n of
the
the
ite
ion
1 be
of
es;
=e
on-
led
int
CCD
rs.
> of
88)
lly
ern
ga
ods
the
nce
and
ich
nts
er.
es,
on.
ect
sed
res
get
g &
ate
the
nes
ate
ing
the
For
ray
ted
of
rue
of
tes
als
ion
rue
| of
era
ith
tal
of
the camera lens was 25 mm. On each station, 3*3
frames of image were captured so as to cover the
whole target array (see the figure). Because
only one CCD camera mount was available, it had
to be mounted on the two theodolites in turn and
the environment kept stable in the meantime. The
accuracy in image capture and processing had
been checked before other operations and found
stable with little (short) warm-up effect.
= Pi m=
| t fx | —— Target
__—— CCD camera
— Theodolite
Theodolite scanning photogrammetry
The theodolite scanning photogrammetry started
with the theodolite orientation using the
horizontal reciprocal pointing method. The five-
point method was also used for comparison. The
camera was then mounted on the left theodolite
which was followed by the camera calibration
using the two-target camera-on-theodolite method
where the target to theodolite distances were
obtained through theodolite triangulation
integrated in the five-point orientation
process. 5*5*2 equivalent targets were generated
in this method. Two affine and one radial
distortion parameters were used. Immediately
after the calibration, the 3*3 frames of image
were captured as the theodolite was swung to the
3*3 pre-specified directions where the
theodolite readings were also recorded (as shown
in the figure). The same process was then
repeated with the right theodolite station.
Afterwards, the target images were located and
matched automatically by the image processing
software. The three dimensional coordinates of
the targets were then determined simply by
photogrammetric space intersection as all the
camera parameters needed had been derived from
the result of calibration and theodolite
orientation and the theodolite readings.
The accuracy of calibration alone was assessed
by using check point back ray tracing method and
found to be about 0.06 pixel in sample and 0.03
in line. The following tables present the
residuals in the transformation from the
determined coordinates to the true coordinates.
The RMS of the residuals at the bottom of the
tables are the measures of 3-D determination
accuracy of the system.
6. EXPERIENCE AND FUTURE WORK
As is seen in the tables, the relative accuracy
of the theodolite scanning photogrammetry with
3*3 frames of image is more than twice as high
as a single pair of image would achieve. As to
the theodolite orientation, the horizontal
reciprocal pointing method gave almost the same
accuracy as the five-point method in this
experiment. They are both easy to use and could
be automated. Testing with the possible longest
The accuracy of theodolite scanning
photogrammetry assessed with check points
Final Transformation Values:
