-
Notifications
You must be signed in to change notification settings - Fork 5
/
dataaugmentationlib.py
311 lines (222 loc) · 10.8 KB
/
dataaugmentationlib.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
import math
import numpy as np
THETA_1 = math.pi/2
THETA_2 = math.pi
THETA_3 = (3*math.pi)/2
SIGMA_1 = 0.0005
SIGMA_2 = 0.001
SIGMA_3 = 0.002
def rotate(signals, labels):
"""
This function creates a list containing three rotated copies of each element in signals. Rotations are done by
90°, 180° and 270°.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes each one representing a rotated signal.
"""
# Rotated signal matrix B is obtained by multiplication of T transformation matrix with A original signal matrix.
#
# B = TxA
#
# T is defined as shown below.
#
# | cos(theta) -sin(theta) |
# | sin(theta) cos(theta) |
def T(theta):
return np.array([
[np.cos(theta), -np.sin(theta)],
[np.sin(theta), np.cos(theta)]
])
T_1 = T(THETA_1)
T_2 = T(THETA_2)
T_3 = T(THETA_3)
# list containing rotated signals
rotated_signals = []
rotated_labels = []
# for each signal B = TxA
for i in range(0, len(signals)):
A = signals[i]
l = labels[i]
# rotate by THETA_1
B = np.matmul(T_1, A)
rotated_signals.append(B)
rotated_labels.append(l)
# rotate by THETA_2
B = np.matmul(T_2, A)
rotated_signals.append(B)
rotated_labels.append(l)
# rotate by THETA_3
B = np.matmul(T_3, A)
rotated_signals.append(B)
rotated_labels.append(l)
return rotated_signals, rotated_labels
def rotate_and_concatenate_with_signals(signals, labels):
"""
This function creates a list containing three rotated copies of each element in signals, concatenated with the
given signals list. Rotations are done by 90°, 180° and 270°.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes each one representing a rotated signal.
"""
rotated_signals, rotated_labels = rotate(signals, labels)
return np.concatenate((signals, rotated_signals)), np.concatenate((labels, rotated_labels))
def flip(signals, labels, direction):
"""
This function creates a list containing a single flipped (horizontally or vertically) copy of each element in
signals.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
direction: 'horizontal' | 'vertical'
Returns:
numpy.array of 2x128 matrixes each one representing a flipped signal.
"""
flipped = []
for signal in signals:
I = signal[0]
Q = signal[1]
if direction == "horizontal":
flipped.append([-I, Q])
if direction == "vertical":
flipped.append([I, -Q])
flipped = np.array(flipped)
return flipped, labels
def horizontal_flip(signals, labels):
"""
This function creates a list containing a single horizontally flipped copy of each element in signals.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes each one representing an horizontally flipped signal.
"""
return flip(signals, labels, "horizontal")
def horizontal_flip_and_concatenate_with_signals(signals, labels):
"""
This function creates a list containing a single horizontally flipped copy of each element in signals,
concatenated with the given signals list.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes representing horizontally flipped signals concatenated with given signals.
"""
hflipped_signals, hflipped_labels = horizontal_flip(signals, labels)
return np.concatenate((signals, hflipped_signals)), np.concatenate((labels, hflipped_labels))
def vertical_flip(signals, labels):
"""
This function creates a list containing a single vertically flipped copy of each element in signals.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes each one representing an vertically flipped signal.
"""
return flip(signals, labels, "vertical")
def vertical_flip_and_concatenate_with_signals(signals, labels):
"""
This function creates a list containing a single vertically flipped copy of each element in signals,
concatenated with the given signals list.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes representing vertically flipped signals concatenated with given signals.
"""
vflipped_signals, vflipped_labels = vertical_flip(signals, labels)
return np.concatenate((signals, vflipped_signals)), np.concatenate((labels, vflipped_labels))
def horizontal_and_vertical_flip_and_concatenate_with_signals(signals, labels):
"""
This function creates a list containing a vertically and an horizontal flipped copy of each element in signals,
concatenated with the given signals list.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes representing flipped signals concatenated with given signals.
