Georgia Tech is leading a spectral phenomenology research effort as a component of a Multi-University Research Initiative; these efforts are focused on studying the impact of environmental processes on electro-optical signatures. In particular, this program is conducting phenomenological studies on hyperspectral and polarimetric signatures of landmines and backgrounds in the visible and infrared wavebands. Research studies have focused on the impact of various environmental factors and processes (e.g., subsurface processes) on the resultant spectral infrared signatures. A variety of approaches have been employed in this research to gain a better understanding of the impact of the environment on the spectral and polarimetric characteristics of soil and landmine signatures. These approaches include theoretical analyses, physics-based signature modeling, field measurements, and laboratory studies. Results from these continuing studies will be presented that underscore the importance of incorporating the environmental processes into the signature analyses and analyze the impact of these processes on detection algorithm development. The results of these analyses have been propagated to algorithm developers to permit the creation of more robust processing techniques.

Commercially marketed airborne laser swath mapping (ALSM) instruments currently use laser rangers with sufficient energy per pulse to work with return signals of thousands of photons per shot. The resulting high signal to noise level virtually eliminates spurious range values caused by noise, such as background solar radiation and sensor thermal noise. However, the high signal level approach requires laser repetition rates of hundreds of thousands of pulses per second to obtain contiguous coverage of the terrain at sub-meter spatial resolution, and with currently available technology, affords little scalability for significantly downsizing the hardware, or reducing the costs. A photon-counting ALSM sensor has been designed by the University of Florida and Sigma Space, Inc. for improved topographic mapping with lower power requirements and weight than traditional ALSM sensors. Major elements of the sensor design are presented along with preliminary simulation results. The simulator is being developed so that data phenomenology and target detection potential can be investigated before the system is completed. Early simulations suggest that precise estimates of terrain elevation and target detection will be possible with the sensor design.

The Swedish Defence Research Agency (FOI) has presented several approaches to temporal analysis of thermal IR data in the application of mine detection during the years. Detection by classification is performed using a number of detection algorithms with varying, in general good, results. The FOI temporal analysis method is tested on images randomly chosen from a diurnal sequence. The test sequence show very little contrast. The reference features are taken from a known object in the scene or from a numerical model of the object of interest. In this paper variations of the method are evaluated on the same test data. Focus is on the question if increased number of data collection times affects the detection rate and false alarm rate. The ROC curves show performance better than random for all of the tested cases, and excellent for some. Detection rate increases and false alarm rate decreases with increased number of images used for some of the tested cases.

The overall objective of this paper is to improve the understanding of thermodynamic mechanisms around buried objects. The purpose is to utilize most favourable conditions for detection and also to enhance and evaluate other detection methods shown in a companion paper. This paper focuses on physical based models and simulations with measured data as boundaries for different situations of buried objects. For numerical models some assumptions of the real environment and boundaries have to be made, this paper shows the effects of different approaches of these assumptions. The investigations are carried out using a FEM approach with measured weather data as well as different sub models for the boundaries. All modelling works are carried out very in close connections with experiments with the purpose to achieve high accordance between measured and simulated values. This paper shows experimental and simulated results and discusses also the temporal analysis of thermal IR data.

It is important to understand remote sensing systems and associated platforms in the context of autonomous or semi-autonomous designs for (robotic & mechatronics) that may be affect the motion control or stabilization aspects of the imagery, scan lines or fixed points scanned. This need can be most easily conceived as being related to the reduction of risks associated with false detection as well as the risks associated with hardware and software failure and risks associated with the actual operation of sensor and platform in dangerous environments. Thus safety is ultimately our concern when it comes to risk assessment. This paper will describe (a) remote sensing systems, (b) platforms (fixed and mobile, as well as to demonstrate (c) the value of thinking in terms of scalability as well as modularity in the design and application of new systems and (d) creation of synthetic signatures obtained for detection of targets in the aquatic environment. New systems - sensing systems as well as autonomous or semiautonomous robotic and mechatronic systems will be essential to secure domestic preparedness for humanitarian reasons as well as for demining and UXO detection. These same systems hold tremendous value, if thoughtfully designed for other applications which include environmental monitoring and surveillance.

DRDC Suffield and Itres Research have jointly investigated the use of visible and infrared hyperspectral imaging for landmine detection since 1988. There has been considerable success detecting surface-laid landmines by classification of their visible/near infrared (VNIR - 400 to 1000 nm wavelength) spectral signatures, but it has not been possible to find VNIR spectral characteristics that would generically distinguish anthropogenic objects from natural features such as rocks, vegetation, soil, etc. Preliminary studies in 1998 suggested that it might be possible to develop such a generic classifier in the short wave infrared (SWIR) and that detection performance might improve. Because of a lack of available SWIR hyperspectral imagers with adequate performance for mine detection, a prototype pushbroom SWIR hyperspectral imager was developed and completed in summer 2002. The now commercially available instrument, sasi, has 160 bands over a spectral range of 850 to 2450 nm, signal to noise ratio of 400:1 with f/1.8 fore-optics, and 600 pixels over a 37.7° field of view. A number of mission flights have been carried out and excellent imagery obtained. In October 2003, Itres and DRDC Suffield personnel obtained field SWIR hyperspectral imagery in the DRDC Suffield Mine Pen of numerous surface-laid mines, one buried mine, other surface-laid human-made items, background materials and people from a horizontally scanning personnel-lift at an altitude of roughly 5 m. Preliminary indications are that a simple generic classification decision boundary should be able to distinguish surface-laid landmines from many human-made artifacts and natural materials. The buried mine was not detected, but the mine had been buried for several years and hence there would be no residual surface disturbance. Furthermore, the small sample size and limited observation time make it difficult to generalize about SWIR performance for buried mines. The instrument is described and the preliminary results of the trial, planned improvements and future research are discussed.

Near Infra-Red (NIR) offers enhanced contrast of man-made objects against vegetation. Shape detection algorithms for identifying side-attack mines in sequences of NIR imagery are described. These algorithms use morphological representations of features of the object in a network that learns features and classification simultaneously. A training set was constructed using NIR images of side attack mines. Testing sets were constructed using pairs of sequences of NIR images. Each pair of sequences contains a sequence containing a side attack mine and another sequence of the same scene with no side attack mine. Testing results from these sequences are presented.

There are three steps to performing landmine clearing operations. Mine disposal experts first detect suspect devices, discriminate the objects, and finally neutralize the landmines. Can lasers be useful in accelerating and perhaps even improving this process? They can obviously be used for direct radiation of surface landmines. We tested it successfully on five different types of landmine. For underground landmines, it is inefficient; laser energy is wasted in heating the ground, and is not applied to destroying the mine. After consideration of the traditional method for clearing landmines, it was concluded that the use of mechanical probes could provide a feasible solution to bring a laser beam directly to the mine. Since mechanical probes are in contact with the mine, they could be equipped with an optical fiber that can deliver the laser to the mine and destroy it without having to dig it out. In one single step the mine could be burned right after it is located, reducing the duration and risk of the operation and preventing its terrorist reuse. The idea of an "optical laser probe" was pushed further and a single low-cost probe was designed and patented. The investigation of hyperspectral discrimination for underground objects based on an adaptation of our device has studied in collaboration with the COPL, Quebec. We conclude that the use of an optical probe to locate a suspect object, discriminate and destroy it in the same operation signicantly improves the speed and security of landmine clearing operations.

Laser-induced breakdown spectroscopy (LIBS) spectra were collected under laboratory conditions and compiled in a library for a suite of plastic landmine casings and a variety of non-mine plastic materials on two occasions during 2004 using a Nd-YAG laser and a high-resolution broadband spectrometer to collect the full 200-980 nm LIBS spectrum.. The landmine casings examined included a broad selection of anti-personnel and anti-tank mines from different countries of manufacture. Two 'blind' tests were conducted in which LIBS spectra for the landmine casings and plastics were compared with a previously-constructed material spectral library. Using a linear correlation software, 'mine/no mine' determinations were correctly made for >90% of the samples in both tests.

Electromagnetic induction (EMI) sensing of a highly conducting and permeable metallic object buried inside a permeable medium is studied. The numerical technique is based on the method of auxiliary sources (MAS) and combined MAS/ thin skin depth approximation (MAS/TSA). The effect of the air/soil interface is accounted for via image theory, tailored for the quasi-magnetostatic case. First, the electromagnetic field inside a permeable medium originating from a state of the art EMI sensor is modeled using image theory. Image theory is then expanded to treat multi-layered cases. An analytical expression is derived for determining a half space magnetic permeability from EMI data, and is applied to measured data. The MAS/TSA is used for solving the full EMI scattering problem for a heterogeneous, highly conducting and permeable metallic object in a permeable medium. Several numerical examples are designed to show how the geological soil’s magnetic permeability can affect the signal from a buried metallic object.

In this paper we evaluate mine level detection performance of the human operator using high resolution mid-wave infrared (MWIR) imagery and compare it with the performance of automatic target recognition (ATR) like RX detector. Previous studies have shown that the anomaly detectors like the RX detector and even more sophisticated ATR techniques fall short of the performance achieved by human analyst for mine and minefield detection. There are three main objectives of the paper. First, we seek to establish performance bounds for mine detection using a single MWIR sensor under different conditions. Second, we evaluate the conditions under which the human visual system contributes significantly over and above RX anomaly detector. Third, we seek to qualitatively study the visual processes and mental models employed by the human operators to detect mines. A graphical user interface (HILgui) was developed using MATLAB to evaluate mine level detection performance for the operator. This interface is used to conduct a series of experiments examining performance for twenty subjects. The mine images varied systematically based on the time of day the images were collected, the type of terrain and type of mines. All the experiments were video-recorded and post-experiment interviews were conducted for qualitative analysis. Both qualitative and quantitative research techniques were used to gather and analyze the data. Results from different quantitative analysis including the accuracy of mine detection, propensity of false alarms and the time taken by the operator to mark individual targets are discussed. The mental models developed by the subjects for detection of mine targets are also discussed. Limitations of the current experiments and plans for future work are discussed. It is hoped that this systematic evaluation of a human operator in airborne mine detection will help in developing new and better ATR techniques and help identify critical features required in the operator interface for the warfighter-in-the-loop (WIL) minefield detection.

