The focus of this project was to design and demonstrate an application for processing and analyzing electromagnetic induction (EMI) for the purpose of successfully classifying military munitions that are buried in the ground and not visually identifiable. The primary objective was to create a software environment and capabilities for analyzing data acquired by advanced EMI sensors that is transparent, modular, and commercially available, and demonstrate the software’s capabilities.

Technology Description

UX-Analyze provides the means and methods to analyze EMI sensor data for the purposes of classifying unexploded ordnance. It is transparent, modular, and commercially available and can be used for dynamic and static survey data. At a high level, data from classification-grade EMI sensors are first characterized by means of a solver, then classified based on quantitative model parameters associated with the characterization.

UX-Analyze is embedded into Oasis montaj and activated using their commercial licensing scheme. Data handling, processing, analysis, and documentation tools are included. Specific capabilities include (1) support for, and inversion of, multi-coil, multi-axis EMI data, (2) a modularized workflow, (3) single and multi-source solvers, (4) quality control checks and standard products, and (5) online help & documentation. Customized Graphical User Interfaces (GUIs) streamline the inversion and classification phases. Once processed, the analyst utilizes an information rich, analysis environment to review the decision rationale.

Demonstration Results

The primary technical performance metric of correctly classifying all target of interest (TOI) was achieved. The clutter rejection rate at the analysts’ threshold was 86%, which easily surpassed the desired rejection rate of 50%. Less than 1% of the anomalies were classified as Cannot Analyze, which also passed the desired benchmark. The XY offsets for TOI had a standard deviation of 0.17m. The XY offsets for all sources, regardless of TOI or not, was 0.36m. The performance metric objective of 0.15m for the horizontal standard deviation or less was not achieved.

Implementation Issues

The project did not run into any major implementation issues. Although major changes have been made to modularize the code and simplify the process, the workflow presents many options during analysis and is still perhaps a bit too verbose. Most of these relate to making the machine work harder so that the analyst can focus on assessment of the decisions and data quality. As design-lead of the UX-Analyze software suite, the project team wants to strike a balance between providing options versus unnecessary complications.