Defence R&D Canada (DRDC) has an active research and development program on detection of landmines using nuclear methods. They are intended for confirmation by detection of characteristic radiation or imaging of back scattered intensity distributions. Both vehicle-mounted and person-portable systems are being developed. Research on thermal neutron analysis (TNA) was initiated in 1994 to provide a confirmation detector for the DRDC developed multisensor, teleoperated, vehicle-mounted, landmine detection system. A version is now commercially available and four units have been fielded by the Canadian Land Forces. A prototype next generation TNA, which uses an electronic neutron generator as a source, has been constructed. Preliminary tests have shown improved performance. Research is now ongoing to investigate the addition of a fast neutron analysis capability to the next generation TNA. Characterization studies and software improvements are being conducted. Related research is investigating whether fast inorganic scintillator materials can provide an improvement in energy resolution. For person-portable applications, both neutron and photon irradiation processes are being investigated. A prototype landmine detector based on neutron moderation imaging has been completed and preliminary images of antipersonnel mine simulants obtained. It consists of a novel thermal neutron imaging system, a unique neutron source to uniformly irradiate the underlying ground and hardware and software for image generation and enhancement. Simulations show that it should provide a significant improvement over non-imaging neutron backscatter systems. X-ray backscatter imaging research is concentrating on non-collimated approaches to enable it to be person-portable. One such method, coded aperture imaging, is being investigated and extensive simulations using Geant4 have demonstrated its merits. Initial joint experiments with UC San Diego, using their HEXIS detector, have been conducted.

The International Atomic Energy Agency has initiated a Coordinated Research Project (CRP) for the development of nuclear techniques for landmine detection. Out of the fourteen institutes participating in the CRP, twelve are working on neutron-based techniques. Small isotope neutron sources and D-T neutron generators have been used by the researchers. The techniques used include neutron scattering by explosives as well as gamma spectroscopy following the interaction of neutrons with explosives. Neutrons are readily thermalized by hydrogen in explosives and backscattered. Cape Town University, South Africa, and Delft University, Netherlands, have developed instruments based on this principle. Both are portable units and laboratory tests prove their capability to detect dummy landmines (100 g explosive simulant) buried 3-6 cm below dry soil. Further improvements are in progress. Another device, PELAN, developed by the Western Kentucky University, U.S. is based on pulsed fast and thermal neutron activation and has reached a fairly advanced stage of development. The equipment was tested with real mines in a test field in Croatia. In this first series of tests, PELAN could detect antitank mines (5.6 kg explosive) buried 7.5 cm below soil, and antipersonnel mines (200 g explosive) buried 5 cm below soil. More field tests and methods for improving performance are being pursued. The research groups are investigating different facets of the problem such as detector development, Monte Carlo calculations, spectrum unfolding, detector shielding and data analysis.

The neutron backscattering technique may be applied to search for non-metallic land mines in relatively dry soils. A novel, ergonomic detector system has been constructed. Tests with real land mines in a realistic environment show that anti-tank mines can reliably be found, but that anti-personnel mines may escape detection. The performance could be improved when an image of the mine signal could be obtained. One approach is to use an array of position sensitive ³He detectors placed close to the soil. A first test with a pulsed neutron generator shows that further improvements can be made by applying a time window on the neutron transit time. The possibilities of neutron backscattering imaging systems are investigated using Monte Carlo simulations with GEANT. A neutron backscattering imaging device with a 2D sensitive detection plane is currently under development.

An X-ray mine imaging system (XMIS) that uses a new form of backscatter x-ray radiography developed at the University of Florida was successfully field-tested at Fort A.P. Hill, Virginia in October, 2001. The XMIS obtained high quality images of both AP and AT mines on several of the Fort A.P. Hill test lanes. For high resolution imaging at a power level of 750 watts, total time for scanning and for processed image acquisition was about 60 s for a 0.5 x 0.5 m area. The very good imaging results obtained from the initial field tests at Fort Hill with the XMIS demonstrate the excellent capabilities of this system as a confirmation sensor for land mine detection. These initial field tests showed that some fairly simple modifications could significantly improve the performance of the XMIS. The total cost of components for an XMIS field demonstration system that includes these modifications is about $60 K and includes about $24 K for the x-ray generator and about $16 K for the specially-made detector assemblies. Further field-testing of the XMIS needs to be performed, but this should be done following implementation of the indicated modifications. With the modifications, high resolution scanning of a 0.5 m x 0.5 m area can be done in 20 to 30 seconds at a power level of 300 watts to 400 watts.

The high-resolution sonar is one of the principal sensors used by the Navy to detect and classify sea mines in minehunting operations. For such sonar systems, substantial effort has been devoted to the development of automated detection and classification (D/C) algorithms. These have been spurred by several factors including (1) aids for operators to reduce work overload, (2) more optimal use of all available data, and (3) the introduction of unmanned minehunting systems. The environments where sea mines are typically laid (harbor areas, shipping lanes, and the littorals) give rise to many false alarms caused by natural, biologic, and man-made clutter. The objective of the automated D/C algorithms is to eliminate most of these false alarms while still maintaining a very high probability of mine detection and classification (PdPc). In recent years, the benefits of fusing the outputs of multiple D/C algorithms have been studied. We refer to this as Algorithm Fusion. The results have been remarkable, including reliable robustness to new environments. This paper describes a method for training several multivariate Gaussian classifiers such that their And-ing dramatically reduces false alarms while maintaining a high probability of classification. This training approach is referred to as the Focused- Training method. This work extends our 2001-2002 work where the Focused-Training method was used with three other types of classifiers: the Attractor-based K-Nearest Neighbor Neural Network (a type of radial-basis, probabilistic neural network), the Optimal Discrimination Filter Classifier (based linear discrimination theory), and the Quadratic Penalty Function Support Vector Machine (QPFSVM). Although our experience has been gained in the area of sea mine detection and classification, the principles described herein are general and can be applied to a wide range of pattern recognition and automatic target recognition (ATR) problems.

Over the past several years, Raytheon Company has adapted its Computer Aided Detection/Computer-Aided Classification (CAD/CAC)algorithm to process side-scan sonar imagery taken in both the Very Shallow Water (VSW) and Shallow Water (SW) operating environments. This paper describes the further adaptation of this CAD/CAC algorithm to process SW side-scan image data taken by the Battle Space Preparation Autonomous Underwater Vehicle (BPAUV), a vehicle made by Bluefin Robotics. The tuning of the CAD/CAC algorithm for the vehicle's sonar is described, the resulting classifier performance is presented, and the fusion of the classifier outputs with those of three other CAD/CAC processors is evaluated. The fusion algorithm accepts the classification confidence levels and associated contact locations from the four different CAD/CAC algorithms, clusters the contacts based on the distance between their locations, and then declares a valid target when a clustered contact passes a prescribed fusion criterion. Four different fusion criteria are evaluated: the first based on thresholding the sum of the confidence factors for the clustered contacts, the second and third based on simple and constrained binary combinations of the multiple CAD/CAC processor outputs, and the fourth based on the Fisher Discriminant. The resulting performance of the four fusion algorithms is compared, and the overall performance benefit of a significant reduction of false alarms at high correct classification probabilities is quantified. The optimal Fisher fusion algorithm yields a 90% probability of correct classification at a false alarm probability of 0.0062 false alarms per image per side, a 34:1 reduction in false alarms relative to the best performing single CAD/CAC algorithm.

A novel sea mine computer-aided-detection / computer-aided-classification (CAD/CAC) processing string has been developed. The overall CAD/CAC processing string consists of pre-processing, adaptive clutter filtering (ACF), normalization, detection, feature extraction, feature orthogonalization, optimal subset feature selection, classification and fusion processing blocks. The range-dimension ACF is matched both to average highlight and shadow information, while also adaptively suppressing background clutter. For each detected object, features are extracted and processed through an orthogonalization transformation, enabling an efficient application of the optimal log-likelihood-ratio-test (LLRT) classification rule, in the orthogonal feature space domain. The classified objects of 4 distinct processing strings are fused using the classification confidence values as features and logic-based, “M-out-of-N”, or LLRT-based fusion rules. The utility of the overall processing strings and their fusion was demonstrated with new shallow water high-resolution sonar imagery data. The processing string detection and classification parameters were tuned and the string classification performance was optimized, by appropriately selecting a subset of the original feature set. A significant improvement was made to the CAD/CAC processing string by utilizing a repeated application of the subset feature selection / LLRT classification blocks. It was shown that LLRT-based fusion algorithms outperform the logic based and the “M-out-of-N” ones. The LLRT-based fusion of the CAD/CAC processing strings resulted in up to a nine-fold false alarm rate reduction, compared to the best single CAD/CAC processing string results, while maintaining a constant correct mine classification probability.

Defence R&D Canada-Suffield and the University of California, San Diego, have recently begun a collaborative effort to develop a coded aperture based X-ray backscatter imaging detector that will provide sufficient speed, contrast and spatial resolution to detect antipersonnel landmines and improvised explosive devices. While our final objective is to field a hand-held detector, we have currently constrained ourselves to a design that can be fielded on a small robotic platform. Coded aperture imaging has been used by the observational X-ray and gamma ray astronomy community for a number of years, which has driven advances in detector design that is now being realized in systems that are substantially faster, cheaper and lighter than those only a decade ago. With these advances, a coded aperture hand-held imaging system has only recently become a possibility. One group at the Center for Astrophysics and Space Sciences, University of California, San Diego, has had a longterm programme developing the CZT based HEXIS detector as the detection element of a coded aperture imager. Designed as a satellite payload, this low-power system is ruggedized and light-weight, all necessary qualities for incorporation into the envisioned portable imaging system. This paper will begin with an introduction to the landmine and improvised explosive device detection problem, followed by a discussion of the HEXIS detector. We will then present early results from our proof-of-principle experiments, and conclude with a discussion on future work.

Nuclear Quadrupole Resonance (NQR) is being researched as a confirmatory sensor for use in mine detection as part of the research carried out by the Defence Science and Technology Laboratory (Dstl) for the UK MOD Applied Research Programme. NQR is a radio frequency (RF) spectroscopy technique used at close range to detect explosives, typically TNT and RDX, found in anti-tank and anti-personnel landmines. Detection is carried out by averaging NQR data until the signal to noise ratio increases enough for the signal to be distinguished from RF noise and interference. Environmental RF noise dominates the received signal because NQR signals are, in comparison, extremely low in magnitude. Therefore, RF interference, which varies depending on the time of day, environment, and frequency of the radiation, directly affects detection times. Methods of reducing RF interference such as antenna design, signal processing and phase cycling are reviewed and discussed. Results are presented from research undertaken to enhance the signal to noise ratio, taken in various environments.

The Streak Tube Imaging Lidar (STIL) is a three-dimensional electro-optic sensor that provides photographic quality images and was developed to identify objects of interest on the ocean bottom. STIL sends out a pulsed fan of light through the water column with the photon time of flight returns measured, giving range information that allows water backscatter to be separated from the bottom return. Although three-dimensional, the current choice of operation to display STIL data is to render it into two-dimensional contrast and range maps (images). This paper describes a fully automated process (single executable) that renders the three-dimensional STIL data cube into contrast and range maps, removes inherent system noise, and enhances low contrast regions for optimal display to the operator.

Spatial-spectral anomaly detection (the “RX Algorithm”) has been exploited on the USMC's Coastal Battlefield Reconnaissance and Analysis (COBRA) Advanced Technology Demonstration (ATD) and several associated technology base studies, and has been found to be a useful method for the automated detection of surface-emplaced antitank land mines in airborne multispectral imagery. RX is a complex image processing algorithm that involves the direct spatial convolution of a target/background mask template over each multispectral image, coupled with a spatially variant background spectral covariance matrix estimation and inversion. The RX throughput on the ATD was about 38X real time using a single Sun UltraSparc system. A goal to demonstrate RX in real-time was begun in FY01. We now report the development and demonstration of a Field Programmable Gate Array (FPGA) solution that achieves a real-time implementation of the RX algorithm at video rates using COBRA ATD data. The approach uses an Annapolis Microsystems Firebird PMC card containing a Xilinx XCV2000E FPGA with over 2,500,000 logic gates and 18MBytes of memory. A prototype system was configured using a Tek Microsystems VME board with dual-PowerPC G4 processors and two PMC slots. The RX algorithm was translated from its C programming implementation into the VHDL language and synthesized into gates that were loaded into the FPGA. The VHDL/synthesizer approach allows key RX parameters to be quickly changed and a new implementation automatically generated. Reprogramming the FPGA is done rapidly and in-circuit. Implementation of the RX algorithm in a single FPGA is a major first step toward achieving real-time land mine detection.

Multispectral, hyperspectral, and polarization filters have been shown to provide additional discriminants when searching for mines and other obstacles, but they demand more illumination for the sensing system. Conventional CCD video cameras, when used through such filters, fail at sunset or soon after. It is tempting to employ an automatic-gain intensified camera to push this time deeper into the night (especially with artificial illumination) but relating the response between different images or different channels, possibly taken at different gain, is not as straightforward as it is with a bare-silicon CCD. Over the last several years, Coastal Systems Station has developed a set of simple system characterization and calibration procedures that enable using an intensified video camera as a serviceable imaging radiometer. Parameters from this calibration procedure are easily inserted into predictive models and images are directly comparable using them. These methods will be described, especially as they apply to the camera used in the recent Airborne Laser Diode Array Illuminator (ALDAI) tests. Minimum data that must be monitored in the camera will also be listed.

The objective of the Office of Naval Research (ONR) Rapid Overt Reconnaissance (ROR) program and the Airborne Littoral Reconnaissance Technologies project's Littoral Assessment of Mine Burial Signatures (LAMBS) contract is to determine if electro-optical spectral discriminants exist that are useful for the detection of land mines located in littoral regions. Statistically significant buried mine overburden and background signature data were collected over a wide spectral range (0.35 to 14 μm) to identify robust spectral features that might serve as discriminants for new airborne sensor concepts. The LAMBS program further expands the hyperspectral database previously collected and analyzed on the U.S. Army's Hyperspectral Mine Detection Phenomenology program [see "Detection of Land Mines with Hyperspectral Data," and "Hyperspectral Mine Detection Phenomenology Program," Proc. SPIE Vol. 3710, pp 917-928 and 819-829, AeroSense April 1999] to littoral areas where tidal, surf, and wind action can additionally modify spectral signatures. This work summarizes the LAMBS buried mine collections conducted at three beach sites - an inland bay beach site (Eglin AFB, FL, Site A-22), an Atlantic beach site (Duck, NC), and a Gulf beach site (Eglin AFB, FL, Site A-15). Characteristics of the spectral signatures of the various dry and damp beach sands are presented. These are then compared to buried land mine signatures observed for the tested background types, burial ages, and environmental conditions experienced.

