163
4. Maximising the usable image in-formation density
The usable image information density is defined as the
interpretable information quantity stored per unit area on
a remote sensing data carrier; it is differentiated accor
ding to the geometric resolution (or spatial frequency,
radiometric resolution (grey scale, temperature steps
etc.), and spectral resolution (bandwidth per channel). In
photographic systems, the maximum is given by the number of
tonal density levels attainable per film area and by the
maximum spatial frequency in resolved lines per mi 11imetre;
it is related to an optimization of exposure, among other
factors /7/.
The resolving power of modern aerial films with their
greatly reduced emulsion thicknesses is only slightly con
trast-dependent. Consequently, full utilization of the
potential blackness or density range in the image should be
aimed at. This means about 1.6 density units in the linear
part of the characteristic curve of ORWO VF-45 film or
KODAK PANATOMIC X-Aerographic II contrary to the former
constant limitation to 0.85.
This extension of the density range, in combination with
the details explained in the preceding chapter, improved
differential exposure metering and the use of microcompu
ters on the basis of improved computing algorithms, should
be capable of maximizing the image information density
recorded with photographic systems.
The processes of aerial film development must also be
considered in this context, because the above measures call
for a film development to a current nominal gradient which
must be controllable within a film. This requires a modi
fied continuous developing machine in which the travelling
speed of the film is controlled automatically by a data
carrier that is scanned synchronously with the film and may
be contained on the film itself /2/.
5. Necessary preparation of scanner data
Practical tests with push-broom scanners by means of visual
navigation and without autopilot equipment in small
aeroplanes have shown that the position and attitude
accuracy as well as position stability are not sufficient.
Considerable geometric errors occur.
The motion-conditioned distortion within one line is
negligible, but between neighbouring lines it is imperative
to take it into account because of the occurrence of
scanninq gaps (Fig. 3), in particular when a small exposure
time is chosen in relation to the line cycle frequency in
favour of a small pixel bluring.
Overlapping or multiple scanning as shown in the example in
Figure 3, can be corrected afterwards, but missing object
details with dimensions smaller than or equal to the
respective scanning gap cannot be regained in the
subsequent image processing, which means that an on-line
compensation procedure has to be applied at least for that.
The overlaps and gaps remain then sufficiently small.
