-
Notifications
You must be signed in to change notification settings - Fork 0
/
distortion_correction.py
235 lines (202 loc) · 7.69 KB
/
distortion_correction.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
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
import numpy
import math
import sys
_EPS = numpy.finfo(float).eps * 4.0
def get_frame(G, g):
G = numpy.array(G)
G_original = G
g = numpy.array(g)
g_original = g
Gx, Gy, Gz = numpy.sum(G, axis=1)
centroid_1 = numpy.array([[Gx], [Gy], [Gz]])/len(G[0])
G = G - centroid_1 #center around origin
gx, gy, gz = numpy.sum(g, axis=1)
centroid_2 = numpy.array([[gx], [gy], [gz]])/len(g[0])
g = g - centroid_2 #center around origin
xx, yy, zz = numpy.sum(G * g, axis=1)
xy, yz, zx = numpy.sum(G * numpy.roll(g, -1, axis=0), axis=1)
xz, yx, zy = numpy.sum(G * numpy.roll(g, -2, axis=0), axis=1)
N = [[xx+yy+zz, yz-zy, zx-xz, xy-yx],
[yz-zy, xx-yy-zz, xy+yx, zx+xz],
[zx-xz, xy+yx, yy-xx-zz, yz+zy],
[xy-yx, zx+xz, yz+zy, zz-xx-yy]]
w1, v1 = numpy.linalg.eig(N)
max_index = numpy.argmax(w1)
q = v1[:,max_index]
n = numpy.dot(q, q)
if n < _EPS:
R = numpy.identity(3)
else:
rot_matrix = [[math.pow(q[0], 2) + math.pow(q[1], 2) - math.pow(q[2], 2) - math.pow(q[3], 2), 2*(q[1]*q[2] - q[0]*q[3]), 2*(q[1]*q[3] + q[0]*q[2])],
[2*(q[1]*q[2] + q[0]*q[3]), math.pow(q[0], 2) - math.pow(q[1], 2) + math.pow(q[2], 2) - math.pow(q[3], 2), 2*(q[2]*q[3] - q[0]*q[1])],
[2*(q[1]*q[3] - q[0]*q[2]), 2*(q[2]*q[3] + q[0]*q[1]), math.pow(q[0], 2) - math.pow(q[1], 2) - math.pow(q[2], 2) + math.pow(q[3], 2)]]
R = numpy.array(rot_matrix)
t = numpy.dot(R.T, centroid_2) - centroid_1
return Frame(R.T, -t)
class Frame:
def __init__(self, rotation = None, translation = None):
if rotation is None:
self.rotation = [numpy.identity(3)] #the identity matrix should be default
else:
self.rotation = rotation
if translation is None:
self.translation = numpy.array([0, 0, 0]) #default translation should be zero
else:
self.translation = translation
def set_rot(self, rotation):
self.rotation = rotation
def get_rot(self):
return self.rotation
def set_trans(self, translation):
self.translation = translation
def get_trans(self):
return self.translation
def B(k, v):
# print(v)
nCk = math.factorial(5) / (math.factorial(k)*math.factorial(5 - k))
return nCk*((1 - v)**(5 - k))*(v**(k))
def get_coeff(C_expected, C):
F = []
C_norm = []
# C_max = numpy.linalg.norm(numpy.amax(C, axis=0))
# C_min = numpy.linalg.norm(numpy.amin(C, axis=0))
C_max = []
C_min = []
C = numpy.array(C)
for i in range(3):
C_max.append(numpy.max(C[:,i]))
C_min.append(numpy.min(C[:,i]))
for i in range(len(C)):
to_append = []
for j in range(3):
to_append.append((float(C[i][j]) - float(C_min[j]))/(float(C_max[j]) - float(C_min[j])))
C_norm.append(to_append)
# C_expected_norm.append((C_expected[i])/(C_max))
# C_norm = C
for data_point in range(len(C_norm) - 10):
F.append([])
for i in range(6):
for j in range(6):
for k in range(6):
F[data_point].append(B(i, C_norm[data_point][0])*B(j, C_norm[data_point][1])*B(k, C_norm[data_point][2]))
# print()
return numpy.linalg.lstsq(numpy.array(F), numpy.array(C_expected[:-10]))[0], C_min, C_max
def correct_distortion(coeffs, q, q_min, q_max):
G_corrected = []
q_norm = []
# q_max = []
# q_min = []
# q = numpy.array(q)
# for i in range(3):
# q_max.append(numpy.max(q[:,i]))
# q_min.append(numpy.min(q[:,i]))
for i in range(len(q)):
to_append = []
# For each vector, take each component and normalize them
for j in range(3):
to_append.append((float(q[i][j]) - float(q_min[j]))/(float(q_max[j]) - float(q_min[j])))
q_norm.append(to_append)
F = []
for data_point in range(len(q_norm)):
# print(q_norm[data_point])
# sum = [0, 0, 0]
F.append([])
for i in range(6):
for j in range(6):
for k in range(6):