We are developing a novel GPR system for stand off landmine detection. This is a bistatic GPR system, which uses a TEM horn antenna for a transmitter and a passive optical electric field sensor as a receiver. A small size passive optical sensor will be scanned on the ground surface, and the acquired signal is used for synthetic aperture processing. Since the receiver is very small, it is suitable for scanning on the ground, where landmines can be buried.

Ground Penetrating Radar has been applied for several years to the problem of detecting both anti-personnel and anti-tank landmines. Most of the evaluation effort has focused on obtaining the end-to-end performance metrics (e.g. Pd and pfa ) of complete detection systems. This is the third in a series of papers in which we focus on the specific performance of one critical component of GPR systems: the antenna subsystem. In this paper, we examine several free-space characteristics of Planning Systems Inc. Archimedean Spiral Antennas. Specifically, we (1) investigate a spurious signal response observed with a large metal plate reflecting target, (2) determine gain and phase properties of these antennas, (3) calculate the antennas' impulse response, and (4) image several targets to validate our approach.

A project for developing a compact size GPR system for landmine detection was started. It will be based on stepped-frequency radar system for wider application in various kinds of soil conditions. We have developed a prototype stepped-frequency GPR system for fundamental evaluation of the system. This system uses an array of broadband Vivalidi antennas and operates at 2-6GHz. The system was tested in laboratory and could be used for imaging buried mine-line targets by high resolution. Series of test was carried out by using sand with rough surface and inhomogeneous soil. Array signal processing is useful for reduction of clutter from the rough grounds surface.

Reducing the false alarm rate of vehicular and hand-held mine detection systems has been a goal of most countermine detection programs. No thorough investigation into the causes of false alarms has been conducted to date. We present here an investigation into the sources of persistent ground-penetrating radar (GPR) false alarms that occurred during testing of a vehicular mine detection system. Data collected with this system was used to identify false alarms that persisted over several tests conducted over a two-year period over the same simulated roadway. A dig list was generated and several sites were excavated. Soil samples were collected at the sites and analyzed in the lab. The results of the excavation will be presented.

Single-polarity, synthetic aperture radar (SAR) data collected in spotlight mode is examined as part of an effort to identify surface land mines in high-frequency radar imagery. A measurements program was recently conducted using a Ku-band (16 GHz) radar. In this experiment, metal and plastic mines were placed on smooth dirt lanes and in tall and short grass areas adjacent to these lanes. The collected data set consisted of magnitude-only data for several different passes over a common target area that included various reference reflectors as well as the landmines. The metal and plastic mines on the dirt lanes were clearly visible in the processed radar imager, while the mines in the grass areas were not observable - even after applying multi-look averaging. (Multi-look averaging exploits the circular symmetry of the mines to enhance the contrast between the mines and the background clutter). To investigate these effects, we used rigorous moment method-based electromagnetic solvers to compute the backscatter from the metal and plastic mines in a variety of backgrounds. The model results were shown to be consistent with the measurement data for metal mines on the dirt lanes. The plastic mines were not consistent with the data, however. We believe that the difference is due to uncertainty in the mine dielectric constant. The model results also showed that a significant focusing effect (or “glory wave”) could be seen in the plastic mines at low depression angles. Finally, the model demonstrated the highly absorptive nature of the grass, as shown by the significantly reduced radar cross section of mines placed in a three-layer grass model.

This paper describes the development of an affordable mine detector, MINETECT, specifically designed for humanitarian use. The project was sponsored by the UK Department for International Development and was developed by ERA Technology. Using a radically different approach from conventional GPR designs, in terms of the man machine interface, MINETECT offers simplicity of use and affordability, both key factors in humanitarian demining operations. The ground penetrating radar employs novel operator audio interface techniques embodied in European patent number 99306164.7. This paper describes the design concept, summarises the trials carried out and provides the conclusions as to requirements for GPR performance. Further development work, after trials in the terrain of Southern Lebanon, showed that mine classification is feasible with the GPR technology.

Under the Office of Naval Research's Organic Mine Countermeasures Future Naval Capabilities (OMCM FNC) program, Lite Cycles, Inc. is developing an innovative and highly compact airborne active sensor system for mine and obstacle detection in very shallow water (VSW), through the surf-zone (SZ) and onto the beach. The system uses an innovative LCI proprietary integrated scanner, detector, and telescope (ISDT) receiver architecture. The ISD tightly couples all receiver components and LIDAR electronics to achieve the system compaction required for tactical UAVintegration while providing a large aperture. It also includes an advanced compact multifunction laser transmitter; an industry-first high-resolution, compact 3-D camera, a scanning function for wide area search, and temporally displaced multiple looks on the fly over the ocean surface for clutter reduction. Additionally, the laser will provide time-multiplexed multi-color output to perform day/night multispectral imaging for beach surveillance. New processing algorithms for mine detection in the very challenging surf-zone clutter environment are under development, which offer the potential for significant processing gains in comparison to the legacy approaches. This paper reviews the legacy system approaches, describes the mission challenges, and provides an overview of the ROAR system architecture.

Electro-optic identification sensors provide photographic quality images and were developed to identify objects of interest on the ocean bottom. Two of these high-resolution sensors are currently in use: one based on Streak Tube Imaging Lidar (STIL) technology and the other based on Laser Line Scan (LLS) technology. Both of these sensors produce high fidelity imagery that is unparalleled in quality for underwater imaging systems. They differ in that LLS sensors produce two-dimensional (2-D) contrast images only while STIL produces three-dimensional (3-D) data that can be rendered into 2-D contrast and range maps (images). Although still an emerging technology, recent advances have begun to point to significant advantages with the supplementary range information (3-D information) in identifying objects of interest on the sea floor. This paper discusses some of these advantages of range information for 3-D visual display, computer aided identification and target recognition, modeling, and the general identification process.

The Airborne Littoral Reconnaissance Technologies (ALRT) Project has demonstrated a nighttime operational minefield detection capability using commercial off-the-shelf high-power Laser Diode Arrays (LDAs). Historically, optical aerial detection of minefields has primarily been limited to daytime operations but LDAs promise compact and efficient lighting to allow for enhanced reconnaissance operations for future mine detection systems. When combined with high-resolution intensified imaging systems, LDAs can illuminate otherwise unseen areas. Future wavelength options will open the way for active multispectral imaging with LDAs. The Coastal Systems Station working for the Office of Naval Research on the ALRT project has designed, developed, integrated, and tested both prototype and commercial arrays from a Cessna airborne platform. Detailed test results show the ability to detect several targets of interest in a variety of background conditions. Initial testing of the prototype arrays, reported on last year, was completed and further investigations of the commercial versions were performed. Polarization-state detection studies were performed, and advantageous properties of the source-target-sensor geometry noted. Current project plans are to expand the field-of-view coverage for Naval exercises in the summer of 2003. This paper describes the test collection, data library products, array information, on-going test analysis results, and future planned testing of the LDAs.

During the recent nighttime testing of the Airborne Laser Diode Array Illuminator (ALDAI) system in a humid and aerosol-rich environment, it was found that the laser backscatter competed with the ground-reflected signal at higher altitudes. Such an effect was stronger in the polarization channel parallel to the outgoing illumination. Theoretical calculations indicated that most of the scattering was taking place in the first few feet of overlap between the outgoing illumination and the field of regard. To test this, the laser was pointed into a geometry which had no target field, that scattered solely into the air. Backscattered returns were reduced and plotted versus source-receiver separation distance.

This work presents the Mueller matrix elements for scattering from dielectric targets buried beneath 2-D random rough surfaces (3-D scattering problem). The fully polarimetric scattering matrix S is computed for hundreds of computer generated random rough surface realizations and hence the Mueller matrix elements are obtained. The numerical results show that if one relies only on the co- or cross-polarized intensities (i.e. the magnitude of the four elements of the polarimetric scattering matrix S); it is very difficult to sense the buried objects. However investigating all the sixteen Mueller matrix elements greatly helps in detecting these objects.

In this work, radar images of dielectric targets buried beneath 2-D rough ground surfaces are simulated. These images are based on the amplitude and phase of all polarization of incident and scattered waves. Both bistatic and backscatter intensity-based images are considered in this work to compare the image quality in each case. All simulations are obtained using the integral equation based technique; the Steepest Descent Fast Multiple method (SDFMM). The numerical results show that images based on the Covariance matrix are very unique when the surrounding illuminated areas have almost the same rough surface profile. However, these images deteriorate as the surface profile greatly changes in the surrounding ground areas.

The surf zone is a challenging environment for conducting mine countermeasures operations. The performance of acoustic sensors in this environment is extremely limited. Airborne LIDAR sensors have significantly better prospects for successfully working in this environment. However, the complex environment will be a driving factor limiting their performance. The environmental factors influencing the performance of airborne LIDAR sensors will be examined in this paper. These factors can be highly dynamic. Breaking surf action causes bottom sediment resuspension and the formation of bubbles and foam. The resuspended sediments then begin the process of settling, while the bubbles and foam begin to dissipate. All of these phenomena impact the optical properties of the water, which, in turn, impact the performance of the LIDAR system. An experiment was designed and conducted to study the impact of these dynamic processes on the optical properties of the water. The experiment was conducted in September 2002 at the Army Corp of Engineers Field Research Facility in Duck, North Carolina. Preliminary results from the analysis of this data are presented here. This work is being conducted by the Airborne Littoral Reconnaissance Technology (ALRT) project under ONR sponsorship.

The Anti-Invasion Mine Signature Measurement and Assessment for Remote Targeting (AIMSMART) program has undertaken a lidar mine signature data collection for ONR to characterize electro-optic (EO) signatures of anti-invasion mines and environmental factors affecting their detection in the littorals. Two lidar sensors, one 3-D and one polarimetric, both developed by Arete, were fielded at the FRF test facility in Duck, NC. Data were collected with these sensors over a wide variety of mine targets, obstacles, backgrounds, water quality, and wave movements. The principle goal of this analysis is to characterize lidar signature features, especially 3-D, of in-water mines and correlate those features to physical processes in the VSW and SZ environments. This paper describes the approach to characterizing these mine signatures and presents initial results from the analysis.

Research has shown that naturally occurring light outdoors and underwater is partially linearly polarized. The polarized components can be combined to form an image that describes the polarization of the light in the scene. This image is known as the degree of linear polarization (DOLP) image or partial polarization image. These naturally occurring polarization signatures can provide a diver or an unmanned underwater vehicle (UUV) with more information to detect, classify, and identify threats such as obstacles and/or mines in the shallow water environment. The SHallow water Real-time IMaging Polarimeter (SHRIMP), recently developed under sponsorship of Dr. Tom Swean at the Office of Naval Research (Code 321OE), can measure underwater partial polarization imagery. This sensor is a passive, three-channel device that simultaneously measures the three components of the Stokes vector needed to determine the partial linear polarization of the scene. The testing of this sensor has been completed and the data has been analyzed. This paper presents performance results from the field-testing and quantifies the gain provided by the partial polarization signature of targets in the Very Shallow Water (VSW) and Surf Zone (SZ) regions.

Airborne multispectral imagery was collected over various targets on the beach and in the water in an attempt to characterize the surf zone environment with respect to electro-optical system capabilities and to assess the utility of very low cost, small multispectral systems in mine counter measures (MCM) and intelligence, surveillance and reconnaissance applications. The data was collected by PAR Government Systems Corporation (PGSC) at the Army Corps of Engineers Field Research Facility at Duck North Carolina and on the beaches of Camp Pendleton Marine Corps Base in Southern California. PGSC flew the first two of its MANTIS (Mission Adaptable Narrowband Tunable Imaging Sensor) systems. Both MANTIS systems were flown in an IR - red - green - blue (700, 600, 550, 480 nm) configuration from altitudes ranging from 200 to 700 meters. Data collected has been lightly analyzed and a surf zone index (SZI) defined and calculated. This index allows mine hunting system performance measurements in the surf zone to be normalized by environmental conditions. The SZI takes into account water clarity, wave energy, and foam persistence.

We investigate both two-dimensional (2-D) and three-dimensional (3-D) synthetic aperture radar (SAR) imaging techniques for a forward-looking ground penetrating radar (FLGPR) system. In particular, we consider SAR imaging using the delay-and-sum (DAS), phase-shift migration, and spectral estimation (joint APES (Amplitude and Phase EStimation) and RCB (Robust Capon Beamforming)) approaches with the PSI (Planning Systems Inc.) FLGPR Phase II system. For the DAS and phase-shift migration approaches, we use shading in both frequency and cross-track aperture dimensions to reduce sidelobe leakages and clutter. We perform both coherent and non-coherent multi-look processing as well as smoothing to improve the SAR imaging quality and landmine detection capability of the system. The effectiveness of the approaches are demonstrated with an experimental data set collected by the PSI FLGPR Phase II system.

In order to detect buried land mines in clutter, Planning Systems Incorporated has adapted its Ground Penetrating Synthetic Aperture Radar (GPSAR) technology for forward-looking applications. The Forward Looking GPSAR (FLGPSAR), is a wide-band stepped-frequency radar operating over frequencies from 400 MHz to 4 GHz. The FLGPSAR system is based on a modified John Deere E-Gator turf vehicle that is capable of remote control. Custom Archimedean spiral antennas are used to populate the GPSAR array. These antennas are designed and built by PSI and have exceptional broad-band radiation characteristics. The FLGSPAR system has been used to detect plastic and metallic landmines at U.S. Army test facilities and at PSI's engineering center in Long Beach Mississippi. Multi-look SAR processing has been shown to significantly improve the quality of FLGPSAR imagery.

Ground Penetrating Radar (GPR) is an established technology for detecting anomalies beneath the surface of the ground. GPR systems currently in use tend to be hand held or trolley mounted devices that can be moved smoothly over the surface with little or no stand off from the ground and normally have a single transmit/receive antenna pair. However, these properties are quite different from the requirements of a vehicle mounted system such as track width coverage, variable ground clearance and a noisier environment. This paper, based on Countermine research carried out by the UK Defence Science and Technology Laboratory (Dstl), details the development and application of a military, vehicle mounted GPR system. Requirements of a vehicle mounted system are outlined and research towards creating a multi-antenna, vehicle mounted technology demonstrator is discussed. The paper also examines methods of data representation for GPR systems and the advantages that can be gained in this area using a multi-antenna array such as enhanced imagery and three dimensional reconstruction of objects beneath the surface.

For the past three years, we have been systematically exploring the issues involved in using ground penetrating radar (GPR) for anti-personnel (AP) landmine detection. Our focus has been on testing and understanding the basic issues using existing commercial GPR. We have investigated the following factors affecting landmine detection: mine characteristics, soil physical properties, soil water content, surface roughness, antenna height and signal polarization. Field testing in controlled conditions and numerical techniques have been used to parametrically study response factors. Based on our research, the AP landmine fabrication characteristics are critical in determining the magnitude and response character, spatial processing is essential to see the targets against background variability, optimal spectral bandwidth is 500 to 2000 MHz and the practical issues of deploying sensors in rough field conditions are a major challenge.

