-
Notifications
You must be signed in to change notification settings - Fork 0
/
2.feature_extra.py
221 lines (194 loc) · 10.1 KB
/
2.feature_extra.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
# 第二步,输入文件是已经打好label的文件
import numpy as np
import pandas as pd
import os
# -------------------------------------------------------------------------------------------------
# 把加速度传感器的值从对手机坐标系转换成对地坐标系,具体转换公式可以参考Android的getRotationMatrix源码
# 需要重力和磁感应的三轴数值
#计算错误弃用了
def getRotationMatrix(R, gravity, geomagnetic):
Ax = gravity[:, 0].T
Ay = gravity[:, 1].T
Az = gravity[:, 2].T
Ex = geomagnetic[:, 0].T
Ey = geomagnetic[:, 1].T
Ez = geomagnetic[:, 2].T
Hx = Ey * Az - Ez * Ay
Hy = Ez * Ax - Ex * Az
Hz = Ex * Ay - Ey * Ax
normH = np.sqrt(Hx * Hx + Hy * Hy + Hz * Hz)
invH = 1.0 / normH
Hx *= invH
Hy *= invH
Hz *= invH
invA = 1.0 / np.sqrt(Ax * Ax + Ay * Ay + Az * Az)
Ax *= invA
Ay *= invA
Az *= invA
Mx = Ay * Hz - Az * Hy
My = Az * Hx - Ax * Hz
Mz = Ax * Hy - Ay * Hx
R[:, 0] = Hx
R[:, 1] = Hy
R[:, 2] = Hz
R[:, 3] = Mx
R[:, 4] = My
R[:, 5] = Mz
R[:, 6] = Ax
R[:, 7] = Ay
R[:, 8] = Az
def dev():
os.chdir('test')
data = pd.read_csv("dirty.csv")
print(data.shape)
gravity = np.asarray([data['g_x'], data['g_y'], data['g_z']]).T
magnetic = np.asarray([data['m_x'], data['m_y'], data['m_z']]).T
rotate = np.zeros(shape=(len(gravity), 9))
acc_o = np.asarray([data['acc_x'], data['acc_y'], data['acc_z']]).T
acc = np.zeros(shape=(len(acc_o), 3))
getRotationMatrix(rotate, gravity, magnetic)
print("data", data.shape)
print("acc", acc.shape)
acc[:, 0] = rotate[:, 0] * acc_o[:, 0] + rotate[:, 1] * acc_o[:, 1] + rotate[:, 2] * acc_o[:, 2]
acc[:, 1] = rotate[:, 3] * acc_o[:, 0] + rotate[:, 4] * acc_o[:, 1] + rotate[:, 5] * acc_o[:, 2]
acc[:, 2] = rotate[:, 6] * acc_o[:, 0] + rotate[:, 7] * acc_o[:, 1] + rotate[:, 8] * acc_o[:, 2] - 9.807
label = np.asarray(data['label'])
time = np.asarray(data['time'])
print(label.shape)
label = label.reshape(label.shape[0], 1)
time = time.reshape(time.shape[0], 1)
print(label.shape)
label = pd.DataFrame(label)
acc = np.hstack((time, acc))
acc_xy = pd.DataFrame(np.sqrt(np.square(acc[:, 1]) + np.square(acc[:, 2])))
acc_xyz = pd.DataFrame(np.sqrt(np.square(acc[:, 1]) + np.square(acc[:, 2]) + np.square(acc[:, 3])))
acc = pd.DataFrame(acc)
# 输出格式['time', 'acc_x', 'acc_y', 'acc_z']
# --------------------------------------------------------------------------------------------------
# # 对o_w,o_x,o_y,o_z进行坐标系的转换,具体公式可以参考ubicomp2018第一名的公式
orientation = np.asarray([data['o_w'], data['o_x'], data['o_y'], data['o_z']])
orien = orientation.T
rn0 = np.asarray(1 - 2 * (np.square(orien[:, 2]) + np.square(orien[:, 3])))
rn1 = 2 * (orien[:, 1] * orien[:, 2] - orien[:, 0] * orien[:, 3])
rn2 = 2 * (orien[:, 1] * orien[:, 3] + orien[:, 0] * orien[:, 2])
rn3 = 2 * (orien[:, 1] * orien[:, 2] + orien[:, 0] * orien[:, 3])
