forked from ElvinC/gyroflow
-
Notifications
You must be signed in to change notification settings - Fork 0
/
calibrate_video.py
1277 lines (916 loc) · 42.4 KB
/
calibrate_video.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
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
import json
from datetime import date
import numpy as np
import cv2
import glob
from _version import __version__
from tkinter import Tk
from tkinter.filedialog import askopenfilename
from scipy.spatial.transform import Rotation
import sys
# https://www.imatest.com/support/docs/pre-5-2/geometric-calibration/projective-camera
def inverse_cam_mtx(K):
# inverse for zero skew case
if K.shape != (3,3):
raise ValueError("Not 3x3 matrix")
fx = K[0,0]
fy = K[1,1]
px = K[0,2]
py = K[1,2]
Kinv = np.array([[fy, 0, -px*fy],
[0, fx, -py*fx],
[0, 0, fx*fy]])
Kinv /= fx * fy
return Kinv
class FisheyeCalibrator:
"""Class for calculating camera matrix and distortion coefficients
from images or videoframes
Mostly based on https://stackoverflow.com/a/50876130
9x6 chessboard by default: https://github.com/opencv/opencv/master/doc/pattern.png
"""
def __init__(self, chessboard_size=(9,6)):
self.chessboard_size = chessboard_size
# termination criteria
self.subpix_criteria = (cv2.TERM_CRITERIA_EPS +
cv2.TERM_CRITERIA_MAX_ITER, 30, 0.001)
self.calib_criteria = (cv2.TERM_CRITERIA_EPS +
cv2.TERM_CRITERIA_MAX_ITER, 30, 1e-6)
self.calibration_flags = (cv2.fisheye.CALIB_RECOMPUTE_EXTRINSIC + # cv2.fisheye.CALIB_CHECK_COND +
cv2.fisheye.CALIB_FIX_SKEW)
# prepare object points, like (0,0,0), (1,0,0), (2,0,0) ....,(6,5,0)
self.objp = np.zeros((chessboard_size[0] * chessboard_size[1],3), np.float32)
self.objp[:,:2] = np.mgrid[0:chessboard_size[0],0:chessboard_size[1]].T.reshape(-1,2)
# Arrays to store object points and image points from all the images.
self.objpoints = [] # 3d point in real world space
self.imgpoints = [] # 2d points in image plane.
# num images loaded
self.num_images = 0
# num images used in last calibration
self.num_images_used = 0
self.num_processed_images = 0
self.first_image_processed = False
# Equal if no stretching is applied
self.orig_dimension = np.array([0, 0])
self.calib_dimension = np.array([0, 0])
# K & D (camera matrix and and distortion coefficients)
self.K = np.zeros((3, 3))
self.D = np.zeros((4, 1))
# RMS error in pixels. Should be <1 after successful calibration
self.RMS_error = 100
# Horizontal stretching factor
self.input_horizontal_stretch = 1
self.data_from_preset_file = False
# when loading a preset file
self.extra_cam_info = None
def new_calibration(self):
self.objpoints = []
self.imgpoints = []
self.num_images = 0
self.num_images_used = 0
self.num_processed_images = 0
def set_horizontal_stretch(self, new_stretch = 1):
