Feature-based UXO Detection and Discrimination

Abstract : With current technology and survey practices, unexploded ordnance (UXO) site characterization is inefficient and incomplete. Not all buried UXO are routinely detected, and those that are cannot be routinely distinguished from other items in the ground that pose no risk (e.g., scrap metal objects and other artifacts), which leads to an enormous amount of expensive digging. Over the past decade, DoD has invested heavily in developing survey data analysis and processing techniques for use with commercial sensors that can improve UXO detection and discrimination between UXO and clutter. These techniques include characterization procedures for estimating target features from survey data (size, shape, depth of burial, orientation, etc.) and feature-based classification procedures to aid decision making. Our technical approach promotes the selection of potential UXO targets using quantitative evaluation criteria and transparent decision-making processes. As such, we developed UX-Analyze, an analysis framework within Oasis montaj (trademark) that integrates quantitative analysis algorithms and custom-designed visualization schemes. The analysis algorithms assume a dipolar source and derive the best set of induced dipole model parameters that account for the spatial variation of the signal as the sensor is moved over the object. The model parameters are target location and depth, three dipole response coefficients corresponding to the principal axes of the target (electromagnetic induction [EMI] only), and the three angles that describe the orientation of the target. The source's size can be estimated using empirical relationships between either the dipole moment for magnetic data or the sum of the targets' response coefficients. The objective of this technology demonstration was to discriminate 4.2 inch mortars from native clutter at Camp Sibert, Alabama, by characterizing and classifying anomalies identified in EMI and magnetic survey data.