Project Members:
Partner:
-
Prof. Dr. Andreas Kirsch, Universität Karlsruhe (TH), Germany
- Dr. Thomas Wonik
, Institut für Geowissenschaftliche Gemeinschaftsaufgaben (GGA), Hannover, Germany
- Dipl.-Geophys. Andeas Donat
, Universität zu Köln, Germany
- Prof. Dr. Hartmut Ewald
, Universität Rostock, Germany
- Dipl.-Phys. Hartmut Eigenbrod
, Fraunhofer-Gesellschaft (FhG), Stuttgart, Germany
Funding: BMBF, 754.000 € (of 4.200.000 € total)
Duration: 04/2004 - 03/2007
Project Description:
Anti-personnel land mines are one of the war legacies with the
most serious consequences for the population and industries of
countries which have been at war. Therefore the improvement of
metal detectors for humanitarian demining is still an area of
major interest for these countries. In this project,
special software applications for metal detectors were presented
which minimize the false alarm rate by using noise reduction
methods, and enable identification of mines by means of self-learning algorithms,
such as neural networks and vector supporting
machines (CI methods). The major advantage of these methods stems from the fact that
geologically and meteorologically based anomalies of the soil
structure can be identified as non-relevant information,
incomplete data sets can be completed, and system parameters,
hidden up to now, can be identified. As our investigations
demonstrate, our system allows to calculate the position and depth of the mines,
identify mine-like structures, and deal with
effects caused by the soil structure.
Contact E-Mail:
matthias.reuter [at] tu-clausthal.de