This paper addresses the detection of mine-like objects in stepped-frequency ground penetrating radar (SF-GPR) data as a function of object size, object content, and burial depth. The detection approach is based on a Selective Independent Component Analysis (SICA). SICA provides an automatic ranking of components, which enables the suppression of clutter, hence extraction of components carrying mine information. The goal of the investigation is to evaluate various time and frequency domain ICA approaches based on SICA. The performance comparison is based on a series of mine-like objects ranging from small-scale anti-personal (AP) mines to large-scale anti-tank (AT) mines. Large-scale SF-GPR measurements on this series of mine-like objects buried in soil were performed. The SF-GPR data was acquired using a wideband monostatic bow-tie antenna operating in the frequency range 750 MHz - 3.0 GHz. The detection and clutter reduction approaches based on SICA are successfully evaluated on this SF-GPR dataset.

Electro-optic identification (EOID) sensors have been demonstrated as an important tool in the identification of bottom sea mines and are transitioning to the fleet. These sensors produce two and three-dimensional images that will be used by operators and algorithms to make the all-important decision regarding use of neutralization systems against sonar contacts classified as mine-like. The quality of EOID images produced can vary dramatically depending on system design, operating parameters, and ocean environment, necessitating the need for a common scale of image quality or interpretability as a basic measure of the information content of the output images and the expected performance that they provide. Two candidate approaches have been identified for the development of an image quality metric. The first approach is the development of a modified National Imagery Interpretability Rating Scale (NIIRS) based on the EOID tasks. Coupled with this new scale would be a modified form of the General Image Quality Equation (GIQE) to provide a bridge from the system parameters to the NIIRS scale. The other approach is based on the Target Acquisition Model (TAM) that has foundations in Johnson’s criteria and a set of tasks. The following paper presents these two approaches along with an explanation of the application to the EOID problem.

When surveying an area for sea mines with a sidescan sonar, the ability to find the same object in two different sonar images is helpful to determine the nature of the object. The main problem with matching two sidescan sonar images is that a scene changes appearance when viewed from different viewpoints. This paper presents a novel approach for matching two sidescan sonar images. The method first registers the two images to ground, then uses the cross correlation of the object positions on the seabed to find the correct displacement between the two images. In order to correct any minor displacements of the relative objects position as a result of the ground registration, the object position is given an area of influence. The method is compared to an existing method for matching sidescan sonar images based on hypothetical reasoning. The two methods are compared on a number of real sidescan sonar images in which the displacement is already known, as well as on images taken of a scene from two different viewpoints. We conclude that the proposed method has fewer variables to tune in order to get satisfactory results, and that it gives better or equal results compared to the hypothetical reasoning method.

The demands of the unmanned airborne multispectral surf-zone mine counter-measures (MCM) mission require high spatial resolution. Weight, volume and power constraints preclude stabilized operation of the cameras for this application. Further, the system is to be flown on a rotary-winged platform, with its attendant vibration characteristics. Thus, the MTF needs to be measured in flight to insure it meets the resolution requirements. We apply the slanted-edge MTF method to the in-flight characterization of airborne high-resolution cameras, analyzing images of orthogonal slanted edges to estimate the motion and vibration contributions to the MTF, and show that the system meets its requirements. We also apply a methodology for scaling to other altitudes and speeds to show that the system will have excellent imaging performance throughout its operational envelope. For our application, the slanted-edge method is more accurate and reproducible than the alternative of placing MTF bar targets under the aircraft flight path. Further, the slanted-edge targets are much easier to deploy and recover, and ease the navigation tolerances.

This paper indicates how sea=test data collected by N independent sensors - or alternatively, data collected by a single sensor, but processed through N independent processing strings - can be used to model, estimate, and predict the performance of a mine identification system. The proposed procedure exploits the information supplied by the sensors/processes (namely, the locations of their individual detection reports), to approximate the probabilities of detection and false alarm in terms of the ratios of the numbers of reports, as seen by the various combinations of sensors. A constrained Least-Squares procedure, fitting the products of these ratios as dictated by their independence equivalencies, is then used to estimate the individual sensor/process probabilities of detection, of false alarm caused by mine-like objects, and of false alarm due to noise. We can then obtain the corresponding probabilities that can be expected after fusing the data with an M-out-of-N voting process.

This paper presents a study on using polarimetric ground penetrating radar (GPR) for the identification of plastic antipersonnel mines. In general, the polarimetric radar response of a surface-laid or buried object depends on the orientation of the object with respect to the transmitting and receiving antennas. Hence, in order to make identification possible, it is crucial to measure the full scattering matrix and transform the data into the target frame, in which the response is orientation independent. In this paper, we present an impulse ultrawideband ground penetrating radar with a polarimetric antenna system. Using this radar, the scattering matrices for a set of surface-laid targets with different shape and internal structure have been measured. The measurements were done for different target orientations. Transformation of the measured response into the target frame was achieved by matrix diagonalization in the frequency domain. The eigenvalues obtained by matrix diagonalization constitute a set of orientation invariant features and have been studied as possible target discriminators. In particular, we addressed the problem of classifying targets with respect to shape (rotationally symmetric versus elongated). The results suggest the possibility to distinguish between targets by looking at how the eigenvalues change as a function of frequency. Moreover, matrix diagonalization yielded an angle of orientation and the significance of this angle for small minelike targets and elongated targets is discussed. The analysis was repeated for scattering matrices acquired over buried targets and the results are compared against those obtained for the surface-laid objects.

The results of the measurement campaign, which has been held recently at the test facilities for landmine detection systems located at TNO-FEL (The Hague, the Netherlands), are presented. The test facilities give an opportunity to evaluate system performance in different environment (such as grass, sand, clay, etc.) under controlled conditions. The test lanes contain various types of antipersonnel and antitank mines. In this campaign we used the Video Impulse Ground Penetrating Radar that has recently been developed in the IRCTR. The design of the radar allows us to perform simultaneous full-polarimetric measurements in two ultra-wide frequency bands. Furthermore, the scattered from subsurface electromagnetic field is measured in quasi-monostatic and essentially bistatic antenna configurations. The acquired during the measurement campaign data are of high quality in terms of time stability, radar positioning and signal-to-noise ratio. This has allowed to extract full-polarimetric target responses and to analyze them. The obtained results are of importance for target classification.

In this paper we use a sequential probability ratio test (SPRT) on ground penetrating radar (GPR) data to detect buried antipersonnel land mines and to reject clutter objects. Detection is performed for both fixed-depth and variable-depth cases. We use high-dimensional analysis of variance (HANOVA) to window the GPR data before SPRT analysis. Our algorithm uses a library of mine and clutter objects and performs a series of SPRTs using the object library for each unknown image. We also evaluate the performance of our fixed-depth and variable-depth detection algorithms versus noise and SPRT threshold value.

An empirical performance model for the Mine Hunter/Killer (MH/K) Ground Penetrating Radar (GPR) was developed and used to analyze the performance of this GPR as a function of soil type, soil moisture, mine casing and mine depth. The empirical modeling approach used can be modified to evaluate the performance of other GPRs if adequate data are collected. All of the data were reprocessed with the final MH/K automatic target recognition (ATR) algorithm so that performance variations due to environmental conditions could be characterized independently of ATR changes. The model estimates Probability of Detection (Pd) and False Alarm Rate (FAR) for buried mines as a function of ATR confidence, estimated soil moisture content (dry, moist or wet), mine casing (metal or plastic), burial depth (shallow or deep) and soil type (dirt or gravel). Time Domain Reflectometry (TDR) moisture probe measurements at one location augmented with qualitative observations of the soil conditions characterized the soil moisture content. The performance model was created from 52 alarm files collected at a temperate US Army test site over a total of 4 weeks during a 13-month period. The results show that for the MH/K GPR performance against plastic mines in dirt improves as soil moisture increases and performance in gravel is better overall than in dirt.

Measurements of the acoustic impedance of a VS 2.2 anti-tank plastic landmine reveal significant resonances in the frequency range between 80 and 650 Hz. The top surface resonances are due to its complicated mechanical structure vibrating in air. The lowest mode of the landmine results in a high Q simple harmonic oscillator resonance of the top surface, which behaves like a rigid mass. At higher frequencies the top surface behaves like thin circular plat acoustic modes. When these landmines are buried in soils, the modes are mass loaded. Resonances from measurements of the normal component of the acoustically induced soil surface particle velocity (due to sufficient acoustic-to-seismic coupling) are used for detection schemes. Since the interface between the top plate and the soil responds to pressure fluctuations nonlinearly, characteristics of landmines, the soil, and the interface are rich in nonlinear physics and allow for new methods of buried landmine detection not previously exploited. Here, the structure of a family of resonant tuning curves for relatively low amplitude, but nonlinear drive levels, reveals the “nonclassical” nonlinear resonant behavior of the soil-landmine oscillator.

Representative data pertaining to various critical aspects of air-coupled ultrasonic Doppler sensing of ground vibrations are presented. The behavior of an ultrasonic sensor is systematically compared with that of commercial laser vibrometers. The inherent drawbacks and advantages of both techniques are discussed and evaluated in systematic experiments. The experiments are designed so as to synthesize various scenarios that may be encountered in practice. Thus the vibration sensing capability of ultrasonic vibrometers is investigated in cases including flat and grass-covered surfaces, granular media, with and without ambient air motion. The work is supported by the Office of Naval Research.

The resonant behavior of landmines has been exploited by an acoustic detection technique to find buried mines. The resonance of the buried mine is induced by broadcasting an acoustic wave, which couples into the ground. The resonating mine causes the soil above it to vibrate and this vibration is measured with either a laser Doppler vibrometer (LDV) or a geophone. A set of resonance frequencies, which can be attributed to the design, material, and dimensions of the mine, is exhibited when the mine, sitting on a rigid surface above the ground, is excited by an acoustic wave. These resonance frequencies shift when the mine is buried. Acoustic models have been developed to predict these burial effects on mine resonant frequency behavior. This paper will discuss measurements made of several mines of the same type buried at various depths and will compare these measurements to predictions made by a lumped element model.

Polarimetric scattering models are developed to predict the detectability of surface-laid landmines. A specular polarimetric model works well only under the condition that there is either no sunlight or the sun is not close to the specular reflection direction. Moreover, this model does not give insight why certain man-made objects like landmines give a higher polarimetric signature than natural background. By introducing a polarimetric bidirectional reflectance distribution function (BRDF) the specular model is extended. This new model gives a better prediction of the polarimetric signature and gives a close match to the measurements of landmines with different casings as well as the sand background. The model parameters indicate that the landmines have a lower surface roughness and a higher refractive index, which is the reason why these objects are detectable from the background based on their polarimetric signature.

Polarization measurements in the IR region, especially in the 8-12 μm wavelength region (LW), are useful to detect man-made object in a natural environment. An example is surface laid mines partly covered by grass, dirt or sand. It has been shown that polarization measurements have improved the possibility to detect partly covered objects. This makes reconnaissance and surveillance sensors and warners more efficient. The equipment used in the measurements reported here is a IR Thermovision camera, in the region 8-12 μm. By applying a linear polarizing filter in front of the camera, it is possible to measure the Stokes parameters, “Degree of Linear Polarization” (DoLP) and “Theta”. The circular polarization component is regarded as small and is being neglected in these measurements. This article reports the results of surface scattering properties measurements on personal mines by using polarization. Measurements and simple model calculations performed on covered surface landmines in natural environment have been reported earlier. A more systematic investigation is reported here. The results indicate that IR polarization is a powerful tool to detect mines. Further, the emissivity as a function of emission angle has been measured for different coverage of the mine, and the DoLP has been calculated for different emission angle.

This paper gives a comparison of two vehicle-mounted infrared systems for landmine detection. The first system is a down-ward looking standard infrared camera using processing methods developed within the EU project LOTUS. The second system is using a forward-looking polarimetric infrared camera. Feature-based classification is used for this system. With these systems data have been acquired simultaneously of different test lanes from a moving platform. The performance of each system is evaluated using a leave-one-out method. On the training set the polarimetric infrared system performs better especially for low false alarm rates. On the independent evaluation set the differences are much smaller. On the ferruginous soil test lane the down-ward looking system performs better at certain points whereas on the grass test lane the forward-looking system performs better at certain points.

Autonomous detection of tripwires using optical systems is of great interest. This paper describes methods for detection of tripwires using an image processing algorithm based on the diffusion equation. A video camera with sensitivity in the near infrared (IR) band records the target scene and the digital images are then transported to a computer to apply an image processing algorithm to determine if a tripwire is present. In this paper, we show that coherence enhancing diffusion filtering can recover broken edges and smooth background without smoothing coherent structures. A comparison of detection results is given with and without diffusion filtering.

The US Army is interested in technologies that will enable it to maintain the free flow of traffic along routes such as Main Supply Routes (MSRs). Mines emplaced in the road by enemy forces under cover of darkness represent a major threat to maintaining a rapid Operational Tempo (OPTEMPO) along such routes. One technique that shows promise for detecting enemy mining activity is Airborne Change Detection, which allows an operator to detect suspicious day-to-day changes in and around the road that may be indicative of enemy mining. This paper presents an Airborne Change Detection that is currently under development at the US Army Night Vision and Electronic Sensors Directorate (NVESD). The system has been tested using a longwave infrared (LWIR) sensor on a vertical take-off and landing unmanned aerial vehicle (VTOL UAV) and a midwave infrared (MWIR) sensor on a fixed wing aircraft. The system is described and results of the various tests conducted to date are presented.

Acoustic-Seismic methods for landmine detection are under intensive investigation. Data collected by the University of Mississippi have by processed by a variety of investigators with excellent results in many cases. Increases in performance are sought based on an understanding of the physical principles leading to the differences between the vibrational velocities of soil over buried landmines and over locations not above landmines. Donskoy suggested modeling the physical system using damped harmonic oscillators. This model suggests a use of magnitude and phase information in image processing algorithms for detecting. In this paper, some methods for incorporating magnitude and phase into image processing algorithms are described and demonstrated. Previous algorithms relied on magnitude only. Increased performance is achieved by incorporating phase into the algorithms.

