-
Notifications
You must be signed in to change notification settings - Fork 0
/
calibrate.backup.py
252 lines (204 loc) · 7.98 KB
/
calibrate.backup.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
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
import numpy as np
import pandas as pd
import cv2
import os
board_w = 8 # horizontal enclosed corners on chessboard
board_h = 6 # vertical enclosed corners on chessboard
square = 2.74
# prepare object points, like (0,0,0), (1,0,0), (2,0,0) ....,(board_w-1,board_h-1,0)
objp = np.zeros((board_h*board_w, 3), np.float32)
objp[:, :2] = np.mgrid[0:board_w, 0:board_h].T.reshape(-1, 2)*square
# Arrays to store object points and image points from all the images.
object_points = [] # 3d point in real world space
imgL_points = [] # 2d points in image plane.
imgR_points = [] # 2d points in image plane.
R = None
t = None
distance = 0
fs_read = cv2.FileStorage('./exp-0/Intrinsics.xml', cv2.FILE_STORAGE_READ)
intrinsic = fs_read.getNode('Intrinsics').mat()
fs_read.release()
fs_read = cv2.FileStorage('./exp-0/Distortion.xml', cv2.FILE_STORAGE_READ)
distCoeff = fs_read.getNode('DistCoeffs').mat()
fs_read.release()
count = 0
capture = cv2.VideoCapture(0)
capture.set(cv2.CAP_PROP_FPS, 15)
capture.set(3, 640)
capture.set(4, 360)
cv2.namedWindow('Raw')
cv2.namedWindow("Undistorted")
raw = {
"isMeasuring": False,
"p1": np.asarray([-1, -1]),
"p2": np.asarray([-1, -1])
}
undistorted = {
"isMeasuring": False,
"p1": np.asarray([-1, -1]),
"p2": np.asarray([-1, -1])
}
saving = False
side = "None"
goodResult = False
criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 30, 0.001)
def mouse_callback(event, column, line, flags, params):
if event == 1: # left button in my mac
params["isMeasuring"] = not params["isMeasuring"]
if (params["isMeasuring"]):
# first point
params["p1"] = np.asarray([column, line])
params["p2"] = np.asarray([-1, -1])
else:
# second point
p1 = params["p1"]
p2 = np.asarray([column, line])
params["p2"] = p2
def drawLine(img, data, color):
p1 = data["p1"]
p2 = data["p2"]
if (p2[0] > 0):
p1_3D = project3D(p1)
p2_3D = project3D(p2)
dist = np.linalg.norm(p2-p1)
dist3D = np.linalg.norm(p2_3D-p1_3D)
if img.size <= 640 * 480:
img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
cv2.line(img, tuple(p1), tuple(p2), color, 2)
print("p1, p2: {}, {}".format(p1, p2))
h, w, c = img.shape
cv2.putText(img, "{} pixels".format(dist), (10, h-20),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 1, cv2.LINE_AA)
cv2.putText(img, "{} cm".format(dist3D), (10, h-40),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 1, cv2.LINE_AA)
print("dist:{}".format(dist))
print("dist3D:{}".format(dist3D))
return img
def calcP():
P = np.zeros((4, 4), np.float32)
K = np.zeros((3, 4), np.float32)
K[:, :-1] = intrinsic
if not distance == 0:
zeros = np.zeros(4)
zeros[-1] = 1
Rt = np.hstack((R, t))
Rt = np.vstack((Rt, zeros))
P = K @ Rt
return P
def project3D(point):
print('**** point', point)
cam = np.ones(3) # x
cam[:-1] = point
P = calcP()
P = np.delete(P, 2, 1) # delete 3rd column
W = np.linalg.inv(P) @ cam
W = W / W[2]
W[2] = 0
# X = X/w, Y = Y/w, Z = 0, w not needed
return W
def calibrateStereo(imgL, imgR):
image_size = imgL.shape
find_chessboard_flags = cv2.CALIB_CB_ADAPTIVE_THRESH | cv2.CALIB_CB_NORMALIZE_IMAGE | cv2.CALIB_CB_FAST_CHECK
left_found, left_corners = cv2.findChessboardCorners(
imgL, (board_w, board_h), flags=find_chessboard_flags)
right_found, right_corners = cv2.findChessboardCorners(
imgR, (board_w, board_h), flags=find_chessboard_flags)
if left_found:
cv2.cornerSubPix(left_img, left_corners, (11,11), (-1,-1), criteria)
if right_found:
cv2.cornerSubPix(right_img, right_corners, (11,11), (-1,-1), criteria)