# For handling anamorphic or squeezed footage.
self.input_horizontal_stretch = new_stretch
def get_stretched_size_from_dimension(self, new_img_dim):
return (round(new_img_dim[0] * self.calib_dimension[0] / self.orig_dimension[0]), round(new_img_dim[1] * self.calib_dimension[1] / self.orig_dimension[1]))
def image_is_stretched(self):
return self.input_horizontal_stretch != 1
def get_stretched_size(self, img):
h, w = img.shape[:2]
if self.input_horizontal_stretch < 0:
new_h = round(h / self.input_horizontal_stretch)
new_w = w
else:
new_w = round(w * self.input_horizontal_stretch)
new_h = h
return (new_w, new_h)
def stretch_image(self, img):
if self.input_horizontal_stretch == 1:
return img
# 16:9 to 4:3 gives input_horizontal_stretch of
# (4/3)/(16/9) = 0.75
h, w = img.shape[:2]
if self.input_horizontal_stretch < 0:
new_h = round(h / self.input_horizontal_stretch)
new_w = w
else:
new_w = round(w * self.input_horizontal_stretch)
new_h = h
return cv2.resize(img, (new_w, new_h))
def add_calib_image(self, img):
"""Add chessboard image for calibration
Args:
img (np.ndarray): Image or video frame
Returns:
(bool, string, np.ndarray): (success, status message, corners)
"""
if self.data_from_preset_file:
raise Exception("Preset already loaded from file")
if self.num_images == 0:
# save the dimensions of the first image [width, height]
self.orig_dimension = img.shape[:2][::-1]
self.calib_dimension = self.get_stretched_size(img)
# check image dimension
if img.shape[:2][::-1] != self.orig_dimension:
return (False, "Image dimension doesn't match previous samples", None)
gray = cv2.cvtColor(self.stretch_image(img),cv2.COLOR_BGR2GRAY)
# Find the chess board corners
ret, corners = cv2.findChessboardCorners(gray, self.chessboard_size, None)
if not ret:
return (False, "Failed to detect chessboard", None)
# If found, add object points, image points (after refining them)
self.num_images += 1
self.objpoints.append(self.objp)
corners2 = cv2.cornerSubPix(gray,corners,(11,11),(-1,-1),self.subpix_criteria)
# print(corners2)
# horizontal scaling
self.imgpoints.append(corners2)
# Draw and display the corners
#img = cv2.drawChessboardCorners(img, (9,6), corners2,ret)
#scaled = cv2.resize(img, (960,720))
#cv2.imshow('img',scaled)
#cv2.waitKey(500)
corners_orig = np.copy(corners2)
if self.image_is_stretched():
# Transform back to original image format
corners_orig[:,:,0] *= self.orig_dimension[0] / self.calib_dimension[0] # x axis
corners_orig[:,:,1] *= self.orig_dimension[1] / self.calib_dimension[1] # y axis
return (True, "Image processed and added", corners_orig)
def remove_calib_image(self):
"""Remove last added calibration image
"""
if self.num_images > 0:
self.objpoints.pop(-1)
self.imgpoints.pop(-1)
self.num_images -= 1
def compute_calibration(self, center_camera=True):
"""Compute camera calibration from loaded images
Args:
center_camera (bool): center camera matrix after calib.
Raises:
Exception: No calibration frames/data
Returns:
float: Calibration RMS pixel error. <1 is great
"""
if self.num_images == 0:
raise Exception("No calibration data")
# recompute only if new images added
# if self.num_images_used == self.num_images:
# return self.RMS_error
num_corners = self.chessboard_size[0]*self.chessboard_size[1]
temp_objpoints = np.asarray(self.objpoints,dtype=np.float64)
temp_objpoints = np.reshape(self.objpoints, (self.num_images, 1, num_corners, 3))
temp_imgpoints = np.asarray(self.imgpoints,dtype=np.float64)
temp_imgpoints = np.reshape(self.imgpoints, (self.num_images, 1, num_corners, 2))
rvecs = [np.zeros((1, 1, 3), dtype=np.float64) for i in range(self.num_images)]
tvecs = [np.zeros((1, 1, 3), dtype=np.float64) for i in range(self.num_images)]
try:
retval, self.K, self.D, rvecs, tvecs = cv2.fisheye.calibrate(temp_objpoints,
temp_imgpoints,
self.calib_dimension,
self.K,
self.D,
rvecs,
tvecs,
self.calibration_flags,
self.calib_criteria)
except:
print("Error computing calibration, remove a frame and try again")
return 100
if center_camera:
self.K[0,2] = self.calib_dimension[0]/2
self.K[1,2] = self.calib_dimension[1]/2
self.RMS_error = retval
self.num_images_used = self.num_images
return self.RMS_error
def get_camera_matrix(self):
"""Get camera matrix from calibration
Returns:
np.ndarray: Camera matrix (K)
"""
self.compute_calibration()
return self.K
def get_inverse_camera_matrix(self):
self.compute_calibration()
return inverse_cam_mtx(self.K)
def get_distortion_coefficients(self):
"""Get distortion coefficients from calibration
Returns:
np.ndarray: distortion coefficients (D)
"""
self.compute_calibration()
return self.D
def get_rms_error(self):
"""Get the calibration rms error
Returns:
float: Calibration RMS pixel error. should be <1.