Acoustic landmine detection (ALD) is a technique for the detection of buried landmines including non-metal mines. An important issue in ALD is the acoustic excitation of the soil. Laser excitation is promising for complete standoff detection using lasers for excitation and monitoring. Acoustic excitation is a more common technique that gives good results but requires an acoustic source close to the measured area. In a field test in 2002 both techniques were compared side by side. A number of buried landmines were measured using both types of excitation. Various types of landmines were used, both anti-tank and anti-personnel, which were buried at various depths in different soil types with varying humidity. Two Laser Doppler Vibrometer (LDV) systems of two different wavelengths for the different approaches were used, one based on a He-Ne laser at 0.633 μm with acoustic excitation and one on an erbium fiber laser at 1.54 μm in the case of laser excitation. The acoustic excitation gives a good contrast between the buried mine and the surrounding soil at certain frequencies. Laser excitation gives a pulse response that is more difficult to interpret but is potentially a faster technique. In both cases buried mines could be detected.

This paper discusses the performance and experimental results of a multiple beam laser Doppler vibrometer designed to locate buried landmines with the laser-acoustic technique. The device increases the speed of landmine detection by simultaneously probing 16 positions on the ground over a span of 1 meter, and measuring the ground velocity at each of these positions. Experimental results are presented from controlled laboratory experiments as well as from landmine test lanes at the University of Mississippi. In the mine lanes, the multiple beam system is raised to a height of 2.5 meters with a forklift, with the 16 beams spread over a 1 meter line along the mine lane. A motor system then allows the 16 beams to be translated across the mine lane, enabling the system to scan a 1 x 1 meter area in a much shorter time than with previous scanning techniques. The effects of experimental parameters such as platform motion, angle of incidence, speckle dropout, and system depth-of-field will be presented and discussed.

A single beam laser Doppler vibrometer (LDV) has been used in acoustic-to-seismic mode [Sabatier, J.M. and Xiang, N. IEEE Trans. Geoscience and Remote Sensing 39, 2001, pp. 1146-1154; Xiang, N. and Sabatier, J.M., J. Acoust. Soc. Am. 113 Mar 2003]. One of the major requirements is the operational scanning speed in the acoustic detection methods. To increase the operational speed, the LDV must move continuously along the ground. An initial effort has demonstrated the feasibility of continuously scanning the ground by controlling the mirrors in a scanning laser vibrometer [Valeau et al., Development of a time-frequency representation for acoustic detection of buried objects, J. Acoust. Soc. Am., 2003 (submitted).]. A continuously scanning LDV on a stationary platform has been employed. This work will discuss systematic investigations using a continuously scanning LDV to obtain field data in Army test lanes.

Landmines buried in the ground can be found acoustically by insonifying the ground and detecting a contrast between the vibratory motion of the ground surface directly above the mine and away from the mine. A technique for the numerical computation of the scattered velocity field is presented here. The mine is assumed to be a rigid cylinder with a compliant top. The ground (soil) is modeled both as an effective fluid and as an elastic effective solid. To discretize the full space model, the computational domain is taken to be a cylindrical waveguide of sufficiently large radius. It is shown that the method converges for the effective fluid case providing qualitative understanding of the field data. However, in the case of an elastic solid, a surface wave propagates radially out from the mine limiting the applicability of the method in its current form. Comparisons with actual field velocity data will also be presented.

This paper presents preliminary results of an investigation into the impact of buried objects on the environmental properties and electro-optical spectral characteristics of terrain features. This study focused on the analyses of various sensor information, including hyperspectral and thermal data, collected under a limited set of circumstances; these analyses include laboratory measurements and theoretical computations. A digital terrain model incorporating the relevant physical processes was also constructed to support these investigations. These analyses are particularly relevant to the detection of landmines and the exploitation of hyperspectral sensor data in this application. Results from these analysis efforts will be presented along with example spectral data and computational results.

Hyperspectral imaging is an important technology for the passive optical detection of surface and buried land mines from an airborne platform. Hyperspectral remote sensing can exploit many different potential mine observables in the visible and infrared portions of the spectrum. The primary surface mine observable is a spectral difference between the mine body and the background. With a high quality VNIR/SWIR hyperspectral sensor, it is possible to detect these mines as spectral anomalies using techniques that have been previously applied to the detection of military targets. Algorithms developed for the military surveillance application can be directly applied to the surface mine problem. In this paper, two different spectral anomaly approaches are explored. The first is a local spectral anomaly detection algorithm, which examines the color of each pixel for differences with its surroundings. The second is a global spectral anomaly detection algorithm that measures the color of each pixel relative to its occurrence in the whole scene. Both algorithms were developed for the problem of detecting military targets in complex backgrounds and are applied here to the problem of detecting surface mines.

The Rapid Route Area Mine Neutralisation System (RRAMNS) has been constructed by the Defence Science and Technology Organization (DSTO) as a concept demonstrator to investigate issues of dynamically detecting land mines. Three main sensors are mounted on a mobile platform to acquire target information. The passive imaging sensor suite centers on the visible and infrared wavelengths while the ground penetrating radar and EM instrument constitute the active sensor arrays. Through a process of sensor fusion, which takes place in a following vehicle, the detection signals of each sensor are fused to enhance the rate of detection. This paper will discuss the RRAMNS electro optical sensor array from a systems point of view. In its current configuration, the imaging system (IS) is showing considerable promise as a mobile optical detector. Early system trials have shown surface deployed anti-personnel mines are readily detectable as are larger classes of mines. Buried mines have also been detected and can be distinguished from buried shrapnel fragments and rocks. The imaging system has also been designed as a diagnostic tool to investigate phenomena associated with the detection of mines in a range of environmental conditions.

This paper answers in the affirmative the question: will it ever be feasible to predict useful infrared buried mine detection performance? The infrared (IR) is essentially blind at certain hours, but can have excellent vision at other times. The trick to making the IR a tactically useful tool is to plan mine detection operations during its best time of utility. Rather than use thermal models with their difficulty in representing IR imagery, we used a matched filter detector on IR video, in combination with prediction techniques using neural nets and weather data, to show that weather conditions can be successful in predicting IR mine detection performance. Prediction using mine detection models and weather data, correlated using neural nets and then predicted using weather data alone is not only theoretically feasible, but is also practical. Feasibility was demonstrated in Train A/Test A mode, where the neural nets achieved 100% prediction accuracy for both AP and AT mines. Practicality was demonstrated using single day Train A/Test B results, where 98% to 88% accuracy was achieved for AT mines from 2.5 to 12.5 hours forward, respectively. The technique is expected to be limited only by the accuracy of the short-term weather forecast.

A technique for the detection of buried landmines, which uses a seismic probing signal in conjunction with a non-contact radar-based surface displacement sensor, has been studied for several years at Georgia Tech. Laboratory experiments and numerical models have indicated that this technique shows great promise for imaging a large variety of mine types and burial scenarios. In order to develop a detection system based on this technique, recent studies have focused on transitioning the experimental work from laboratory models to realistic field environments, which poses several challenges for system development. Unknown soil properties at field sites as well as the presence of local inhomogeneities, vertical stratification, and surface variations make the propagation and the modal content of the seismic probing signal more difficult to predict. This complicates the processing required to image buried mines. The small-scale surface topography and naturally-occurring ground cover impede the function of the system's non-contact sensor, which must be capable of looking through the ground cover and spatially averaging its measurement over the irregular surface. A prototype detection system has been tested at several field sites with widely disparate soil properties. Problems were encountered that required modifications to the system sensor, scanning technique, and signal processing algorithms. Following these changes, system performance comparable to that observed in laboratory models was demonstrated during field testing.

For several years a system has been under development at Georgia Tech that uses seismic surface waves to detect and image buried landmines. The details of this system have been previously reported in the literature. Current work involves the transition from a laboratory experimental system to a field-operable experimental system with the ultimate goal of creating an integrated field-operable prototype. Several issues have arisen in the transition to field testing. One of these is the nature and magnitude of the noise levels that limit system performance at field sites and the relevance of these for predicting noise that might be encountered in a realistic demining scenario. Noise introduced to the system sensor (a radar-based, non-contact displacement sensor) can arise from many sources (both natural and manmade). It may be received through a variety of mechanisms in addition to the sensor's primary transduction mechanism. Moreover, even noise which is received through the primary transduction mechanism need not involve purely seismic motion of the ground that is being interrogated. It might instead represent motion of the sensor's support structure or the purely local coupling of airborne noise into surface motion. To understand these effects, measurements have been made using ground contacting sensors at four field sites where other system-related measurements have also been made. The nature of the noise measurements has required that refinements be made to both the sensors themselves (triaxial geophones) and to the data acquisition system used for the measurement of the system’s seismic interrogation signals (a 12-bit, PC-based digitizer).

The use of a laser Doppler vibrometer (LDV) to sense the acoustic-to-seismic coupling ratio for buried landmine detection has previously been demonstrated. During these experiments, the LDV is mounted on a fixed platform and the beam moves continuously across the ground. Experiments show that fixed mounted LDV can achieve scanning speeds up to 3.6 km/h for successful detection of buried landmines in outdoor ground. The problems associated with taking a fixed-mount, scanning LDV and transitioning to a mobile system involve such issues as vehicle vibration, additional Doppler bandwidth due to vehicle speed, speckle noise, and sample time vs. spatial averaging. This paper presents the results of field tests with the moving platform on U.S. Army mine lanes showing that many of these issues can be overcome with an appropriately designed moving platform. The testing involved scanning different types of mines at varying depths and different speeds. Different aspects of the experiment are also discussed.

Landmine detection research demonstrates that acoustically-induced vibrations into the ground creates a vibrational response originating from landmines which can be differentiated from that of background. Field tests utilizing acoustic technology performed under static conditions has yielded high probabilities of detection coupled with low false alarm rates. Current research has proven acoustic mine detection can be performed on a forward moving platform. The speed limitations have not yet been discovered though. This paper will present the results of a series of field tests in which a laser doppler vibrometer suspended from a moving trolley is passed over buried land mine targets that are excited by an acoustic source. The paper will discuss the experimental protocol, the results and the interpretation of these results. This paper will also discuss our future efforts at acoustic land mine detection.

A hand-held mine detector has two modes of operation: search and localization. In search mode, the goal is to identify areas where a buried mine might be located. Since minimizing the number of misses is a top priority, many regions identified in this mode may contain clutter. To separate the clutter from the mines, the detector can be switched into the localization mode during which a more thorough interrogation of the region is performed. Because causality is not required in localization mode, the analyzed signal is not limited to a single data point, but instead can consist of the response across an entire spatial "region". Previous work has demonstrated that so called "region processing" can potentially improve the localization performance of the detector. We have used the Minelab F1A4 metal detector, an EMI-based system, to collect regional data for a variety of objects including buried mines, metallic and non-metallic clutter, and short-circuited copper loops in free space. Several physics-based processing algorithms were developed and used to predict discrimination performance. Analysis of the loops, whose physical properties were known, indicated that discrimination between objects might be possible using a feature extracted from the detector output. Subsequently, this feature was used as the basis of an algorithm which, when used to process the mine/clutter data, significantly decreased the false alarm rate. This algorithm and its performance were further enhanced by incorporating information about the entire regional response of each object.

Magnetic Singularity Identification (MSI) is a technique that is used to determine the unique natural resonances of metallic objects such as land mines. Dyadic MSI theory is extended herein to rotationally symmetric metal mines, including formalization of an averaged data waveform of individual measurements taken over multiple illumination polarizations. Identification of threat target objects and discrimination against innocuous metal clutter items by pole extraction techniques in real time is viable, but is highly computationally intensive and pushes the state-of-the-art of real-time computing systems. Simpler and faster alternative identification/discrimination algorithms are sought. One promising candidate is the E-pulse technique, or a modification thereof called the ξ-pulse. This note addresses the specific application of the E-pulse approach to metallic targets where three distinct exponential decay terms can be extracted using MSI techniques. These decay terms, obtained from calibration runs in non-real time, are used to build a library of known threat targets and corresponding E-pulse waveforms. These are used in turn to provide a mine discrimination metric in real time in the field.

Using broadband magnetoresistive sensors, Quantum Magnetics is developing a metal detector for landmine/UXO detection and discrimination. When completed, this active system will be incorporated into a passive manportable gradiometer system being developed under funding from the Strategic Environmental Research and Development Program. The resulting system will be a handheld passive/active magnetic tensor gradiometer detector capable of detecting, locating, and discrimination buried targets. In this paper, we discuss these developments and recent results.

This paper describes a prototype three-dimensional electromagnetic induction (EMI) sensor system that has the potential to measure directly the multiple components of buried metal targets' magnetic polarizability tensor without the need to invert spatial data from single-axis EMI sensors. This novel sensor is called a three-dimensional steerable magnetic field (3DSMF) sensor system. The 3DSMF sensor is a high-time resolution, wide-bandwidth time-domain EMI system combined with a 3-axis magnetic field generator (3AMFG) and magnetic field receivers. The 3AMFG differs from previous 3-axis magnetic field generators in a number of ways: the projected magnetic field is relatively uniform in space and is steerable. These two features offer the potential to greatly improve target classification. This paper discusses the 3DSMF sensor system design philosophy and modeling results.

Chemical sensing for buried landmines is a complex phenomenon that includes mine chemical emissions, soil chemical transport/degradation, and detection at the ground surface. The technology to assess soil chemical transport has evolved and now provides a complex systems analysis capability using high fidelity computational simulation tools. Data requirements to evaluate a chemical sensing scenario include soil chemodynamic properties, micrometeorological conditions, and mine chemical emissions. Mine chemical emission tests were performed on four antipersonnel landmines using whole landmines in soil flux chambers. Soil flux chambers are simple containers that surround landmines with dry soil that act as an adsorbent. After a certain soak time, residue analysis of the soil provides the total chemical emission - a combination of both permeation and leakage. An evaluation of permeation differences into wet soil versus dry soil was also completed using thin polymer coupon sections.

The United States Army has expressed an interest in developing sensors that are capable of detecting explosives found in buried landmines. Techniques under development often detect explosive molecules that have migrated out of buried landmines, through the soil, and to the soil’s surface. Since the success of detectors using the above method depends on the presence of explosives at the soil surface, it is critical to have an understanding of the nature of the explosive signature that is being detected. Many factors affect the migration of explosives from the landmine through the soil. These factors include, but are not limited to, soil moisture, terrain, mine type, and explosive type. This is a complex system to study. The experiment presented here attempts to monitor the explosive signature above twenty-seven landmines that have been buried for a number of years in a temperate environment. There are nine mine types represented in the experiment. Five of the mine types are anti-tank mines and four of the mine types are anti-personnel mines. Soil samples have been collected above and around these twenty-seven mines and analyzed using gas chromatography coupled with an electron capture detector (GC-ECD). Samples were collected in June 2001, October 2001, February 2002, and June 2002. Results of the GC-ECD analysis of these samples are presented in this paper.

Chemical signatures from buried landmines vary widely due to landmine and environmental conditions. The simulation model T2TNT was developed to evaluate the nature of chemical transport in the soil surrounding a buried landmine. This model uses landmine chemical emission, soil physics, soil-chemical interaction, and surface weather data to estimate surface and subsurface concentrations to help understand the phenomenology of landmine trace chemical detection. While T2TNT compares favorably to controlled laboratory experiments for a buried source of DNT, field data-model comparisons are needed to further increase confidence in T2TNT predictions. The only multi-season landmine soil residue data are from a long-term monitoring project at the DARPA-developed Ft. Leonard Wood Site in Missouri, USA. About 1000 soil residue samples had been taken over six sampling events spanning 21 months since landmine burial. This effort compares the soil residue data from two landmine types to T2TNT model predictions. A one-dimensional model was used to represent the situation and used actual weather data from the site during this period, landmine flux data specific for the mines buried, and temperature and moisture-content dependent degradation rates. Spatial and temporal predictions of chemical concentrations in the soil compare favorably with the soil residue data from Ft. Leonard Wood, increasing confidence in the utility of T2TNT estimates of landmine signature chemicals for other locations.

Areas subject to military conflict or military training sometimes contain unexploded ordinance and/or explosives residues resulting from detonations. A variety of past field and modeling studies have investigated the behavior of explosives in soils in warm climates, but the behavior in cold climates, including frozen soil and snow, has been less studied. In northern areas of military conflict, and at Army military training grounds in cold regions, winter weather exists for many months of the year. The impact of temperature and moisture changes in the soil, due to changing weather conditions, can have a large effect on the fate and transport of the explosives. The basic transport parameters for the behavior of the contaminant in frozen soil are unknown, yet these parameters are needed for problem assessment both for simple estimates and full numerical studies. In this paper we discuss sample results of controlled laboratory experiments performed to investigate the diffusion of 2,4-DNT through sand, under two conditions each of temperature and moisture. Based on the experimental data we present preliminary effective diffusion coefficients for the conditions. The concentrations show clear transport of the contaminant due to vapor diffusion and sorption. Sorption is a controlling feature of diffusion of explosives in sand. Diffusion rates and concentration on the particles increase with temperature. In both frozen and unfrozen sand, higher moisture content causes faster diffusion rates but lower particle concentration levels than in the corresponding dry cases.

This paper gives an overview of the activities on research and development in the technology area for landmine detection in the Netherlands. The main players, their projects and the long term and short term project goals are presented. The projects cover the range from military applications to humanitarian demining. In the “conventional” detection systems area the activities on Metal detection, Ground Penetrating Radar and Thermal Infrared are covered. Signal processing and Sensor fusion are key activities in this area and examples of these activities are shown as well. The focus for these techniques is on vehicle mounted and airborne multi-sensor systems. The activities are supported by more fundamental modeling of the interaction of sensors with the landmines and especially the effects of the environment of the mines on this interaction. In the area of more future oriented techniques the following techniques are discussed: forward looking Polarised Infrared for moving platforms, Neutron Backscattering techniques and Laser Vibrometry for acoustic detection.

Defence R&D Canada (DRDC), an agency within the Department of National Defence, has been conducting research and development (R&D) on the detection of landmines for countermine operations and of unexploded ordnance (UXO) for range clearance since 1975. The Canadian Centre for Mine Action Technologies (CCMAT), located at DRDC Suffield, was formed in 1998 to carry out R&D related to humanitarian demining. The lead group responsible for formulating and executing both countermine and humanitarian R&D programs in detection is the Threat Detection Group at DRDC Suffield. This paper describes R&D for both programs under the major headings of remote minefield detection, close-in scanning detection, confirmation detection and teleoperated systems. Among DRDC's achievements in landmine and UXO detection R&D are pioneering work in electromagnetic and magnetic identification and classification; the first military-fielded multisensor, teleoperated vehicle-mounted landmine detection system; pioneering use of confirmation detectors for multisensor landmine detection systems; the first fielded thermal neutron activation landmine confirmation sensor; the first detection of landmines using a real-time hyperspectral imager; electrical impedance imaging detection of landmines and UXO and a unique neutron backscatter landmine imager.

This is a keynote address surveying the field of Humanitarian Demining (HD) from the viewpoint of a participating company. The controlling bodies, funding structures and some of the important sources of R&D relevant to HD are identified. The various techniques and technologies in common use as also technologies freshly put into field use are mentioned. The way in which they all fit into the demining toolbox is explained. Finally a view of future technologies that are potentially able to change HD efficiency and safety is discussed.

We have created a new architecture for the detection and location of specific buried targets. The system uses a combination of acoustic vibrations and electromagnetic waves to achieve highly specific target recognition, and a multistatic configuration to determine target location. The mechanical vibration resonance properties of the constituent elements of the targets constitute a signature which can be identified in clutter. In order to better detect these vibrations, continuous-wave radar signals are used rather than acoustic reflections, as in sonar-based systems. The energy stored in resonant vibrating elements is not directly detected, but rather modulates the radar signal. The received signals are sampled at high resolution to facilitate target signature recognition by cross-correlation and phase measurement. Location is accomplished by travel time determination for each receiver using absolute phase measurements at multiple frequencies. The phase measurements provide multiple sets of confocal elliptical lines of position, whose intersection identifies the target location. The region in which a mine can be precisely located is a subset of the region in which its presence can be detected.

Controlled-source electromagnetic (CSEM) spatial profiles and 2-D conductivity maps were obtained on the Brazos Valley, TX floodplain to study the fractal statistics of geological signals and effects of man-made conductive targets using Geonics EM34, EM31 and EM63. Using target-free areas, a consistent power-law power spectrum (|A(k)| ~ k ^^-β) for the profiles was found with β values typical of fractional Brownian motion (fBm). This means that the spatial variation of conductivity does not correspond to Gaussian statistics, where there are spatial correlations at different scales. The presence of targets tends to flatten the power-law power spectrum (PS) at small wavenumbers. Detection and localization of targets can be achieved using short-time Fourier transform (STFT). The presence of targets is enhanced because the signal energy is spread to higher wavenumbers (small scale numbers) in the positions occupied by the targets. In the case of poor spatial sampling or small amount of data, the information available from the power spectrum is not enough to separate spatial correlations from target signatures. Advantages are gained by using the spatial correlations of the fBm in order to reject the background response, and to enhance the signals from highly conductive targets. This approach was tested for the EM31 using a pre-processing step that combines apparent conductivity readings from two perpendicular transmitter-receiver orientations at each station. The response obtained using time-domain CSEM is influence to a lesser degree by geological noise and the target response can be processed to recover target features. The homotopy method is proposed to solve the inverse problem using a set of possible target models and a dynamic library of responses used to optimize the starting model.

A discretely loaded resistive vee dipole is designed and realized for use in pulse radiation applications, such as ground-penetrating radars. The resistive vee dipole is capable of radiating a broadband pulse whose shape is simply related to the input signal. In addition, it mostly eliminates the multiple reflections between the surface of the ground and the antenna because of its low radar cross section. Other investigators have studied the resistive vee dipoles using continuous loading. The antenna presented in this paper is printed on a circuit board and discretely loaded with off-the-shelf surface-mount chip resistors, making it easy, inexpensive to build, and mechanically stable. The characteristics of the antenna on a circuit board are measured and compared with the characteristics of the antenna in free space, which is numerically modeled using the method of moments code. The effects of the balun on the performance of the antenna are also presented.

The double-y balun, transitioning from an unbalanced coplanar waveguide (CPW) to a balanced coplanar strip (CPS), was originally developed for use with balanced mixers. In this paper, the feasibility of this balun for feeding pulsed antennas was investigated via time-domain pattern measurements of a resistively loaded V-dipole. The need for a balun when feeding a symmetric antenna is illustrated via time-domain pattern measurements of the resistively loaded V-dipole with and without the double-y balun.

This paper addresses continuous wave (CW) electromagnetic induction (EMI) measurements of wire loops (q-coils) and low metallic (LM) content landmines. Most importantly, measurement errors are addressed. A method based on the measurement of a q-coil with an analytically calculable response is used to produce an error correction function. When this function is used to correct EMI CW measurements of other objects cross over frequency errors are reduced relative to the uncorrected case on average by a factor of 2. Insights into the sources of systematic errors are provided through the analysis of a simple equivalent circuit model of the EMI measurement system.

We use the time-frequency analysis techniques for buried plastic landmine detection with a forward-looking Ground Penetrating Radar (GPR) system. Several time-frequency distributions are considered to characterize and interpret the scattering phenomena of both targets and clutter. An ambiguity function based detector is also proposed, which employs principal component analysis for data dimensionality reduction and linear discriminant analysis for feature selection. Experimental results based on the SRI (Stanford Research Institute) experimentally measured forward-looking GPR data are presented, showing a significant detection performance improvement over the conventional detector.

Digital signal processing algorithms for the detection of landmines using ground penetrating radar are computationally intensive if not due to algorithmic complexity, then due to the vast quantity of data which must be processed in real-time. As a result of this, surface area coverage rates using general purpose computers are limited without an additional investment in multiple central processing units and the parallelization of the executable. This results in an excess of unused resources with the associated cost both in terms of monetary cost and power consumption. The increase in power consumption alone also causes an increase cost in cooling and the requirement for larger prime power and/or reduced battery life. Field programmable gate array (FPGA) hardware devices are reconfigurable in seconds and they can be reprogrammed in the field using relatively standard equipment such as a laptop computer. A secondary advantage of re-configurable dedicated hardware is the flexibility it affords in terms of the specific signal processing algorithm being executed on the re-configurable computing device. As an example of this type of hardware optimization of an algorithm, this paper describes an implementation of volumetric (3D) template matching using re-configurable digital hardware, namely an FPGA. This is a viable alternative for the acceleration of digital signal processing and directly results in an increase in mine detection area coverage rates for a relatively small investment. This also results in a more compact, fieldable real-time implementations of landmine detection algorithms and a common mine detector whose hardware is standard but whose optimized algorithms are downloaded into the FPGA for the particular minefield to be cleared. In this paper we give a quantitative analysis of the increase in execution speed achieved by performing cross correlation of large template sizes on large data.

The efficiency of unexploded ordnance (UXO) remediation is currently limited by the inadequate discrimination capability of present detection technologies, such as single sensing coil inductive sensors. While these methods often detect all relevant metal objects, they generally cannot discriminate between harmful objects and harmless clutter. False indications continue to far outnumber verified detections. To help address the need for a fieldable detection and clutter suppression capability, high resolution inductive arrays are being developed for UXO imaging. This development effort leverages existing MWM-Array sensor and instrumentation technology used in nondestructive testing to create quantitative images of geometric and material property variations. This program is being funded by SERDP and JENTEK Sensors. This paper reviews the MWM-Array technology and its extension to UXO detection and discrimination. The technology uses unique designs for electromagnetic induction sensor arrays incorporating a single drive with multiple sense elements. The drive creates a shaped magnetic field pattern that concentrates the field energy into longer wavelength spatial modes for deeper and “focused” penetration into the ground. Arrays of small inductive coils, placed throughout the shaped field, sense the response from conducting or magnetic UXO and clutter. Images obtained from scans over buried objects provide a basis for spatial filtering and signal processing. Multiple sensor arrays placed at different positions within the drive provide different “views” of buried objects and clutter. Model-based grid measurement methods are also reviewed as a real-time method for multiple property measurements, and real-time data analysis/image generation.

It may not be feasible to completely survey large tracts of land suspected of containing minefields. It is desirable to develop a characterization protocol that will confidently identify minefields within these large land tracts if they exist. Naturally, surveying areas of greatest concern and most likely locations would be necessary but will not provide the needed confidence that an unknown minefield had not eluded detection. Once minefields are detected, methods are needed to bound the area that will require detailed mine detection surveys. The US Department of Defense Strategic Environmental Research and Development Program (SERDP) is sponsoring the development of statistical survey methods and tools for detecting potential UXO targets. These methods may be directly applicable to demining efforts. Statistical methods are employed to determine the optimal geophysical survey transect spacing to have confidence of detecting target areas of a critical size, shape, and anomaly density. Other methods under development determine the proportion of a land area that must be surveyed to confidently conclude that there are no UXO present. Adaptive sampling schemes are also being developed as an approach for bounding the target areas. These methods and tools will be presented and the status of relevant research in this area will be discussed.

Current state-of-the-art electromagnetic induction (EMI) metal detector research systems have shown the potential to detect low metal content buried targets as well as discriminate the type of target as a mine or clutter. However, further research is needed to investigate metal target discrimination potential for closely spaced metal targets. A series of experiments designed to investigate the spatial and time decay responses of multiple metal targets were conducted using a spatial scanning, time-domain EMI metal detector. Time decay signatures were taken of two calibration targets placed over varying distances with the objective of analyzing target identification and spatial resolution. This paper presents results of these experiments.

Due to an object's unique combination of several physical characteristics, including conductivity and permeability, detection systems using the principles of electromagnetic induction (EMI) can be used to detect and classify a characteristic shape or signature for various targets. Subsequently, a library of signatures for targets of interest may be generated for classification purposes to reduce the false alarm rate associated with remediation. However, targets of interest are rarely isolated in the subsurface environment; metallic clutter and/or other targets of interest may be located in close proximity, thereby altering the returned signature and creating the possibility of false alarms. In this presentation, we will present data that was collected on mine-like targets located in close proximity to each other using a prototype frequency-domain EMI sensor, the Geophex GEM-3. We will show that weighted combinations of each object's independent signature can represent the resulting EMI response from two objects, located in close proximity. We will also present simple algorithms to detect the presence of overlapping objects and analyze their performance as a function of object separation and other relevant parameters. The creation and use of target recognition algorithms that consider multiple closely spaced objects will also be discussed.

Virtually all signal processing strategies for discrimination of buried UXO are clutter limited. Most buried UXO are to be found in the top meter of soil, and therefore produce detectable electromagnetic responses. However they also typically reside in settings with widespread metallic clutter from detonated ordnance or other sources. While generally smaller than the UXO, metallic fragments can be numerous and may be shallower than the UXO we seek. Thus clutter signals may be stronger than those from UXO, especially locally, and may cause either highly localized or diffuse obscuration of signatures. They may mask crucial UXO frequency and temporal response patterns, and may distort the otherwise revealing spatial variations of response. To deal with this, first an analytical physical model of electromagnetic induction (EMI) scattering from widespread metallic clutter is formulated and tested. The dependence of signal magnitude on antenna elevation is determined for both thin surface layers and volume layers of clutter. This dependence is different from that of a single UXO size target. In treatment of UWB EMI measurements, this difference is exploited to elicit evidence of the UXO-like target when it is screened from the sensor by a surface layer of small metallic objects. Inversions are also performed for characterizing the geometry of a UXO-like target beneath a surface layer of clutter. The test cases compare a simple least squares (SLS) and a Bayesian-inspired statistical (BIS) approach. As target depth is increased and signal to clutter ratio decreases, the BIS generally produces more consistent and accurate results.

