In the formulas for s x and s y some minor changes have been made in the last term under the square
root sign in comparison with previous publication of the formulas in reference 1.342:1 upon suggestion
of Dr. D. Harkin, GIMRADA Ft Belvoir Ya USA.
In previous investigations and tests of the formulas (see ref. 1.342:1) the basic standard errors of unit
weight s' 0 p and s' oc were assumed to be equal because of insufficient investigation of their magnitude
under operational conditions. In the research reported above, particular attention has been paid to the
determination of these two factors. See 1.62, 3.2 and 3.3. As shown in fig. 1.62:9 the factor s' oc varies
considerably with location in the image and this fact should be taken into account in the derivation of
the formulas for the final accuracy. So far, however, there have been too few investigations of this
fundamental factor. Therefore, a mean value of the standard error will be used.
The ratio / , > " c will be computed for values which may be expected in practice.
S op
In the actual tests where the measurements have been made in a stereocomparator the standard errors
of unit weight of the y-parallaxes which will be used are shown in table 3.3:1. The mean value 5.8 ^m
(which is representative for the area where the control and check points are located) will be used as the
standard error of unit weight of the image coordinates.
Table 4.4:1. Tests of the theoretical accuracy of analytical photogrammetry. Aerial photography: Aviogon c=153 mm.
Oland test field. Flying altitude 1200 m. Five control points in each model.
Camera
s ol5
«m
s m
«m
Root mean square values of
discrepancies in check points,
mm in the terrain.
Theoretical root mean square values according
to s ol5 and confidence limits (5 % level),
mm in the terrain.
X
y
z
X
y
z
Ag3
4.1
3.1
40
48
74
30< 41< 72
31< 43< 76
42< 59< 104
Ag29
4.4
3.3
52
46
87
31< 44< 78
33< 46< 81
45< 63< 111
Ag38
5.2
3.1
64
67
70
36< 52< 91
39< 55< 96
52< 75< 131
Ag41
4.9
4.5
53
54
90
34< 49< 87
36< 52< 91
49< 71< 124
Agl96
6.7
4.3
53
53
125
47< 66< 121
49< 70< 128
67< 96< 173
s G15 is the standard error of unit weight of y-parallaxes from 15 points and s m is the root mean square
value of residual y-parallaxes in other points. Regular errors of the image coordinates according to 1.42
and with respect to fiducial marks were corrected in the computing program. The true values of the root
mean square discrepancies are all well within the confidence limits and the theoretical derivation of the
accuracy appears to be confirmed.
Results from two of the models, shown in table 4.4:1, were selected to show the consequences of no
correction for regular errors of the image coordinates. See table 4.4:2.
Table 4.4:2. Analytical photogrammetry applied to two models without correction for regular errors of image coordinates.
Rms-values of discrepancies
Corresponding theoretical values according to
s ol5
V
in
check points.
s ol5
and confidence limits.
//III
urn
mm in the terrain
mm in the terrain
X
y
Z
X
y
z
Ag3
5.6
5.3
139
155
134
39< 56< 101
41< 59 <107
56< 81 <145
Ag38
9.7
6.9
138
124
103
68< 97< 165
71< 102< 185
98< 140< 251
The regular errors of the image coordinates cause a clear decrease of accuracy of the final coordinates.
The confidence limits of the theoretical values are in some cases exceeded.