"""
return self.compute_calibration()
def undistort_image(self, img, fov_scale=1.0):
"""Undistort image using the fisheye camera model in OpenCV
Args:
img (np.ndarray): Input image
fov_scale (float, optional): Virtual camera focal length divider. Defaults to 1.
Returns:
np.ndarray: Undistorted image
"""
self.compute_calibration()
img_dim = img.shape[:2][::-1]
scaled_K = self.K * img_dim[0] / self.calib_dimension[0]
scaled_K[2][2] = 1.0
new_K = cv2.fisheye.estimateNewCameraMatrixForUndistortRectify(scaled_K, self.D,
img_dim, np.eye(3), fov_scale=fov_scale)
self.new_K = new_K
#print("FOV BEFORE: {}".format(scaled_K[0,0]))
#print("FOV EFTER: {}".format(new_K[0,0]))
map1, map2 = cv2.fisheye.initUndistortRectifyMap(scaled_K, self.D, np.eye(3), new_K, img_dim, cv2.CV_16SC2)
undistorted_image = cv2.remap(img, map1, map2, interpolation=cv2.INTER_LINEAR,
borderMode=cv2.BORDER_CONSTANT)
return undistorted_image
def get_maps(self, fov_scale = 1.0, output_dim = None, new_img_dim = None, update_new_K = True, quat = None, focalCenter = None, original_stretched = True):
"""Get undistortion maps
Args:
fov_scale (float, optional): Virtual camera focal length divider. Defaults to 1.
new_img_dim (tuple, optional): Dimension of new image
Returns:
(np.ndarray,np.ndarray): Undistortion maps
"""
if new_img_dim and self.image_is_stretched() and original_stretched:
# new_img_dim is dimension of unstretched image
new_img_dim = (round(new_img_dim[0] * self.calib_dimension[0] / self.orig_dimension[0]), round(new_img_dim[1] * self.calib_dimension[1] / self.orig_dimension[1]))
img_dim = new_img_dim if new_img_dim else self.calib_dimension
out_dim = output_dim if output_dim else self.calib_dimension
focalCenter = focalCenter if focalCenter is not None else np.array([self.calib_dimension[0]/2,self.calib_dimension[1]/2])
R = np.eye(3)
if type(quat) != type(None):
quat = quat.flatten()
#R = Rotation([-quat[1],-quat[2],quat[3],-quat[0]]).as_matrix()
R = Rotation([quat[1],quat[2],quat[3],quat[0]]).as_matrix()
R[[0,0,1,2],[1,2,0,0]] *=-1
#final_rotation = np.eye(3)
#final_rotation[0,0] = -1
#R = np.linalg.multi_dot([np.linalg.inv(final_rotation), R, final_rotation])
img_dim_ratio = img_dim[0] / self.calib_dimension[0]
scaled_K = self.K * img_dim_ratio
scaled_K[2][2] = 1.0
new_K = np.copy(self.K)
new_K[0][0] = new_K[0][0] * 1.0/fov_scale
new_K[1][1] = new_K[1][1] * 1.0/fov_scale
new_K[0][2] = (self.calib_dimension[0]/2 - focalCenter[0])* img_dim_ratio/fov_scale + out_dim[0]/2
new_K[1][2] = (self.calib_dimension[1]/2 - focalCenter[1])* img_dim_ratio/fov_scale + out_dim[1]/2
if update_new_K:
self.new_K = new_K
if original_stretched and self.image_is_stretched():
map1, map2 = cv2.fisheye.initUndistortRectifyMap(scaled_K, self.D, R, new_K, out_dim, cv2.CV_32FC1)
# Rescale input and convert to int for speed
map1, map2 = cv2.convertMaps(map1 * self.orig_dimension[0] / self.calib_dimension[0], map2 * self.orig_dimension[1] / self.calib_dimension[1], cv2.CV_16SC2)
return map1, map2
else:
map1, map2 = cv2.fisheye.initUndistortRectifyMap(scaled_K, self.D, R, new_K, out_dim, cv2.CV_16SC2)
return map1, map2
def undistort_points(self, distorted_points,new_img_dim = None):
img_dim = new_img_dim if new_img_dim else self.calib_dimension
scaled_K = self.K * img_dim[0] / self.calib_dimension[0]
scaled_K[2][2] = 1.0
pts = cv2.fisheye.undistortPoints(distorted_points, scaled_K, self.D, None, scaled_K)
return pts if type(pts) != type(None) else np.array([])
def decompose_homography(self, H, new_img_dim = None):
img_dim = new_img_dim if new_img_dim else self.calib_dimension
scaled_K = self.K * img_dim[0] / self.calib_dimension[0]
scaled_K[2][2] = 1.0
return cv2.decomposeHomographyMat(H, scaled_K)
def recover_pose(self, pts1, pts2, new_img_dim = None):
""" Find rotation matrices using epipolar geometry
Args:
pts1 (np.ndarray): Initial points
pts2 (np.ndarray): Resulting points
new_img_dim (tuple, optional): New image dimension. Defaults to None.
Returns:
[type]: [description]
"""
# https://answers.opencv.org/question/31421/opencv-3-essentialmatrix-and-recoverpose/
img_dim = new_img_dim if new_img_dim else self.calib_dimension
scaled_K = self.K * img_dim[0] / self.calib_dimension[0]
scaled_K[2][2] = 1.0
E, mask = cv2.findEssentialMat(pts1, pts2, scaled_K, cv2.RANSAC, 0.999, 0.1) # cv2.LMEDS or cv2.RANSAC
#retval, R, t, mask = cv2.recoverPose(E, pts1, pts2, scaled_K)
try:
R1, R2, t = cv2.decomposeEssentialMat(E)
except:
# Can't figure it out, assume no rotation
return np.eye(3), np.eye(3), np.array([0,0,0])
return R1, R2, t
def get_rotation_map(self, img, quat):
"""Get maps for doing perspective rotations
WORK IN PROGRESS. Currently for testing
"""
# https://stackoverflow.com/a/12293128
# https://en.wikipedia.org/wiki/Homography_(computer_vision)
rotXval = 0
rotYval = 0
rotZval = 0
rotX = (rotXval)*np.pi/180
rotY = (rotYval)*np.pi/180
rotZ = (rotZval)*np.pi/180
rot_mat = np.eye(4)
#print(Rotation([quat[0,1],quat[0,2],quat[0,3],quat[0,0]]).as_euler('xyz'))
quat = quat.flatten()
eul = Rotation([quat[1],quat[2],quat[3],quat[0]]).as_euler('xyz')
combined_rotation = np.eye(4)
#combined_rotation[0:3,0:3] = Rotation.from_euler('xyz', [eul[0], eul[1], -eul[2]], degrees=False).as_matrix()
combined_rotation[0:3,0:3] = Rotation([-quat[1],-quat[2],quat[3],-quat[0]]).as_matrix()
rot_mat = combined_rotation
img_dim = img.shape[:2][::-1]
# Scaled 3x4 camera matrix
K = np.zeros((3,4))
K[:3,:3] = self.K
# should make the rotation match fov change
K[0,0] = self.new_K[0,0]
K[1,1] = self.new_K[1,1]
K *= img_dim[0] / self.calib_dimension[0]
K[2][2] = 1.0
# compute inverse camera matrix using scaled K
Kinv = np.zeros((4,3))
Kinv[0:3,0:3] = inverse_cam_mtx(K[:3,:3])
Kinv[3,:] = [0, 0, 1]
H = np.linalg.multi_dot([K, rot_mat, Kinv])
#transform = self.K * trans
outimg = cv2.warpPerspective(img,H,(img.shape[1],img.shape[0]))
return outimg
def get_calibration_data(self):
calibration_data = {
"name": self.extra_cam_info.get("name", ""),
"note": self.extra_cam_info.get("note", ""),
"calibrated_by": self.extra_cam_info.get("calibrated_by", "N/A"),
"camera_brand": self.extra_cam_info.get("camera_brand", "N/A"),
"camera_model": self.extra_cam_info.get("camera_model", "N/A"),
"lens_model": self.extra_cam_info.get("lens_model", "N/A"),
"camera_setting": self.extra_cam_info.get("camera_setting", "N/A"),
"calibrator_version": self.extra_cam_info.get("calibrator_version", "N/A"),
"date": self.extra_cam_info.get("date", "N/A"),
"calib_dimension": {
"w": self.calib_dimension[0],
"h": self.calib_dimension[1]
},
"orig_dimension": {
"w": self.orig_dimension[0],
"h": self.orig_dimension[1]
},
"input_horizontal_stretch": self.input_horizontal_stretch, # to de-stretch anamorphic/linearly stretched video.
