TABLE OF CONTENTS The following Section provides case studies on a variety of approaches to mine detection.
Dr Martin Fritzsche
This contribution discusses the application of ground penetrating radar sensors to the detection and classification of buried landmines. A novel detection and classification scheme is described, which combines image processing of horizontal plane views for object localisation with signal processing and pattern recognition techniques for object classification. Based on the modelling of transient scattering from spheres it is shown how internal object structure determines scattering characteristics. Various types of time-frequency transforms can thus be used to extract salient signal features and to associate each radar signal to one possible category, i.e. as being caused by a mine, by any type of nonhazardous object or simply by the soil background. The investigated techniques are applied to a large data set obtained in field trials and show quite promising results.
The first part of this Section deals with the modelling of transient scattering from spheres, buried in realistic soils. The intention is to illustrate how the characteristics of scattered signals are affected by the internal object structure as well as by soil parameters. In the secondpart, a novel detection algorithm is presented, which is based on the application of image processing techniques to horizontal plane views, obtained with a projection approach. Detections, thus obtained, serve as regions of interest in the subsequent feature extraction and classification steps, which are performed for...