Scale = 0.97961914
X Shift = -0.78éó mm
Y Shift 0.8056 mm
Z Shift = -1.2856 mm
X Rotation -0.00089 Gon
Y Rotation = 0.00229 Gon
Z Rotation = -0.84044 Gon
esiduals in mm
£ Point x Y z
1 K2H -0.033
2 C3H 0.
3 C3AH 0.
4 CD3
5 C4H 0
6 X11H -0.
7 X14H -0.
8 X21H -0.
9 X23H -0.
10 X24H -0.
11 X12H -0.
12 CSH 0.
18 L2H -0.
14 CDS a.
15 K3H -0.
16 D3H -0.
17 D4H -0. :
18 DES 0. .
19 DSH 0. :
20 X41H 0. 0. 0.
21 X42H -0. 0.3428 0.
22 X44H 0.2711 . -0.1222 -0.0941
23 D7H 0.1875 -0.1979 0.1207
24 L3H 0.0363 0.4035 -0.0919
25 E3H 0.3612 0.3687 0.0231
26 K4H -0.2794 -0.0702 -0.0482
27 EF3 0.3847 0.3115 -0.0655
28 E3AH 0.0301 -0.7128 -0.0224
29 F1H 0.0536 -0.8699 0.3725
30 J3H -0.2177 -0.2811 -0.2338
31 F2H -0.0073 -0.6748 0.1425
32 E4H 0.0773 0.0846 0.0348
32 EF4 -0.1400 0.0662 0.1251
34 EFS 0.0086 0.7348 0.0103
35 F3H 0.1351 0.6980 . -0.1747
86 J5H -0.2393 0.3747 -0.2765
37 ESH 0.0982 -0.1994 0.0408
38 EF6 -0.1276 -1.0336 0.1189
39 L4H -0.1633 0.2552 -0.1671
40 FSH 0.1533 -0.1432 0.1029
41 JéH -0.0338 0.6355 -0.5732
42 C4BH 0.2575 0.1718 -0.0087
43 DSH 0.1121 0.1600 0.0571
44 DES 0.1454 -0.7418 0.1903
45 E4BH 0.2272 0.5950 0.0174
0.1911 0.4820 0.1479
Total RMS = 0.5392 mm
END OF LOCAL TO OBJECT TRANSFORMATION
Final Transformation Values:
Scale = 0.99995179
X Shift = -1.1885 mm
Y Shift = 0.6615 mm
2 Shift = -2.0723 mm
X Rotation = -0.01098 Gon
Y Rotation = 0.00890 Gon
Z Rotation = 0.05627 Gon
esiduals in mm
£ Point X Y 2
1 K2H -0.0129 -0.2835 -0.2575
2 KSH -0.1621 0.1479 -0.3040
3 K4H 70.2571 -0.2240 -0.1733
4 C3H 0.1841 -0.7782 0.1004
5 CD3 0.1386 -0.6063 -0.0706
6 C4H 0.3143 0.4694 0.2141
7 CSH -0.1012 -0.1485 0.2353
8 CDS 0.0977 0.4266 0.2185
9 C3AH 0.3588 1.4840 0.3133
10 C4BH -0.0042 -0.3519 0.0995
11 D3H 0.0622 0.3977 0.0407
12 DES -0.1975 -0.6149 0.0664
13 DSH 0.1327 0.7174 -0.0483
14 DSH -0.1574 -0.2580 0.0031
15 D4H 0.1107 0.0844 0.1558
16 E3H 0.2232 0.0069 70.0317
17 EF3 -0.0160 -0.4773 0.1012
18 E4H 0.0977 0.0273 0.0717
19 EF4 -0.1156 -0.0863 0.1673
20 ESH 0.1662 0.1916 0.0699
21 E3AH 0.2171 0.1678 -0.0775
22 E4BH 0.0565 0.5388 0.0492
23 EFé -0.1734 -0.4841 0.1487
24 EFS -0.2475 70.1259 0.1619
25 F3H -0.0462 0.1396 70.1264
26 FSH -0.0704 71.0015 0.3871
27 F2H -0.0847 -0.2639 -0.1167
28 F4H -0.0762 -0.4510 0.1204
29 J3H 70.1276 -0.4928 -0.2958
30 J4H -0.2580 0.3279 -0.4141
31 JSH -0.070? 0.2502 -0.4413
32 JéH 70.2361 -0.0019 -0.3843
83 X12H 0.1722 0.7160 0.0972
34 X14H -0.2135 -0.2796 0.0581
35 X21H -0.0305 0.3641 0.0168
36 X23H 0.0605 0.2489 -0.0212
37 X41H 0.1920 0.0032 -0.0948
38 X42H -0.2553 0.5358 -0.0556
39 X44H 0.0048 -0.1790 -0.0962
40 X11H 0.0326 -0.7902 0.0269
41 X24H 0.0661 0.0184 0.1770
42 L4H 0.2267 0.6347 -0.0890
0.1644 0.4801 0.1871
Total RMS = 0.5409 mm
END OF LOCAL TO OBJECT TRANSFORMATION
focus lenses would reveal the practical accuracy
potential of the system. We have not, however,
been able to do this successfully because at the
time of experiment we did not have longer focus
lenses satisfactorily fixable to the theodolite
mount. This will become part of our future work.
