Bruce Barrow

Subsurface discrimination using electromagnetic induction sensors

This paper reviews the problem of subsurface discrimination using electromagnetic induction (EMI) sensors. Typically, discrimination is based on differences in the multiaxis magnetic polarizability between different objects. They review work on frequency and time domain systems, and their interrelationship. They present the results of comprehensive measurements of the multiaxis EMI response of a variety of inert ordnance items, ordnance fragments, and scrap metal pieces recovered from firing ranges. The extent to which the distributions of the eigenvalues of magnetic polarizability for the different classes of objects do not overlap establishes an upper bound on discrimination. For various reasons, the eigenvalues cannot always be accurately determined using data collected above a buried target. This tends to increase the overlap of the distributions, and hence degrade discrimination performance.

Model-based characterization of electromagnetic induction signatures obtained with the MTADS electromagnetic array

A sensor response model based on empirically determined orthogonal response coefficients /spl beta/ is presented for the analysis of electromagnetic induction (EMI) sensor data. The model is applied to high-quality survey data from the multi-sensor towed array detection system (MTADS) program. Data from both a prepared test range and a live-site survey are analyzed. At the test field, the authors show that the model successfully reproduces the measured signatures of ordnance, ordnance simulants, flat plates, and clutter. At the live site, fit statistics were collected showing that for the most abundant ordnance target, 81-mm mortars, the three beta response parameters are distributed log-normally. This allows for a simple method of discrimination. A receiver operating characteristic (ROC) analysis based on this discrimination method is performed and the results discussed.

Time and Frequency Domain Electromagnetic Induction Signatures of Unexploded Ordnance

This paper documents some of the progress that has been made in recent years in the application of electromagnetic induction (EMI) technology to the problem of discriminating between buried unexploded ordnance and clutter items such as exploded ordnance fragments or scrap metal. EMI classification of an unknown subsurface target is based on estimating the targets magnetic polarizability tensor from data collected above the ground. One can discriminate between different targets only if their polarizability tensors are sufficiently different. In this paper, we review the relationship between the time and frequency dependence of the polarizability tensor, evaluate the relative information content of the time and frequency domain signatures, and discuss how the sensor response affects signature measurement fidelity. Our analysis centers on simple parametric representations for the time and frequency domain EMI signatures that have been synthesized from extensive empirical studies of the EMI response of ordnance and clutter items. There are three basic parameters (amplitude factor, time constant and demagnetization factor) for each of the targets principal axes. They are related to the physical characteristics of the target and can be used for classification and discrimination.

Source separation using sparse-solution linear solvers

An algorithm is proposed to enumerate, locate and characterize individual signal sources given observation of their combined signals. No a-priori estimate for the number of sources is required. We assume a forward model exists, and that superposition holds, i.e. coupling between sources is ignored. A system of linear equations y=Ax is set up in which columns of matrix A contain expected signals from a large number of hypothesized sources, and y contains the observed signal. Recently-developed solvers designed for linear systems with sparse non-negative solutions make this approach feasible even when large numbers of sources are involved. With each iteration, the collection of hypothesized sources is refined using a Harmony Search algorithm. Application is demonstrated on the problem of locating multiple buried conductors based on electromagnetic induction (EMI) signals observed at ground surface.

Characterization of a GEM-3 array for UXO classification

We have designed and built a non-synchronous, sequential array of GEM-3 sensors for use with the Multi-sensor Towed Array Detection System (MTADS) with support from ESTCP. The roughly 2-m square array consists of three, 96-cm diameter GEM-3s in a triangular configuration. The GEM drive electronics have been modified to produce a substantially higher transmit moment, and thus increased sensitivity, than the standard GEM-3. The individual sensors transmit a composite waveform made up of ten frequencies from 30 Hz to 48 kHz for a single 1/30 s base period. Sequential operation allows two of these base periods for deconvolution and output of the frequency-dependent response from each GEM-3. After allowing for a short coil settling time between sensors, we achieve an array sampling rate of just over 9 Hz. Coupled with our standard survey speed of 3 mph, this results in a down-track sampling spacing of ~15 cm. The cross-track spacing is 50 cm. We have characterized these sensors at our Blossom Point test site. The static and dynamic response of the array to a variety of ordnance, ordnance simulants, and scrap is presented with consideration given to both detection and classification.

Correcting GPS measurement errors induced by system motion over uneven terrain

Many cart- and vehicular-based UXO detection systems employ GPS receivers to accurately determine the systems position. However, the unevenness of the terrain often causes the system to tilt during the data collection, introducing errors in the GPS measurements. In this paper, two approaches are considered to correct the errors in the GPS measurements caused by the tilting of the system; low-pass filtering and adaptive filtering using a hidden Markov model (HMM). The low-pass filter smooths the data collection path recorded by the GPS receiver. Although this filter does not explicitly model the system motion, it does remove dramatic, and unrealistic, jumps in the GPS measurements. In contrast, the movement of the system can be explicitly modeled by an HMM. The HMM characterizes the cart motion so that the subsequent filtering is appropriate for the type of motion encountered. The error correction techniques are first applied to simulated data, in which both the sources of error and the ground truth are known so that the performance of the algorithms can be compared. The algorithms are then applied to measured data collected with a cart-based system to evaluate the robustness of their performance.

UXO and UXO Sensor Technology

Advanced electromagnetic induction sensors have been developed under the SERDP and ESTCP Munitions Response program to find and identify buried unexploded ordnance. These sensors consist of multiple transmit and receive coil configurations that collect sufficient data for inverting the time decaying, dipole polarizations of a buried metallic object. These polarizations can be used to identify the object as UXO or metallic debris. These sensor platforms have been deployed in both dynamic survey modes to locate and identify and stationary “cued” modes to just identify at target locations. ESTCP has sponsored a number of Live Site Demonstrations and these systems have been found to be very effective in finding UXO. An approach has been developed for advanced EMI survey data that applies a model based detection filter to locate metallic targets and a dipole inversion to identify the targets as UXO or clutter based on the inverted polarizations. Analysis of the Live Site data has found a significant fraction of the target locations have multiple target signals present. To address this, an N-dipole inversion is being applied to all target locations. This inversion inverts for a specified (N) number of targets. Fit results are returned for N = 1, 2, and 3 possible targets at each location. The problem arises that quite often all of the multi-target fits represent the data equally well. We will present these results and some strategies taken to insure that all possible targets of interest are selected from the multiple fit results. Work is also in progress to try and evaluate when the local target density is too high for valid analysis.