"num_images": self.num_images_used,
"use_opencv_fisheye": True,
"fisheye_params": {
"RMS_error": self.RMS_error,
"camera_matrix": self.K.tolist(),
"distortion_coeffs": self.D.flatten().tolist()
},
# For (potential) use with the standard cv2.calibrateCamera
"use_opencv_standard": False,
"calib_params": {}
}
return calibration_data
def load_calibration_data(self, cal_data, printinfo = False):
try:
if not cal_data["use_opencv_fisheye"]:
raise Exception("Preset not for OpenCV fisheye lens model")
self.data_from_preset_file = True
if printinfo:
print("Preset name: {}".format(cal_data["name"]))
print("Note: {}".format(cal_data["note"]))
print("Made with {} frames using calibrator version {} on date {}"
.format(cal_data["num_images"],
cal_data["calibrator_version"],
cal_data["date"]))
if cal_data["calibrator_version"] != __version__:
print("Note: Versions don't match. Calibrator: {}, Preset: {}. Should be fine though."
.format(__version__, cal_data["calibrator_version"]))
cal_width = cal_data["calib_dimension"]["w"]
cal_height = cal_data["calib_dimension"]["h"]
self.calib_dimension = (cal_width, cal_height)
# Added in 0.3.0
if "orig_dimension" in cal_data:
orig_w = cal_data["orig_dimension"]["w"]
orig_h = cal_data["orig_dimension"]["h"]
self.input_horizontal_stretch = cal_data["input_horizontal_stretch"]
self.orig_dimension = (orig_w, orig_h)
else:
self.input_horizontal_stretch = 1
self.orig_dimension = self.calib_dimension
self.num_images = self.num_images_used = cal_data["num_images"]
self.RMS_error = cal_data["fisheye_params"]["RMS_error"]
self.K = np.array(cal_data["fisheye_params"]["camera_matrix"])
self.D = np.array(cal_data["fisheye_params"]["distortion_coeffs"])
#if presets["calibrator_version"].split(".")[0:1] != ["0","1"]:
# version 0.1.x doesn't have cam information
fixed_name = " ".join(cal_data.get("name").replace("_", " ").split())
self.extra_cam_info = {
"name": fixed_name,
"note": cal_data.get("note"),
"calibrated_by": cal_data.get("calibrated_by", "N/A"),
"camera_brand": cal_data.get("camera_brand", "N/A"),
"camera_model": cal_data.get("camera_model", "N/A"),
"camera_setting": cal_data.get("camera_setting", "N/A"),
"lens_model": cal_data.get("lens_model", "N/A"),
"calibrator_version": cal_data.get("calibrator_version"),
"date": cal_data.get("date"),
"width": self.orig_dimension[0],
"height": self.orig_dimension[1],
"aspect": self.orig_dimension[0]/self.orig_dimension[1],
"num_images": self.num_images
}
except ZeroDivisionError:
raise KeyError("Error loading preset file")
return self.extra_cam_info
def save_calibration_json(self, filename="calibration.json", calib_name="Camera name", camera_brand="", camera_model="", lens_model="", camera_setting="", note="", calibrated_by=""):
"""Save camera calibration parameters as JSON file
Args:
filename (str, optional): Path and name of file. Defaults to "calibration.json".
calib_name (str, optional): Calibration name in file. Defaults to "Camera name".
note (str, optional): Extra note, calibration setup, calibrator name etc.