Recent investigations into the use of new electro-optical sensing modalities for the detection of landmines and other objects in complex backgrounds have led to the need to understand the optical properties of background materials (e.g., soils) in more detail. In particular, the use of spectral and polarimetric signatures in the optical and infrared domains has been the subject of much study; an understanding of soil, foliage, and other background optical properties and their variations is critical to evaluating the utility of these signatures. Our research examined soil emissive and polarimetric signature characteristics in the context of a real world environment; specifically, we examined the spectral properties of landmines and soils within a complex radiative environment. A modified Hapke radiative transfer model was employed to compute these properties. This paper will present a brief overview of those modifications and results of the optical property computations for several scenarios.

To improve the success of electromagnetic induction (EMI) metal detectors in identifying anti-personnel land mines buried in slightly ferromagnetic natural soils, we need to know what range of soil physical properties must be dealt with. We have therefore built a laboratory instrument for measuring complex magnetic susceptibility in inch-sized samples over a frequency range from 100 Hz to ~ 70 kHz with errors of a few percent of the sample susceptibility in a sample of ~1 milli-SIU volume susceptibility, (i.e. ~30 micro-SIU). The instrument is a symmetrical, six coil, induction spectrometer. A pair of transmitter coils in Helmholtz configuration generates a uniform magnetic field over the sample region. The magnetic moment induced in the sample is detected (mainly) by a pair of receiver coils which are closer to the sample than the transmitter pair and also (nearly) in Helmholtz configuration, so as to provide uniform sensitivity over the whole sample region. The coupling of the main receiver pair to the transmitter pair is annulled with a second pair of coils (called the reference receiver pair) situated outside the transmitter pair. The transmitter coils are energized with a wideband current. Data acquisition is by a PC computer with a 192 kHz, 24 bit, 2 channel sound card using software in written in MatLab. Although our instrument is still a prototype and its design continues to evolve, we have measured susceptibility spectra of some samples from de-mining projects in areas where false alarms are a problem and have found dispersive susceptibilities.

The detectability of buried non-conductive objects with high frequency dielectric methods depends strongly on the contrast between the dielectric properties of the object and the surrounding soil. In this study we report on effects of dielectric relaxation phenomena on the dielectric “constant” of five texturally different soils. We found that fine textured soils (loam, silt and clay) exhibit significant decreases in permittivity between 100 and 1000 MHz. It was also found that the strength of the decrease depends linearly on the soil water content and that the high-frequency permittivity of the soils follow the Topp-curve. The changes in permittivity, however, do not significantly increase the detectability of buried non-conductive low-permittivity objects in soil.

The Italian VS 1.6 anti-tank landmine pressure plate was considered in a high fidelity finite element modal analysis using the ANSYS multi-physics application. The paper describes the approach taken to obtain material properties, namely density and Young's modulus, of the plastic components of the mine. It also discusses and provides illustrations of the three-dimensional finite element domain model with boundary conditions. The paper also describes the method by which the model was verified in a laboratory setting.

Many soil physical and chemical properties interfere with landmine detection signals. Since prior knowledge of these property distributions would allow appropriate technology selection and efficient demining operations, rapid mapping of these properties over wide areas are considered for meeting military and economic constraints. As soil electrical conductivity (EC) interferes with widely used detection systems, such as metal detectors and ground penetrating radar, we have started with developing a rapid mapping technique for EC using remote sensing. Electromagnetic surveys are proven methods for mapping EC, but do not provide all information required for demining. Therefore, EC prediction by imaging of soil moisture change using radar satellite imagery acquired by RADARSAT is being tested in eastern Alberta (Canada) and northern Mississippi (U.S.A.). Areas of little soil moisture change with time are associated with high moisture retention and high clay content, suggesting higher EC. These soil characteristics are also associated with trafficability. RADARSAT soil moisture change detection images for eastern Alberta identified five areas with possible high moisture retention characteristics. Validation by soil and trafficability maps verified the predictions for more than half of the areas. Lack of some prediction accuracy is considered due to image acquisition timing and lack of physical property knowledge of some soil constituents.

In recent years it has become apparent that the performance of detection sensors for land mines and UXO may be seriously hampered by the magnetic behavior of soils. In tropical soils it is common to find large concentrations of iron oxide minerals, which are the predominant cause for soil magnetism. However, a wide range of factors such as parent material, environmental conditions, soil age, and drainage conditions control soil development. In order to predict whether magnetic-type iron oxide minerals are present it is important to understand the controlling factors of soil development. In this paper we present a conceptual model for predicting magnetic soil characteristics as a function of geological and environmental information. Our model is based on field observations and laboratory measurements of soils from Hawaii, Ghana, and Panama. The conceptual model will lead to the development of pedotransfer functions that quantitatively predict the occurrence and nature of magnetism in soils.

In this paper we present the results of a study of some soil magnetic properties in Ghana. The soils sampled formed in different parent materials: Granites, Birimian rocks, and Voltaian sandstones. We discuss the role of environmental controls such as parent material, soil drainage, and precipitation on the magnetic properties. The main conclusion of this reconnaissance study is that the eight different soil types sampled have their own unique magnetic signature. Future research will have to confirm whether this conclusion holds for other soils in Ghana. If it does, the measurement of magnetic soil properties may become a viable complement for the investigation of soil erosion, land degeneration, and pedogenesis. The magnetic soil properties measured would probably not pose any limitations for the use of electromagnetic sensors for the detection of land mines and UXO.

Magnetic soils can seriously hamper the performance of electromagnetic sensors for the detection of buried land mines and unexploded ordnance (UXO). Soils formed on basaltic substrates commonly have large concentrations of ferrimagnetic iron oxide minerals, which are the main cause of soil magnetic behavior. Previous work has shown that viscous remanent magnetism (VRM) in particular, which is caused by the presence of ferrimagnetic minerals of different sizes and shapes, poses a large problem for electromagnetic surveys. The causes of the variability in magnetic soil properties in general and VRM in particular are not well understood. In this paper we present the results of laboratory studies of soil magnetic properties on three Hawaiian Islands: O’ahu, Kaho’olawe, and Hawaii. The data show a strong negative correlation between mean annual precipitation and induced magnetization, and a positive correlation between mean annual precipitation and the frequency dependent magnetic behavior. Soil erosion, which reduces the thickness of the soil cover, also influences the magnetic properties.

The intrinsic frequency dependence to electromagnetic induction (EMI) measurements of a metallic object depends on the size, shape, and metal type. This frequency dependence, referred to as the spectral response, can be exploited in identifying or classifying the object, as well as suppressing spurious EMI responses to geology. In order to acquire the necessary multi-frequency data, a sophisticated EMI sensor is needed. The GEM-3 utilizes a continuous wave frequency-domain mode in which a hybrid current waveform is digitally generated containing energy at typically ten logarithmically spaced frequencies from a few hundred Hertz up to 48 kHz. The concentric-coil configuration suppresses the primary field at the receiver coil using a dual transmitter coil scheme. A digital Fourier transform is performed on the receiver coil emf at the selected frequencies, providing inphase and quadrature measurements. The detection channel, selected from several options, combines the multi-frequency channels in a way that suppresses the response to magnetic soil and emphasizes metallic targets. Once a target is detected, the spectral character is compared to training data spectra stored in a library for the anticipated mines, and if the goodness-of-fit to the best matching library item is within a threshold, a mine is declared. The matching algorithm essentially compares the shape of the inphase and quadrature spectra, but modifications have been incorporated to account for amplitude reasonableness, to suppress noise induced from magnetic soil, and to allow for larger percentage errors for weak-response low-metal mines.

The Geophex GEM-3 sensor was tested at a government test site comprised of 980 1-m squares containing buried landmines and clutter (metallic debris). Electromagnetic (EM) induction spectroscopy (EMIS) was used to discriminate between the landmines and clutter items. Receiver-operator characteristics (ROC) were constructed based on the results of the analysis. Approximately 92% of the landmines were correctly identified as such, with a false alarm rate of 12%. In this report, we present a comparison of our identification results against the ground truth. The EMIS method works well for high-metal mines for which the misfit threshold can be easily established, yielding a correct declaration in all cases without false alarms. For medium-metal mines, even though the misfit differences between the mines and clutter are not as clear as those for the high-metal mines, these mines were still identified at very low false alarm rates with the GEM-3 sensor. The low-metal mines may be discriminated from clutter if they yield reliable signals, but often at a much higher false alarm rate. The primary reason for this is that the EM signals from the low-metal mines are intrinsically weak and thus more subject to distortion by noise. There are several possibilities for improving the low-metal mine identification, including (1) increasing the upper limit of the frequency band to obtain a stronger signal and better defined spectra; (2) decreasing the size of the sensing head to further localize the region of sensitivity of the sensor; (3) displaying the spectral curves and performing the identification in real time to allow operator inspection of the spectral match; and (4) defining a generalized misfit that incorporates signal amplitude and possibly other spectral features such as the quadrature peak.

This paper discusses the ability of time and frequency domain electromagnetic induction systems to discriminate unexploded ordnance from clutter. Toward this end, time and frequency domain electromagnetic induction systems were built and the responses of a wide variety of targets including loops, spheres, cylinders and inert UXOs were measured. Also, time and frequency responses of test targets are numerically modeled using finite element methods to validate the experimental work. Target information is more distinct in the frequency domain than time domain. Moreover, discrimination performance of the frequency domain electromagnetic induction system was enhanced by almost a factor of two when the usual the low frequency spectrum (30 Hz to 24 kHz) was extended down to extremely low frequencies (1 Hz to 30 Hz). However, data acquisition at extremely low frequencies is a time consuming process especially if data averaging is required to achieve acceptable SNR. Therefore, in practice, it would be better to have two operating modes when using a frequency domain electromagnetic induction system; one with very few operating frequencies and the other operating in the entire band (1 Hz to 24 kHz). Once a target location is marked using the first mode, the system can be used as a “cued” sensor in the second mode, thus improving the discrimination.

Last year, we reported on a preliminary evaluation of GE’s frequency-domain EMI prototype sensor capable of measuring the wideband response of simulant and inert low metal mines at shallow depths over a frequency range from 100 Hz to 150 kHz. Since then, the prototype sensor has undergone further power and sensitivity improvements and has been taken to the field to collect signature data on targets in a calibration grid located at an Army facility in Virginia. The frequency-domain EMI responses have been analyzed by Duke University utilizing matched subspace detector (MSD) processing. The limited amount of data collected, so far, suggests that MSD processing of the frequency-domain data is a robust technique for target detection and identification. However, more data need to be collected for robust testing.

