Improving Uxo Detection With Four-Dimensional Matched Filters

Abstract

A newly developed three-component, transient-electromagnetic (TEM) system for UXO<br>detection and characterization continuously records full transients 64 times per second. Acquisition at a<br>walking speed of 1 m/sec generates three transients, each with 31 time windows, every 1.5 cm along<br>line. Such high-density over sampling provides an opportunity to improve signal-to-noise ratios with<br>post-acquisition filtering.<br>Tapered stacking and decimation by eight reduces background noise levels by a factor of three,<br>while producing a sampling density sufficient for UXO detection and characterization. For detection, a<br>matched filter derived from UXO modeling projects data onto a scalar profile suitable for target picking.<br>The filter’s spatial shape is based on the response of a compact spherical body given the TEM system’s<br>transmitter and receiver loop geometry. The filter’s temporal properties are matched to an expected<br>polarizability transient shape based on UXO characterization modeling. Using a matched filter tailored<br>to the expected spatial and temporal TEM response character over UXO has proved to be effective at<br>increasing UXO detectability relative to methods using only vertical-component amplitudes.