"""
self.compute_calibration()
calibration_data = {
"name": calib_name,
"note": note,
"calibrated_by": calibrated_by,
"camera_brand": camera_brand,
"camera_model": camera_model,
"lens_model": lens_model,
"camera_setting": camera_setting,
"calibrator_version": __version__,
"date": str(date.today()),
"calib_dimension": {
"w": self.calib_dimension[0],
"h": self.calib_dimension[1]
},
"orig_dimension": {
"w": self.orig_dimension[0],
"h": self.orig_dimension[1]
},
"input_horizontal_stretch": self.input_horizontal_stretch, # to de-stretch anamorphic/linearly stretched video.
"num_images": self.num_images_used,
"use_opencv_fisheye": True,
"fisheye_params": {
"RMS_error": self.RMS_error,
"camera_matrix": self.K.tolist(),
"distortion_coeffs": self.D.flatten().tolist()
},
# For (potential) use with the standard cv2.calibrateCamera
"use_opencv_standard": False,
"calib_params": {}
}
with open(filename, 'w') as outfile:
json.dump(
calibration_data,
outfile,
indent=4,
separators=(',', ': ')
)
def load_calibration_json(self, filename, printinfo = False):
"""Load calibration preset from JSON file
Args:
filename (string): path and filename to load
printinfo (bool, optional): Print extra info from preset file. Defaults to False.
"""
with open(filename, "r") as infile:
presets = json.load(infile)
return self.load_calibration_data(presets, printinfo)
def load_calibration_prompt(self, printinfo = False):
"""Trigger file browser to load calibration preset
Args:
printinfo (bool, optional): Print extra info from preset file. Defaults to False.
"""
Tk().withdraw() # hide root window
# file browser prompt
filename = askopenfilename(title = "Select calibration preset file",
filetypes = (("JSON files","*.json"),))
self.load_calibration_json(filename, printinfo)
def undistort_image_prompt(self, fov_scale=1):
"""Trigger file browser to load and undistort image
Args:
fov_scale (float, optional): Virtual camera focal length divider. Defaults to 1.
"""
self.compute_calibration()
Tk().withdraw()
filename = askopenfilename(title = "Select image to undistort",
filetypes = (("jpeg images","*.jpg"),("png images","*.png")))
raw_img = cv2.imread(filename)
undistorted_img = self.undistort_image(raw_img, fov_scale=1)
for i in range(5):
rotated_img = self.get_rotation_map(undistorted_img,30)
scaled = cv2.resize(rotated_img, (960,720))
cv2.imshow('OpenCV image viewer',scaled)
cv2.waitKey(500)
rotated_img = self.get_rotation_map(undistorted_img,0)
scaled = cv2.resize(rotated_img, (960,720))
cv2.imshow('OpenCV image viewer',scaled)
cv2.waitKey(500)
class StandardCalibrator:
"""Class for calculating camera matrix and distortion coefficients
from images or videoframes
Mostly based on https://stackoverflow.com/a/50876130
9x6 chessboard by default: https://github.com/opencv/opencv/master/doc/pattern.png
"""
def __init__(self, chessboard_size=(9,6)):
self.chessboard_size = chessboard_size
# termination criteria
self.subpix_criteria = (cv2.TERM_CRITERIA_EPS +
cv2.TERM_CRITERIA_MAX_ITER, 30, 0.001)
self.calib_criteria = (cv2.TERM_CRITERIA_EPS +
cv2.TERM_CRITERIA_MAX_ITER, 30, 1e-6)
self.calibration_flags = (cv2.CALIB_SAME_FOCAL_LENGTH +
cv2.CALIB_RATIONAL_MODEL +
cv2.CALIB_FIX_PRINCIPAL_POINT +
cv2.CALIB_USE_INTRINSIC_GUESS +
cv2.CALIB_TILTED_MODEL)
# prepare object points, like (0,0,0), (1,0,0), (2,0,0) ....,(6,5,0)
self.objp = np.zeros((chessboard_size[0] * chessboard_size[1],3), np.float32)
self.objp[:,:2] = np.mgrid[0:chessboard_size[0],0:chessboard_size[1]].T.reshape(-1,2)
