Science Service System

Summary of Proposal LAN0031

TitleApplication of Polarimetric SAR Data for the Verification of Geospatial Databases
Investigator Soergel, Uwe - Leibniz Universitšt Hannover, Germany, Institute of Photogrammetry and GeoInformation
Team Members
Dr. Lohmann, Peter - Leibniz Universitšt Hannover, Germany, Institute of Photogrammetry and GeoInformation
SummaryABSTRACT of the application for the fellowship: The verification of geospatial databases, which means in this context the comparison of the existing vector data to up-to-date imagery, is currently an interesting and widely discussed issue. The motivation for such verification is on the one hand the need for a quality management of this kind of data being used in many applications. On the other hand, in situations where the nature undergoes rapid changes (earthquake, flooding etc.) a quick comparison between the state before the disaster (i.e. the data from the geospatial database) and the new situation is required. Up to now a lot of approaches for the verification of geospatial databases using optical air- or spaceborne imagery exist. Most of these approaches are embedded into an interactive environment. The human operator is supported by the system: the task of the algorithms is to verify as much data as possible. But, as automatic image processing algorithms can not verify the whole dataset (due to insufficient data/classification etc.), the operator has to solve the remaining problems interactively. By this means the overall process of verification should be more efficient compared to a fully manual workflow. The application of active microwave systems (Side Looking Radar, SLR) for the verification of geospatial databases is only tackled marginally in literature, although it has some very interesting properties. SLR-Systems can observe through clouds, dust or smoke and they can be operated during night. One drawback of the SLR-data, more specific of the Synthetic Aperture Radar (SAR), is that human operators are normally not familiar with the interpretation of such data as the imaging principle differs fundamentally from optical systems. The goal of the work proposed in this application is to develop and assess methods for an automatic verification of GIS databases using polarimetric SAR data. Existing approaches for supervised and unsupervised classification of polarimetric SAR data are combined in order to achieve both: a reliable and efficient verification of the existing geospatial data. The following aspects are of special interest : a) mapping of the classes achieved by unsupervised classification to the respective object classes in the database, b) development of a statistically sound strategy to use the existing data to be verified as training data for supervised classification and c) develop an integrated verification scheme using the classified data. It is expected that the resulting approaches can easily be incorporated into existing assessment systems and thus the full range of air- and spaceborne sensors can be utilized for this issue which will surely receive more and more attention in the future. In order to emphasize on the growing importance of this topic also in the framework of international cooperations (for instance data harmonization in the EU, INSPIRE), the work will be carried out in an established French research institution (INRIA).
DATA REQUIREMENT: PolSAR-DATA, ground sample size <= 3m (TerraSAR-X StripMap Mode) Multitemp-data is not necessary, one scene per site is suitable so far.
FUNDING: Fellowship
Deliverable: Science Report

Back to list of proposals

© DLR 2004-2016