-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathmake_data.py
executable file
·194 lines (171 loc) · 8 KB
/
make_data.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
#!/usr/bin/env python
import sys
import os
import time
import argparse
import configparser
import h5py
import numpy as np
import cupy as cp
class DataGenerator():
def __init__(self, config_file):
config = configparser.ConfigParser()
config.read(config_file)
self.size = config.getint('parameters', 'size')
self.num_data = config.getint('make_data', 'num_data')
self.fluence = config.get('make_data', 'fluence', fallback='constant')
self.mean_count = config.getfloat('make_data', 'mean_count')
self.bg_count = config.getfloat('make_data', 'bg_count', fallback=None)
self.rel_scale = config.getfloat('make_data', 'rel_scale', fallback=1000.)
self.dia_params = [float(s) for s in config.get('make_data', 'dia_params').split()]
self.shift_sigma = config.getfloat('make_data', 'shift_sigma')
self.out_file = os.path.join(os.path.dirname(config_file),
config.get('make_data', 'out_photons_file'))
if self.fluence not in ['constant', 'gamma']:
raise ValueError('make_data:fluence needs to be either constant (default) or gamma')
with open('kernels.cu', 'r') as f:
kernels = cp.RawModule(code=f.read())
self.k_slice_gen_holo = kernels.get_function('slice_gen_holo')
self.k_slice_gen = kernels.get_function('slice_gen')
self.object = cp.zeros((self.size, self.size), dtype='f8')
self.object_sum = 0
self.bgmask = cp.zeros_like(self.object)
self.bgmask_sum = 0
def make_obj(self, bg=False):
mask = self.bgmask if bg else self.object
num_circ = 80
mcen = self.size // 2
x, y = cp.indices(self.object.shape, dtype='f8')
for _ in range(num_circ):
rad = (0.7 + 0.3*cp.random.rand(1, dtype='f8')) * self.size / 25.
while True:
cen = cp.random.rand(2, dtype='f8') * self.size / 5. + mcen * 4./ 5.
dist = float(cp.sqrt((cen[0]-mcen)**2 + (cen[1]-mcen)**2) + rad)
if dist < mcen:
break
diskrad = cp.sqrt((x - cen[0])**2 + (y - cen[1])**2)
mask[diskrad <= rad] += 1. - (diskrad[diskrad <= rad] / rad)**2
if bg:
#mask *= self.bg_count / mask.sum()
self.bgmask_sum = float(mask.sum())
else:
#mask *= self.mean_count / mask.sum()
self.object_sum = float(mask.sum())
with h5py.File(self.out_file, 'a') as fptr:
if 'solution' in fptr:
del fptr['solution']
fptr['solution'] = mask.get()
def parse_obj(self, bg=False):
mask = self.bgmask if bg else self.object
dset_name = 'bg' if bg else 'solution'
with h5py.File(self.out_file, 'r') as fptr:
mask = cp.array(fptr[dset_name][:])
if bg:
mask *= self.bg_count / mask.sum()
self.bgmask_sum = float(mask.sum())
self.bgmask = mask
else:
#mask *= self.mean_count / mask.sum()
self.object_sum = float(mask.sum())
self.object = mask
def make_data(self, parse=False):
if self.object_sum == 0.:
if parse:
self.parse_obj()
else:
self.make_obj()
if self.bg_count is not None:
if parse:
self.parse_obj(bg=True)
else:
self.make_obj(bg=True)
mask = cp.ones(self.object.shape, dtype='f8')
x, y = cp.indices(self.object.shape, dtype='f8')
cen = self.size // 2
pixrad = cp.sqrt((x - cen)**2 + (y - cen)**2)
mask[pixrad<4] = 0
mask[pixrad>=cen] = 0
fptr = h5py.File(self.out_file, 'a')
if 'ones' in fptr: del fptr['ones']
if 'multi' in fptr: del fptr['multi']
if 'place_ones' in fptr: del fptr['place_ones']
if 'place_multi' in fptr: del fptr['place_multi']
if 'count_multi' in fptr: del fptr['count_multi']
if 'num_pix' in fptr: del fptr['num_pix']
if 'true_shifts' in fptr: del fptr['true_shifts']
if 'true_diameters' in fptr: del fptr['true_diameters']
if 'true_angles' in fptr: del fptr['true_angles']
if 'bg' in fptr: del fptr['bg']
if 'scale' in fptr: del fptr['scale']
if self.bgmask_sum > 0:
fptr['bg'] = self.bgmask.get()
fptr['num_pix'] = np.array([self.size**2])
dtype = h5py.special_dtype(vlen=np.dtype('i4'))
place_ones = fptr.create_dataset('place_ones', (self.num_data,), dtype=dtype)
place_multi = fptr.create_dataset('place_multi', (self.num_data,), dtype=dtype)
count_multi = fptr.create_dataset('count_multi', (self.num_data,), dtype=dtype)
ones = fptr.create_dataset('ones', (self.num_data,), dtype='i4')
multi = fptr.create_dataset('multi', (self.num_data,), dtype='i4')
#shifts = np.random.random((self.num_data, 2))*6 - 3
#shifts = np.random.randn(self.num_data, 2)*1.
