ioetz
>ti-
(b)
1sed
age
A
11
ar-
zes
ad eur-
oped
sre-
hoto-
hing
But from an interpreters point of view an image should present, apart
from optimum contrast, also:
- clarity of patters;
- relief, or no effects of relief, depending on application;
- information from different seasons;
- correct or specifically coded radiometric densities;
- correct geometry, etc.
Pre-processing may aim at optimizing a given image so that these require-
ments for interpretation are met. The interpreter can then work qualitatively
with images, but many encounter difficulties in various types of quantitative
Work: The eye can differentiate many tens of different gray of color tones
on & gray or color wedge; but this may reduce to only 1o - 20 when the gray-
tones or colours are chaotically mixed, as is the case in remote sensing
images. Digital recording and analysis offers abundant capabilities in this
respect: a system with 8 bits storage per picture element (pixal) permits
to operate quantitatively with 256 gray (color) tones.
An entire new perspective opens up in the area of radiometric and geo-
metric manipulations. Digital approaches improve the flexibility of geome-
tric corrections that have been previously possible, e.g. by photogrammetric
rectification and orthophoto techniques, however, with limitations by opti-
cal and mechanical designs. Radiometric corrections, on the other hand,
were hardly possible at all until the advent of image processing methods.
These, however, have led to impressive capabilities to combine data from
different seasons (Blanc et al., 1977), different sources (Anuta et al.,
1976; Dailey et al., 1978) and permit to control perturbating effects of
the sensing systems (Soha et al., 1977),and of the relief (Kahle et Ale)
1977).
Data compression of multiple images is another important image pre-pro-
cessing step. Digital techniques offer flexibility of applying mathematical .
and statistical methods for this task.
Visual image interpretation is seen as opposed to automatic analysis.
In fact, however, the two may complement or overlap each other in an inter-
active type of operation so that automation would only support some of the
visual interpretation. Such approaches are presently emerging in a wide
variety of interactive hardware and generally involve a digital component.
Digital pre-processing of aerial photography is very uncommon. Experienced
photo-interpreters argue that there is no increase in the diagnostic value
of photography and that apart from dodging no pre-processing techniques are
needed. These considerations, and the effort required for digitization as
well as the potential loss of resolution that could occur, have so far pre-
vented the use of existing image processing capabilities for this purpose.
3. PRE-PROCESSING OF SINGLE IMAGES
According to Anuta (1977) image pre-processing serves to "rectify sen-
sor problems, placing the data in the form desired for analysis, and to
combine, transform or transmit the data". One may groups and discuss these
pre-processing techniques in seven categories, resulting from a sub-division
of categories as proposed by Goetz et al. (1975) and Pratt (1978) (compare
section 1.):