# co_index = 36*i + 6*j + k
# print(sum)
# sum = [sum[0] + coeffs[co_index][0]*B(i, q_norm[data_point][0]), sum[1] + coeffs[co_index][1]*B(j, q_norm[data_point][1]), sum[2] + coeffs[co_index][2]*B(k, q_norm[data_point][2])]
F[data_point].append(B(i, q_norm[data_point][0])*B(j, q_norm[data_point][1])*B(k, q_norm[data_point][2]))
# print(coeffs[36*i + 6*j + k][0]*B(i, q_norm[data_point][0]), coeffs[36*i + 6*j + k][1]*B(j, q_norm[data_point][1]), coeffs[36*i + 6*j + k][2]*B(k, q_norm[data_point][2]))
# print(sum)
# G_corrected.append(sum)
# print(coeffs.shape)
F = numpy.array(F)
# print(F.shape)
G_corrected = numpy.dot(F, coeffs)
# G_corrected = [G_corrected[0] - G_corrected[0], G_corrected[1] - G_corrected[0], G_corrected[2] - G_corrected[0], G_corrected[3] - G_corrected[0]]
# G_corrected = numpy.array(G_corrected) - numpy.array(G_corrected[0])
# print(G_corrected/G_corrected[1])
return numpy.round(G_corrected, 3)
# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# Main file
if (len(sys.argv) != 4):
sys.exit(0)
cal_body = open(sys.argv[1])
cal_readings = open(sys.argv[2])
piv_points = open(sys.argv[3])
index1 = int(sys.argv[1].index("pa1"))
index2 = int(sys.argv[1].index("calbody"))
filename = sys.argv[1][index1:index2]
# cal_readings.readline()
first_line = cal_readings.readline().split(",")
N_d = int(first_line[0].strip())
N_a = int(first_line[1].strip())
N_c = int(first_line[2].strip())
N_frames = int(first_line[3].strip())
d = []
a = []
c = []
F_d = []
F_a = []
C = []
C_expected = []
piv_first_line = piv_points.readline().split(",")
num_markers = int(piv_first_line[0].strip())
G = []
# cal_body
for i in range(N_d):
line = cal_body.readline().split(",")
tpose = numpy.array([float(line[0].strip()), float(line[1].strip()), float(line[2].strip())])
d.append(numpy.array(tpose).T)
for i in range(N_a):
line = cal_body.readline().split(",")
tpose = numpy.array([float(line[0].strip()), float(line[1].strip()), float(line[2].strip())])
a.append(numpy.array(tpose).T)
for i in range(N_c):
line = cal_body.readline().split(",")
c.append([float(line[0].strip()), float(line[1].strip()), float(line[2].strip())])
# cal_readings
for i in range(N_frames):
D = []
A = []
# C.append([])
# C_expected.append([])
# G.append([])
for j in range(N_d):
line = cal_readings.readline().split(",")
tpose = numpy.array([float(line[0].strip()), float(line[1].strip()), float(line[2].strip())])
D.append(numpy.array(tpose).T)
for j in range(N_a):
line = cal_readings.readline().split(",")
tpose = numpy.array([float(line[0].strip()), float(line[1].strip()), float(line[2].strip())])
A.append(numpy.array(tpose).T)
F_d.append(get_frame(numpy.array(D).T, numpy.array(d).T))
F_a.append(get_frame(numpy.array(A).T, numpy.array(a).T))
R_d_i = numpy.linalg.inv(F_d[i].get_rot())
P_d_i = numpy.dot(R_d_i, F_d[i].get_trans())
P_a = F_a[i].get_trans()
R_a = F_a[i].get_rot()
for j in range(N_c):
line = cal_readings.readline().split(",")
tpose = numpy.array([float(line[0].strip()), float(line[1].strip()), float(line[2].strip())])
C.append(numpy.array(tpose).T)
# C[i].append(numpy.array(tpose).T)
inside = numpy.dot(R_a, c[j]) + P_a
square = numpy.dot(R_d_i, inside) - P_d_i
C_expected.append([square[0][0], square[1][1], square[2][2]])
# C_expected[i].append([square[0][0], square[1][1], square[2][2]])
# print(numpy.array(G))
cal_body.close()
cal_readings.close()
piv_points.close()
coeffs, q_min, q_max = get_coeff(C_expected, C)
print(numpy.allclose(correct_distortion(coeffs, C, q_min, q_max), C_expected, rtol=1e+02))
# output = open("OUTPUT/" + filename + "output2.txt", 'w')
# output.write(str(N_c) + ", " + str(N_frames) + ", " + filename + "output2.txt\n")
# for frame in range(N_frames):
# print(numpy.array(F[frame]).shape, numpy.array(c).shape)
# output.write(get_coeff(numpy.array(F[frame]), numpy.array(c)))
# # print("%.2f" % C_expected[i][0] + ", " + "%.2f" % C_expected[i][1] + ", " + "%.2f" % C_expected[i][2])
# output.close()