07-01-2014, 04:31 PM
Exploiting Ground-Penetrating Radar Phenomenology in a Context-Dependent Framework
for Landmine Detection and Discrimination
Exploiting Ground-Penetrating.pdf (Size: 1.2 MB / Downloads: 25)
Abstract
A technique for making landmine detection with a
ground-penetrating radar (GPR) sensor more robust to fluctua-
tions in environmental conditions is presented. Context-dependent
feature selection (CDFS) counteracts environmental uncertain-
ties that degrade detection and discrimination performances by
modifying decision rules based on inference of the environmental
context. This paper utilized both physics-based and statistical
methods for extracting features from GPR data to characterize
surface texture and subsurface electrical properties, and a non-
parametric hypothesis test was used to identify the environmental
context from which the data were collected. The results of proba-
bilistic context identification were then used to fuse an ensemble of
classifiers for discriminating landmines from clutter under diverse
environmental conditions. CDFS was evaluated on a large set of
GPR data collected over several years in different weather and
terrain conditions. Results indicate that our context-dependent
technique improved landmine discrimination performance over
conventional fusion of several currently fielded algorithms from
the recent literature.
INTRODUCTION
DETECTION of landmines is a daunting task for militaries
in times of war and peace alike. Since landmines are
often relics of past conflicts, the vast majority of victims are
civilians, and many are children. The number of casualties
due to landmines is not precisely known but is estimated to
range from 15 000 to 20 000 per year [1]. In recent years, most
landmine-related casualties have been confined to the Middle
East and South Asia but have also been reported from Africa,
Latin America, and eastern Europe.
Large-scale demining operations are very costly and time
consuming, and route clearance patrols may be continuously
under the threat of enemy attack.
NIITEK Radar System
Data were collected for this paper with a vehicle-mounted
GPR system manufactured by NIITEK, Inc. (pictured in Fig. 1)
[39]. The GPR system under consideration in this paper consists
of a 51-channel ultrawideband (200 MHz–7 GHz) downward-
looking antenna array mounted on the front of a manned
vehicle. The 51 antennas are spaced 5 cm apart in a lin-
ear array across the front of the vehicle (referred to as the
crosstrack direction). The antennas are pulsed sequentially,
with each adjacent pair operating as a bistatic radar system.
As the vehicle proceeds forward (referred to as the downtrack
direction), measurements are collected from each channel ev-
ery 5 cm.
CONCLUSION
A context-dependent framework for landmine detection and
discrimination has been developed to exploit GPR phenomenol-
ogy using a feature-based context-identification technique.
Context-identification features were first extracted from raw
GPR data and subsequently used to probabilistically infer the
soil/moisture context of prescreener alarms. Results of MAP
context identification illustrated that the physics-based tech-
nique succeeds in identifying soil/moisture contexts of pre-
screener alarms.