if left_found and right_found:
imgL_points.append(left_corners)
imgR_points.append(right_corners)
object_points.append(objp)
stereocalib_criteria = (cv2.TERM_CRITERIA_MAX_ITER +
cv2.TERM_CRITERIA_EPS, 100, 1e-5)
stereocalib_flags = cv2.CALIB_FIX_ASPECT_RATIO | cv2.CALIB_ZERO_TANGENT_DIST | cv2.CALIB_SAME_FOCAL_LENGTH | cv2.CALIB_RATIONAL_MODEL | cv2.CALIB_FIX_K3 | cv2.CALIB_FIX_K4 | cv2.CALIB_FIX_K5
ret, cameraMatrix1, distCoeffs1, cameraMatrix2, distCoeffs2, R, T, E, F = cv2.stereoCalibrate(
object_points, imgL_points, imgR_points, image_size, criteria=stereocalib_criteria, flags=stereocalib_flags)
def calculateExtrinsics(image):
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
global object_points
global R
global t
global distance
global goodResult
found, corners = cv2.findChessboardCorners(
image, (board_w, board_h), cv2.CALIB_CB_ADAPTIVE_THRESH + cv2.CALIB_CB_NORMALIZE_IMAGE)
# If found corners, refine
if found == True:
corners = cv2.cornerSubPix(
image, corners, (11, 11), (-1, -1), criteria)
if (corners.shape[0] == 48):
goodResult = True
# cv2.solvePnPRansac(objectPoints, imagePoints, cameraMatrix, distCoeffs[, rvec[, tvec[, useExtrinsicGuess[, iterationsCount[, reprojectionError[, minInliersCount[, inliers[, flags]]]]]]]])
# ret, r, t, inliners = cv2.solvePnPRansac(object_points, corners, intrinsic, distCoeff,None, None, True, 500, )
ret, rvec, t = cv2.solvePnP(object_points, corners, intrinsic,
distCoeff, None, None, False, cv2.SOLVEPNP_ITERATIVE)
R, j = cv2.Rodrigues(rvec)
C = np.matmul(np.linalg.inv(-R), t)
distance = np.linalg.norm(C)
# distance = np.linalg.norm(t)
image = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)
image = cv2.drawChessboardCorners(
image, (board_w, board_h), corners, found)
else:
image = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)
return image
while(capture.isOpened()):
_, image = capture.read()
image = cv2.flip(image, 1) # mirrors image
h, w = image.shape[:2]
newcameraintrinsic, roi = cv2.getOptimalNewCameraMatrix(
intrinsic, distCoeff, (w, h), 1, (w, h))
#undistort
color = image
color = cv2.undistort(color, intrinsic, distCoeff, None, newcameraintrinsic)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
dst = cv2.undistort(image, intrinsic, distCoeff, None, newcameraintrinsic)
# # crop the image
# x,y,w,h = roi
# dst = dst[y:y+h, x:x+w]
cv2.setMouseCallback('Raw', mouse_callback, raw)
cv2.setMouseCallback('Undistorted', mouse_callback, undistorted)
image = drawLine(image, raw, (33, 255, 33))
cv2.imshow('Raw', image)
dst = calculateExtrinsics(dst)
dst = drawLine(dst, undistorted, (255, 33, 255))
if (dst.size > 640*480):
h, w, c = dst.shape
else:
print('erro')
h, w = dst.shape
dst = cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)
cv2.putText(dst, "Distance:{} cm".format(distance), (w-200, 20),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 1, cv2.LINE_AA)
cv2.imshow('Undistorted', dst)
k = cv2.waitKey(60) & 0xFF
if k==27: # Esc key to stop
break
if k == ord("l"): # l -> save left
count = 0
saving = True
side = "LEFT"
if k == ord("r"):
count = 0
saving = True
side = "RIGHT"
if k == 32: #space => stop saving
saving = False
if saving and goodResult:
count += 1
goodResult = False
filename = './results/extr-{}-{}-gray-undistorted.png'.format(side, count)
cv2.imwrite(filename, dst)
filename = './results/extr-{}-{}-color-undistorted.png'.format(side, count)
cv2.imwrite(filename, color)
fs_write = cv2.FileStorage(
'./results/Extrinsics-{}-{}.xml'.format(side, count), cv2.FILE_STORAGE_WRITE)
fs_write.write('R', R)
fs_write.write('t', t)
fs_write.write('distance', distance)
fs_write.release()
capture.release()
cv2.destroyAllWindows()