STUDIES ON THE AUTOMATIC CONTROL OF AN AERIAL CAMERA
Prof.Dr. Karl-Heinz Schmelovsky, Dr.-Ing. Stephan Irmscher
and Dr.-Ing. Jochen Winkler
Akademie der Wissenschaften der DDR,Institut für Kosmos-
forschung,Berlin,GDR
Commission I/3
GEOMETRIC-OPTICAL SITUATION
In photographic flights, the pilot keeps the aeroplane at a
constant hight and flies along a predetermined route at a con-
stant speed, the influence of cross-wind being largely elimi-
nated by setting the longitudinal axis of the aircraft at an
angle relative to the flying direction. This angle of attack
is set by the pilot according to verbal feedback with the pho-
graphy navigator. This procedure cannot fully eliminate devi-
ations, which have to be compensated by rotatine the camera.
The situation is shown on the diagram (Fig. 1).
The measuring system required for such rotation consists of
two parallel CCD rows arranged at right angles with the direc-
tion of flight. The CCD rows scan the terrain stripwise through
an optical system. Row I captures a strip ahead of the plane,
row II one behind the plane. After the first cyclic exposure
of row I, row II is exposed repeatedly until it recognizes
the strip previously registered by row I. Until that recog-
nition a certain time has passed during which the plane may
have turned through the angle AK (yaw) so that the two sensor
images are shifted relative to one another, both angularly
and especially, linearly. The digitized brightness profiles,
filed in the memory of the control computer, can then be shif-
ted back realtive to each other until the best possible agreement
is achieved, the amount of shift « being a measure of the
angle AR . Recognition is impaired primarily by parallel shifts
of the plane's longitudinal axis and by the plane's longi-
tudinal axis and by the plane's roll. The error resulting
from the angular offset of the registered brightness pro-
files due to yaw can be kept small by frequent scanning and
optical smearing.
The second variable quantity to be registered is the speed-
-to-height ratio (v/h). It is required for setting the frame
rate so that all photos have the preset degree of overlap
and for the correct compensation of forward motion during
film exposure. Fig. 2 illustrates the process of determining
the v/h ratio. Assuming that row II is exposed at short time
intervals of At, n:At is the time that passes until row II
recognizes the brishtness profile registered by row I n steps
earlier. Then distance covered is ven-4t, and with a simple
proportional equation the dependence v/h = v/h (n^!) ean be
determined. Given the fixed row spacing d, the focal length
of the lens and the known scanning time At, we have
129
