = tam_forma_aux
# After creating the volume, I place it at a random position inside the base volume
# Random position
centro_forma = (rnd.randint(margenes[0]+(tam_forma[0]/2),tam_base[0] - margenes[1] - (tam_forma[0]/2)) ,
rnd.randint(margenes[2]+(tam_forma[1]/2),tam_base[1] - margenes[3] - (tam_forma[1]/2)) ,
rnd.randint(margenes[4]+(tam_forma[2]/2),tam_base[2] - margenes[5] - (tam_forma[2]/2)) )
# Corner of the shape in base volume coordinates
centro_forma = (centro_forma[0]-(tam_forma[0]/2) ,
centro_forma[1]-(tam_forma[1]/2) ,
centro_forma[2]-(tam_forma[2]/2) )
# Addition to base volume
vol_poblar[centro_forma[0]:centro_forma[0]+tam_forma[0],
centro_forma[1]:centro_forma[1]+tam_forma[1],
centro_forma[2]:centro_forma[2]+tam_forma[2]] = vol_forma + vol_poblar[centro_forma[0]:centro_forma[0]+tam_forma[0], centro_forma[1]:centro_forma[1]+tam_forma[1], centro_forma[2]:centro_forma[2]+tam_forma[2]]
# Background label fill
verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0],
centro_forma[1]:centro_forma[1]+tam_forma[1],
centro_forma[2]:centro_forma[2]+tam_forma[2],
0] = label_forma + verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0], centro_forma[1]:centro_forma[1]+tam_forma[1], centro_forma[2]:centro_forma[2]+tam_forma[2], 0]
# Fill class label according to type (single or multiclass)
if (train_single_class == 0):
verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0],
centro_forma[1]:centro_forma[1]+tam_forma[1],
centro_forma[2]:centro_forma[2]+tam_forma[2],
tipo_forma] = label_forma + verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0], centro_forma[1]:centro_forma[1]+tam_forma[1], centro_forma[2]:centro_forma[2]+tam_forma[2], tipo_forma]
# If it is "single class" I fill the "others" class with all non target shapes
elif (train_single_class == label_forma.max()):
verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0],
centro_forma[1]:centro_forma[1]+tam_forma[1],
centro_forma[2]:centro_forma[2]+tam_forma[2],
1] = label_forma + verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0], centro_forma[1]:centro_forma[1]+tam_forma[1], centro_forma[2]:centro_forma[2]+tam_forma[2], 0]
elif (train_single_class != label_forma.max()):
verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0],
centro_forma[1]:centro_forma[1]+tam_forma[1],
centro_forma[2]:centro_forma[2]+tam_forma[2],
2] = label_forma + verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0], centro_forma[1]:centro_forma[1]+tam_forma[1], centro_forma[2]:centro_forma[2]+tam_forma[2], 0]
# Output clipping (safety)
verdad_base = np.clip(verdad_base, 0, 1)
# I invert the background channel to generate a network fiendly class.
verdad_base[:,:,:,0] = 1 - verdad_base[:,:,:,0]
# If it is single class the overlappling zones are deleted from "other" class
if (train_single_class != 0):
dims = verdad_base.shape
for i in range(0,dims[0]):
for j in range(0,dims[1]):
for k in range(0,dims[2]):
if verdad_base[i,j,k,2] == verdad_base[i,j,k,1]:
verdad_base[i,j,k,2] = 0
return (vol_poblar, verdad_base)
def fun_poblar_vol_each(vol_poblar, margenes, tam_limite, transparencia_limite, noise_max, gauss_sigma, repetitions, train_single_class):
# This function generates a volume with one element of each class, used to show performance
if (train_single_class != 0):
Numero_de_canales_verdad = 3
else:
Numero_de_canales_verdad = 6+1
tam_base = vol_poblar.shape
verdad_base = np.zeros((tam_base[0], tam_base[1], tam_base[2], Numero_de_canales_verdad))
# Shapes to be used:
# -- Cube
# -- Ball
# -- Cilinder
# -- Octahedro
# -- Extruded Star
# -- Revolution Star
Numero_de_formas = 6