Near field ( ~ 1 m) electromagnetic induction (EMI) sensing, from 10's of Hz up to 100's of kHz, has been successful in detecting subsurface metallic targets. However, the discrimination of buried unexploded ordinance (UXO) from innocuous objects still remains a challenging problem. The EM fields radiated by both antenna and target fall off very sharply as function ~1/R³, for a combined decay rate of ~ 1/R⁶. Therefore EMI sensors affect different materials and sections of the target differently, and signals depend very strongly on what parts of the target are closest to the sensor. Taking into account proximity effects is particularly important for identification and discrimination of actual UXO. The classification of unseen, buried objects, which in general is an inverse problem, requires very fast and accurate representation of the target response. To address these critical issues and to enhance of UXO identification, this paper presents very fast, rigorous ways to compute EMI scattering from a composite target. The method is based on the hybrid full method of auxiliary source (MAS) and MAS-thin skin depth approximation technique (MAS-TSA), together with modal decomposition and reduced source set techniques. For general excitation, a primary field is decomposed into the fundamental spheroidal modes on a fictious spheroid surrounding a real target. Then the problem is solved for each spheroidal mode, taking advantage of axial symmetry. Finally the total response from the target is reproduced using only a few auxiliary magnetic charges. The numerical results are given and compared with experimental data.

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.

This paper presents an automatic UXO classification system using neural network and fuzzy inference based on the classification rules developed by the OSU. These rules incorporate scattering pattern, polarization and resonance features extracted from an ultra-wide bandwidth, fully polarimetric radar system. These features allow one to discriminate an elongated object. The algorithm consists of two stages. The first-stage classifies objects into clutter (group-A and D), a horizontal linear object (group-B) and a vertical linear object (group-C) according to the spatial distribution of the Estimated Linear Factor (ELF) values. Then second-stage discriminates UXO-LIKE targets from clutters under groups B and C. The rule in the first-stage was implemented by neural network and rules in the second-stage were realized by fuzzy inference with quantitative variables, i.e. ELF level, flatness of Estimated Target Orientation (ETO), the consistency of the target orientation, and the magnitude of the target response. It was found that the classification performance of this automatic algorithm is comparable with or superior to that obtained from a trained expert. However, the automatic classification procedure does not require the involvement of the operator and assigns a unbiased quantitative confidence level (or quality factor) associated with each classification. Classification error and inconsistency associated with fatigue, memory fading or complex features should be greatly reduced.

Approximately 75% of buried UXO cleanup costs are expended excavating false alarm anomalies (i.e., digging on the locations of geophysical anomalies that are not caused by UXO). Although probabilities of detection at documented UXO test sites are commonly >90%, there is little documented discrimination capability. This lack of discrimination capability leads to excessively high false alarm rates for both test site and live site surveys. Despite considerable advances in quantitative interpretation methods for discrimination, the state of practice is qualitative or empirical. The UXO thrust of the Army Engineer Research and Development Center's (ERDC) Environmental Quality Technology Program seeks to develop enhanced detection and discrimination capability for survey data from total field magnetometry, time-domain electromagnetic induction, and frequency-domain electromagnetic induction methods. Enhanced discrimination capability by formal geophysical inversion is demonstrated at documented test sites and live sites. A current emphasis is the development of formal inversion procedures that utilize the information content in multiple geophysical datasets. Two approaches are considered: (1) cooperative or constrained inversion; and (2) joint inversion. Cooperative inversion is the process of using inversion parameters from one dataset to constrain the inversion of other data. In true joint inversion, the target model parameters common to the forward models for each type of data are identified and the procedure seeks to recover the model parameters from all the survey data simultaneously. High-quality datasets acquired at seeded test sites at Former Fort Ord, California, demonstrate the confidence in applying these two approaches to discrimination of UXO from non-UXO targets.

Buried landmines are often detected through their chemical signature in the thin air layer, or boundary layer, right above the soil surface by sensors or animals. Environmental processes play a significant role in the available chemical signature. Due to the shallow burial depth of landmines, the weather also influences the release of chemicals from the landmine, transport through the soil to the surface, and degradation processes in the soil. The effect of weather on the landmine chemical signature from a PMN landmine was evaluated with the T2TNT code for three different climates: Kabul, Afghanistan, Ft. Leonard Wood, Missouri, USA, and Napacala, Mozambique. Results for TNT gas-phase and solid-phase concentrations are presented as a function of time of the year.

Many engineering groups desire to construct instrumentation to replace dog-handler teams in identifying and localizing chemical mixtures. This goal requires performance specifications for an “artificial dog-handler team”. Progress toward generating such specifications from laboratory tests of dog-handler teams has been made recently at the Sensory Research Institute, and the method employed is amenable to the measurement of tasks representative of the decision-making that must go on when such teams solve problems in actual (and therefore informationally messy) situations. As progressively more quantitative data are obtained on progressively more complex odor tasks, the boundary conditions of dog-handler performance will be understood in great detail. From experiments leading to this knowledge, one ca develop, as we do in this paper, a taxonomy of test conditions that contain various subsets of the variables encountered in “real world settings”. These tests provide the basis for the rigorous testing that will provide an improved basis for deciding when biological sensing approaches (e.g. dog-handler teams) are best and when “artificial noses” are most valuable.

A sensor (known as Fido) that utilizes amplification of fluorescence quenching as the transduction mechanism for ultra-trace detection of nitroaromatic compounds associated with landmines has been described previously. Previous sensor prototypes utilized a single band of amplifying polymer deployed inside a capillary waveguide to form the sensing element of the detector. A new prototype has been developed that incorporates multiple, discrete bands of different amplifying polymers deployed in a linear array inside the capillary. Vapor-phase samples are introduced into the sensor as a sharp pulse via a gated inlet. As the vapor pulse is swept through the capillary by flow of a carrier gas, the pulse of analyte encounters the bands of polymer sequentially. If the sample contains nitroaromatic explosives, the bands of polymer will respond with a reduction in emission intensity proportional to the mass of analyte in the sample. Because the polymer bands are deployed serially, the analyte pulse does not reach the bands of polymer simultaneously. Hence, a temporal response pattern will be observed as the analyte pulse traverses the length of the capillary. In addition, the intensity of response for each band will vary, producing a ratiometric response. The temporal and ratiometric responses are characteristic of a given analyte, enhancing discrimination of target analytes from potential interferents. This should translate into a reduction in sensor false alarm rates.

Nitrogen-rich compounds have a large cross section for resonance electron capture at very low incident electron energies. Although this fact has been known for a number of years, full benefit of this ubiquitous property of NOX compounds for explosives detection studies has not been fully implemented. Here we report detection of picogram to femtogram levels of TNT, 2,4-DNT and RDX in soil samples and other complex matrices. Toluene extracts as well as thermally desorbed GC-MS analyses were conducted using a JEOL GCmate II coupled to a Tunable-Energy Electron Monochromator (TEEM). Use of TEEM-GC/MS permitted rapid sweeping of electron energy and tuning of the electron monochromator and ion source while monitoring the electron capture resonance in real time. In addition, Solid-Phase Micro-Extraction (SPME) was used to selectively preconcentrate analytes prior conventional GC/MS analysis. The SPME protocol was able to screen explosives in spiked water, in concentrations below the reported detection limits. Standard solutions of TNT were prepared in the range of interest (0.5-10 ppm) and analyzed using a GC/MSD direct injection. Potential use of developed methodology in landmine environmental studies and sensors development will be discussed.

Inversion algorithms for UXO discrimination using magnetometery have recently been used to achieve very low False Alarm Rates, with 100% recovery of detected ordnance. When there are many UXO and/or when the UXO are at significantly different depths, manual estimation of the initial position and scale for each item, is a laborious and time-consuming process. In this paper, we utilize the multi-resolution properties of wavelets to automatically estimate both the position and scale of dipole peaks. The Automated Wavelet Detection (AWD) algorithm that we develop consists of four-stages: (i) maxima and minima in the data are followed across multiple scales as we zoom with a continuous wavelet transform; (ii) the decay of the amplitude of each peak with scale is used to estimate the depth to source; (iii) adjacent maxima and minima of comparable depth are joined together to form dipole anomalies; and (iv) the relative positions and amplitudes of the extrema, along with their depths, are used to estimate a dipole model. We demonstrate the application of the AWD algorithm to three datasets with different characteristics. In each case, the method rapidly located the majority of dipole anomalies and produced accurate estimates of dipole parameters.

Unexploded ordnance (UXO) discrimination is investigated using the wide band electromagnetic induction (EMI) data. The main focus of this paper is on the practical phenomenological modeling for the induced wideband EMI sensor response from different targets. Modeling for the sensor response provides feature vectors to UXO classification algorithms, and it has been proven to be very important for the improvement of the overall remediation performance. A parametric model is discussed with the emphsis on multiple offset dipole centers. The measured data from several actual targets are utilized to validate the model and to demonstrate the advantage of multiple offset dipole centers vs. single dipole center. We further illustrate the application of the model with multiple dipoles in target classifications by numerical examples. We show that the classification performance might be improved substantially. Finally, we state that the nonlinear EMI dipole model can be decomposed into a linear model. Thus it benefits from the rich literature of linear algebra and signal processing. To report one of our efforts, two methods are proposed to detect the number of dipoles blindly by the information theoretic criteria, namely the Akaike information criterion (AIC) and the minimum description length (MDL). The methods are testified using measured EMI data.

EMI (electromagnetic Induction) sensing has been shown to be a very valuable technique for target identification in UXO detection and discrimination. Numerical simulation for EMI sensing of metallic objects is difficult in part because the electromagnetic fields inside the object decay over very small distances, especially at high frequency. This challenges both numerical and analytical techniques. EMI signal inversion has been hampered heretofore by this lack of tractable solutions for basic alternative target shapes. Recently investigators developed an analytical solution formulation for general spheroids. To deal with evaluation problems in those solutions, a high frequency approximation (SPA) was developed, which turns out to have remarkably broadband applicability for steel objects. In this paper we will study its application for UXO identification. One of the most simple and effective procedures for UXO detection and recognition is "fingerprint" matching, for which we know the potential target(s) we are looking for, and we proceed by matching patterns in measured data against those archived for the specific target(s). Since the location and orientation of the target are unknown, we need to estimate them by solving the inverse problem, as a prerequisite for the matching calculations. Both numerical simulation and measurement data indicate that many realistic targets can be approximated by a representative spheroid. The SPA algorithm was then adopted to provide fast forward solution (EMI response from the representative spheroid).. Results show that in a certain range, the spheroid-based forward model can provide sufficiently accurate representation of the responses for a variety of representative target types. Its application in inversion is shown to be very beneficial for target detection and identification.

Low-metal content landmines can be particularly difficult to detect and classify. Their responses are often less than that of indigenous clutter and the small amounts of asymmetrically distributed metal results in significant changes in the signature of the mine as the sensor to target orientation varies. A number of algorithms have been previously developed in order to aid in target classification and reduce the false-alarm rate. In our work, multiple data sets were collected for each of five targets, of varying metal content, at several sensor to target heights and horizontal displacements using a prototype frequency-domain EMI sensor, the Geophex GEM-3. The data was then evaluated using one of three classification algorithms including a neural network, a matched filter, and a normalized matched filter. Here, a One Class One Network (OCON) architecture in which only one neural network makes a decision was selected for use. We will discuss the training and testing process for this algorithm. We will also show that the neural network performed much better than the matched filter but slightly worse than the normalized matched filter. In addition, the results demonstrate the necessity of training the algorithms with spatially collected data when precise sensor centering is not possible.

Trace explosive detection is a major technological challenge. Spectroscopic characterization of explosive traces is a major step toward explosive detection strategies and sensor development. We report here on white light imaging measurements and Raman microscopy, atomic force microscopy (AFM), scanning electron microscopy (SEM) and energy dispersed X ray analysis (EDX) for the characterization RDX nanoparticles deposited on glass substrates surfaces. The RDX nanoparticles were prepared by exposure of glass substrate surfaces to an aerosol jet containing RDX. An average RDX particle size of 300 nm is determined from the SEM measurements. The spectroscopic signature of the RDX nanoparticles between 750 and 950 cm^-1 is dominated by the ring breathing mode centered at about 877 cm^-1. The smallest particle characterized with vibrational spectroscopy measurements are about 750 nm in size.

Laser Induced Breakdown Spectroscopy (LIBS) is an atomic emission spectroscopic technique that utilizes a pulsed laser to create a microplasma on the target together with an array spectrometer to capture the transient light for elemental identification and quantification. LIBS has certain important characteristics that make it a very attractive sensor technology for military uses. Such attributes include that facts that LIBS (1) is relatively simple and straightforward, (2) requires no sample preparation, (3) generates a real-time response, and (4) only engages a very small sample (pg-ng) of matter in each laser shot and microplasma event, (5) has inherent high sensitivity, and (6) responds to all forms of unknowns, and, therefore, is particularly suited for the sensing of dangerous materials. Additionally, a LIBS sensor system can be inexpensive, configured to be man-portable, and designed for both in-situ point sensing and remote stand-off detection with distances of up to 20-25 meters. Broadband LIBS results covering the spectral region from 200-970 nm acquired at the Army Research Laboratory (ARL) under laboratory conditions for a variety of landmine casings and explosive materials. This data will illustrate the potential that LIBS has to be developed into a hand-deployable device that could be utilized as a confirmatory sensor in landmine detection. The concept envisioned is a backpack-size system in which an eyesafe micro-laser is contained in the handle of a deminer's probe and light is delivered and collected through an optical fiber in the tapered tip of the probe. In such a configuration, analyses can be made readily by touching the buried object that one is interested in identifying.

Nomadics has developed a novel sensing technology that detects the chemical signature of explosives emanating from buried landmines. Canines have demonstrated the ability to detect these signatures, but use of canines for this task presents a number of logistical and physical limitations that can be overcome by use of chemical sensors. Nomadics is the exclusive licensee of novel amplifying fluorescent polymer materials developed by the Massachusetts Institute of Technology (MIT). These materials enable detection of ultra-trace concentrations of nitroaromatic compounds such as TNT, the most commonly utilized explosive in the production of landmines. When vapors of nitroaromatics are presented to the sensor, the fluorescent polymers emit light at a greatly reduced intensity, a property that enables rapid detection of trace quantities of explosives using relatively low-cost electronics and optics. Studies performed by Jenkins et al suggest that the chemical signature of a landmine is heterogeneous and can be dispersed a significant distance from the location of the mine. Because the signature is not highly localized and is not characterized by a well-defined concentration gradient, the sensor may have difficulty indicating the exact position of a mine, especially in high-density minefields. Conversely, if the chemical signature extends some distance from the mine position, the sensor may have utility in detecting the edges of minefields. In combat scenarios, this will allow commanders to select safe paths for personnel and vehicles. This paper will present the latest findings related to minefield edge detection at several test sites.

The research center established by Army Research Office under the Multidisciplinary University Research Initiative program pursues a multidisciplinary approach to investigate and advance the use of complementary analytical techniques for sensing of explosives and/or explosive-related compounds as they occur in the environment. The techniques being investigated include Terahertz (THz) imaging and spectroscopy, Laser-Induced Breakdown Spectroscopy (LIBS), Cavity Ring Down Spectroscopy (CRDS) and Resonance Enhanced Multiphoton Ionization (REMPI). This suite of techniques encompasses a diversity of sensing approaches that can be applied to detection of explosives in condensed phases such as adsorbed species in soil or can be used for vapor phase detection above the source. Some techniques allow for remote detection while others have highly specific and sensitive analysis capabilities. This program is addressing a range of fundamental, technical issues associated with trace detection of explosive related compounds using these techniques. For example, while both LIBS and THz can be used to carry-out remote analysis of condensed phase analyte from a distance in excess several meters, the sensitivities of these techniques to surface adsorbed explosive-related compounds are not currently known. In current implementations, both CRDS and REMPI require sample collection techniques that have not been optimized for environmental applications. Early program elements will pursue the fundamental advances required for these techniques including signature identification for explosive-related compounds/interferents and trace analyte extraction. Later program tasks will explore simultaneous application of two or more techniques to assess the benefits of sensor fusion.

Step-frequency ground penetrating radar (SFGPR) is a prominent sensor in current buried land mine and unexploded ordnance (UXO) detection systems. Often GPR data is presented in its raw form and it is left to the signal processor to condition the signal. Discussed are the basics of SFGPR and how to condition the data with a minimum of a priori information. Qualitative comparison is shown between first order simulations and measured SFGPR data. Upon conclusion, detection and classification features based on the Radon transform are presented.

Many cart- and vehicular-based UXO detection systems employ GPS receivers to accurately determine the system's 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.

Detection and remediation of unexploded ordnance (UXO) represents a major challenge. The detection problem is exacerbated by the fact that on sites contaminated with UXO, extensive surface and sub-surface clutter and shrapnel is also present. Traditional methods used for UXO remediation have difficulty distinguishing buried UXO from these anthropic clutter items as well as from naturally occurring magnetic geologic noise, and thus incur prohibitively high false alarm rates. In this research, model-based statistical signal processing techniques are applied to field data from magnetometer and electromagnetic induction (EMI) sensors in order to determine to what degree such an approach results in false alarm mitigation. Features of the target signatures are extracted by inverting the measured sensor data associated with an anomaly using the associated physical, or forward, model. The statistical uncertainty in the feature space is explicitly treated using statistical processors, including generalized likelihood ratio tests and support vector machines, to discriminate targets from clutter. This approach has been evaluated on data collected in a recent field trial that was performed at JPG. Results are presented for one area in which ground truth was known, and for two others in which the ground truth was not known. Substantial reduction of the false alarm rate is achieved for two different platforms, the GEM-3 and the MTADS system. For example, using data from the GEM-3 in one area, the number of false targets was reduced from 181 to 20 with 100% detection of all UXO objects.

This paper describes the application of a 2-dimensional (2-D) lattice LMS algorithm for anomaly detection using the Wichmann/Niitek ground penetrating radar (GPR) system. Sets of 3-dimensional (3-D) data are collected from the GPR system and these are processed in separate 2-D slices. Those 2-D slices that are spatially correlated in depth are combined into separate “depth segments” and these are processed independently. When target/no target declarations need to be made, the individual depth segments are combined to yield a 2-D confidence map. The 2-D confidence map is then thresholded and alarms are placed at the centroids of the remaining 8-connected data points. Calibration lane results are presented for data collected over several soil types under several weather conditions. Results show a false alarm rate improvement of at least an order of magnitude over other GPR systems, as well as significant improvement over other adaptive algorithms operating on the same data.

The NVESD Mine Lane Facility has recently undergone an extensive renovation. It now consists of an indoor, dry lane portion, a greenhouse portion with moisture-controlled lanes, a control room, and two outdoor lanes. The indoor structure contains six mine lanes, each approximately 2.5m (width) × 1.2m (depth) × 33m(length). These lanes contain six different soil types: magnetite/sand, silt, crusher run gravel (bluestone gravel), bank run gravel (tan gravel), red clay, and white sand. An automated trolley system is used for mounting the various mine detection systems and sensors under test. Data acquisition and data logging is fully automated. The greenhouse structure was added to provide moisture controlled lanes for measuring the effect of moisture on sensor effectiveness. A gantry type crane was installed to permit remotely controlled positioning of a sensor package over any portion of the greenhouse lanes at elevations from ground level up to 5m without shadowing the target area. The roof of the greenhouse is motorized, and can be rolled back to allow full solar loading. A control room overlooking the lanes is complete with recording and monitoring devices and contains controls to operate the trolleys. A facility overview is presented and typical results from recent data collection exercises are presented.

Landmines are buried typically in the top 30 cm of soil. A number of physical, chemical and electromagnetic properties of this near-surface layer of ground will potentially affect the wide range of technologies under development worldwide for landmine detection and neutralization. Although standard soil survey information, as related to conventional soil classification, is directed toward agricultural and environmental applications, little or no information seems to exist in a form that is directly useful to humanitarian demining and the related R&D community. Thus, there is a general need for an information database devoted specifically to relevant soil properties, their geographic distribution and climate-driven variability. A brief description of the various detection technologies is used to introduce the full range of related soil properties. Following a general description of the need to establish a comprehensive soil property database, the discussion is then narrowed to soil properties affecting electromagnetic induction metal detectors - a problem of much restricted scope but of immediate and direct relevance to humanitarian demining. In particular, the complex magnetic susceptibility and, to a lesser degree, electrical conductivity of the host soil influence the performance of these widely used tools, and in the extreme instance, can render detectors unusable. A database comprising these properties for soils of landmine-affected countries would assist in predicting local detector performance, planning demining operations, designing and developing improved detectors and establishing realistic and representative test-evaluation facilities. The status of efforts made towards developing a database involving soil electromagnetic properties is reported.

Ground penetrating radar and thermal sensors hold much promise for the detection of non-metallic land mines. In previous work we have shown that the performance of ground penetrating radar strongly depends on field soil conditions such as texture, water content, and soil-water salinity since these soil parameters determine the dielectric soil properties. From soil physics and field measurements we know that the performance of thermal sensors also strongly depends on soil texture and water content. There is it critical that field soil conditions are taken into account when radar and thermal sensors are employed. The objectives of this contribution are (i) to make an inventory of readily available soil data bases world wide and (ii) to investigate how the information contained in these data bases can be used for derivation of soil dielectric and thermal properties relevant for operation of land mine sensors.

This study aims to move towards a reliable method for synthesizing artificial soils that emulate the effect of real soils on pulse-induction metal detectors. The signals resulting from some mineralized soils can greatly impair the effectiveness of mine detectors. We analysed a sample of one such soil from Cambodia, using resistivity measurement, X-ray diffraction, magnetic susceptibility measurement and electron microscopy. The physical origin of the soil signal was found to be high proportions of small grained magnetite. By mixing a U.K. topsoil with finely crushed magnetite, a synthetic soil with similar magnetic susceptibility to the real Cambodian soil was created. Other problematic soils from around the world were also synthesized. Comparative tests with a pulse induction metal detector showed decay characteristics for the artificial soils to be within 13% of those for the real soil. We achieved close susceptibility matching between artificial and real soils, either for a 2kHz measurement or a pulse-induction measurement. In our experiments it was not possible to match with both measurements at once. We propose that this discrepancy is due to differences in the average properties of magnetic grains between those in real soil and the magnetite concentrate used for the synthesis tests.

Soil magnetic susceptibility is always greater than zero and is detectable using an electromagnetic (EM) induction sensor. When the frequency-domain EM response is affected by magnetic polarization, the in-phase component becomes negative at the low frequency and proportional to the ground magnetic susceptibility. The in-phase measurement can thus be used to compute the apparent magnetic susceptibility. This approach provides a means of detecting a buried object based on it susceptibility contrast to the host medium. For example, an M19 anti-tank mine is physically large (33cm×33cm×9cm) but has so little metal that metal detectors can miss it. When an M19 is buried in soil, it produces a cavity in magnetic susceptibility, which may be detected as a region of low or anomalous apparent susceptibility compared to the surrounding area. We derived a simple formula to compute the apparent magnetic susceptibility from the in-phase data at the resistive limit. The behavior of the apparent susceptibility for layered earth models has been studied using synthetic data. Apparent susceptibility anomalies may be predicted from these studies based on the susceptibility contrast, and geometry of the sensor and target. Finally, we present experimental data obtained using two sensors, a GEM-2 and a GEM-3.

The recently developed physics-based "mean field" formalism for efficiently computing the time-domain response of compact metallic targets is applied to the solution of model inverse problems for remote classification of buried UXO-like targets. The formalism is first used to compute model forward scattering data, in the form of time-domain decay curves as measured by EMI or magnetic field, for a sequence of canonical ellipsoidal target shapes of various geometries. This data is subsequently used as input to a genetic algorithm-based inversion routine, in which the target parameter model space, comprised of target shape, conductivity, location, orientation, etc., is efficiently searched to find the best fit to the data. Global search procedures, such as genetic algorithms, typically require the forward scattering solution for hundreds, or perhaps thousands, of candidate target models. To be practical, these forward solutions must be rapidly computable. Our solution approach has been specifically designed to meet this requirement. Of special interest is the ability of the inversion algorithm to distinguish robustly between UXO-like targets, modelled here as cylindrically shaped prolate spheroids, and, say, flat sheet-like clutter targets, modelled as very thin oblate spheroids.

The augmentation of a currently employed baseline feature set for mine classifier design by “transform” or “moment” derived features, e.g. such as Discrete Cosine Transform and Pseudo-Zernike Moments, results in an aggregate feature set which is large in size. A “traditional” approach to this problem in the context of using multilayer perceptron(MLP) neural networks for classification consists first in the use of feature selection techniques, followed by some cross-validation based training algorithm. In this paper we contrast results obtained using the described “traditional” approach, with those obtained from using the Support Vector Machine(SVM) based framework for classifier design. The SVM approach is regarded as more attractive for large feature sets due to the optimization of a criterion in training, which is closely related to theoretical bounds on classifier generalization ability.

Thermal sensors hold much promise for the detection of non-metallic landmines. However, the prediction of their thermal signatures depends on a large number of factors. In this paper, an analytical solution for temperature propagation through homogeneous and layered soils is presented to predict surface temperatures as a function of soil heat flux amplitude, soil texture, soil water content, and thermal properties and burial depth of the landmine. Comparison with the numerical model HYDRUS-2D shows that the relatively simple analytical solution proposed here is reasonably accurate. The results show that an increase in soil water content has a significant effect on the thermal signature, as well as on the phase shift of the maximum temperature difference. Different soil textures have relatively little effect on the temperature at the surface. The thermal properties of the mine itself can play a significant role. It is shown that for most soils 10 cm is the maximum burial depth to produce a significant thermal signature at the surface.

Factors controlling the distribution and intensity of soil magnetic susceptibility (MS) and electrical conductivity (EC) were investigated. The purpose was to determine the factors to be considered in predicting MS and EC characteristics of soils in landmine-affected areas and in developing effective landmine detection systems and strategies. Results indicate that knowledge of bedrock geology, soil weathering and transportation (wind and water) history is essential to predict soil MS and EC characteristics. These factors determine the distribution, concentration and mineral type (e.g. ferromagnetic and clay minerals) in soil. For example, fluctuating water tables in tropical climates could produce soils rich in ferromagnetic minerals at the surface, even though their source (bedrock) may have low iron content. Also, subsequent weathering may change these minerals to high or low MS values. Although high clay concentrations homogeneously distributed may not produce high soil EC values, a low clay content concentrated in a single layer may produce extremely high EC values. These suggest that bedrock geology, agricultural soil, air photo and airborne geophysical survey maps can be used for predicting soils MS and EC characteristics of landmine-affected areas. Laboratory and surficial geophysical surveys are techniques for use in validation.

The inversion of surface wave propagation measurements to determine soil properties within a few meters of the surface is being investigated to facilitate the development and simulation of seismic landmine detection techniques. Knowledge of soil types, soil material properties, inhomogeneities, stratification, water content, and nonlinear mechanisms in both the propagation path and the source-to-surface coupling can be used to validate and improve both numerical and experimental models. The determination of the material properties at field test sites is crucial for the continued development of numerical models, which have shown a strong dependency on the assumed soil parameter variations in elastic moduli and density as a function of depth. Field experiments have been conducted at several test sites using both surface and sub-surface sensors to measure the propagation of elastic waves in situ with minimal disruption of the existing soil structure. Material properties have been determined from inversion of surface wave measurements using existing spectral analysis of surface waves (SASW) techniques. While SASW techniques are computer-intensive, they do not disturb the existing soil structure during testing as do borehole and trench techniques. Experimental data have been compared to results from 3-D finite-difference time-domain (FDTD) modeling of similar soil structures and measurement methods.

Tens of millions of mines are currently buried in a number of countries around the world. They cause injuries to civilians and economic damage to war-torn countries by restricting the civilian access to huge agricultural lands. Rapid Route and Area Mine Neutralisation System (RRAMNS) is a Capability Technology Demonstrator (CTD) conducted by Defence Science and Technology Organisation (DSTO) in Australia. The detection system consists of three sensors: a metal detector array, an array of ground penetrating radar (GPR), and forward looking infrared and visual imaging systems. The Kalman filter-based detection technique has previously been shown to be a powerful tool for detection of landmines from metal detector data. In this paper scalar Kalman filter-based detection algorithm has been extended to the multi-dimensional case. The new version of the detection technique has been successfully implemented in RRAMNS real-time mine detection system.

The use of near, mid wavelength and long wavelength infrared imagery for the detection of mines and concealed weapons is demonstrated using several techniques. The fusion algorithms used are wavelet based fusion and Fuzzy Logic Approach (FLA) fusion. The FLA is presented as one of several possible methods for combining images from different sensors for achieving an image that displays more information than either image separately. Metrics are suggested that could rate the fidelity of the fused images, such as, an entropy metric.