Soils that are conductive or magnetic or both can adversely affect the operation of induction metal detectors widely used to detect buried landmines. Although this effect has been known for a long time, it is only recently that efforts to rigorously characterize and quantify it has been initiated. The work reported in this paper is a part of on-going studies to clarify which properties of soil are important and to what extent they affect the performance of metal detectors which operate on the principle of electromagnetic induction. The electromagnetic response of a buried small metallic sphere is analyzed and computed. The results are used to investigate the influence that electrical conductivity and magnetic susceptibility of the host soil have on the signal produced by a target and hence on its detectability by a metal detector. The burial medium is modelled as a half-space. While soil electrical conductivity has been assumed to be real and independent of frequency, its magnetic susceptibility has been modelled as complex and frequency-dependent in general. Results include three specific cases of practical importance, namely, non-conducting soil with constant susceptibility, non-conducting soil with frequency-dependent susceptibility and non-magnetic soil with constant conductivity.

Electromagnetic induction (EMI) has prominent technique in UXO detection and discrimination research. Fast forward solutions are needed for target discrimination, which is essentially an inverse problem. We have previously developed a physically complete modeling system that includes all effects of the heterogeneities and their interactions within the object, in both near and far fields. Since the problem is highly ill-conditioned, the high order excitation modes were truncated and only the solutions of low and dominant modes were solved for. In this paper we introduce a two step approach for extracting the model parameters for both low and high order excitation modes. In the first step, high order modes are truncated. Solutions for low order modes are inferred from the measured data that mostly contain low order mode excitations. In the second step, solutions for both low and high order modes are solved by giving more weight to measurements that contains high order modes, and using the first step solutions as prior information. In the cases investigated here the two step approach provides a more accurate forward model and is still fast enough for inversion calculations in UXO detection and discrimination.

Near field (~1 m) electromagnetic induction (EMI) sensing, from 10's of Hz up to 100's of kHz, has shown significant success in detecting subsurface metallic targets. However, the discrimination of buried unexploded ordinance (UXO) from innocuous objects still remains a challenging and very expensive problem. The problem is particularly complicated in many field surveys where the data are highly contaminated with noise and clutter. In EMI data the noise and clutter are generated by the sensor, surrounding media (magnetic soil), sensor operation (motion and rotation) etc. Understanding and taking into account noise associated with the ambient environment are particularly important for developing a new generation of geological electromagneticc induction sensors as well for identification and discrimination of UXO. To address these critical issues, this paper investigates EMI scattering from a highly permeable and conducting objects subject to the state of the art of sensors placed in an infinite permeable non-conducting medium. The numerical calculation is done via the method of auxiliary sources combined with thin skin depth approximation algorithm (MAS-MAS/TSA). Using the image theory, the formulation is extended for magnetic half spaces. First the accuracy of the proposed method is checked against available analytical data for a sphere. Then several numerical results are shown and analyzed to assess the permeable soils effect on object responses, including object-soil surface interation effects and surface roughness effects. Ultimately, a user friendly EMI body of revolution code is put forward that combines these two features. It is available in the public domain, for the solution of EMI problems with single and multi (heterogeneous) objects buried inside an infinite magnetic space or in magnetic half space, subject to state of the art of sensor excitation. The code produces results in both time and frequency domains.

Mass spectroscopy methods are promising tools for the detection of trace amounts of explosive materials in a number of physical environments. The purpose of this study is to establish the kinetic energy distribution of NO as a product of photo-fragmentation of nitro-compounds like TNT and RDX on soil substrates surfaces using femto second laser pulses for molecular dissociation and subsequent mass spectrometry measurements as a function of time. We have successfully performed NO TOF measurements on TNT deposits photo decomposed with 130 femtosecond laser pulses with a 400 nm wavelength. The data is modeled with a modified Boltzman distribution. NO kinetic energy distributions resulting from TNT and RDX on soil substrate surfaces are compared.

Electromagnetic induction (EMI) has become a promising technique for UXO detection and discrimination. In most studies the effect of the ground itself is assumed small and neglected. This assumption holds up relative to ground conductivity and corresponding induced electric currents. However experience shows that magnetic effects may sometimes be significant. Here we consider the case when the ground itself is mildly permeable, a common condition. Magnetic (i.e. permeable) soil could conceivably affect the EMI response of buried metallic targets in three ways: (1) the half space of soil itself produces a scattered field, dependent on the position of the sensor, which becomes part of the background; (2) The incident field that reaches the target and the response that reaches the sensor are altered by the air-ground interface; and (3) the frequency response of the target may be altered by changes in the ratio of its magnetic permeability to that of the ground in which it is buried. Regarding the first factor, analysis shows that the response of a half space to an above-ground dipole source should be flat across the EMI spectrum. By describing our actual sensor in terms of a collection of infinitesimal dipoles, we are thus able to calculate the response due to the ground alone as a function of antenna elevation and tilt. This can then be subtracted from the data as background. Examination of realistic ground parameters at UXO sites and reference to basic magneto-quasistatic solutions allows to discount the effects of the second and third factors. We then construct a forward model which takes the soil effect into account via the first factor, and apply the model in a pattern matching approach for UXO discrimination. Example results show that the effect of soil is important in some cases, and neglecting soil effect may cause quite significant difficulty or error in UXO discrimination.

Electromagnetic induction (EMI) sensors and magnetometers have successfully detected surface laid, buried, and visually obscured metallic objects. Potential military activities could require detection of these objects at some distance from a moving vehicle in the presence of metallic clutter. Results show that existing EMI sensors have limited range capabilities and suffer from false alarms due to clutter. This paper presents results of an investigation of an EMI sensor designed for detecting large metallic objects on a moving platform in a high clutter environment. The sensor was developed by the U.S. Army RDECOM CERDEC NVESD in conjunction with the Johns Hopkins University Applied Physics Laboratory.

A new technique is presented for canceling the coupling between the coils of an electromagnetic induction sensor while using simple dipole detection coils. A secondary bucking transformer is used to cancel the coupling between the coils. The technique allows the cancellation that can be obtained using a quadrupole receive coil while maintaining the depth sensitivity and simple detection zone of a dipole coil. Simple circuit models for the sensor with some of the important parasitic effects are developed. An experimental model is developed and used to demonstrate the technique. Experimental results are presented that demonstrate more than 75 dB of cancellation up to 100 kHz and the response of the sensor to a few targets.

The objective of this paper is to study the advantage of multi-axis (vector) data over scalar one-dimensional data in the electromagnetic induction (diffusion) regime in both frequency and time domains for discriminating unexploded ordnance (UXO). Particular attention is given to the time domain. Traditional magnetometers and coil-based electromagnetic induction sensors measure only one component of the scattered magnetic field. They provide high sensitivity, but one-component magnetic field measurements provide limited information about the electromagnetic signatures of buried items, particularly for target localization and determination of target parameters. Recently much effort has been directed at developing next-generation electromagnetic geophysical sensors to collect vector data; for example, Geophex has built a new 3D GEM-3 sensor, with one transmitter and three (all Hx, Hy, Hz) receiver coils, and similar capabilities exist in the time domain. In this paper a surface magnetic charge (SMC) model, in conjunction with a differential evolution (DE) algorithm, is used to treat multi-axis data to advance, motivated by potential application to discrimination of buried UXO’s. In the SMC model the scattered magnetic field is produced by a set of magnetic charges distributed mathematically around the target location. The amplitudes of these charges is determined by matching to measured magnetic fields at a selected set of points. When the charge amplitudes are normalized by the corresponding normal component of the primary field at each location, their sum is regarded as an indication of the magnetic capacity of the object and is used as a discriminant. Once the amplitude of this normalized source set is found for each object, it can be stored for subsequent use in a discrimination algorithm. Time domain SMCs are developed for highly permeable and metallic objects buried inside a magnetic half-space. Air/magnetic ground interface effects are taken into account using image theory. Examples of synthetic electromagnetic induction data sets in the time domain are designed to show the advantage of vector over scalar data. The numerical tests for inversion of an object’s location and position from the multi-axis data and single component data will are discussed and analyzed in detail.

This paper presents an application of a combined differential evolution (DE) and surface magnetic charge (SMC) model to discriminate objects of interest, such as unexploded ordnance (UXO), from innocuous items. In entire electromagnetic induction (EMI) sensing considered here (tens of Hertz up to several hundreds of kHz), the scattered magnetic field outside the object can be represented in terms of scalar magnetic potential, from which one can obtain all scattered magnetic fields. Such fields are appropriately and readily produced mathematically by equivalent magnetic charges. The amplitudes of these charges are determined from measurement data. The surface magnetic charge model takes into account the scatterer's heterogeneity and near- and far-field effects. It is very fast and simple to implement in EMI inverse scattering algorithms. For simplification of discrimination algorithms, the frequency spectrum of the total normalized equivalent charge is investigated here as a discriminant. Two inversions scenarios are discussed: 1. Simple, when we assume that a buried object's location and orientation are known but its identity is not; and 2. General when both identity and all positional parameters are unknown. In the first case, because the task is only to identify the object, only the SMC model is required and this serves as a test of it alone. In the second case the combined DE and SMC model approach is required for identifying the object as well as its location and orientation. In this case an iterative two-step inversion procedure is used together with measured data. One step calculates an object's location and orientation, and the other calculates the amplitudes of the responding fictitious magnetic charges. Once the object's location, orientation, and spectrum of total magnetic charge are all determined, then that spectrum is compared to cataloged library data for UXO's of interest. To illustrate the applicability of the combined DE and SMC algorithm for UXO discrimination, first a simple inversion methodology is given for an actual UXO and then a general inversion approach is tested for a single object.

An improved 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, optimal subset feature selection, feature orthogonalization, 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 either “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. Two significant improvements were made to the CAD/CAC processing string by employing sub-image adaptive clutter filtering (SACF) and utilizing a repeated application of the subset feature selection / feature orthogonalization / LLRT classification blocks. A new nonlinear (Volterra) feature LLRT fusion algorithm was developed. It was shown that this Volterra feature LLRT fusion of the CAD/CAC processing strings outperforms the “M-out-of-N” and baseline LLRT algorithms, yielding significant improvements over the best single CAD/CAC processing string results, and providing the capability to correctly call all mine targets while maintaining a very low false alarm rate.

