However, usually there is distortion in the photographs. In that case
freedom from vertical parallaxes is no prove that the relative orienta
tion of the two photographs is re-established. If the correct p-tilt
between the two photographs should be obtained it is necessary to in
sert vertical parallaxes, determined as above stated.
Determination of <p-convergence by adjusting a third photograph
Three photographs are orientated relatively and absolutely as is shown
in Fig. 4. The first stereo model is absolutely orientated and the third
photograph is orientated relatively to the second one, which gets a tilt
b
relative the horizon plane of the instrument of ^ . The nadir point N
gets a height error
b->
Ah = - — (5)
If the photographs have radial distortion the nadir directions of the
b
two photographs have a convergence of p + 2 da and the height
error in N 3 will be
b
Ah = -
b <¥
+ 2 da)
(6)
A h
b
or 2 da — —
b
— R
(7)
If A h is positive the lens has a distortion, which causes a ^-diver
gence, which is greater than, and of opposite sign to the effect of the
curvature of the earth.
Determination of <p-convergence in bridging
In bridging we wish to receive a constant scale and heights, which
correspond to the geodetic system.
In order to obtain this as close a possible the second photograph of
b
the preceding model must be given a correction in <p of — ( +2 da)
K.
before the next photograph is connected. This correction factor can be
determined as follows, if we have an elevation control point in the
strip.
The strip will be triangulated with the classical bridging method. If
the number of models or bases is n + 1 the height error, fh a , will
propagate over n bases: