44 
control points. The scales of the individual models are only approximate- 
ly determined, for instance, with the aid of the flying altitude and the 
calibrated focal length of the camera. Model coordinates are determined 
in transfer points which must be well identified in adjacent models. It 
is very suitable to use artificial transfer points as was indicated above 
(point 1.21) for stereo-radial triangulation. With the aid of simple 
coordinate transformations the individual models are then transformed 
into one and the same scale, for instance, the approximate scale of the 
first model. Then the entire strip is transformed into the coordinate 
system of the ground with the aid of at least two control points. 
The problems which are to be treated here are 
à. the error accumulation in cantilever extension of the triangulation 
b. the error distribution along the strip after bridging. 
2.1. The basic error expressions 
Evidently the errors of the coordinates x and y in the transfer points 
of the individual models are of basic importance for the investigations 
of the error accumulation and distribution. The errors of the coordinates 
x and y are assumed to depend mainly upon the errors of the relative 
orientation and the setting of the floating mark in the actual points. 
The relations between the errors dx and dy of the model coordinates and 
the errors of the elements of the relative orientation are wellknown. 
For dependent relative orientation we have (fig. 8) 
xy x (x — b) z { (x — b)2 + A 
de = — - p dre = dbz, + - bh dq, — 
x (x — b) 4 
M y / do, (148) 
dby, x—b y x l\y 
di — 2 d 72 eed b dx, — 3 — 2 j, dz, "^ 
[ (x — b? --»? x—b | y fu? +h  (x—b)y | dw, 
* | Ps t: 2 E dq. | o 7 5 | h (149) 
We assume two transfer points to be used as demonstrated in fig. 5. 
In the model à — 1,7 the points 4 and 6 have the coordinates 
Za c Ô qb 
(150) 
yı=d Yu=—d