In 1999 Raytheon adapted its shallow-water Side-Looking Sonar (SLS) CAD/CAC algorithm to process side-scan sonar data obtained with the Woods Hole Oceanographic Institute's Remote Environmental Monitoring Units (REMUS) autonomous underwater vehicle (AUV). To date, Raytheon has demonstrated the ability to effectively execute mine-hunting missions with the REMUS vehicle through the fusion of its CAD/CAC algorithm with several other CAD/CAC algorithms to achieve a high probability of correct classification while maintaining a low false alarm rate. Raytheon recently reported CAD/CAC algorithm enhancements that demonstrated a significant improvement in overall CAD/CAC performance across a diverse set of environments. Additional algorithm enhancements that further improve performance over this same set of environments are described herein. The paper also presents results obtained from processing this diverse environmental data set with the enhanced Raytheon CAD/CAC algorithm, and the performance achieved by fusing the Raytheon CAD/CAC outputs with those of the other CAD/CAC algorithms.

In high-resolution sonar imagery, large variations in image background can make it difficult to reliably find targets. These variations result from irregular illumination of the sea floor, which is caused by sonar platform motion. Contributing to this problem is the fact that the spatial frequencies of the varying background can be similar to those of the target. Consequently, image-processing methods that attempt to segment image regions associated with target highlight (or shadow), are often fooled by bright (or dark) target-size patches of background. This typically results in an increase of the number of false alarms. This paper describes two image normalization methods: the Cross-Range Forward-Backward filter and the Serpentine Forward-Backward filter. Results are presented that show the impact of the image normalization on reducing false alarms.

Automatic detection of sea mines in coastal regions is a difficult task due to the highly variable sea bottom conditions present in the underwater environment. Detection systems must be able to discriminate objects which vary in size, shape, and orientation from naturally occurring and man-made clutter. Additionally, these automated systems must be computationally efficient to be incorporated into unmanned underwater vehicle (UUV) sensor systems characterized by high sensor data rates and limited processing abilities. Using noncommutative group harmonic analysis, a fast, robust sea mine detection system is created. A family of unitary image transforms associated to noncommutative groups is generated and applied to side scan sonar image files supplied by Naval Surface Warfare Center Panama City (NSWC PC). These transforms project key image features, geometrically defined structures with orientations, and localized spectral information into distinct orthogonal components or feature subspaces of the image. The performance of the detection system is compared against the performance of an independent detection system in terms of probability of detection (P_d) and probability of false alarm (P_fa).

The Airborne Littoral Reconnaissance Technology (ALRT) program has successfully demonstrated the Wide-Field Airborne Laser Diode Array Illuminator (ALDAI-W). This illuminator is designed to illuminate a large area from the air with limited power, weight, and volume. A detection system, of which the ALDAI-W is a central portion, is capable of detecting surface-laid minefields in absolute darkness, extending the allowed mission times to night operations. This will be an overview report, giving processing results and suggested paths for additional development.

Down-looking ground penetrating radar (DLGPR) has been used extensively for landmine detection. Most operational prototype systems and data collection devices use multiple transmit and receive antennas that are directed downward and mounted at the front or bottom of a moving platform. The resultant 3D datasets generated by these devices have commonalities that lend themselves to systematic analysis. While in-air measurements allow for GPR antenna characterization, it has been difficult to compare the effectiveness of a specific GPR antenna configuration as applied to the landmine detection problem. We have developed software and analysis techniques to bridge the gaps in understanding that exist between GPR antenna characterization and assessment of DLGPR landmine detection systems. We examine several datasets that were collected over an identical set of buried landmines by different DLGPR systems, and compare the mine signatures by using a simple measure of effectiveness. The simplicity of the metric allows one to separate the effects of the system from the algorithms designed to enhance mine detection performance.

In this work we investigate which bandwidth of a ground penetrating radar (GPR) is optimal for time-frequency landmine discrimination. We extracted three time-frequency features of the early-time target response from the Wigner distribution. The features were found to be relatively invariant to target depth for a data acquired with a stepped-frequency ultra-wideband GPR. The frequency sweep was from 0.3 GHz up to 6 GHz. The features allowed discrimination of two different low-metal landmines from a mine-like stone. The results were visualized in the three-dimensional feature space where each point related to a certain target represents a certain GPR scenario. For a number of scenarios we obtained two separated clusters for the landmines and the stone respectively. Numerically the quality of target discrimination can be evaluated with the Mahalanobis distance which estimates the separation between such feature clusters accounting for their size. Here we use the Mahalanobis distance as a criterion of optimality for the GPR bandwidth. Having obtained good results for the large data bandwidth, we reduce it by digital filtering with a small step in changing the cut-off frequencies, then extract the features and compute the Mahalanobis distance between the landmines and the stone. Its maximal value defines the optimal GPR lower and upper frequencies.

Multi-look processing provides a straightforward method for enhancing the quality of grainy (speckle-filled) synthetic aperture radar (SAR) imagery. This improvement in quality results from a reduction in variability of the individual pixel values brought about by non-coherent averaging of multiple, statistically independent views of the same scene. That is, the variance of the average of independent, identically distributed random variables is inversely proportional to the number of independent looks that are averaged. Of course, the quality of the averaged image depends heavily on the accuracy of the algorithm used to align the individual looks prior to averaging. The Army Research Laboratory (ARL) has recently applied multi-look processing techniques to sets of images from independent views to generate look-averaged images. In the course of this processing, we have identified the need to quantify any degradation in multi-look image quality resulting from mismatches in registration of the individual looks used to form the look-averaged image. Since these degradations can affect the performance of automatic target detection algorithms, we have also identified the need to quantify the effects of look-averaged image degradation on the performance of such algorithms. In this paper, we use X-Band SAR data to determine the extent to which translational and rotational errors in the registration of individual looks degrade the quality of the resulting multi-look averaged image. We then examine how this degradation in image quality impacts the automatic detection of small targets in the final, look-averaged image. Finally, we introduce variations in target-to-clutter ratio within each of the individual looks and analyze how these changes affect the resulting look-averaged image.

The measurement of radiation patterns of antennas in air is relatively straightforward. In contrast, the measurement of the underground pattern for ground-penetrating radar (GPR) antennas poses particular challenges. Since GPRs are equipped with transmitting and receiving paths, the combined pattern is the most useful. To measure this pattern, a probe (scatterer) can be used to reflect part of the received signal back to the receiving antenna. However, the processing on the receiving end must determine whether or not that signal comes from the probe (“desired”) or from the soil or other objects (“undesired.”) These two issues can be addressed by using a modulated scatterer, i.e., a scatterer that is modulated at a frequency much less than the carrier frequency. The modulation can be realized either electrically or optically. The advantage of the optical approach is that spurious reflections are greatly reduced since an optical fiber is used instead of current-carrying metallic cables. The electrical approach, however, allows for deeper modulation levels, which increases the level of “desired” signal at the receiver. Another issue is related to the bandwidth of the scatterer. Since GPRs are generally very broadband, it is of interest to measure their broadband radiation patterns. The scatterers in the present work are successfully made broadband by resistively loading them. The results and trade-offs resulting from this technique are shown. In summary, the modulated scatterer technique is verified to be useful for these purposes. Experiments are realized in air and underground and the corresponding radiation patterns of a set of GPR antennas are shown.

An array of compact, inexpensive radars has been developed that utilizes commercial Doppler radar transceivers and box horn array antennas operating in the near field. The 25 x 25 radar unit array acts as a set of vibrometers that detect the acoustic mechanical resonances of buried land mines by sensing the ground motion above them. Operating with the antenna apertures 35 cm above the ground, spatial resolution, grid spacing, and location accuracy are 10 cm. The outputs of each radar unit, voltages directly proportional to the amplitude of the ground vibration displacement, include in-phase and quadrature components that are digitized to 18 bits resolution. A portable PC based LabVIEW program serves as the data processor to provide 25 simultaneous DFT spectra that isolate the mine resonances. Tests showed that the radar derived spectra are clearly discernible but, to some extent, subject to fluctuations due to coherent scattering from features on rough ground surface. One partially successful attempt to ameliorate these fluctuations is discussed.

SAR-GPR is a sensor system composed of a GPR and a metal detector for landmine detection. The GPR employs an array antenna for advanced signal processing for better subsurface imaging. This system combined with synthetic aperture radar algorithm, can suppress clutter and can image buried objects in strongly inhomogeneous material. SAR-GPR is a stepped frequency radar system, whose RF component is a newly developed compact vector network analyzers. The size of the system is 30cm x 30cm x 30cm, composed from 6 Vivaldi antennas and 3 vector network analyzers. The weight of the system is less than 30kg, and it can be mounted on a robotic arm on a small unmanned vehicle. The field test of this system was carried out in March 2005 in Japan, and some results on this test are reported.

This paper describes the further engineering development and performance of the MINEHOUND affordable humanitarian mine detector, sponsored by the UK Department for International Development and developed by ERA Technology. Using a radically different patented approach from conventional ground penetrating radar (GPR) designs, in terms of the man machine interface, MINEHOUND offers simplicity of use and affordability, both key factors in humanitarian demining operations. Trials were carried out during the period 2002-2004 and have been reported at SPIE 2002 and SPIE 2004. MINEHOUND has the capability of detecting completely non-metallic mines and offers an affordable solution to hand held mine detection. The GPR is a time-domain radar transmitting 1ns duration impulses at a repetition frequency of 1MHz. The GPR transmitter- receiver and associated control and signal processing is mounted on a compact purpose designed printed circuit board 220mm by 100mm. A dedicated state of the art “Blackfin” DSP processor is used to provide all control and signal processing functions. Trials of batches of MINEHOUND are planned for 2005 in the Cambodia and Angola as well as the Balkans.

