Full text: Fusion of sensor data, knowledge sources and algorithms for extraction and classification of topographic objects

International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 7-4-3 W6, Valladolid, Spain, 3-4 June, 1999 
Dietmar Kunz 
Institute of Photogrammetry and Remote Sensing, University of Karlsruhe, Englerstr. 7, D-76128 Karlsruhe, Germany 
KEYWORDS: Image Analysis, ATKIS-DLM 200, Symbolic Scene Description, Disjoint Topographic Objects, Semantic 
Classification, Semantic Network 
The aim of this approach is to obtain an improved and automated satellite image analysis with regard to the determination of landuse 
classes by means of specific knowledge, which is represented in digital topographic databases. Because of the relatively poor 
geometric resolution of common satellite sensors (e.g. Landsat TM with 30 m) this project concentrates on main area-based landuse 
classes like ’settlement’, forest’, ’water’ or ’agriculture’. By using integrated knowledge processing (data fusion) in feature extraction, 
building of disjoint geometric objects and a classification process, the resulting semantic scene description is more accurate than with 
classical methods of satellite image analysis. The pixel-oriented satellite image and the vector-oriented topographic database are two 
different sources of knowledge. Both present a description of the same scene, but from different perspectives. For the comparison of 
the two different knowledge bases, it is necessary to introduce methods on higher symbolic level. The results of feature extraction 
supply this symbolic description of the satellite image, while the topographic database is already on this higher level. By means of an 
intersection of both geometric scene descriptions from segmentation and topographic database, we obtain a new unambiguous scene 
description with disjoint objects. The classification process is performed in a semantic network, which is able to process general and 
specific knowledge about the topographic and disjoint objects. The use of a semantic network for satellite image analysis is new. 
After the classification, disjoint objects with the same semantic meaning and a common border are merged. The result is a complete 
semantic description of the scene. Experiences with the represented knowledge based analysis process confirm the efficiency of this 
concept by a better separability of object classes. The determination of a good valuation function of spectral as well as non-spectral 
features in the decision process that is performed in the semantic network has however proved to be a very complex and time 
consuming task. 
The main emphasis of this paper is placed on the 
investigation and use of synergy effects between satellite 
image analysis and digital topographic database. Since 
several years many countries all over the world have been 
building-up digital topographic databases for planning, 
modelling and visualization purposes. In Germany, the 
surveying administration has designed and created the 
topographic database ATKIS (Authoritative Topographic 
Cartographic Information System) (AdV, 1989). ATKIS is 
divided into three different landscape models DLM25, 
DLM200 and DLM1000, which contain mainly the objects 
of the official topographical maps at scales 1:25,000, 
1:200,000 and 1:1,000,000 respectively. For updating such a 
given topographic database, the detail of image information 
should be higher. Because of common satellite sensors with 
resolution of approximately 10m to 30m, this project 
concentrates on the DLM200 map information. The use of 
such topographic databases can support and improve the 
satellite image analysis, not only for a knowledge based 
feature extraction, but also in the following process of 
semantic modelling of the topographic objects contained in 
the scene. 
Fig. 1. Flowchart of the analysis process. 
real world 
digital database satellite image 

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.