-
Notifications
You must be signed in to change notification settings - Fork 0
/
FillSmoothC3D.py
237 lines (182 loc) · 5.88 KB
/
FillSmoothC3D.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
236
237
from ezc3d import c3d
import numpy as np
import matplotlib.pyplot as plt
from pykalman import KalmanFilter
import sys, os
#
# I make no appologies for using hard tabs to indent. %$*# the Python style guide,
# hard-tabs are the _right_ approach.
#
#
# Function to plot a track
#
def Plot( points, resids ):
x = points[0,:,:]
y = points[1,:,:]
z = points[2,:,:]
w = resids[0,:,:]
print(x.shape)
for jc in range( x.shape[0] ):
plt.subplot(4,1,1)
plt.plot( x[jc,:] )
plt.subplot(4,1,2)
plt.plot( y[jc,:] )
plt.subplot(4,1,3)
plt.plot( z[jc,:] )
plt.subplot(4,1,4)
plt.plot( w[jc,:] )
plt.show()
#
# Function to run a Kalman smoother given noise estimates.
#
def KalmanSmoothTrack( trk, tNoise, mNoise ):
tm = np.array( [ [1,0,0, 1,0,0, 1,0,0],
[0,1,0, 0,1,0, 0,1,0],
[0,0,1, 0,0,1, 0,0,1],
[0,0,0, 1,0,0, 1,0,0],
[0,0,0, 0,1,0, 0,1,0],
[0,0,0, 0,0,1, 0,0,1],
[0,0,0, 0,0,0, 1,0,0],
[0,0,0, 0,0,0, 0,1,0],
[0,0,0, 0,0,0, 0,0,1] ] )
iState = [ trk[0,0],
trk[0,1],
trk[0,2],
trk[1,0] - trk[0,0],
trk[1,1] - trk[0,1],
trk[1,2] - trk[0,2],
0,
0,
0 ]
kf = None
if tNoise < 0 and mNoise < 0:
kf = KalmanFilter(transition_matrices=tm,
em_vars=['transition_covariance', 'observation_covariance'],
initial_state_mean = iState,
n_dim_obs=3)
kf.em(trk)
elif tNoise < 0 and mNoise >= 0:
mCov = mNoise * np.eye(3, dtype=np.float32)
kf = KalmanFilter(transition_matrices=tm,
em_vars=['transition_covariance'],
observation_covariance = mCov,
initial_state_mean = iState,
n_dim_obs=3)
kf.em(trk)
elif tNoise >= 0 and mNoise < 0:
tCov = tNoise * np.eye(9, dtype=np.float32)
kf = KalmanFilter(transition_matrices=tm,
em_vars=['observation_covariance'],
transition_covariance = tCov,
initial_state_mean = iState,
n_dim_obs=3)
kf.em(trk)
else:
tCov = tNoise * np.eye(9, dtype=np.float32)
mCov = mNoise * np.eye(3, dtype=np.float32)
kf = KalmanFilter(transition_matrices=tm,
transition_covariance = tCov,
observation_covariance = mCov,
initial_state_mean = iState,
n_dim_obs=3)
#print( kf.transition_covariance )
#print( kf.observation_covariance )
res = kf.smooth( trk )
print(res[0].shape)
#plt.plot(trk[:,2], label="trk")
#plt.plot(res[0][:,2], label="res")
#plt.legend()
#plt.show()
#exit(0)
return res[0]
#
# Linearly interpolate if we're missing values.
#
def HandleMissing( points, resids ):
print( points.shape, resids.shape )
npoints = points.copy()
nresids = resids.copy()
firstValid = 0
finalValid = points.shape[2]
for kpc in range( points.shape[1] ):
# find the first valid frame
minfc = 0
while resids[0, kpc, minfc] <= 0:
minfc += 1
firstValid = max(firstValid, minfc)
lastValid = minfc
invalid = False
for fc in range( minfc, points.shape[2] ):
if resids[0,kpc,fc] > 0:
if invalid:
for ifc in range( lastValid+1, fc ):
v = (ifc-lastValid) / (fc-lastValid)
for k in range(3):
npoints[k,kpc,ifc] = points[k, kpc, lastValid] + v * (points[k,kpc,fc] - points[k,kpc,lastValid])
npoints[3,kpc,ifc] = 1.0
nresids[0,kpc,ifc] = 1.0
lastValid = fc
invalid = False
else:
invalid = True
finalValid = min( finalValid, lastValid )
return (npoints, nresids, firstValid, finalValid )
def FindFiles(path):
print(f"[INFO] - Searching for files beneath: {path}")
retFiles = []
for root, dirs, files in os.walk(path):
for file in files:
if file.endswith(".c3d") and file.find("-filled") < 0 and file.find("-smoothed") < 0:
pth = os.path.join(root,file)
retFiles.append(pth)
return retFiles
if len(sys.argv) < 2:
print("This is a tool to fill, smooth and plot a .c3d file using a Kalman filter.")
print("Usage: ")
print(sys.argv[0], " True ")
print(sys.argv[0], " False <trans noise> <obs noise> < file00.c3d > [ file01.c3d] ... [ file##.c3d ] ")
print(" - or - ")
print("output will be to:")
print("file00.c3d -> (file00-filled.c3d, file00-smoothed.c3d)")
exit(0)
assert( sys.argv[1] == "True" or sys.argv[1] == "False" )
files = []
tNoise = 0.01
obsNoise = 15.0
if sys.argv[1] == "True":
import config
files = FindFiles( config.PATH )
tNoise = config.KALMAN_TRANS_NOISE
obsNoise = config.KALMAN_OBS_NOISE
else:
assert( len(sys.argv) > 4 )
tNoise = float( sys.argv[2] )
obsNoise = float( sys.argv[3] )
files = sys.argv[4:]
print(f"[INFO] - Got: {len(files)} files")
for f in files:
print(f"[INFO] - Processing: {f}")
track = c3d( f )
points = track['data']['points']
resids = track['data']['meta_points']['residuals']
(fpoints, fresids, firstValid, lastValid) = HandleMissing( points, resids )
track['data']['points'] = fpoints
track['data']['meta_points']['residuals'] = fresids
a = f.rfind('.c3d')
filledFilename = f[:a] + "-filled.c3d"
print("writing: ", filledFilename )
track.write( filledFilename )
kpoints = fpoints.copy()
for kpc in range( points.shape[1] ):
trk = kpoints[:3,kpc,firstValid:lastValid].transpose()
trkSmooth = KalmanSmoothTrack( trk, tNoise, obsNoise )
kpoints[:3,kpc,firstValid:lastValid] = trkSmooth.transpose()[:3,:]
smoothedFilename = f[:a] + "-smoothed.c3d"
track['data']['points'] = kpoints
track['data']['meta_points']['residuals'] = fresids
print("writing: ", smoothedFilename )
track.write( smoothedFilename )
#Plot( points, resids )
#Plot( fpoints, fresids )
#Plot( kpoints, fresids )
#exit(0)