'ABLE OF
€ exists gross
in the object
ted with. The
tions. It is not
OSS errors in a
issumed that at
The following
(16)
sis.
(17)
calar. Z, isa
oss error. The
sung that there
vation.
, the following
othesis (Koch,
(18)
J
. is a co-factor
a square of
s the sum of a
»ss error when
S à gross error,
rejected (it is
fies the testing
ted for a gross
vork reliability
n), it is judged
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004
6. EXPERIMENTS
6.1 Experimental environment
To examine (1) precision, (2) sensitivity and (3) reliability for a
typical slope measurement, a slope model of 1,100mm wide and
S00m high was made. This was about 1/30 scale of a typical
construction site of a slope along a highway. Figure 2 shows
one of images exposed in the following experiment displayed
on the authors’ measurement system named “SUBARU”.
A total of 34targets of 5 mm in diameter were placed at 200mm
intervals. This interval is a standard one between anchor bolts
on road and tunnel slopes. Three points are placed on a board
at the central part, displacement of which is controllable. The
other 31 points were stationary. A fixture in the right centre is a
special device for automatic orientation (Hattori 2002).
It is usually sufficient if a 2 mm displacement is detected over a
30 m in width. Thus, the object coordinate precision of
0.050mm is envisaged to be at least necessary as practical to
satisfy the other criteria. The experiments were conducted for
the case of no displacements and for the case where the board
was shifted by 0.050mm in the Y direction (downward in
Figure 3). The camera used was a Nikon D100 (3K X 2K
pixels) with a 20 mm lens. A total of 20 images were taken.
Four were taken at cach of five stations at a distance of 1,000
mm, as shown in Figure 1. The four images were taken by
rotating the camera by 90 degrees around the optical axis. This
set of images is called Set 20. The set without displacements is
called Set 20 00, while one with displacements is called Set
20 50. Similarly the set of 12 images taken at three stations, i.e.
left, center and right is called Set 12. And the set of eight
images exposed at two stations, i.e. left and right is called Set 8.
Magnitude of the displacement is denoted by 00 and 50 in the
same way. There were no differences in lighting or other
physical conditions in these sets.
The most probable values of interior orientation parameters and
their variance-covariance matrices used through experiments
are values obtained for simultaneous adjustment of Set 20 00
and Set 20 50.
6.2 Precision of coordinate measurements
The precision (standard deviation) obtained from the adjustment
calculations for Sets 20 00, 12 00 and 8 00 are shown in Table
l. The increasing tendency of standard deviation almost
satisfies the error propagation law. The standard deviation for
the image coordinates of the targets was about 0.0003 mm.
However, an empirical value would be about 0.0005 mm even
under good photo-taking conditions such as exposure, etc.,
based on the authors’ experiences. Thus, it is thought difficult
to achieve higher precision from the actual measurements.
Furthermore, an ideal value obtained from the self-calibration
was used for the prior calibration value of the Camera’s internal
orientation element. Thus, the results shown below should be
discounted and regarded to be about two to three times higher.
However, even so, Set 8 would ensure 1 mm precision at real
scale.
Lynne 500 1000 15.1500 “2000 2500 3000
firareots <{ >| target 7 2e] Mod j zum [camera D100, lens 20 mm
som Of: com of CT Hn Manual] BEBE DU | 3/0]
ER Lx [:9] 1 J[99 [22 esectif Kouhoi Zentrod Zenpod. d [ exce
XoYo.2z0| -535.-140.
Figure 2. Model a slope displayed on a screen of the
measurement system "SUBARU"
Table 1. Measurement precision of object coordinates for
three exposure configurations
Set 20 00 Set12 00 Set 8 00
X (mm) 0.0050 0.0006 0.0078
Y (mm) 0.0048 0.0058 0.0072
Z (mm) 0.0148 0.0163 0.0179
Average (mm) 0.0095 0.0106 0.0120
Table 2. Measurement movement of the three points
(Set20 00 and Set20. 50)
101 102 103
X (mm) -0.0054 -0.0080 -0.0019
Y (mm) 0.0340 0.0438 0.0353
Z (mm) 0.0042 0.0053 0.0127
Table 3. F-test for displacement monitoring
(Set 8 00- Set 8. 50)
Z T, (9574) Sy ô for T (20%)
101 8.93 2.61 2.29 3.3
101.102 10.08 2.10 3.18 37
101,102,103 6.04 1.89 2.45 4.0
(Set 12. 00- Set 12. 50)
T T, (9596) À ô for n (20%)
101 8.93 2.61 2.99 3.3
101,102 10.08 2.10 3.18 3.7
101,102,103 9.54 1.89 3.09 4.0
(Set 20. 00- Set 20. 50)
T T, (9594) So ô for T, (20%)
101 13.1 2.61 3.62 3.3
101,102 16.4 2.10 4.05 3.7
101,102,103 | 14.7 1.89 3.84 4.0