Summary
International Symposium on Antennas and Propagation
2006
Session Number:2B1b
Session:
Number:2B1b-6
A Three-Dimensional Feature Vector for Identification of Buried Landmines Using GPR Data
Masahiko Nishimoto, Yusuke Kimura, Takaaki Tanaka,
pp.1-5
Publication Date:2006/11/2
Online ISSN:2188-5079
Summary:
Identification of shallowly buried landmines using ground penetrating radar (GPR) data is studied. Three kinds of features for target identification are proposed: (a) time interval between two pulses reflected from top and bottom sides of the objects, (b) normalized waveform correlation, and (c) dispersion of arrival time of target responses. Since the identification considered here is reduced to a classification problem of a desired target and other clutter objects, a support vector machine (SVM) is employed as a classifier. In order to evaluate the identification performance, we carry out a Monte Carlo simulation using dataset generated by a twodimensional finite difference time domain (FDTD) method. The results show that good identification performance is obtained, and thus we can confirm that the proposed features are useful for discrimination of landmines from confusing clutter objects.