-
Notifications
You must be signed in to change notification settings - Fork 3
/
utils.py
175 lines (145 loc) · 5.61 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
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
import re
import cv2
import math
import random
import string
import pathlib
import numpy as np
from PIL import ImageDraw, Image
from confusables.confusables import Confusables
def get_similar_char(string):
confusables = Confusables('./confusables/confusables.txt')
cpattern = confusables.confusables_regex(string)
return cpattern.replace("[","").replace("]","").split("\\")
# https://stackoverflow.com/questions/73426687/change-ascii-text-font-to-unicode-font-in-python
def random_printable_unicode():
def very_random_chars():
out = np.random.choice(np.random.randint(1,65533))
return out
def random_chars():
cyrillic_range = (int('0410', 16), int('0450', 16))
jappo_range = (int('3040', 16), int('309F', 16))
greek_range = (int('0370', 16), int('03FF', 16))
arr1 = np.random.randint(*cyrillic_range)
arr2 = np.random.randint(*jappo_range)
arr3 = np.random.randint(*greek_range)
out = np.stack((arr1, arr2, arr3))
out = np.stack((arr1, arr2, arr3))
out = np.random.choice(out)
return out
while True:
# i = random_chars()
i = very_random_chars()
c = chr(i)
if c.isprintable():
return c
# should add another conditional break
# to avoid infinite loop
def get_text(text_path):
# get text and sanitize
file = open(text_path, 'rU', encoding='utf8')
text = file.readlines()
text = ' '.join(text)
return text
def create_chunks(text, word_range):
text = text.split(' ')
idx_prev = 0
text_list = list()
while idx_prev <= len(text):
# get vertical_motion random step between 1 and set word_range
idx = np.random.randint(*word_range)
# if between the two indexes we found vertical_motion \n, split and update idx
for pos, word in enumerate(text[idx_prev: idx + idx_prev]):
if word.find('\n') != -1 or False: # or word.find(',') != -1:
idx = pos + 1
break
text_list.append(' '.join(text[idx_prev: idx + idx_prev]))
idx_prev += idx
return text_list
def make_chaos(text_chunk, f_percent):
# get len chunk
len_chunk = len(text_chunk)
if len_chunk < 4:
f_percent = 1 + (len_chunk - 1) * (-0.13)
# modify chunk with len == 0
if len_chunk is 0:
len_chunk += 1
text_chunk += " "
# set number of letters to modify
n_chaos_letters = np.random.randint(0, math.ceil(len_chunk * f_percent))
# random number of letters that we change
pos_letters = [0] * (len_chunk - n_chaos_letters) + [1] * n_chaos_letters
pos_random_letters = random.sample(pos_letters, len(pos_letters))
# random number (0 to 3) of extra letters in random place
random_chars = ''.join([random_printable_unicode() for _ in range(40)])
replace = ''.join(random.choices(random_chars + string.ascii_uppercase + string.digits, k=len_chunk))
# append chunks
new_text_chunk = ''
for i, pos in enumerate(pos_random_letters):
new_text_chunk += text_chunk[i] if pos is 0 else replace[i]
## alternative if similar chars will ever works
# new_text_chunk = ''
# for letter in text_chunk:
# similar_chars = get_similar_char(letter)
# similar_char = random.choices(similar_chars, k=1)[0]
# new_text_chunk += similar_char
return new_text_chunk
def find_max_chunk(text_list, pad_img, unicode_font):
max_chunk = max(text_list, key=len)
draw = ImageDraw.Draw(Image.new("RGB", (1, 1)))
w, h = draw.textsize(max_chunk, font=unicode_font)
# add padding
w += pad_img[0]
h += pad_img[1]
# make size divisible in block of 16 pixels
h = h - h % 16
w = w - w % 16
return w, h
def add_space_btw_chunks(text_list):
ext_text_list = list()
for word in text_list:
ext_text_list.append(np.random.randint(0, len(max(text_list, key=len))) * " ")
ext_text_list.append(word)
return ext_text_list
def create_blur_(f):
"""
in: img 4 dimensions
in function: 2 dimensions
out: same
"""
# Suppress/hide the warning
np.seterr(invalid='ignore')
f = f[:,:,0]
f = f/f.max() # normalize
# F(u,v), image in frequency domain
F = np.fft.fft2(f)
# H(u,v), motion blur function in frequency domain
# Create matrix H (motion blur function H(u,v))
M,N = F.shape
H = np.zeros((M+1,N+1), dtype=np.complex128) # +1 to avoid zero division
# Motion blur parameters
exposure_duration = 0.5 # duration of exposure
vertical_motion = 0 # vertical motion
horizontal_motion = 0.05 # horizontal motion
# Fill matrix H
for u in range(1,M+1):
for v in range(1,N+1):
s = np.pi*(u*vertical_motion + v*horizontal_motion)
H[u,v] = (exposure_duration/s) * np.sin(s) * np.exp(-1j*s)
# index slicing
H = H[1:,1:]
# G(u,v), blurred image in frequency domain
G = H * F
# g(x,y), blurred image in spatial domain
g = np.fft.ifft2(G)
g = np.abs(g)
return np.repeat(g[:, :, np.newaxis], 4, axis=2)
def create_blur(img, size=30):
# generating the kernel
kernel_motion_blur = np.zeros((size, size))
kernel_motion_blur[int((size-1)/2), :] = np.ones(size)
kernel_motion_blur = kernel_motion_blur / size
# applying the kernel to the input image
output = cv2.filter2D(img, -1, kernel_motion_blur)
output = np.uint8(output / 2)
return np.asarray(output)