A multi-static ground-penetrating radar (GPR) has been developed to investigate the potential of multi-static inversion algorithms. The GPR consists of a linear array of six resistively-loaded vee dipoles (RVDs), a network analyzer, and a microwave switch matrix all under computer control. The antennas in the array are spaced 12cm apart so the spacing between the transmitter and the receiver pairs in the measurements are from 12cm to 96cm in 12cm increments. The size of the array is suitable for the landmine problem and scaled measurements of the buried structure problem. The RVD is chosen as an array element because it is very "clean" in that it has very little self clutter and a very low radar cross section to lessen the reflections between the ground and the antenna. The shape and the loading profile of the antenna are designed to decrease the reflection at the drive point of the antenna while increasing the forward gain. The antenna and balun are made in a module, which is mechanically reliable without significant performance degradation. The multi-static GPR operation is demonstrated on targets buried in clean sand and targets buried under the ground covered by rocks. The responses of the targets are measured by each transmitter-receiver pair. A synthetic aperture, multi-static GPR imaging algorithm is extended from conventional monostatic back-projection techniques and used to process the data. Initial images obtained from the multi-static data are clearer than those obtained from bistatic data.

An experimental polygon (control-test site) is represented, equipped by a complex of polarimetric, combined, short pulse scatterometer-radiometer systems of S-, K_u-, and K_a-band of frequencies, for bare soil, soil vegetation and land snow cover microwave reflective and emissive characteristics simultaneous and spatially coincident measurements. The polygon equipped as well by facilities for microwave devices absolute calibration, by spatially distributed stations for insitu measurements of soil moisture and temperature, and has a local meaning small weather station. This paper has an aim to attract attentions of researchers interested in such kind measurements and to invite them to perform their own or joint measurements using available facilities.

The height variation of ground surface and incorrect velocity will affect imaging processing of landmine. To eliminate these effects, ground surface topography and velocity model are needed. For effective detection of landmines, a stepped-frequency continuous-wave array antenna ground penetrating radar system, called SAR-GPR, was developed. Based on multi-offset common middle point (CMP) data acquired by SAR-GPR, we describe a velocity model estimation method using velocity spectrum technique. Also after pre-stack migration, the ground surface can be identified clearly. To compensate landmine imaging for the effect created by height variation, the ground surface displacement, a kind of static correction technique, is used based on the information of ground surface topography and velocity model. To solve the problem of incorrect velocity, we present a continuous variable root-mean-square velocity based on the velocity model. The velocity is used in normal moveout correction (NMO) to adjust the time delay of multi-offset data, and also applied to migration for reconstruction of landmine image. After the application of ground surface topography and velocity model to data processing, we could obtain good landmine images in experiment.

This paper presents a method to improve the SAR image of mine-type targets from wideband radar data that contain spectral gaps. The key concept that is exploited here is the correct identification of where the radar energy gaps map into in the ground-plane spectral domain of a targetis SAR image as a function of the targetis coordinates (particularly, its depression angle) as well as the radar system parameters. Once this is established, we apply an iterative method to recover the missing data in the targetis spectral signature. This scheme is capable of not only suppressing the undesirable side-lobes but also recovering subtle coherent (phase) information in the targetis SAR signature. Results using data from a low-frequency ultra-wideband synthetic aperture radar (SAR) designed by Mirage System will be presented.

The Army Research Laboratory has recently collaborated with Raytheon to determine the effects of various target signature phenomena on the performance of a detection pre-screening processing chain. These signature phenomena for plastic mines were predicted by ARL's high fidelity electromagnetic models and then observed in airborne X-band synthetic aperture radar (SAR) data. The agreement of the modeled results with experimental data was then used to guide pre-screener design. In this paper we present predicted plastic mine signatures generated by ARL and compare the results with actual target samples extracted from X-Band SAR data. We then briefly describe the new prescreener algorithm and examine modeling results for other frequency bands in an effort to determine if similar notions can be exploited in these bands as well.

Synthetic aperture radar (SAR) imaging with ground penetrating radar (GPR) has proven to be an effective nondestructive testing tool. Most GPR SAR imaging algorithms are based on delay-and-sum beamforming, which relies heavily on the travel path calculation in the media where targets are buried. In this paper, a computationally very efficient and convergence guaranteed algorithm is proposed for determining refraction points between boundaries of multiple planar media, which is vital to GPR SAR imaging.

Electromagnetic sensors such as ground penetrating radar and electromagnetic induction sensors are among the most widely used methods for the detection of buried land mines and unexploded ordnance. However, the performance of these sensors depends on the dielectric properties of the soil, which in turn are related to soil properties such as texture, bulk density, and water content. To predict the performance of electromagnetic sensors it is common to estimate the soil dielectric properties using models. However, the wide variety of available models, each with its own characteristics, makes it difficult to select the appropriate one for each occasion. In this paper we present an overview of the available methods, ranging from phenomenological Cole-Cole and Debye models to volume-based dielectric mixing models, and (semi-) empirical pedotransfer functions.

Most of the research on developing automatic target recognition (ATR) algorithms for acoustic-seismic landmine detection platforms has been focused on using geometric features, such as size and shape, of anomaly to distinguish between mines and clutter. This approach has achieved some success especially in detecting larger anti-tank mines. However, for smaller anti-personnel landmines, the difference in geometric features between mines and clutter can be very small, if any. To improve the detection vs. false alarm rates, it is necessary to incorporate other features into the ATR process. It has been observed from the collected acoustic data that areas with buried mines reveal more complicated surface vibration structures, such as the ring-like pattern, at certain frequencies than what a one-dimensional lumped mass-spring-dashpot model can describe. In this paper, we utilize the distributed mine/soil interaction model developed by the University of Mississippi to describe the surface vibration patterns. We develop a modified Hankel transform to extract features from areas under interrogation. Under such transform, concentration of energy is closely related to an object's physical properties. The frequency at which the energy concentration occurs corresponds to the object's natural frequency, while the corresponding Bessel basis captures its mode shape. After de-noising the transformed data, we use the frequencies, Bessel bases, and magnitudes of the energy concentrations, together with other geometric features, to form the feature vectors. We tested these features on a dataset consisting of anti-tank and anti-personnel mines as well as blank areas and metallic and non-metallic clutter. Classifiers designed based on the combined geometric and model-based features perform significantly better than those based on the geometric features alone.

Understanding the variability of the grounds acoustic properties will lead to a reduction in false alarms associated with acoustic landmine detection. Experimental measurements of the acoustic-to-seismic transfer functions performed at a US Army eastern temperate site reveal frequency modulation scales in the acoustic-to-seismic transfer function. These modulations have different spatial dependencies along and across the mine lanes. It was hypothesized that these are due to spatial dependencies of the acoustic parameters in the ground layers. It also was speculated that downward gradients in these parameters are due to additional soil strain produced by the wheels of vehicles repeatedly moving down the lane. The measured transfer functions for a few sites were analyzed. It is shown that an elastic layered model of the ground with downward gradients of sound speed in the ground layers successfully models the features observed in the experimental data. Direct time-of-flight measurements of sound speeds in and out of the wheeled tracks confirm the results obtained from the acoustic-to-seismic transfer function analysis.

A technique has been developed that exploits remote seismic sources and local measurement of the surface displacement of the ground for the detection of buried landmines. Most of the previously reported investigation of this technique has focused on non-contact displacement sensors in order to ensure the safety of the operators of both handheld and vehicle-based systems. This is not inherently a constraint that requires a non-contact sensor, but rather one requiring a sensor that is non-intrusive (i.e. its presence does not alter the measured quantity). Current research is directed toward the development of autonomous and semi-autonomous robotic systems based on this technique. Here both unit cost and power consumption are issues of comparable importance to the survival of the sensor platform. Non-intrusive surface-contacting vibrometers are therefore a reasonable alternative. Several configurations have been studied for suitable vibrometers. The configuration that has shown the most promise is based on a commercial accelerometer coupled to the ground with a small normal force and isolated from the backing structure that is used to reposition it between measurements. It is a relatively simple matter to detect seismic motion with an accelerometer. The major issue in an effective implementation of the technique is to combine reproducibility with fidelity in the measurement. These are competing goals in that reproducibility is easily achieved with large normal forces, but fidelity requires that these be small. Sufficient reproducibility for imaging purposes has been achieved with normal forces that pose no danger of landmine detonation. Unlike reproducibility, fidelity is linked to both the nature of the imposed force and to its magnitude through the nonlinearity of the soil’s elasticity. Both continuous and incremental motions of the sensor platform have been studied, although incremental movement shows the most promise for the intended application.

Acoustic detection of the landmines, which is based on the analysis of both spatial and frequency dependencies of the acoustic-to-seismic transfer function (A/S TF), exploits the difference between the mine impedance and the impedance of the surrounding ground. However, some deeply-buried mines and some types of the mines are hard to detect due to the natural variability of the ground. This work addresses the problem of false alarms and clutter (high values of the A/S TF in some frequency bands) that mimic the physics of a buried landmine. A time-scale, linear method (wavelet analysis) was utilized for improving the probability of landmine detection. Wavelet analysis of the measured signals resulted in typically stable characteristics for the undisturbed ground, the disturbed ground, and the ground with a mine. These characteristics may be used for the discrimination of false alarms and as an additional criterion to find mines that are hard to locate by traditional methods. The advantages of the suggested technique are illustrated using the experimental data.

The vibration interaction between the top-plate of buried VS 1.6 and VS 2.2 plastic, anti-tank landmines and the soil above it appears to exhibit similar characteristics to the nonlinear mesoscopic/nanoscale effects that are observed in geomaterials like rocks or granular materials. In nonlinear detection schemes, airborne sound at two primary frequencies f₁ and f₂ (chosen several Hz apart on either side of resonance) undergo acoustic-to-seismic coupling. Interactions with the compliant mine and soil generate combination frequencies that, through scattering, can effect the vibration velocity at the surface. Profiles at f₁, f₂, f₁-(f₂-f₁) and f₂+(f₂-f₁) exhibit a single peak while profiles at 2f₁-(f₂-f₁), f₁+f₂ and 2f₂+(f₂-f₁) are attributed to higher order mode shapes. Near resonance the bending (softening) of a family of increasing amplitude tuning curves (involving the surface vibration over the landmine), exhibits a linear relationship between the peak particle velocity and corresponding frequency. Subsequent decreasing amplitude tuning curves exhibit hysteresis effects. New tuning curve results for buried M 14 and VS 50 plastic anti-personal landmines along with experiments with a buried “plastic drum head” mine simulant behave similarly. Slow dynamics explains the amplitude difference in tuning curves for first sweeping upward and then downward through resonance, provided the soil modulus drops after periods of high strain.

Acoustic-to-seismic coupling-based technology using a multi-beam laser Doppler vibrometer (LDV) as a vibration sensor has proved itself as a potential confirmatory sensor for buried landmine detection. The multi-beam LDV simultaneously measures the vibration of the ground at 16 points spread over a 1-meter line. The multi-beam LDV was used in two modes of operation: stop-and-stare, and continuously scanning beams. The noise floor of measurements in the continuously scanning mode increased with increasing scanning speed. This increase in the velocity noise floor is caused by dynamic speckles. The influence of amplitude and phase fluctuations of the Doppler signal due to dynamic speckles on the phase locked loop (PLL) demodulated output is discussed in the paper. Either airborne sound or mechanical shakers can be used as a source to excite vibration of the ground. A specially-designed loudspeaker array and mechanical shakers were used in the frequency range from 85-2000 Hz to excite vibrations in the ground and elicit resonances in the mine. The efficiency of these two methods of excitation has been investigated and is discussed in the paper. This research is supported by the U. S. Army Research, Development, and Engineering Command, Night, Vision and Electronic Sensors Directorate under Contract DAAB15-02-C-0024.

Time-reversal focusing is studied in the context of an elastic-wave landmine-detection system. Time-reversal focusing has been previously applied to the system and proven to be useful in focusing energy to targets in inhomogeneous media with discrete non-uniform scattering objects. In earlier experiments, these scattering objects took the form of multiple rocks buried throughout the region of interest. In this study, the performance of time-reversal focusing is evaluated for the case of uniform discrete scattering objects. Cylindrical and spherical scatterers are buried below the surface to provide uniform scattering. In other media, high concentrations of uniform scatterers have been observed to produce super-resolution of the time-reversal focus point. In this paper, the super-resolution phenomenon is examined in the context of the elastic wave landmine detection system operating in a soil medium.

The detection of land mines using acoustic and seismic excitation is problematic due to the small amplitude of vibration that can be induced in the soil. Increasing this level reduces the requirement on a sensor’s noise floor and may be useful for nonlinear detection. For these experiments, an array of loudspeakers broadcast orthogonal noise signals to excite ground vibrations. A contacting geophone measures the system’s vibration response to all signals. We then correlate an excitation signal with the measured vibration response to approximate the system impulse response between a loudspeaker and the geophone. Time reversing the impulse response generates a pre-filter for each loudspeaker. Subsequent signals transmitted through the pre-filter and loudspeaker tend to be temporally focused at the receive location as well as greater in amplitude. Results compare vibration amplitude with and without the time reversal process for spatial locations near the mine.

Several non-contact vibrometers have been investigated for use in a seismic landmine-detection system developed at Georgia Tech; however, these non-contact vibrometers are relatively complex and expensive compared to commercially available microphones. This makes the commercial microphones an appealing alternative in applications where reduced surface-standoff distances are permissible (such as small autonomous systems or hand-held mine detectors that exploit seismic techniques). The seismic wave field involves multiple modes of propagation. Among these, the Rayleigh wave has been found to be particularly effective for the interrogation of near-surface soil where landmines are likely to be found. Thus the seismic system currently under development preferentially excites this wave type. The acoustic pressure in the air that results from a Rayleigh wave’s surface displacement can only be sensed close to the ground because Rayleigh waves are subsonic in most soils and produce evanescent acoustic fields in the air. Experimental measurements in a laboratory model have shown that buried pressure-fused landmines can be detected by measurement of the acoustic pressure within about five centimeters of the ground’s surface. Signal processing efforts including planar near-field acoustic holography, k-space filtering, and mode extraction have been used to amplify the effects of the Rayleigh wave. The signal-to-noise ratio of microphone measurements can also be improved by decreasing the microphone’s height above the soil surface or by improving the coupling of the microphone to the evanescent field with a waveguide or a horn. Experimental measurements made with the microphone compare well with direct measurements of surface displacement made using a radar-based non-contact vibrometer that has been described in previous papers.

Previous results have shown the potential of acoustic-to-seismic coupling with Laser Doppler Vibrometry for the detection of buried landmines. An important objective of the present technology is to improve the spatial resolution and the speed of the measurement. In this paper, MetroLaser reports on a whole-field digital vibrometer (WDV) that measures an entire one meter area with sub-centimeter spatial resolution in just a few seconds. The WDV is based on a dual-pulsed laser such that each pulse illuminates a one meter area on the ground, and the temporal separation between the two laser pulses can be adjusted to match the ground excitation frequency. By sweeping this excitation frequency, a displacement map of the ground at each frequency can be quickly generated. In addition, an innovative speckle repositioning strategy allows for movement of the measurement platform at reasonable speeds while still obtaining measurements with interferometric precision. This paper describes the WDV instrument and presents preliminary experimental results obtained with this system. This research is being supported by the U.S. Army RDECOM CERDEC NVESD under Contract W909MY04-C-0004.

In recent years, the acoustic technology for landmine detection has demonstrated success in field testing. Acoustic-to-seismic responses of buried landmines are exploited for locating the targets. Field experiments have demonstrated that different burial conditions and different landmines show different linear frequency responses. Therefore, the landmine detection system uses broad-band excitations. Until now, the research work for acoustic landmine detection has primarily focused on demonstrating a high probability of detection and low false alarm rate through systematic field experiments, such as blind field tests, especially for anti-tank mines. However, the speed of detection has not yet been shown to meet operational requirements. In designing a moving platform, one must know how fast an acoustic detector can acquire high-quality data, and what factors limit increased moving speed. Based upon field test results, this paper investigates the relationship between the bandwidth of the pseudo-random excitation, frequency resolution of linear response measurements, speckle noise, and reliable moving speeds of acoustic/seismic sensors.

Researchers in academia have successfully demonstrated acoustic landmine detection techniques. These typically employ acoustic or seismic sources to induce vibration in the mine/soil system, and use vibration sensors such as laser vibrometers or geophones to measure the resultant surface motion. These techniques exploit the unique mechanical properties of landmines to discriminate the vibration response of a buried mine from an off-target measurement. The Army requires the ability to rapidly and reliably scan an area for landmines and is developing a mobile platform at NVESD to meet this requirement. The platform represents an initial step toward the implementation of acoustic mine detection technology on a representative field vehicle. The effort relies heavily on the acoustic mine detection cart system developed by researchers at the University of Mississippi and Planning Systems, Inc. The NVESD platform consists of a John Deere E-gator configured with a robotic control system to accurately position the vehicle. In its present design, the E-gator has been outfitted with an array of laser vibrometers and a bank of loudspeakers. Care has been taken to ensure that the vehicle’s mounting hardware and data acquisition algorithms are sufficiently robust to accommodate the implementation of other sensor modalities. A thorough discussion of the mobile platform from its inception to its present configuration will be provided. Specific topics to be addressed include the vehicle’s control and data acquisition systems. Preliminary results from acoustic mine detection experiments will also be presented.

A new generation of Scanning Laser Doppler Vibrometer (SLDV) has been realized; based on experience and results of a former proof-of-concept design and a number of field tests. This new device SLDV III comprises a number of technical improvements in the transmitter and receiver section as well as in the evaluation of the recorded vibration signals. The subsequent paper summarizes the main features of this instrument.

In support of the International Test and Evaluation Programme for landmine detection techniques and procedures, an advisory report on the set-up of an ideal landmine test site has been prepared and submitted. This paper reviews the report and highlights the tests and site requirements that will need to be considered when testing GPR technology. The emphasis of the proposed procedures and test sites will be to evaluate the field performance of GPR systems in a realistic variety of situations and to obtain a measure of the bounds of detector performance.

The Army is interested in determining the explosive signatures of different types of munitions ranging from landmines to artillery projectiles. While a significant amount of work has been performed to determine the explosive signature of landmines, a relatively little amount of research has focused on artillery projectiles. This paper focuses on determining the levels of 2,4,6-trinitrotoluene (TNT) existing on the exterior surface of M107 artillery projectiles. The hypothesis is that there will be detectable levels of TNT on the surfaces of these projectiles due to their manufacture as well as their storage conditions. It is believed that this surface contamination provides one source of TNT that can then contaminate the surrounding environment. It is the goal of this research to determine whether or not projectiles that are manufactured and stored in similar fashions will exhibit a predictable range of TNT concentrations. This data can then be used to predict the level of environmental contamination that would occur if the projectile were present. Initially, the problem of sample collection is addressed. Specifically, quantifying the collection efficiency of different sampling techniques is investigated. This experimental aspect is crucial in determining the total quantity of TNT found of the surfaces of the projectiles. Considerations such as total amount of TNT removed compared to total amount of TNT present (on control samples) as well as the method's ease of use in the field are addressed. Data collected from M107 projectiles being stored at an Army test facility will then be analyzed and discussed.

Detection and removal of antipersonnel and antitank landmines is a great challenge and a worldwide enviromental and humanitarian problem. Sensors tuned on the spectroscopic signature of the chemicals released from mines are a potential solution. Enviromental factors (temperature, relative humidity, rainfall precipitation, wind, sun irradiation, pressure, etc.) as well as soil characteristics (water content, compaction, porosity, chemical composition, particle size distribution, topography, vegetation, etc), have a direct impact on the fate and transport of the chemicals released from landmines. Chemicals such as TNT, DNT and their degradation products, are semi-volatile, and somewhat soluble in water. Also, they may adsorb strongly to soil particles, and are susceptible to degradation by microorganisms, light, or chemical agents. Here we show an experimental procedure to quantify the effect of the above variables on the spectroscopic signature. A number of soil tanks under controlled conditions are used to study the effect of temperature, water content, relative humidity and light radiation.

The detection of trace level of explosives is a challenging field of great importance to national security and landmine detection. Chemical signatures of buried landmines are in a very complex environment. External physical conditions that affect explosive vapors and particles in soil can affect the explosive chemical signature. The chemical spectroscopic signature of the RDX in clay soil environments has been investigated by means of reflectance FT-IR microspectroscopy. The soil obtained from the University of Puerto Rico at Mayaguez was treated using the textural mechanical method in order to separate the clay from all the other components in the soil. B3LYP/6-311G** calculations performed on the low energy conformers of RDX helped to determine its most stable conformations, their symmetry, and vibrational spectra. The FT-IR technique confirmed the existence of two different RDX solid phases, known as the α-RDX and β-RDX, which have different symmetries and revealed significant differences in their spectra. The IR microspectroscopic study showed that the RDX-Clay mineral complex and its interactions can be detected using the FT-IR technique at a low concentration of 1000 part-per-millions. Variations in the clay's pH revealed changes in the RDX-Clay complex spectroscopic signature. These results also indicate that the interaction between the RDX and the clay minerals affects mainly the ring breathing, the C-N vibrations and the NO2 groups of the explosive molecules. It is suggested that the electron donor nitrogen atoms from RDX are interacting with the electron acceptor oxygen atoms of the edge sites of the clay's surface.

