390
PRELIMINARY REMARKS AND POSING THE PROBLEM
In the FRG do exist about 65,000 sheets of the German Basic
map 1:5,000. They are printed in two colours, black for the
planimetry representation and the characters and sepia-
brown for the contour-lines. For each sheet in the modern
edition we have three cartographic originals, one for the
planimetry representation, one for the characters and one
for the contour-lines.
In some areas, especially in city areas, we can get digital
geometric data by the coordinates of the cadastral boundary
points and of the house corners, which are stored for our
digital cadastral map. But in the larger part of the FRG we
have no topographic data in digital form. So we have to
digitize existing topographic map originals.
For the quantity of map sheets at the scale 1 :5,000 exclu
sive manual digitization has no future. At that time we
would prefer a combination of manual and automatic digi
tization. So it is the topic of our investigations to look
for practical solutions with existing hard- and software.
A well-known problem of the application of automatic digi
tization by means of a raster-scanner is the subsequent
conversion from raster-to-vector - if vector data are
desired - and the feature-tagging. It was our main objective
to perform these processes as much as possible in batch
processing. But if there are economic or time-consuming
advantages in a manual or interactive procedure, we would
like to integrate it.
Because of the great importance for several purposes there
do exist a great number of software systems (partly hard
ware-systems too) for the automatic raster digitization of
contour-line-originals and the subsequent raster-vector
conversion. So it was not a problem to scan the contour-
line-original of our test sheet of the German Basic map at
the scale 1:5,000 and to convert the data from raster to
vector form. We used the ARLIP- System (Woetzel 1977).
With this software-system developed by the Gesellschaft fur
Mathematik und Datenverarbeitung (GMD) - Bonn, FRG - some
years ago, a workable program has been made available for
topological skeletting and line- and node-extraction from
binary images. Experiments made with feature-se1ected
cartographic originals turned out quite promising. A very
important restriction is, however, that different qualities
of lines and areas are not recognised automatically. For
this reason atuomatic digitizing by means of raster-
scanners is confined to so-called feature-se1ected graphic
representations which are graphic scanning originals
containing only one line quality (e.g. only contour-lines
or only boundary-lines of the same quality). This restric
tion is a great handicap for the general usage for carto
graphic scanning originals because - apart from a few
exceptions (e.g. contour-line-original) - cartographic