-
Notifications
You must be signed in to change notification settings - Fork 2
/
utils.py
129 lines (108 loc) · 4.11 KB
/
utils.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
import numpy as np
import cv2
import pywt
from cfg import par
def patch_generator_color(im, par):
patch_t = []
for s in par.scales:
im_s = cv2.resize(im, (int(im.shape[0]*s), int(im.shape[1]*s)), interpolation=cv2.INTER_CUBIC)
for i in range(0, im_s.shape[0]-par.patch_size+1, par.stride):
for j in range(0, im_s.shape[1]-par.patch_size+1, par.stride):
x = np.divide(im_s[i:i+par.patch_size, j:j+par.patch_size, :], 255).transpose(2, 0, 1) # C W H
for k in range(par.aug_time):
x_aug = data_aug_color(x, mode=np.random.randint(0, 8))
patch_t.append(x_aug)
return patch_t
def patch_generator_gray(im, par):
patch_t = []
for s in par.scales:
im_s = cv2.resize(im, (int(im.shape[0]*s), int(im.shape[1]*s)), interpolation=cv2.INTER_CUBIC)
for i in range(0, im_s.shape[0]-par.patch_size+1, par.stride):
for j in range(0, im_s.shape[1] - par.patch_size + 1, par.stride):
x = np.divide(im_s[i:i + par.patch_size, j:j + par.patch_size], 255) # W H
for k in range(par.aug_time):
x_aug = data_aug_gray(x, mode=np.random.randint(0, 8))
patch_t.append(x_aug)
return patch_t
def data_aug_color(img, mode=0):
if mode == 0:
return img
elif mode == 1:
return np.flipud(img)
elif mode == 2:
return np.rot90(img, axes=(1, 2))
elif mode == 3:
return np.flipud(np.rot90(img, axes=(1, 2)))
elif mode == 4:
return np.rot90(img, k=2, axes=(1, 2))
elif mode == 5:
return np.flipud(np.rot90(img, k=2, axes=(1, 2)))
elif mode == 6:
return np.rot90(img, k=3, axes=(1, 2))
elif mode == 7:
return np.flipud(np.rot90(img, k=3, axes=(1, 2)))
def data_aug_gray(img, mode=0):
if mode == 0:
return img
elif mode == 1:
return np.flipud(img)
elif mode == 2:
return np.rot90(img)
elif mode == 3:
return np.flipud(np.rot90(img))
elif mode == 4:
return np.rot90(img, k=2)
elif mode == 5:
return np.flipud(np.rot90(img, k=2))
elif mode == 6:
return np.rot90(img, k=3)
elif mode == 7:
return np.flipud(np.rot90(img, k=3))
def color_tensor_generator(noisy, target):
tensor_n = []
tensor_t = []
for i in range(par.img_channel):
coeffs_n = pywt.dwt2(noisy[i, :, :], wavelet=par.wave_base)
cA, (cH, cV, cD) = coeffs_n
tensor_n.append(cA)
tensor_n.append(cH)
tensor_n.append(cV)
tensor_n.append(cD)
coeffs_t = pywt.dwt2(target[i, :, :], wavelet=par.wave_base)
cA, (cH, cV, cD) = coeffs_t
tensor_t.append(cA)
tensor_t.append(cH)
tensor_t.append(cV)
tensor_t.append(cD)
return tensor_n, tensor_t
def gray_tensor_generator(noisy, target):
tensor_n = []
tensor_t = []
coeffs_n = pywt.dwt2(noisy, wavelet=par.wave_base)
cA, (cH, cV, cD) = coeffs_n
tensor_n.append(cA)
tensor_n.append(cH)
tensor_n.append(cV)
tensor_n.append(cD)
coeffs_t = pywt.dwt2(target, wavelet=par.wave_base)
cA, (cH, cV, cD) = coeffs_t
tensor_t.append(cA)
tensor_t.append(cH)
tensor_t.append(cV)
tensor_t.append(cD)
return tensor_n, tensor_t
def gray_reconstruction(tensor):
cA, cH, cV, cD = tensor[0, :, :], tensor[1, :, :], tensor[2, :, :], tensor[3, :, :]
coeffs = cA, (cH, cV, cD)
img_denoised = pywt.idwt2(coeffs, wavelet=par.wave_base)
return img_denoised
def color_reconstruction(tensor):
img_denoised = []
for i in range(par.img_channel):
channel_w = tensor[4*i: 4*(i+1), :, :]
cA, cH, cV, cD = channel_w[0, :, :], channel_w[1, :, :], channel_w[2, :, :], channel_w[3, :, :]
coeffs = cA, (cH, cV, cD)
channel = pywt.idwt2(coeffs, wavelet=par.wave_base)
img_denoised.append(channel)
img_denoised = np.stack(img_denoised, axis=0).astype(np.float32)
return img_denoised