rn4 = 1 - 2 * (np.square(orien[:, 1]) + np.square(orien[:, 3]))
rn5 = 2 * (orien[:, 2] * orien[:, 3] - orien[:, 0] * orien[:, 1])
rn6 = 2 * (orien[:, 1] * orien[:, 3] - orien[:, 0] * orien[:, 2])
rn7 = 2 * (orien[:, 2] * orien[:, 3] + orien[:, 0] * orien[:, 1])
rn8 = 1 - 2 * (np.square(orien[:, 1]) + np.square(orien[:, 2]))
o1 = np.asarray([data['o_x'], data['o_y'], data['o_z']])
o_x = pd.DataFrame(rn0 * o1[0] + rn1 * o1[1] + rn2 * o1[2])
o_y = pd.DataFrame(rn3 * o1[0] + rn4 * o1[1] + rn5 * o1[2])
o_z = pd.DataFrame(rn6 * o1[0] + rn7 * o1[1] + rn8 * o1[2])
pitch = pd.DataFrame(np.arctan(rn7 / rn8))
roll = pd.DataFrame(np.arcsin(-rn6))
yaw = pd.DataFrame(np.arctan(rn3 / rn0))
ori = pd.concat((o_x, o_y, o_z, pitch, roll, yaw), axis=1)
print(ori.shape)
# 输出格式为['o_x', 'o_y', 'o_z', 'pitch', 'roll', 'yaw']
# -----------------------------------------------------------------------------------------------
# 对m_x,m_y,m_z取平方和之后开根号,作为新的列值
magnetic = np.asarray([data['m_x'], data['m_y'], data['m_z']]).T
ma = np.sqrt(np.square(magnetic[:, 0]) + np.square(magnetic[:, 1]) + np.square(magnetic[:, 2]))
magnetic = pd.DataFrame(magnetic)
print("magnetic", magnetic.shape)
ma_t = pd.DataFrame(ma)
print(ma_t.shape)
ma = pd.concat((ma_t, magnetic), axis=1)
# 输出格式为['ma','m_x', 'm_y', 'm_z']
# -----------------------------------------------------------------------------------------------
remain = pd.DataFrame(np.asarray([data['gy_x'], data['gy_y'], data['gy_z'],
data['g_x'], data['g_x'], data['g_x'],
data['l_x'], data['l_x'], data['l_x'],
data['pressure']
]).T)
fin = pd.concat((acc, acc_xy, acc_xyz, ori, ma, remain, label), axis=1)
print(fin.shape)
fin.to_csv("raw_data.csv", index=False, header=['time', 'acc_x', 'acc_y', 'acc_z', 'acc_xy', 'acc_xyz',
'o_x', 'o_y', 'o_z', 'pitch', 'roll', 'yaw',
'magnetic', 'm_x', 'm_y', 'm_z',
'gy_x', 'gy_y', 'gy_z',
'g_x', 'g_y', 'g_z',
'l_x', 'l_y', 'l_z',
'pressure',
'label'])
def train():
os.chdir('data')
data = pd.read_csv("dirty.csv")
print(data.shape)
gravity = np.asarray([data['g_x'], data['g_y'], data['g_z']]).T
magnetic = np.asarray([data['m_x'], data['m_y'], data['m_z']]).T
rotate = np.zeros(shape=(len(gravity), 9))
acc_o = np.asarray([data['acc_x'], data['acc_y'], data['acc_z']]).T
acc = np.zeros(shape=(len(acc_o), 3))
getRotationMatrix(rotate, gravity, magnetic)
print("data", data.shape)
print("acc", acc.shape)
acc[:, 0] = rotate[:, 0] * acc_o[:, 0] + rotate[:, 1] * acc_o[:, 1] + rotate[:, 2] * acc_o[:, 2]
acc[:, 1] = rotate[:, 3] * acc_o[:, 0] + rotate[:, 4] * acc_o[:, 1] + rotate[:, 5] * acc_o[:, 2]
acc[:, 2] = rotate[:, 6] * acc_o[:, 0] + rotate[:, 7] * acc_o[:, 1] + rotate[:, 8] * acc_o[:, 2] - 9.807
label = np.asarray(data['label'])
time = np.asarray(data['time'])
print(label.shape)
label = label.reshape(label.shape[0], 1)
time = time.reshape(time.shape[0], 1)