# Arrays to store object points and image points from all the images.
self.objpoints = [] # 3d point in real world space
self.imgpoints = [] # 2d points in image plane.
# num images loaded
self.num_images = 0
# num images used in last calibration
self.num_images_used = 0
self.first_image_processed = False
self.calib_dimension = np.array([0, 0])
# K & D (camera matrix and and distortion coefficients)
self.K = np.zeros((3, 3))
self.D = np.zeros((4, 1))
# RMS error in pixels. Should be <1 after successful calibration
self.RMS_error = 100
self.data_from_preset_file = False
def add_calib_image(self, img):
"""Add chessboard image for calibration
Args:
img (np.ndarray): Image or video frame
Returns:
(bool, string, np.ndarray): (success, status message, corners)
"""
if self.data_from_preset_file:
raise Exception("Preset already loaded from file")
if not self.first_image_processed:
# save the dimensions of the first image [width, height]
self.calib_dimension = img.shape[:2][::-1]
# check image dimension
if img.shape[:2][::-1] != self.calib_dimension:
return (False, "Image dimension doesn't match previous samples", None)
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
# Find the chess board corners
ret, corners = cv2.findChessboardCorners(gray, self.chessboard_size, None)
if not ret:
return (False, "Failed to detect chessboard", None)
# If found, add object points, image points (after refining them)
self.num_images += 1
self.objpoints.append(self.objp)
corners2 = cv2.cornerSubPix(gray,corners,(11,11),(-1,-1),self.subpix_criteria)
self.imgpoints.append(corners2)
# Draw and display the corners
#img = cv2.drawChessboardCorners(img, (9,6), corners2,ret)
#scaled = cv2.resize(img, (960,720))
#cv2.imshow('img',scaled)
#cv2.waitKey(500)
return (True, "Image processed and added", corners2)
def remove_calib_image(self):
"""Remove last added calibration image
"""
if self.num_images > 0:
self.objpoints.pop(-1)
self.imgpoints.pop(-1)
self.num_images -= 1
def compute_calibration(self, center_camera=True):
"""Compute camera calibration from loaded images
Args:
center_camera (bool): center camera matrix after calib.
Raises:
Exception: No calibration frames/data
Returns:
float: Calibration RMS pixel error. <1 is great
"""
if self.num_images == 0:
raise Exception("No calibration data")
# recompute only if new images added
# if self.num_images_used == self.num_images:
# return self.RMS_error
num_corners = self.chessboard_size[0]*self.chessboard_size[1]
temp_objpoints = np.asarray(self.objpoints,dtype=np.float64)
temp_objpoints = np.reshape(self.objpoints, (self.num_images, 1, num_corners, 3))
temp_imgpoints = np.asarray(self.imgpoints,dtype=np.float64)
temp_imgpoints = np.reshape(self.imgpoints, (self.num_images, 1, num_corners, 2))
rvecs = [np.zeros((1, 1, 3), dtype=np.float64) for i in range(self.num_images)]
tvecs = [np.zeros((1, 1, 3), dtype=np.float64) for i in range(self.num_images)]
self.K = np.array(
[
[
1000,
0.0,
self.calib_dimension[0]/2
],
[
0.0,
1000,
self.calib_dimension[1]/2
],
[
0.0,
0.0,
1.0
]
]
)
retval, self.K, self.D, rvecs, tvecs = cv2.calibrateCamera(temp_objpoints,
temp_imgpoints,
self.calib_dimension,
self.K,
self.D,
rvecs,
tvecs,
self.calibration_flags,
self.calib_criteria)
if center_camera:
self.K[0,2] = self.calib_dimension[0]/2
self.K[1,2] = self.calib_dimension[1]/2
self.RMS_error = retval
self.num_images_used = self.num_images
return self.RMS_error
def get_camera_matrix(self):
"""Get camera matrix from calibration
Returns:
np.ndarray: Camera matrix (K)
"""
self.compute_calibration()
return self.K
def get_inverse_camera_matrix(self):
self.compute_calibration
return inverse_cam_mtx(self.K)
def get_distortion_coefficients(self):
"""Get distortion coefficients from calibration
Returns:
np.ndarray: distortion coefficients (D)
"""
self.compute_calibration()
return self.D
def get_rms_error(self):
"""Get the calibration rms error
Returns:
float: Calibration RMS pixel error. should be <1.