for rep in range(0,repetitions):
for i in range(0,Numero_de_formas):
tipo_forma = i+1
# Random transparency
transp_forma = rnd.uniform(transparencia_limite[0], transparencia_limite[1])
# Shape generation
(tam_forma, vol_forma, label_forma) = fun_generar_vol_random(tipo_forma, tam_limite, transp_forma )
# Random Rotation
eje_rot = rnd.randint(0,2)
grados_rot = rnd.uniform(-90,90)
vol_forma_aux = fun_rotar_vol(vol_forma, eje_rot, grados_rot, 1)
label_forma_aux = fun_rotar_vol(label_forma, eje_rot, grados_rot, 1)
tam_forma_aux = vol_forma_aux.shape
# Chek bounds after rotations, and abort rotation in case it violates any bound
if ( (tam_base[0] - margenes[1] - (tam_forma_aux[0]/2)) - (margenes[0]+(tam_forma_aux[0]/2)) ) >= 1 and ( (tam_base[1] - margenes[3] - (tam_forma_aux[1]/2)) - (margenes[2]+(tam_forma_aux[1]/2)) ) >= 1 and ( (tam_base[2] - margenes[5] - (tam_forma_aux[2]/2)) - (margenes[4]+(tam_forma_aux[2]/2)) ) >= 1:
vol_forma = vol_forma_aux
label_forma = label_forma_aux
tam_forma = tam_forma_aux
# After creating the volume, I place it at a random position inside the base volume
# Random position
centro_forma = (rnd.randint(margenes[0]+(tam_forma[0]/2),tam_base[0] - margenes[1] - (tam_forma[0]/2)) ,
rnd.randint(margenes[2]+(tam_forma[1]/2),tam_base[1] - margenes[3] - (tam_forma[1]/2)) ,
rnd.randint(margenes[4]+(tam_forma[2]/2),tam_base[2] - margenes[5] - (tam_forma[2]/2)) )
# Corner of the shape in base volume coordinates
centro_forma = (centro_forma[0]-(tam_forma[0]/2) ,
centro_forma[1]-(tam_forma[1]/2) ,
centro_forma[2]-(tam_forma[2]/2) )
# Addition to base volume
vol_poblar[centro_forma[0]:centro_forma[0]+tam_forma[0],
centro_forma[1]:centro_forma[1]+tam_forma[1],
centro_forma[2]:centro_forma[2]+tam_forma[2]] = vol_forma + vol_poblar[centro_forma[0]:centro_forma[0]+tam_forma[0], centro_forma[1]:centro_forma[1]+tam_forma[1], centro_forma[2]:centro_forma[2]+tam_forma[2]]
# Background label fill
verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0],
centro_forma[1]:centro_forma[1]+tam_forma[1],
centro_forma[2]:centro_forma[2]+tam_forma[2],
0] = label_forma + verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0], centro_forma[1]:centro_forma[1]+tam_forma[1], centro_forma[2]:centro_forma[2]+tam_forma[2], 0]
# Fill class label according to type (single or multiclass)
if (train_single_class == 0):
verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0],
centro_forma[1]:centro_forma[1]+tam_forma[1],
centro_forma[2]:centro_forma[2]+tam_forma[2],
tipo_forma] = label_forma + verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0], centro_forma[1]:centro_forma[1]+tam_forma[1], centro_forma[2]:centro_forma[2]+tam_forma[2], tipo_forma]
# If it is "single class" I fill the "others" class with all non target shapes
elif (train_single_class == label_forma.max()):
verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0],
centro_forma[1]:centro_forma[1]+tam_forma[1],
centro_forma[2]:centro_forma[2]+tam_forma[2],
1] = label_forma + verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0], centro_forma[1]:centro_forma[1]+tam_forma[1], centro_forma[2]:centro_forma[2]+tam_forma[2], 0]
elif (train_single_class != label_forma.max()):
verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0],
centro_forma[1]:centro_forma[1]+tam_forma[1],
centro_forma[2]:centro_forma[2]+tam_forma[2],
2] = label_forma + verdad_base[centro_forma[0]:centro_forma[0]+tam_forma[0], centro_forma[1]:centro_forma[1]+tam_forma[1], centro_forma[2]:centro_forma[2]+tam_forma[2], 0]
# Output clipping (safety)
verdad_base = np.clip(verdad_base, 0, 1)
