-
Notifications
You must be signed in to change notification settings - Fork 61
/
simulate_paper.py
174 lines (145 loc) · 6.95 KB
/
simulate_paper.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
import cv2
import numpy as np
import math
constant_length = 1000
def Gassian(size, mean = 0, var = 0):
norm = np.random.randn(*size)
denorm = norm * np.sqrt(var) + mean
return np.uint8(np.round(np.clip(denorm,0,255)))
def Getline(distribution, length):
period = distribution.shape[0]
if length < constant_length: # if length is too short, lines are Aligned
patch = Gassian((2*period, length), mean=250, var = 3)
begin = 0
end = 1
for i in range(period):
patch[i]=Gassian((1,length), mean=distribution[i,0], var=distribution[i,1])
else: # if length is't too short, lines is't Aligned
patch = Gassian((2*period, length+4*period), mean=250, var = 3)
begin = Gassian((1,1), mean=2.0*period, var=2*period)
# egin = Gassian((1,1), mean=2.0*period, var=0)
begin = np.uint8(np.round(np.clip(begin,0,4*period)))
begin = int(begin[0,0])
end = Gassian((1,1), mean=2.0*period, var=2*period)
# end = Gassian((1,1), mean=2.0*period, var=0)
end = np.uint8(np.round(np.clip(end,1,4*period+1)))
end = int(end[0,0])
real_length = length+4*period-end-begin
for i in range(period):
patch[i,begin:-end]=Gassian((1,real_length), mean=distribution[i,0], var=distribution[i,1])
patch = Attenuation(patch, period=period, distribution=distribution,begin=begin, end=end)
patch = Distortion(patch, begin=begin, end=end)
return np.uint8(np.round(np.clip(patch,0,255)))
def Attenuation(patch, period, distribution, begin, end):
order = int((patch.shape[1]-begin-end)/2)+1
radius = (period-1)/2
canvas = Gassian((patch.shape[0], patch.shape[1]), mean=250, var=3)
patch = np.float32(patch)
canvas = np.float32(canvas)
for i in range(begin, patch.shape[1]-end+1):
for j in range(period):
a = np.abs((1.0-(i-begin)/order)**2)/3
b = np.abs((1.0-j/radius)**2)*1
patch[j,i] += (canvas[j,i]-patch[j,i])*np.sqrt(a+b)/1.5
# patch[j,i] += 0.75*(canvas[j,i]-patch[j,i]) * (np.abs((1.0-(i-begin)/order)**2))**0.5
return np.uint8(np.round(np.clip(patch,0,255)))
def Distortion(patch,begin,end):
height = int(patch.shape[0]/2)
length = patch.shape[1]
patch = np.float32(patch)
patch_copy = patch.copy()
# central = ((length-begin-end)/2+begin) + np.random.randn()*length/30
central = ((length-begin-end)/2+begin)
if length>100:
radius = length**2/(4*height)
else:
radius = length**2/(2*height)
for i in range(length):
offset = ((central-i)**2)/(2*radius)
int_offset = int(offset)
decimal_offset = offset-int_offset
for j in range(height):
if j>int_offset:
patch[j,i]=int(decimal_offset*patch_copy[j-1-int_offset,i]+(1-decimal_offset)*patch_copy[j-int_offset,i])
else:
patch[j,i]= np.random.randn() * np.sqrt(3) + 250
patch_copy = patch.copy()
if length>100:
for i in range(length):
offset = ((central-i)**2)/(2*radius)
int_offset = int(offset)
decimal_offset = offset-int_offset
for j in range(patch.shape[0]):
if j>int_offset:
patch[j,i]=int(decimal_offset*patch_copy[j-1-int_offset,i]+(1-decimal_offset)*patch_copy[j-int_offset,i])
else:
patch[j,i]= np.random.randn() * np.sqrt(3) + 250
# else:
# radius = length**2/(4*height)
# for i in range(length):
# offset = ((central-i)**2)/(2*radius)
# int_offset = int(offset)
# decimal_offset = offset-int_offset
# for j in range(patch.shape[0]):
# if j>int_offset:
# patch[j,i]=int(decimal_offset*patch_copy[j-1-int_offset,i]+(1-decimal_offset)*patch_copy[j-int_offset,i])
# else:
# patch[j,i]= np.random.randn() * np.sqrt(3) + 250
return np.uint8(np.round(np.clip(patch,0,255)))
def GetParallel(distribution, height, length, period):
if length<constant_length: # constant length
canvas = Gassian((height+2*period,length), mean=250, var = 3)
else: # variable length
canvas = Gassian((height+2*period,length+4*period), mean=250, var = 3)
distensce = Gassian((1,int(height/period)+2), mean = period, var = period/5)
# distensce = Gassian((1,int(height/period)+1), mean = period, var = 0)
distensce = np.uint8(np.round(np.clip(distensce, period*0.8,period*1.25)))
begin = 0
for i in np.squeeze(distensce).tolist():
newline = Getline(distribution=distribution, length=length)
h,w = newline.shape
# cv2.imshow('line', newline)
# cv2.waitKey(0)
# cv2.imwrite("D:/ECCV2020/simu_patch/Line3.jpg",newline)
if begin < height:
m = np.minimum(canvas[begin:(begin + h),:], newline)
canvas[begin:(begin + h),:] = m
begin += i
else:
break
return canvas[:height,:]
def ChooseDistribution(period, Grayscale):
distribution = np.zeros((period,2))
c = period/2.0
difference = 250-Grayscale
for i in range(distribution.shape[0]):
distribution[i][0] = Grayscale + difference*abs(i-c)/c
distribution[i][1] = np.cos((i-c)/c*(0.5*3.1415929))*difference
# distribution[i][0] -= np.cos((i-4)/4.0*(0.5*3.1415929))*difference
# distribution[i][1] += np.cos((i-4)/4.0*(0.5*3.1415929))*difference
return np.abs(distribution)
if __name__ == '__main__':
np.random.seed(100)
canvas = Gassian((500,500), mean=250, var = 3)
# distribution = np.array([[245,31],[238,27],[218,48],[205,33],[214,38],[234,24],[240,42]])
###################################################
###################################################
###################################################
period = 7
Grayscale = 128
H,L = (100,200)
###################################################
###################################################
###################################################
distribution = ChooseDistribution(period=period, Grayscale=Grayscale)
print(distribution)
patch = GetParallel(distribution=distribution, height=H, length=L, period=period)
(h,w) = patch.shape
# patch = GetOffsetParallel(offset=4, distribution=distribution, patch_size=(40,200), period_mean=distribution.shape[0], period_var=1)
# (h,w) = patch.shape
canvas[250-int(h/2):250-int(h/2)+h,250-int(w/2):250-int(w/2)+w] = patch
# cv2.imshow('Parallel', patch[:, 2*distribution.shape[0]:w-2*distribution.shape[0]])
cv2.imshow('Parallel', canvas)
cv2.waitKey(0)
cv2.imwrite("D:/ECCV2020/simu_patch/Parallel4.jpg",patch)
print("done")