print(label.shape)
label = pd.DataFrame(label)
acc = np.hstack((time, acc))
# acc = pd.DataFrame(acc)
acc_xy = pd.DataFrame(np.sqrt(np.square(acc[:, 1]) + np.square(acc[:, 2])))
acc_xyz = pd.DataFrame(np.sqrt(np.square(acc[:, 1]) + np.square(acc[:, 2]) + np.square(acc[:, 3])))
acc = pd.DataFrame(acc)
# 输出格式['time', 'acc_x', 'acc_y', 'acc_z']
# --------------------------------------------------------------------------------------------------
# # 对o_w,o_x,o_y,o_z进行坐标系的转换,具体公式可以参考ubicomp2018第一名的公式
orientation = np.asarray([data['o_w'], data['o_x'], data['o_y'], data['o_z']])
orien = orientation.T
rn0 = np.asarray(1 - 2 * (np.square(orien[:, 2]) + np.square(orien[:, 3])))
rn1 = 2 * (orien[:, 1] * orien[:, 2] - orien[:, 0] * orien[:, 3])
rn2 = 2 * (orien[:, 1] * orien[:, 3] + orien[:, 0] * orien[:, 2])
rn3 = 2 * (orien[:, 1] * orien[:, 2] + orien[:, 0] * orien[:, 3])
rn4 = 1 - 2 * (np.square(orien[:, 1]) + np.square(orien[:, 3]))
rn5 = 2 * (orien[:, 2] * orien[:, 3] - orien[:, 0] * orien[:, 1])
rn6 = 2 * (orien[:, 1] * orien[:, 3] - orien[:, 0] * orien[:, 2])
rn7 = 2 * (orien[:, 2] * orien[:, 3] + orien[:, 0] * orien[:, 1])
rn8 = 1 - 2 * (np.square(orien[:, 1]) + np.square(orien[:, 2]))
o1 = np.asarray([data['o_x'], data['o_y'], data['o_z']])
o_x = pd.DataFrame(rn0 * o1[0] + rn1 * o1[1] + rn2 * o1[2])
o_y = pd.DataFrame(rn3 * o1[0] + rn4 * o1[1] + rn5 * o1[2])
o_z = pd.DataFrame(rn6 * o1[0] + rn7 * o1[1] + rn8 * o1[2])
pitch = pd.DataFrame(np.arctan(rn7 / rn8))
roll = pd.DataFrame(np.arcsin(-rn6))
yaw = pd.DataFrame(np.arctan(rn3 / rn0))
ori = pd.concat((o_x, o_y, o_z, pitch, roll, yaw), axis=1)
print(ori.shape)
# 输出格式为['o_x', 'o_y', 'o_z', 'pitch', 'roll', 'yaw']
# -----------------------------------------------------------------------------------------------
# 对m_x,m_y,m_z取平方和之后开根号,作为新的列值
magnetic = np.asarray([data['m_x'], data['m_y'], data['m_z']]).T
ma = np.sqrt(np.square(magnetic[:, 0]) + np.square(magnetic[:, 1]) + np.square(magnetic[:, 2]))
magnetic = pd.DataFrame(magnetic)
print("magnetic", magnetic.shape)
ma_t = pd.DataFrame(ma)
print(ma_t.shape)
ma = pd.concat((ma_t, magnetic), axis=1)
# 输出格式为['ma','m_x', 'm_y', 'm_z']
# -----------------------------------------------------------------------------------------------
remain = pd.DataFrame(np.asarray([data['gy_x'], data['gy_y'], data['gy_z'],
data['g_x'], data['g_x'], data['g_x'],
data['l_x'], data['l_x'], data['l_x'],
data['pressure']
]).T)
fin = pd.concat((acc, acc_xy, acc_xyz, ori, ma, remain, label), axis=1)
print(fin.shape)
fin.to_csv("raw_data.csv", index=False, header=['time', 'acc_x', 'acc_y', 'acc_z', 'acc_xy', 'acc_xyz',
'o_x', 'o_y', 'o_z', 'pitch', 'roll', 'yaw',
'magnetic', 'm_x', 'm_y', 'm_z',
'gy_x', 'gy_y', 'gy_z',
'g_x', 'g_y', 'g_z',
'l_x', 'l_y', 'l_z',
'pressure',
'label'])
if __name__ == '__main__':
train()