# I invert the background channel to generate a network fiendly class.
verdad_base[:,:,:,0] = 1 - verdad_base[:,:,:,0]
# If it is single class the overlappling zones are deleted from "other" class
if (train_single_class != 0):
dims = verdad_base.shape
for i in range(0,dims[0]):
for j in range(0,dims[1]):
for k in range(0,dims[2]):
if verdad_base[i,j,k,2] == verdad_base[i,j,k,1]:
verdad_base[i,j,k,2] = 0
# Noise addition and filtering
vol_poblar = filters.gaussian(vol_poblar, sigma = gauss_sigma, preserve_range = True)
vol_noise = np.random.rand(tam_base[0], tam_base[1], tam_base[2])
vol_noise = np.multiply(vol_noise, noise_max)
vol_poblar += vol_noise
return (vol_poblar, verdad_base)
#-----------------------------------------------------------------------------#
#--------------- Multiple Random Volume generation ---------------------------#
#-----------------------------------------------------------------------------#
def fun_gen_batch(cant_cubos, tam_vol_poblar, noise_max, cantidad_elementos, margenes, tam_limite, transparencia_limite, gauss_sigma, train_single_class):
if (train_single_class != 0):
# Backgound - Target class - Others Class
Numero_de_canales_verdad = 3
else:
# One clase per shape + Background
Numero_de_canales_verdad = 6+1
# Output allocation
batch_vols = np.zeros((cant_cubos, tam_vol_poblar[0], tam_vol_poblar[1], tam_vol_poblar[2]))
batch_labels = np.zeros((cant_cubos, tam_vol_poblar[0], tam_vol_poblar[1], tam_vol_poblar[2], Numero_de_canales_verdad))
# Generation loop
for i in range(0,cant_cubos):
# Volume to populate
vol_poblar = np.zeros((tam_vol_poblar[0], tam_vol_poblar[1], tam_vol_poblar[2]))
# Volume population
(batch_vols[i,:,:,:], batch_labels[i,:,:,:,:]) = fun_poblar_vol_random(vol_poblar, cantidad_elementos, margenes, tam_limite, transparencia_limite, train_single_class)
# Noise addition and filtering
batch_vols[i,:,:,:] = filters.gaussian(batch_vols[i,:,:,:], sigma = gauss_sigma, preserve_range = True)
vol_noise = np.random.rand(tam_vol_poblar[0], tam_vol_poblar[1], tam_vol_poblar[2])
vol_noise = np.multiply(vol_noise, noise_max)
batch_vols[i,:,:,:] += vol_noise
return (batch_vols, batch_labels)
#-----------------------------------------------------------------------------#
#--------------- Random Volume Transformation --------------------------------#
#-----------------------------------------------------------------------------#
# http://nbviewer.jupyter.org/gist/lhk/f05ee20b5a826e4c8b9bb3e528348688
# http://www.lfd.uci.edu/~gohlke/code/transformations.py.html
def unit_vector(data, axis=None, out=None):
if out is None:
data = np.array(data, dtype=np.float64, copy=True)
if data.ndim == 1:
data /= math.sqrt(np.dot(data, data))
return data
else:
if out is not data:
out[:] = np.array(data, copy=False)
data = out
length = np.atleast_1d(np.sum(data*data, axis))
np.sqrt(length, length)
if axis is not None:
length = np.expand_dims(length, axis)
data /= length
if out is None:
return data
def rotation_matrix(angle, direction, point=None):
sina = math.sin(angle)
cosa = math.cos(angle)
direction = unit_vector(direction[:3])
# rotation matrix around unit vector
R = np.diag([cosa, cosa, cosa])
R += np.outer(direction, direction) * (1.0 - cosa)
direction *= sina
R += np.array([[ 0.0, -direction[2], direction[1]],
[ direction[2], 0.0, -direction[0]],
[-direction[1], direction[0], 0.0]])
M = np.identity(4)
M[:3, :3] = R
if point is not None:
# rotation not around origin
point = np.array(point[:3], dtype=np.float64, copy=False)
M[:3, 3] = point - np.dot(R, point)
return M
def rotate_mesh(input_mesh, size_fov, angle, direction):
# stack the meshgrid to position vectors, center them around 0 by substracting dim/2
xyz=np.vstack([input_mesh[0].reshape(-1)-float(size_fov[0])/2, # x coordinate, centered
input_mesh[1].reshape(-1)-float(size_fov[1])/2, # y coordinate, centered
input_mesh[2].reshape(-1)-float(size_fov[2])/2, # z coordinate, centered
np.ones((size_fov[0],size_fov[1],size_fov[2])).reshape(-1)]) # 1 for homogeneous coordinates
# create transformation matrix
mat=rotation_matrix(angle,direction)
# apply transformation
transformed_xyz=np.dot(mat, xyz)