The transport of the chemical signature compounds from buried landmines in a three-dimensional (3D) array has been numerically modeled using the finite-volume technique. Compounds such as trinitrotoluene, dinitrotoluene, and their degradation products, are semi volatile and somewhat soluble in water. Furthermore, they can strongly adsorb to the soil and undergo chemical and biological degradation. Consequently, the spatial and temporal concentration distributions of such chemicals depend on the mobility of the water and gaseous phases, their molecular and mechanical diffusion, adsorption characteristics, soil water content, compaction, and environmental factors. A 3D framework is required since two-dimensional (2D) symmetry may easily fade due to terrain topography: non-flat surfaces, soil heterogeneity, or underground fractures. The spatial and temporal distribution of the chemical-signature-compounds, in an inclined grid has been obtained. The fact that the chemicals may migrate horizontally, giving higher surface concentrations at positions not directly on top of the objects, emphasizes the need for understanding the transport mechanism when a chemical detector is used. Deformation in the concentration contours after rainfall is observed in the inclined surface and is attributed to both: the advective flux, and to the water flux at the surface caused by the slope. The analysis of the displacements in the position of the maximum concentrations at the surface, respect to the actual location of the mine, in an inclined system, is presented.

In this work, we describe femtosecond laser-induced breakdown spectroscopy (LIBS) to detect trace amounts of explosive-related compounds (ERCs). A high-power pulsed laser is used in LIBS to form a plasma on the material surface and the optical radiation from the plasma is spectrally analyzed to determine the material composition. LIBS is minimally destructive because only a minute amount of material is consumed in the process. LIBS also enables remote analysis because only optical access to the material is needed. Femtosecond LIBS results for TNT on brass and molybdenum substrates, and RDX on molybdenum substrates are reported. We will also show the effect of detection gate delay and gate width on the enhancement of spectral information provided by LIBS.

The IST-2000-25348-BIOSENS project carried out a number of studies to assess the use of explosive detection technology for humanitarian demining. This paper presents sampling/collection technology developed, test methodology and results including comparisons with dogs and soil sampling. Findings are presented in terms of the detection of explosive from mines in the environment and demining.

Advances in the engineering of Quadrupole Resonance (QR) sensors for landmine detection have resulted in improved performance, as well as massive reductions in power, size and weight. The next generation of vehicle-mounted QR confirmation sensors is over an order of magnitude smaller and more power efficient than the system fielded in 2002 and 2003. Early prototypes have also demonstrated a significant improvement in TNT sensitivity, and similar improvements are anticipated in RDX sensitivity during Q1 2005. Blind test results from 2003 confirm the radio frequency interference and piezo-electric ringing immunity of the Quantum Magnetics QR Confirmation Sensor (QRCS).

Why is it so difficult to detect concealed shallow buried landmines while it is relatively easy to image and detect cancers within the human body? One reason is that in medical x-ray imaging, the source is on one side of the body and the detectors are on the other. This is back-illumination, the optimal orientation for x-ray imaging. Can back-illumination be used in landmine detection? That is, is it possible to generate sufficient xrays 10 or more cm below the soil surface so that suitable detectors above ground could be used to image shallow buried objects including landmines? In an x-ray tube, high voltage electron beams produce x-rays by electron deceleration (bremsstrahlung) and induced orbital transitions. It may be possible to produce 1000 amp short pulses of electrons at 30 MeV using an electron gun with multiple field emitters. (This is a section of an antiballistic missile device proposed at SPIE Defense and Security 2004.) Electron beams of such energy have range of approximately 100 m in air and 10-15 cm in soil. This 5-10 m tall device could be carried by balloon, helicopter or land vehicle. X-ray production efficiency at 30 MeV is over 50 fold higher compared to medical x-ray tube efficiency. Such a device would produce a bright isotropic source of x-rays in a subsurface plume that might be usable in landmine detection.

The energies of gamma rays produced by either fast or captured neutrons are unique to each element. The elements in the explosive of a buried mine are carbon, nitrogen, oxygen, and hydrogen. This paper analyzes data taken on buried explosive simulants. The gamma detector was a high purity germanium crystal with excellent energy resolution and the neutrons were produced by a compact deuterium-tritium accelerator, which generated 14 MeV neutrons. The gamma rays from the explosive must be separated from the large soil background. The utility for explosive detection of each element is separately analyzed in detail. Issues of normalization and gain shifts are also addressed.

A potential strategy for wide area airborne mine/minefield detection is to identify localized areas of soil that have been disturbed due to mine emplacement amidst the undisturbed soil. Disturbed and undisturbed soils are rough in varying degrees and this roughness affects the backscattering behavior at the microwave frequencies. We investigated the feasibility of using high-frequency radar (8-18 GHz) backscatter measurements to detect the residual surface disturbances caused by recent mine emplacement. Radar backscatter measurements from recently buried landmines were obtained at a government minefield data collection site. Case studies of radar backscatter from landmines buried in dirt and gravel for varying incident angles are presented. These results demonstrate that the surface roughness contrast between disturbed and undisturbed soils can be exploited to assist in mine detection operations. The maximum radar backscatter contrast between the disturbed and undisturbed soils was observed at normal incidence. The minimum contrast (radar backscatter crossover angle) occurred between 15 and 30 degree incident angles. These experimental results are shown to be consistent with rough surface scattering assumptions.

This paper outlines an approach to extrude two-dimensional existing information to support the prediction of subsurface discrete objects. The goal of this study is to develop one part of a toolkit to generate realistic, simulated subsurface material and property distributions supporting detection of surface and subsurface explosives. A sample one-meter cubic grid of heterogeneous media is simulated along with expected deviations. We explain the method used to generate a volume of soil, number of geologic features included (if any), and location. This method is expected to be used to depict a larger surface area based on representative limited visual and laboratory data. The soil volume will be used to exercise models providing a signature of subsurface media allowing the simulation of detection by various sensors of buried and surface ordnance.

Soil moisture is highly variable in space and time and affects the performance of electromagnetic sensors through its effects on thermal and dielectric properties. This research focused on characterizing soil moisture variability at spatial scales relevant to the sensing of small targets. Surface moistures of the top 6 cm of soil were collected on regular grids with an impedance probe. Measurements were made at 0.1-m resolution over 3- × 4-m and 3- × 5-m grids at a short grass site on silt loam. Tall grass and bare soil sites on gravelly silt loam were sampled at 1.0-m resolution over 20- × 30-m and 10- × 30-m grids. Exponential models fit to sample variograms of the 0.1-m resolution data show that soil moistures were spatially dependent over a distance of 0.5 m. Maximum variances (variogram sill), for data collected over a four-day span following a rainfall event, increased linearly with decreased mean moisture level as the soil dried. The revealed structures can be exploited to simulate soil moisture variation temporally and spatially. The impedance probe’s ability to reproduce variation in volumetric water content observed with conventional oven drying methods was demonstrated prior to the field experiment. Separate tests demonstrated that the probes can be used interchangeably. The impact of sparse surface grass on the moisture variation measured with the probe was also demonstrated to be small under the conditions tested.

To fully assess the environmental regime, it is necessary to identify and account for the effects of the surrounding and underlying geology that influence landscape development and geophysical tools. The geologic environment is the result of a complex network of past and present processes, including deposition and weathering by wind, water, and ice, which produce subsurface conditions difficult to interpret without expert geologic characterization and laboratory analysis. The assessment of geologic history consists of a thorough literature review, site reconnaissance and characterization, and laboratory and microscopic analyses.

The ability to detect buried land mines under a wide variety of environmental conditions is an important Army requirement. Both for interpreting signatures of mines and to ensure appropriate modeling of mine and background signatures, it is important to understand the phenomena that result in different signature patterns. The dynamic signatures can change quickly in time due to changing meteorological conditions and their impact on the mine, the soil, and on the mine-soil interaction. In field tests, infrared measurements of surface and near surface mines have shown anomalous concentric thermal signatures around the mine. The cause of these irregularities is not known. We conduct numerical multidimensional finite element calculations to investigate interactions between the meteorological conditions, the mine, and the nearby soil to elucidate the cause for these signatures. Both in-situ temperature measurements and model results show that thermal interactions between the mine and the soil are responsible for the signatures. The warm area around the mine in the nearby soil is predominant primarily at night. The warm ring effect is most likely to exist in dry soil and for mines whose heat capacity exceeds that of the soil, resulting in thermal dominance of the mine in the coupled mine-soil thermal regime. Wet soils are less likely to display the thermal contrast of the warm ring. Improved understanding of physical interactions between the mine and the background may facilitate improved discrimination between signatures of mines and of false alarms.

A high-resolution, ground-based 3D laser scanner was recently evaluated for terrestrial site characterization of variable-surface minefield sites and generation of surface and terrain models. The instrument used to conduct this research was a Leica HDS3000 3D laser scanner. Two study sites located in the mid-western United States were used for this analysis. A very dense vegetation site (Grass Site) and a bare soil site (Dirt Site) with intermittent rocks and sparse vegetation were selected for data collection to simulate both obscured and semi-obscured minefield sites. High-density scans (0.2 cm to 2.0 cm) were utilized for Cyra target acquisition and were commensurate with size and distance to target from scanner location. Medium-density scans (2.0 cm to 5.0 cm) were chosen for point cloud generation of each site with approximately 10 percent overlap between field scans. To provide equivalent, unobstructed viewing perspectives from all scan locations at each site, the scanner was positioned on a trailer-mounted, chain-driven lift and raised to a scan height of 7.62 m above the ground. Final registration to UTM projected coordinate system of the multiple scan locations for the Dirt Site and Grass Site produced mean absolute errors of 0.014 m and 0.017 m, respectively. The laser scanner adequately characterized surface roughness and vegetation height to produce contour and terrain models for the respective site locations. The detailed scans of the sites along with the inherent, natural vegetation characteristics present at each site provide real-time discrimination of site components under contrasting land surface conditions.

