ResearchCurrent projects
Erstellung eines Feinkonzeptes für ein Verfahren zur automatischen Generierung und Aktualisierung von Geo-Daten hoher Auflösung aus Fernerkundungsdaten (1999)

Concept for automatic generation and update of high resolution geo-data with remote sensing imagery (1999)

Team:  Kian Pakzad
Year:  1999
Lifespan:  Januar 1999 – Oktober 1999
Is Finished:  yes

Cooperation with Uni BW München, ESG-GmbH
Research Groups: Automatic Image Analysis, Thematic Image Processing, Sensor Fusion

Background and goal
Aerial and satellite images are used for a long time to map the earth's surface and are meanwhile state of the art. The recently developed high resolution sensors enable the acquisition and update of more and more topographic objects from space. Because the need for GIS-Data is increasing efforts are made to automate the work of human operators and thereby to increase the productivity.
The goal of this project was to create an overview of methods for automatic and semi-automatic object extraction, to examine the potential of different sensors on this matter and to determine which features from the GIS feature catalogue could be extracted automatically.

Task und methods
Different parts of the task were examined: In order to put together and assess current approaches and methods for the automatic object extraction the current literature was reviewed.
Further the given feature catalogue has been gone through  and all features have been assessed regarding their visibility from the air, their recognizability and their potential for an automatic extraction. After that the visible features were examined by human operators in different sensors, regarding their detectablility and recognizability. In addition automatic methods were also examined

An overview about the state of the art of automated object extraction from high resolution remote sensing data was worked out. The overview shows that beyond the standard methods of multispectral classification there are several promising approaches for model based object extraction. On the other hand these approaches are still far away from practical use. The development of semi automatic methods seems also to be promising in order to raise the productivity of data acquisition and update.