# extract coordinates, don't use transformed_xyz[3,:] that's the homogeneous coordinate, always 1
x=transformed_xyz[0,:]+float(size_fov[0])/2
y=transformed_xyz[1,:]+float(size_fov[1])/2
z=transformed_xyz[2,:]+float(size_fov[2])/2
x=x.reshape((size_fov[0],size_fov[1],size_fov[2]))
y=y.reshape((size_fov[0],size_fov[1],size_fov[2]))
z=z.reshape((size_fov[0],size_fov[1],size_fov[2]))
return [x,y,z]
def ripple_mesh(input_mesh, size_fov, max_ripple, wave_peak, wave_periode):
# Create a volumetric rippled fov
# Aux Volume
rip_vol = np.zeros((size_fov[0], size_fov[1], size_fov[2]))
# Create an uniform grid
ax = OrderedDict()
ax[0] = np.arange(size_fov[0]).astype(np.float32)
ax[1] = np.arange(size_fov[1]).astype(np.float32)
ax[2] = np.arange(size_fov[2]).astype(np.float32)
# First implement the sine across all axis
aux_sins = OrderedDict()
for i in range(0,3):
aux_sins[i] = np.multiply(wave_peak[i],np.sin(ax[i]/wave_periode[i]))
# X-axis
x_xpand = np.expand_dims(np.expand_dims(aux_sins[0], axis=(1)), axis=(1))
x_rip_vol = np.tile(x_xpand, (1,240,155))
# Y-axis
y_xpand = np.expand_dims(np.expand_dims(aux_sins[1], axis=(1)), axis=(0))
y_rip_vol = np.tile(y_xpand, (240,1,155))
# X-axis
z_xpand = np.expand_dims(np.expand_dims(aux_sins[2], axis=(0)), axis=(0))
z_rip_vol = np.tile(z_xpand, (240,240,1))
# Compose
rip_vol = np.multiply(np.multiply(x_xpand,y_xpand),z_xpand)
# Re-scale
rip_vol = np.divide(rip_vol,rip_vol.max())
rip_vol = np.multiply(rip_vol,max_ripple)
# apply
input_mesh[0] += rip_vol
input_mesh[1] += rip_vol
input_mesh[2] += rip_vol
x=input_mesh[0].reshape((size_fov[0],size_fov[1],size_fov[2]))
y=input_mesh[1].reshape((size_fov[0],size_fov[1],size_fov[2]))
z=input_mesh[2].reshape((size_fov[0],size_fov[1],size_fov[2]))
return [x,y,z]
def random_ripple_and_rotation(input_vol, max_ripple_range, wave_periode_range, spline_order, num_cores=multiprocessing.cpu_count(), multi_channel=False):
# Random max ripple
ripple_val = rnd.uniform(max_ripple_range[0], max_ripple_range[1])
# Random wave peaks
wave_peak = np.zeros((3))
wave_peak[0] = rnd.uniform(0,1)
wave_peak[1] = rnd.uniform(0,1)
wave_peak[2] = rnd.uniform(0,1)
# Random wave periods
wave_period = np.zeros((3))
wave_period[0] = rnd.uniform(max_ripple_range[0], max_ripple_range[1])
wave_period[1] = rnd.uniform(max_ripple_range[0], max_ripple_range[1])
wave_period[2] = rnd.uniform(max_ripple_range[0], max_ripple_range[1])
# Random rotate angle
rot_angle = rnd.uniform(-np.pi/4,np.pi/4)
# Random rotate direction (the unity norm is handled by "unit_vector" )
rot_dir = np.zeros((3))
rot_dir[0] = rnd.uniform(max_ripple_range[0], max_ripple_range[1])
rot_dir[1] = rnd.uniform(max_ripple_range[0], max_ripple_range[1])
rot_dir[2] = rnd.uniform(max_ripple_range[0], max_ripple_range[1])
# Get input shape
size_fov = input_vol.shape
if multi_channel:
channels = size_fov[0]
size_fov = size_fov[1:4]
# Create an uniform grid
ax_x = np.arange(size_fov[0]).astype(np.float32)
ax_y = np.arange(size_fov[1]).astype(np.float32)
ax_z = np.arange(size_fov[2]).astype(np.float32)
aux_mesh = np.meshgrid(ax_x,ax_y,ax_z)
# Rotate the mesh
aux_mesh = rotate_mesh(aux_mesh, size_fov, rot_angle, rot_dir)
# Ripple the mesh
aux_mesh = ripple_mesh(aux_mesh, size_fov, ripple_val, wave_peak, wave_period)
# the coordinate system seems to be strange, it has to be ordered like this
aux_mesh=[aux_mesh[1],aux_mesh[0],aux_mesh[2]]
# Re-sample
if multi_channel:
# To make this faster we paralellize the operation
new_vol = np.zeros((channels,size_fov[0],size_fov[1],size_fov[2]))
par_results = Parallel(n_jobs=num_cores)(delayed(fun_multiple_map_coordinates)
(size_fov=size_fov,
input_vol=input_vol,
aux_mesh = aux_mesh,
spline_order = spline_order,
num_vol = i) for i in range(0, channels))
# Collect results
for idx_ch in range(channels):
new_vol[idx_ch,:,:,:] = par_results[idx_ch]
else:
new_vol=scipy.ndimage.map_coordinates(input_vol,aux_mesh, order=spline_order)
return new_vol
def fun_multiple_map_coordinates(size_fov, input_vol, aux_mesh, spline_order, num_vol):
# https://bic-berkeley.github.io/psych-214-fall-2016/map_coordinates.html
new_vol = np.zeros((size_fov[0],size_fov[1],size_fov[2]))
new_vol= ndimage.map_coordinates(input_vol[num_vol,:,:,:],aux_mesh, order=spline_order)
return new_vol