"""
hflipped_signals, hflipped_new_labels = horizontal_flip(signals, labels)
vflipped_signals, vflipped_new_labels = vertical_flip(signals, labels)
return np.concatenate((signals, hflipped_signals, vflipped_signals)),\
np.concatenate((labels, hflipped_new_labels, vflipped_new_labels))
def add_gaussian_noise(signals, labels):
"""
This function creates a list containing three noised copies of each element in signals. Noise added is Gaussian,
and the standard deviation used for the three copies is 0.0005, 0.001 and 0.002.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes each one representing a noised signal.
"""
disturbed_with_noise_signals = []
disturbed_labels = []
for i in range(0, len(signals)):
signal = signals[i]
label = labels[i]
# noise with SIGMA_1
noise = np.random.normal(0, SIGMA_1, (signal.shape[1:]))
disturbed_with_noise_signals.append(signal + noise)
disturbed_labels.append(label)
# noise with SIGMA_2
noise = np.random.normal(0, SIGMA_2, (signal.shape[1:]))
disturbed_with_noise_signals.append(signal + noise)
disturbed_labels.append(label)
# noise with SIGMA_3
noise = np.random.normal(0, SIGMA_3, (signal.shape[1:]))
disturbed_with_noise_signals.append(signal + noise)
disturbed_labels.append(label)
disturbed_with_noise_signals = np.array(disturbed_with_noise_signals)
return disturbed_with_noise_signals, disturbed_labels
def add_gaussian_noise_and_concatenate_with_signals(signals, labels):
"""
This function creates a list containing three noised copies of each element in signals. Noise added is Gaussian,
and the standard deviation used for the three copies is 0.0005, 0.001 and 0.002.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes representing noised signals concatenated with given signals.
"""
gnoised_signals, gnoised_labels = add_gaussian_noise(signals, labels)
return np.concatenate((signals, gnoised_signals)), np.concatenate((labels, gnoised_labels))
def rotate_flip_add_gaussian_noise(signals, labels):
"""
This function creates a list containing three rotated, two flipped (horizontally and vertically) and three
noised copies of each element in signals. Rotations are done by 90°, 180° and 270°, noise added is Gaussian and
the standard deviation used for the three copies is 0.0005, 0.001 and 0.002.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes representing rotated, flipped (horizontally and vertically) and noised signals.
"""
rotated_signals, rotated_new_labels = rotate(signals, labels)
hflipped_signals, hflipped_new_labels = horizontal_flip(signals, labels)
vflipped_signals, vflipped_new_labels = vertical_flip(signals, labels)
gnoised_signals, gnoised_new_labels = add_gaussian_noise(signals, labels)
signals_result, labels_result = np.concatenate((rotated_signals, hflipped_signals)), np.concatenate(
(rotated_new_labels, hflipped_new_labels))
signals_result, labels_result = np.concatenate((signals_result, vflipped_signals)), np.concatenate(
(labels_result, vflipped_new_labels))
signals_result, labels_result = np.concatenate((signals_result, gnoised_signals)), np.concatenate(
(labels_result, gnoised_new_labels))
return signals_result, labels_result
def rotate_flip_add_gaussian_noise_and_concatenate_with_signals(signals, labels):
"""
This function creates a list containing three rotated, two flipped (horizontally and vertically) and three
noised copies of each element in signals. Rotations are done by 90°, 180° and 270°, noise added is Gaussian and
the standard deviation used for the three copies is 0.0005, 0.001 and 0.002.
Args:
signals: numpy.array of 2x128 matrixes each one representing a signal.
labels: labels for each signal.
Returns:
numpy.array of 2x128 matrixes representing rotated, flipped (horizontally and vertically) and noised signals,
concatenated with given signals.
"""
signals_result, labels_result = rotate_flip_add_gaussian_noise(signals, labels)
return np.concatenate((signals, signals_result)), np.concatenate((labels, labels_result))