-
Notifications
You must be signed in to change notification settings - Fork 2
/
utils.py
130 lines (104 loc) · 3.49 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
130
import numpy as np
import torch
from skimage import draw
def get_masks(paths):
masks = torch.zeros(len(paths), 128, 128)
for i, p in enumerate(paths):
masks[i] = torch.fft.fftshift(torch.load(p, map_location='cpu').squeeze())
return masks
def get_mask_diffs(masks, diffs):
Dm = torch.zeros(len(diffs), 128, 128)
for i, diff in enumerate(diffs):
Dm[i] = masks[diff[0]] - masks[diff[1]]
Dm = (Dm - Dm.min()) / (Dm.max() - Dm.min())
Dm -= 0.5
return Dm
def radial_band(
radius_in,
radius_out,
mask_size=(128, 128),
mask_value=1.,
p=2
):
# get the center of the image
c_x = mask_size[0]//2
c_y = mask_size[1]//2
# band mask init
mask = torch.ones(mask_size, dtype=bool)
x = torch.arange(0, mask_size[0])
y = torch.arange(0, mask_size[0])
x_dist = torch.abs(x - c_x)
y_dist = torch.abs(y - c_y)
xx, yy = torch.meshgrid(x_dist, y_dist)
if p==2:
# L2:
dist = torch.sqrt(xx**2 + yy**2)
elif p==1:
# L1:
dist = (torch.abs(xx) + torch.abs(yy))
else:
raise ValueError("Invalid choice for p")
if radius_out is not None:
mask[dist >= radius_out] = False
mask[dist < radius_in] = False
mask = mask * mask_value
return mask
def circular_band(
angle1,
angle2,
mask_size=(128, 128),
mask_value=1.
):
# get the center and max radius considered
r0, c0 = mask_size[0]//2, mask_size[1]//2
R = np.sqrt(mask_size[0]**2 + mask_size[1]**2)
theta0 = np.deg2rad(angle1)
theta1 = np.deg2rad(angle2)
r1, c1 = r0 - 1.5 * R * np.sin(theta0), c0 + 1.5 * R * np.cos(theta0)
r2, c2 = r0 - 1.5 * R * np.sin(theta1), c0 + 1.5 * R * np.cos(theta1)
mask_circle = torch.zeros(mask_size, dtype=bool)
mask_poly = torch.zeros(mask_size, dtype=bool)
rr, cc = draw.disk((r0, c0), R, shape=mask_circle.shape)
mask_circle[rr, cc] = 1
rr, cc = draw.polygon(
[r0, r1, r2, r0],
[c0, c1, c2, c0],
shape=mask_poly.shape
)
mask_poly[rr, cc] = 1
mask = mask_circle & mask_poly
mask = mask * mask_value
return mask
def bandmasks(num_bands, band_intensities):
radii = torch.linspace(0, np.sqrt(2*64**2), num_bands+1)
angles = torch.linspace(0, 360, num_bands+1)
rmasks = []
amasks = []
for i in range(num_bands):
r = radial_band(radii[i], radii[i+1], mask_value=band_intensities[i])
a = circular_band(angles[i], angles[i+1], mask_value=band_intensities[i])
if i==0:
rmasks.append(r)
amasks.append(a)
else:
r[r==rmasks[i-1]] = 0.
a[a==amasks[i-1]] = 0.
if i==num_bands-1:
r[r==rmasks[0]] = 0.
a[a==amasks[0]] = 0.
rmasks.append(r)
amasks.append(a)
return rmasks, amasks
def energies(img, num_bands=32):
img = img/torch.norm(img, p=2)
rmasks, amasks = bandmasks(num_bands, torch.ones(num_bands))
radial = torch.zeros(num_bands)
angular = torch.zeros(num_bands)
for i in range(num_bands):
radial[i] = torch.norm(rmasks[i] * img, p=2)
angular[i] = torch.norm(amasks[i] * img, p=2)
return radial, angular
def get_sectors(num_bands):
band_intensities = np.linspace(0, 1, num_bands)
radial_masks, angular_masks = bandmasks(num_bands, band_intensities)
return sum(radial_masks), sum(angular_masks)