Our group has been using Ground
Penetrating Radar (GPR) for detecting and locating buried
objects. GPR is a non-invasive sensor and has been applied
for clearance of Unexploded Ordnance (UXO) buried under
permeable iron-rich volcanic soil in Hawaii [15].
Conventional magnetic sensors were found to produce very
poor detection performance in permeable environments due to
severe background noise caused by high permeability of such
soils. GPR systems, however, can provide desired features
even in permeable soils where other magnetic sensors (e.g.
EMI and magnetometers) could not function properly.
An ultra-wide bandwidth (UWB),
dual-polarized GPR system and an advanced classification
algorithms have been developed by our group (see Fig. 1a).
Based on initial test results, it is shown that the system
can provide target information such as position,
orientation, depth, length, linearity or inclination from
data collected at multiple passes over the target (see Fig.
1b). Also, a UWB GPR with proper processing algorithms can
reject a large amount of small metallic debris that presents
serious problems in classification and data analysis. Proper
(UWB, low-antenna clutter, multiple-positions) GPR data also
has superior spatial and temporal (or depth) resolutions
that allow one to determine the presence of interfering
objects or to separate the individual responses
We will continue to investigate the scattering
characteristics from various type of UXO under different
soil conditions and develop an effective UXO classification
method using GPR. Based on FDTD simulation results and
experimental measurements in a GPR test-bed and actual
fields, special scattering features discriminating UXO from
clutters can be extracted and applied to detection
algorithm. Effective post-processing algorithms enhancing
reliability of the classification will also be developed
using the artificial intelligent technologies. Eventually,
autonomous GPR UXO classification system, which minimizes
detection error and increases classification speed, will be
developed.