1. Basic Considerations
The photographic image is still the most important tool for Remote Sensina,
and there is no reason today to expect a quick change. The 6 x 6 cm
format photography can be considered as a standard for many applications.
It directly provides high density information (especially when colour
film is used) for low cost. Scanner imagery needs specially equipped
planes and sophisticated digital processing to receive an image suitable
for further investigation. Consequently digital scanner imagery 1s about
ten times moreexpensive than analog conventional photoaraphy.
Conventional photography is ideally appropriated for visual interpre-
tation. Quantitative investigations can routinely be applied as far as
geometry is concerned; semantic information however can be extracted
visually rather than by "measurement". One reason for this situation is
the difficulty to describe the image grey values as a function of the
object characteristics properly. There exist too many factors which
influence the original information, for example illumination, air light,
instrument and film characteristics, as well as film development (see
J. SIEVERS /6/). Principally two different approaches are possible in
order to obtain nevertheless quantitative results from semantic image
analysis:
1. The physical properties of the mentioned disturbing factors, which
superimpose the original reflection from the object surface, are
analysed one by one, and their particular contribution to the regi-
stered grey values is taken for digital or analog data correction.
2. The physical composition of the data is more or less neglected, and a
statistical analysis of known surfaces ("training-sets") provide para-
meters for unknown objects in the image.
Approach No 2 is today widely applied and known as "Maximum - Likelihood -
Classification". The disadvantage of that method is the need of training -
sets, which often can not be defined in an appropriate manner (see
DENNERT-MULLER /2/). Approach No 1 up to now does not lead directly to
operational results for object classification. At the other hand, Maxi-
mum - Likelihood - Classification will yield optimal results only if
the data is rigorously preprocessed for compensation of physical disturb-
ing factors. Therefore the behavior of these factors is important, and
from that point of view many theoretical investiaations within that field
get a practical objective.
This contribution concentrates on a single physical phenomenon, the
light fall - off by a camera lens as well as its compensation using a
reflection standard, which fills the field of view of the camera
totally ("full - frame - reflector").
2. Analysis of the Problem
In August, 1974 pre - flights for the National German Aircraft - Program
took place at the Jade test site (North Sea Coast near Wilhelmshaven).
Coordinated by the Institut für Photogrammetrie und Ingenieurvermessungen
(IPI) of the Technical University of Hannover, HASSELBLAD MK 70 Colour
Photography was obtained from a prepared private CESSNA airplane, as well
as radiometer profiles from ship. At that time the objective of the
investigations was to find appropriate test fields for monitorina tidal
flow, sediment transport etc. as well as best suited spectral channels.