Numerous detection algorithms, using various sensor modalities, have been developed for the detection of mines in cluttered and noisy backgrounds. The performance for each detection algorithm is typically reported in terms of the Receiver Operating Characteristic (ROC), which is a plot of the probability of detection versus false alarm as a function of the threshold setting on the output decision variable of each algorithm. In this paper we present multi-sensor decision fusion algorithms that combine the local decisions of existing detection algorithms for different sensors. This offers, in certain situations, an expedient, attractive and much simpler alternative to "starting over" with the redesign of a new algorithm which fuses multiple sensors at the data level. The goal in our multi-sensor decision fusion approach is to exploit complimentary strengths of existing multi-sensor algorithms so as to achieve performance (ROC) that exceeds the performance of any sensor algorithm operating in isolation. Our approach to multi-sensor decision fusion is based on optimal signal detection theory, using the likelihood ratio. We consider the optimal fusion of local decisions for two sensors, GPR (ground penetrating radar) and MD (metal detector). A new robust algorithm for decision fusion is presented that addresses the problem that the statistics of the training data is not likely to exactly match the statistics of the test data. ROC's are presented and compared for real data.

This paper investigates the use of landmine templates in the GPR data to improve the detection accuracy for a hand-held mine detection unit. The proposed algorithm applies to the discrimination operating mode after the initial detection from the search mode. The proposed template matching-based algorithm extracts the mine templates from the data acquired during the first few sweeps, and correlates the templates from the data at subsequent sweeps to enhance the detection of landmine. The proposed technique does not have a time lag in producing detection values and a detection value is generated at each sample location. Experimental results show that the proposed template matching-based detector is able to increase the detection especially for low-metal anti-personnel mines. Based on the experiment performed over the data set collected at a test site, at 95% Pd, the proposed algorithm reduces the probability of false alarms by 66% for the low-metal anti-personnel mines and 30% for the low-metal anti-tank mines.

In this paper, the multigrid technique was considered for the solution of two dimensional Helmholtz equation with radiation boundary conditions. To achieve h-independent convergence of the iterative method, the original Helmholtz equation was transformed by using so-called “regularizing” plane wave. The recently developed “black-box” type multigrid method was adopted as a solver. This multigrid approach uses matrix dependent prolongation operator. The convergence of proposed algorithm was investigated on several test problems. Numerical results for realistic ranges of parameters in soil and mine-like targets are presented.

We have developed path-planning techniques and tools to look for paths through minefields. Our techniques seek to balance the length and risk associated with different routes through a minefield. Our methods are intended to provide battlegroup commanders powerful new tools to evaluate alternative routes while searching for low risk paths. It is well known how to find the path of shortest distance, and well known how to find the path of least risk. However, to optimize both criteria at once is a challenging problem. We have developed and compared two methods for multi-criteria path planning. We have found that the better method uses a linear combination of the criteria with a user selected risk tolerance parameter. We describe algorithms for finding the fasted bounded-risk path. We also describe the novel use of dynamic graph algorithms to quickly find new paths as the risk is changed due to the neutralization (deletion) of mines and the discovery (insertion) of mines. We will also describe an algorithm used to convert paths to a set of straight-line paths for ship navigation. Our tool allows a commander to find a path of acceptable length and risk, to explore the effect of eliminating mines, and to obtain a set of waypoints for minefield navigation.

Detection of tripwires is an active area of investigation. Researchers at the University of Missouri and the University of Florida are jointly pursuing numerous approaches to detect both the trip wires and the mines to which they are connected. Utilization of robotic vehicles capable of performing this task is one of the goals of this project. In this paper, we discuss issues related to the embedding of current versions of our tripwire detection algorithms into a small and inexpensive robot testbed for real-time experimentation. The robot is based a simple remote-controlled truck where the remote control unit has been replaced by a standard microcontroller. Sensors are added to assist navigation tasks, handled by the microcontroller, and the tripwire detection algorithms are implemented on a laptop PC with video input. There are several sophisticated algorithms that are being investigated for robust tripwire detection. The current detection algorithm that has been pruned down to run in real-time on the robotic platform consists of a Hough transform to find candidate lines followed by post-processing to score the candidate lines for the likelihood that they correspond to a trip wire. Upon detection, the robot is given a command to stop. Results of several experiments both indoors and outside in a variety of settings are described and analyzed.

Over the last decade there has been study of separating ground objects from background using multispectral imagery in the reflective spectrum from 400-2500nm. In this paper we explore using two broadband spectral modalities; visible and ShortWave InfraRed (SWIR), for detection of minelike objects, obstacles and camouflage. Whereas multispectral imagery is sensed over multiple narrowband wavelengths, sensing over two broadband spectrums has the advantage of increased signal rsulting from integrated energy over larger spectrums. Preliminary results presented here show that very basic image fusion processing applied to visible and SWIR imagery produces reasonable illumination invariant segmentation of objects against background. This suggests the use of a simplified compact camera architecture using visible and SWIR sensing focal plane arrays for performing detection of mines and other important objects of interest.

A variety of sensors have been investigated for the purpose of detecting buried landmines in outdoor environments. Mines with little or no metal are very difficult to detect with traditional mine detection systems. Ground Penetrating Radar (GPR) sensors have shown great promise in detecting low metal mines and can easily detect metal mines. Unfortunately, it can still be difficult to detect low-metal mines with GPR due to very low contrast between the mine and the surrounding medium. Acoustic-seismic systems were proposed by Sabatier et.al. and have also shown great promise in detecting low metal mines. There are now a wealth of references that discuss these systems and algorithms for processing data from these systems. Therefore, they will not be discussed in detail here. In fact, low-metal mines are easier to detect than metal mines with this acoustic-seismic systems. Low metal mines that are difficult for a GPR to detect can be quite easy to detect with acoustic-seismic systems. Sensor fusion with these sensors is of interest since together they can find a broader range of mines with relative ease. The algorithmic challenge is to determine a strategy for combining the multi-sensor information in a way that can increase the probability of detection without increasing the false alarm rate significantly. In this paper, we investigate fusion of information obtained from GPR and acoustic-seismic on real data measured from a mine lane containing three types of buried landmines and also areas containing no landmines. Algorithms are applied to data acquired from each sensor and confidence values are assigned to each location at which a measurement is made by each sensor. The GPR is used as a primary sensor. At each location at which the GPR algorithm declares an alarm, a modified likelihood-based approach is used to increase the GPR derived confidence if the likelihood that a mine is present, defined by the acoustic-based confidence, is larger than the likelihood that no mine is present. If the acoustic-derived confidence is very high, then a declaration is made even if there is no GPR declaration. The experiments were conducted using data acquired from the sensors at different times. The acoustic-seismic system collected data over a subset of the region at which the GPR collected data. Results are given only over those regions for which both sensors collected data.

As in many application areas, performance of landmine detection algorithms is judged in terms of detection and false alarm rates. It is widely accepted that single sensors cannot simultaneously achieve both high detection rates and low false alarm rates, since every sensor has its advantages and disadvantages when dealing with a large variety of landmines, from large metal-cased mines to small plastic-cased mines. The recent development of high quality sensors in conjunction with statistical signal processing algorithms has shown that there are sensors that can not only discriminate targets from clutter, but can also identify subsurface or obscured targets. Here, we utilize this identification capability in addition to contextual information in a multi-modal adaptive algorithm where the identification capabilities of multiple sensors are utilized to modify the prior probability density functions associated with statistical models being utilized by other sensors. In general, every sensor modality is associated with a specific physics-based feature set that is extracted from the sensor data. Often, the statistics describing these features are assumed to follow a Gaussian mixture density, where in many cases the individual Gaussian distributions that make up the mixture result from different target types or target classes. We utilize identification information from one sensor to modify the weights associated with the probability density functions being utilized by algorithms associated with other sensor modalities. Using both simulated and real data, this approach is shown to be improve sensor performance by reducing the overall false alarm rate.

In this submission, we report on the successful field demonstration of the LOTUS landmine detection system that took place in August 2002 near the village of Vidovice, in the Northeast of Bosnia and Herzegovina.

Under contact to the United States Army, EG&G Technical Services currently is conducting field tests of the Ground Standoff Mine Detection System (GSTAMIDS) Block 0 Engineering, Manufacturing and Development (EMD) systems. GSTAMIDS is a spiral development effort designed to provide the war fighter an incremental, near-term capability to execute on-road countermine missions. GSTAMIDS is being developed in three distinct blocks. The primary mission for GSTAMIDS Block 0 is route clearance, automatically detecting and marking all metallic and non-metallic Anti-Tank (AT) mines. It consists of a Mine Detection and marking System (MDS) mounted on a teleoperated Mine Detection Vehicle (MDV) and a Main Computer System (MCS) mounted in a Mine Protected Clearance Vehicle (MPCV). Both vehicles have overpass capability for AT mines, as well as armor anti-mine blast protection. The MPCV mounted MCS receives sensor data, along with inertial navigation data, from the MDS via an RF PCM telemetry link, automatically processes and fuses the data for mine detection and sends mine marking commands back to the MDV. The MDS sensors provide a three-meter detection swath and include nine (9) Ground Penetrating Radars (GPR), nine (9) Pulsed Magnetic Induction (PMI) metal detectors, and (as an option) two (2) long-wave infrared (LWIR) cameras. Contractor testing includes raw sensor data collection and sensor evaluation, performance (Pd and FAR), operating and storage environment, and EMI/EMC radiated emissions and susceptibility, as well as maintenance demonstrations. Testing has been conducted at a number of test mine lanes in different climates under a wide range of weather conditions over the past year and a half. This paper will present contractor test results to date.

RDX, a high power explosive used as the main charge in some landmines, was investigated in our laboratory in order to determine the spectroscopic signature to be used in its identification by means of ion mobility spectrometry (IMS) and FTIR. Density functional theory (DFT) was also used to predict structural parameters and vibrational frequencies. It was confirmed that RDX has two conformers known as the β and α-RDX phases. There are several conformers depending on the position of the NO₂ groups with respect to the triazine ring. This is important in order to determine whether RDX will have affinity for soil and/or the different materials in the ground or will be carried out by water once it starts to leak from the container holding the explosives. Different amounts of RDX were deposited on soil, aluminum plates, glass, and vinyl polymeric films. For IMS studies, the surfaces were rubbed with filter paper and the RDX was desorbed directly from the filter to the instrument inlet port. In the case of the FT-IR studies the samples were examined using an ATR coupled FTIR system. The FTIR spectra showed significant differences between the α and β phases of RDX.

In this paper, a signal processing approach for wide-bandwidth ground-penetrating-radar imagery from Non-Intrusive Inspection Technology, Incorporated (NIITEK) vehicle-mounted landmine detection sensor is investigated. The approach consists of a sequence of processing steps, which include signal filtering, image enhancement and detection. Filtering strategies before detection aid in image visualization by reducing ground bounce, systematic effects and redundant signals. Post-filter image processing helps by enhancing landmine signatures in the NIITEK radar imagery. Study results from applying this signal processing approach are presented for test minefield lane data, which were collected during 2002 from an Army test site.

The utility of Constant False Alarm Rate (CFAR) algorithms is that the selection of a detection threshold may be made independently of image intensity. However, wide application of the algorithms shows that detection values are highly dependent on scene characteristics. A threshold selection algorithm is presented for a CFAR detection algorithm. Fitting the output of the detection algorithm with a model of a portion of the theoretical results allows for background independent threshold selection.

The inferred line-spread function is an easy technique for measuring orthogonal components of the two-dimensional modulation transfer function (MTF), even from the air. However, it has been most commonly used for cameras for which the resolution is nowhere near the Nyquist frequency. The purpose of such limitation is so that the pixel sampling does not have a serious consequence on the measurement of the MTF. The binning capability of the purely digital DIPOL camera is used to demonstrate that using this method even in moderately oversampled systems does not impact results as long as certain averaging techniques are used. A brief tutorial of the normalization and pitfalls of the method will also be given so that this powerful and simple measurement will become more widely used. Example images will also be shown of mine simulators, together with polarization-product images.

This paper presents a novel deconvolution algorithm designed to estimate the impulse response of buried objects based on ground penetrating radar (GPR) signals. The impulse response is a rich source of information about the buried object and therefore very useful for intelligent signal processing of GPR data. For example, it can be used in a target classification scheme to reduce the false alarm rate in demining operations. Estimating the target impulse response from the incident and scattered radar signals is a basic deconvolution problem. However, noise sensitivity and ground dispersion prevent the use of simple deconvolution methods like linear least squares deconvolution. Instead, a new deconvolution algorithm has been developed that computes estimates adhering to a physical impulse response model and that can be characterized by a limited number of parameters. It is shown that the new algorithm is robust with respect to noise and that it can deal with ground dispersion. The general performance of the algorithm has been tested on data generated by finite-difference time-domain (FDTD) simulations. The results demonstrate that the algorithm can distinguish between different dielectric and metal targets, making it very suitable for use in a classification scheme. Moreover, since the estimated impulse responses have physical meaning they can be related to target characteristics such as size and material properties. A direct application of this is the estimation of the permittivity of a dielectric target from its impulse response and that of a calibration target.

The acoustic-seismic mine detection concept is based on the principle that an area with a buried object shows different dynamic response to acoustic excitation from that of soil. In this paper, we attempt to model and identify the dynamic behavior of a landmine under acoustic excitation for the purpose of automatic mine detection. A linear distributed model is used to model the two-dimensional vibration patterns of landmines. According to modal analysis of the model, it is shown that locations of the poles remain invariant throughout the area where a mine is buried underneath, and can be used as important features for distinguishing mines from clutter. A time-domain method that utilizes the acoustic pressure measured by a microphone as the input and the ground velocity measured by a laser Doppler vibrometer (LDV) as the output was employed to identify the model parameters including the poles. Based on the invariant property of the poles, the identified poles from neighboring measurements were combined to separate any area that show features in the spatial-spectral domain that correspond to presence of a mine.

Feature extraction is applied to mine detection data from a downward-looking ground penetrating radar (GPR) array. GPR signals have low signal-to-clutter ratio, are non-stationary in space, and vary with humidity, temperature, and soil moisture. To enhance mine-like signals and suppress false alarms, overlapping sensors allow one dimensional sensor fusion and adjacent sensors allow two dimensional sensor fusion. Maximum likelihood estimation followed by template matching perform confident detections in discriminating suspicious locations. The algorithm includes a training phase and a testing phase. In the training phase, local clutter features and their largest ten eigenvectors are extracted from known clean data using principal component analysis. In the testing phase, local background clutter is first removed from the raw data using a moving-average filter. Secondly, the de-cluttered data is projected on the significant clutter eigenvectors developed in training phase. A binary decision is made at each pixel according to template matching distances and geometric sensor structure. Receiver operating curve evaluations against test bed ground truth show improvement as singular value decomposition is enhanced by template matching and 1-D and 2-D sensor fusion.

The Geneva International Centre for Humanitarian Demining studied the needs of landmine clearance groups for equipment to carry out specific functions of mine clearance. This was done on a global level, and useful results were obtained, which will provide the basis for further analysis.