"""
return self.compute_calibration()
def undistort_image(self, img, fov_scale=1.0):
"""Undistort image using the fisheye camera model in OpenCV
Args:
img (np.ndarray): Input image
fov_scale (float, optional): Virtual camera focal length divider. Defaults to 1.
Returns:
np.ndarray: Undistorted image
"""
self.compute_calibration()
img_dim = img.shape[:2][::-1]
scaled_K = self.K * img_dim[0] / self.calib_dimension[0]
scaled_K[2][2] = 1.0
new_K, _ = cv2.getOptimalNewCameraMatrix(scaled_K, self.D,
img_dim, 1.3, img_dim)
self.new_K = new_K
#print("FOV BEFORE: {}".format(scaled_K[0,0]))
#print("FOV EFTER: {}".format(new_K[0,0]))
map1, map2 = self.get_maps()
undistorted_image = cv2.remap(img, map1, map2, interpolation=cv2.INTER_LINEAR,
borderMode=cv2.BORDER_CONSTANT)
return undistorted_image
def get_maps(self, fov_scale = 1.0, new_img_dim = None):
"""Get undistortion maps
Args:
fov_scale (float, optional): Virtual camera focal length divider. Defaults to 1.
new_img_dim (tuple, optional): Dimension of new image
Returns:
(np.ndarray,np.ndarray): Undistortion maps
"""
img_dim = new_img_dim if new_img_dim else self.calib_dimension
scaled_K = self.K * img_dim[0] / self.calib_dimension[0]
scaled_K[2][2] = 1.0
new_K, _ = cv2.getOptimalNewCameraMatrix(scaled_K, self.D,
img_dim, 1.3, img_dim)
self.new_K = new_K
print(new_K)
map1, map2 = cv2.initUndistortRectifyMap(scaled_K, self.D, np.eye(3), new_K, img_dim, cv2.CV_16SC2)
return map1, map2
def undistort_points(self, distorted_points,new_img_dim = None):
img_dim = new_img_dim if new_img_dim else self.calib_dimension
scaled_K = self.K * img_dim[0] / self.calib_dimension[0]
scaled_K[2][2] = 1.0
return cv2.undistortPoints(distorted_points, scaled_K, self.D, None, scaled_K)
def decompose_homography(self, H, new_img_dim = None):
img_dim = new_img_dim if new_img_dim else self.calib_dimension
scaled_K = self.K * img_dim[0] / self.calib_dimension[0]
scaled_K[2][2] = 1.0
return cv2.decomposeHomographyMat(H, scaled_K)
def recover_pose(self, pts1, pts2, new_img_dim = None):
# https://answers.opencv.org/question/31421/opencv-3-essentialmatrix-and-recoverpose/
# Find essential matrix from fundamental matrix
img_dim = new_img_dim if new_img_dim else self.calib_dimension
scaled_K = self.K * img_dim[0] / self.calib_dimension[0]
scaled_K[2][2] = 1.0
E, mask = cv2.findEssentialMat(pts1, pts2, scaled_K, cv2.RANSAC, 0.999, 0.1) # cv2.LMEDS or cv2.RANSAC