AUTOMATIC INTERPRETATION OF DIGITAL MAPS FOR DATA REVISION
Karl-Heinrich Anders and Dieter Fritsch
Institute of Photogrammetry
Stuttgart University
P.O.B. 106037
70049 Stuttgart / Germany
kh.anders@ifp.uni-stuttgart.de
Commission IV, Working Group 3
KEY WORDS: Revision, Updating, GIS, Spatial Data Interpretation, Semantic Modelling, Object Oriented
ABSTRACT:
The amount of spatial data in digital form increases continuously. Governments and companies capture digital spatial
data directly or convert existing analogous map data. The resulting large spatial databases have to be updated continu-
ously; they often include a variety of implicit information which are in most cases not used. Because of the huge efforts
in time and costs the automation of data revision is a topic of growing interest. The automatic interpretation of digital
maps or digital landscape models (DLM) makes it possible to deduce information that is not explicitly stored in the data
model. With this implicit information it is possible to support tasks like computer based revision of digital landscape
models, automatic production of thematic maps and automatic data generalization. This article will concentrate on data
revision based on aerial images, and the revision based on different digital maps. The used object-oriented data model
will be described in detail. Very often a DLM is represented by several data models which might be differ slightly from
each other. For this reason, the corresponding data models have to be converted into a uniform model. Furthermore, a
set of operators for the object recognition in digital maps will be described and how they can be used for data revision.
1 INTRODUCTION puter system has access only to the information about the
geometry of agriculture areas, ground plans of buildings
and text symbols. Therefore, methods have to be found
being able to deduce further information about spatial re-
lationships of stored objects. On the other hand, these
methods should also be able to create new objects from
the existing ones.
Nowadays there exists a huge amount of spatial data in
digital form. All over the world governments and compa-
nies capture spatial data in digital form or convert existing
analogous map data. In Germany, for example, the large
scale database ALK (official digital cadastral map of Ger-
many) and the medium scale database ATKIS (Authora-
tive Topographic Cartographic Information System) have
been initiated and realized. Besides these official basis in-
formation systems the car industry and related vendors
of information technology offer digital map data for traf-
fic management systems. All these information systems
are based on different models of the landscape, ie. they
include only special parts of the landscape, have differ-
ences in the generalization and accuracy of the captured
landscape phenomena. Because of the huge efforts in time
and costs the automation of data revision is a topic of
growing interest. It is important to note, that in the fol-
lowing we will refer to large digital databases and not deal
with scanned raster data. Approaches on the interpreta-
tion of digital raster maps are given by [Illert 1990], [Illert,
A. 1991], [Meng 1993] and [Carosio 1995].
"rums rn rum V s
aU M 3
N | À ZZ m T 1%)
ZZ »
BH
I {ih
SIRE aes,
a ier
5 à ru LE rag o
| ji
zer .
Ah
The automatic interpretation of digital maps or digital
landscape models allows to deduce information that is not
explicitly stored in the data model. For example a hu-
man operator who looks at a part of the digital cadastral
map (ALK) shown in figure 1 has no problems to recognize With this implicit information it should be possible to sup-
Figure 1: Example for the German ALK
roads, crossroads or areas of residential buildings or indus- port tasks like computer based revision of digital landscape
trial areas, although all these kinds of spatial objects are models, automatic production of thematic maps and auto-
not explicitly stored in the ALK. On the contrary a com- matic generalization. This article describes, how implicit
90
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
inf
ces
on
Th
est
tio:
she
on
Th
els
sto
tio
of
int
Th