#shifts = np.zeros((self.num_data, 2))
shifts = np.random.randn(self.num_data, 2)*self.shift_sigma
fptr['true_shifts'] = shifts
if self.fluence == 'gamma':
scale = np.random.gamma(2., 0.5, self.num_data)
else:
scale = np.ones(self.num_data, dtype='f8')
fptr['scale'] = scale
#diameters = np.random.randn(self.num_data)*0.5 + 7.
#diameters = np.ones(self.num_data)*7.
diameters = np.random.randn(self.num_data)*self.dia_params[1] + self.dia_params[0]
fptr['true_diameters'] = diameters
#rel_scales = diameters**3 * 1000. / 7**3
#scale *= rel_scales/1.e3
angles = np.random.random(self.num_data) * 2. * np.pi
#angles = np.zeros(self.num_data)
fptr['true_angles'] = angles
view = cp.zeros(self.size**2, dtype='f8')
rview = cp.zeros_like(view, dtype='f8')
zmask = cp.zeros_like(view, dtype='f8')
model = cp.fft.fftshift(cp.fft.fftn(cp.fft.ifftshift(self.object)))
bsize_model = int(np.ceil(self.size/32.))
stime = time.time()
for i in range(self.num_data):
self.k_slice_gen_holo((bsize_model,)*2, (32,)*2,
(model, shifts[i,0], shifts[i,1], diameters[i], self.rel_scale, scale[i], self.size, zmask, 0, view))
view *= mask.ravel()
view *= self.mean_count / view.sum()
self.k_slice_gen((bsize_model,)*2, (32,)*2,
(view, angles[i], 1., self.size, self.bgmask, 0, rview))
frame = cp.random.poisson(rview, dtype='i4')
place_ones[i] = cp.where(frame == 1)[0].get()
place_multi[i] = cp.where(frame > 1)[0].get()
count_multi[i] = frame[frame > 1].get()
ones[i] = place_ones[i].shape[0]
multi[i] = place_multi[i].shape[0]
sys.stderr.write('\rWritten %d/%d frames (%d) ' % (i+1, self.num_data, int(frame.sum())))
etime = time.time()
sys.stderr.write('\nTime taken (make_data): %f s\n' % (etime-stime))
fptr.close()
def main():
parser = argparse.ArgumentParser(description='Padman data generator')
parser.add_argument('-c', '--config_file',
help='Path to config file (Default: config.ini)',
default='config.ini')
parser.add_argument('-m', '--mask_only',
help='Create mask only and not the data frames',
action='store_true', default=False)
parser.add_argument('-d', '--data_only',
help='Generate data only. Use preexisting mask in file',
action='store_true', default=False)
parser.add_argument('-D', '--device',
help='Device number (default: 0)', type=int, default=0)
args = parser.parse_args()
datagen = DataGenerator(args.config_file)
if not args.data_only:
datagen.make_obj()
if not args.mask_only:
datagen.make_data(parse=args.data_only)
if __name__ == '__main__':
main()