This paper describes the visual, spatial and thermal characteristics, and analysis of dynamic landscape conditions critical to mine detection sensors. The characterization data will be used to develop a geospatial, all-season high fidelity data set to support the modeling of synthetic battlefield environments. Surface and subsurface targets of various materials and sizes were added to natural backgrounds to measure the spectral and thermal changes due to different environmental conditions. The imagery was collected with a four-camera system, each representing the visible near infrared (VNIR), 0.4-1.0 micron spectrum, the near infrared (NIR), 0.9 to 1.7 micron spectrum, the mid-wave infrared (MWIR), 3 to 5 micron spectrum, and the long-wave (LWIR), 8 to 14 micron spectrum. The four imaging systems are mounted on a rotating boom that is raised to approximately 12- meters above ground level to match the nadir aspect airborne imaging systems. Multiple areas within the rotational footprint are selected and measured every 10-minutes through a diurnal cycle. Concurrent meteorological measurements are recorded to identify wind speed and direction, air temperature, surface conditions and relative humidity profiles. The background and target analysis procedure is a process of several steps. First, the regions of interest (ROI's) are selected that identify the target or area to be characterized. Second, the area and statistical values will be calculated for each region of interest. Third, the ROI values are compared to the onsite meteorological station.

Thermal infrared signatures of natural landscapes can vary greatly temporally and spatially. In our research, we describe our preliminary results of the spatial variability of vegetation thermal infrared signatures and vegetation density in a mid-western test minefield characterized by a linear PAR (Photosynthetically Active Radiation) ceptometer and ground-based laser radar. The linear PAR ceptometer consists of a probe about 80-cm long that contains 80 photodiodes that are sensitive to the PAR waveband. The probe calculates leaf area index (LAI, projected leaf area per unit ground area) based on sun zenith angle, leaf angle distribution, and fraction of solar beam radiation. For the 50-m by 175-m test area, the vegetation leaf area index varied from 0.1 to 4.8 and laser measured vegetation height ranged from 0.1-m to 1.9-m. The high-resolution laser radar data are used to estimate high-resolution leaf area index from the coarse PAR ceptometer measurements. These data, combined with local meteorological data, are then used to model the spatial (0.5-m) distribution of surface heat flux under a vegetation canopy.

The goal of the Countermine Computational Testbed Sensor Model Development Program is to design a software simulation for candidate airborne imaging sensors suitable for use in the remote detection of mines. The simulation takes as input several sensor parameters and a time-dependent history of location and orientation of the sensor. A scene sampling module generates an array of query ray origins and directions from the view point through the image plane to obtain radiance values from other testbed modules. Blurring effects, including those from diffraction, aberrations, detector spacing, and digitization are accounted for by using results from the validated NVTherm model. In this way, the sensor system's modulation transfer function is imposed on the image. In addition, atmosphere effects are incorporated through the use of external scattering models. If necessary, the resulting radiance image is re-sampled at desired pixel locations. Finally, the detector response characteristics are applied to the radiance image for computation of signal voltages. A noise voltage is then added and the digitization process simulated to produce the final sensor output synthetic image. The model is implemented in C++ using object-oriented programming techniques that allow for flexible extension of the simulation to different types of sensors and geometries. Model design goals, techniques, components, and specific image synthesis algorithms implementations are discussed along with the presentation of example results.

In the last few decades, the Army has developed and tested vehicular platforms for detecting landmines in roadways. These platforms include ground penetrating radar (GPR), infrared (IR) cameras, electromagnetic induction (EMI) sensors, or some combination of the three. Typically, the sensors are mounted at the front of the vehicle and are directed downward. Detecting surface laid and buried landmines at standoff require that the sensors be forward-looking. Issues of critical importance to the testing and evaluation of forward-looking sensors include geo-location and scoring. We present here detailed descriptions of tests designed to evaluate forward-looking GPR sensors used for landmine detection. We find that careful test design and analysis is necessary to accurately assess the performance of forward-looking GPR as applied to mine detection.

The Engineers Centre of Expertise of the Royal Netherlands Army (RNLA) has conducted a study on countermine in peace operations. This study, finished in 2002, concluded that the final solution to countermine will depend in the first place on better detection of buried low-metal mines, e.g. by direct detection of the explosive components in mines. Until such detection systems are available, intermediate solutions are necessary in order to assure freedom of movement in peace operations. Because countermine operations consist of a number of different activities (area preparation, detection, clearance, etc) and the suitability of the different types of available equipment depends on the scenario, the toolbox concept for countermine equipment was adopted. In 2003 a procurement process was started in order to fill this toolbox with commercial-off-the-shelf and military-off-the-shelf equipment. The paper gives a concise description of the study on countermine operations and the procurement process, and subsequently focuses on the set-up of the tests that were conducted in the framework of the procurement of hand-held mine detection systems, like metal detectors and dual-sensor mine detectors. Programs of requirements for these systems were drawn up, aiming at systems for general use and special purpose systems. Blind tests to check the compliancy to the detection performance requirements were designed and conducted in the short timeframe that was available in the procurement process. These tests are discussed in this paper, including the set-up of the test lanes, the targets used and their depths, and the role of the operator. The tests of the capability of the detectors to discriminate small targets adjacent to large targets were conducted according the guidelines of the CEN Workshop Agreement on metal detector tests. Although the results of the tests are commercially confidential, conclusions and lessons learned from the execution of these tests are presented.

Currently there are a few projects for landmine detection in Afghanistan, which is supported by the Japanese government. Some field test for landmine detection sensors have been carried out in Afghanistan and Japan. We introduce in this paper about the plan of these projects, and its evaluation tests. JST (Japan Science and Technology Agency) which is under the Ministry of Education, Culture, Sports, Science and Technology (MEXT)) is developing unmanned vehicles which are mounted sensors for AP landmine detection. The prototype of the sensors and equipments will be ready by February 2005 and will be tested in a test site in Japan by March 2005. Then, it is planned to be evaluated in Afghanistan in summer 2005. JICS (Japan International Cooperation System) which is under the Ministry of Foreign affairs (MOFA) has a project on "Developing Mine Clearance related equipment in Afghanistan". In this project, we plan to evaluate mine detectors in Afghanistan until March 2005. The evaluation test of JICS project has already started in August-Deecember 2004, in Afghanistan,. In the evaluation the both projects, we are preparing test lanes. Most of the sensors to be evaluated is a combination of a metal detector and GPR, and as for GPR, there has been not many examples of such evaluation tests. In this paper, we introduce the outline of the evaluation test, and also discuss some technical aspects of the evaluation test for the combination sensors of a metal detector and a GPR.

This paper discusses an experimental design method for test and evaluation of anti-personnel landmine detection systems using ground penetrating radar (GPR). Vehicle-mounted mine detection systems to be evaluated here have been developed by six research teams from universities and industries founded by Japan Science and Technology Agency. Sensors used are a GPR and electromagnetic induction (EMI) fused type, which provides underground images to operators. In our basic concept, the systems make no explicit alarm and the final decision whether or not a shadow in the image is a real landmine is left to the operator. This is the same way as medical doctors find cancer by reading CT images. To test these kinds of systems, i.e., to evaluate probability of detection (PD), false detection rate, and other characteristics, seven test lanes using more than 200 landmine surrogates has been designed. Since operators' pre-knowledge of the locations of buried targets significantly influences the detection results in our systems, six out of the seven test lanes are designed to be suitable for blind tests. A comprehensive test and evaluation using the designed experimental lanes is in progress for over one month, and some results obtained from the test are discussed.

With the emergence of commercially available multisensor mine detection systems, the need for standardised test and evaluation procedures becomes more pressing. For metal detectors this already has been established and is laid down in the CEN workshop document CWA14747:2003. The ITEP multisensor working group has taken the first step towards a similar document, by means of a so called "best practice" document, which would ultimately lead to a proper standard. In this paper we address various issues important for multisensor mine detection testing and evaluation and in this way hope to contribute to a draft version of the "best practice" document encouraging other parties to do the same and hence speed up the process of standardisation.

The standards relating to testing of metal detectors for demining operations are developing well, including (but not limited to) CEN Working Agreement CWA14747:2003, UNMAS Mine Action Standards and others. However, for developing multisensor mine detectors there is no agreed standard method of testing. ITEP, the International Test and Evaluation Program for Humanitarian Demining, is currently drawing together several nation's experience of testing multisensor mine detectors into a "best practice" document that could be used as the basis for standardised testing. This paper outlines the test methodology used during recent multisensor mine detector tests and discusses the issues that have arisen and lessons learned. In particular, the requirements for suitable targets, careful site preparation, measurement of appropriate environmental factors and methods of maximising useful results with limited resources are highlighted. Most recent tests have used a combined Metal Detector (MD) and Ground Penetrating Radar (GPR), but other sensor systems will be considered. An emerging test methodology is presented, along with an invitation for feedback from other researchers for inclusion into the "best practice" document.

Ground penetrating radar (GPR) is emerging as viable technology for rapid and accurate landmine detection. Although GPR has been successfully used for landmine and subsurface object detection, the performance of GPR is dependent on the type of medium the subsurface object is buried in. In a previous paper, we compared the imaging response of two antennas in three soils to steel spheres[1]. In this paper, we compare the imaging response of spheres of different materials in different soils and compute energy levels for three regions of interest in the images.

Independent Component Analysis (ICA) is applied to classify unexploded ordnance (UXO) on laboratory UXO test-field data, acquired by stand-off detection. The data are acquired by an Electromagnetic Induction Spectroscopy (EMIS) metal detector and a ground penetrating radar (GPR) detector. The metal detector is a GEM-3, which is a monostatic sensor measuring the response of the environment on a multi-frequency constant wave excitation field (300 Hz 25 kHz), and the GPR detector is a stepped-frequency GPR with a monostatic bow-tie antenna (500 MHz 2.5 GHz). For both sensors the in-phase and the quadrature responses are measured at each frequency. The test field is a box of soil where a wide range of UXOs are placed at selected positions. The position and movement of both of the detectors are controlled by a 2D-scanner. Thus the data are acquired at well-defined measurement points. The data are processed by the use of statistical signal processing based on ICA. An unsupervised method based on ICA to detect, discriminate, and classify the UXOs from clutter is suggested. The approach is studied on GPR and EMIS data, both separately and combined. The potential is an improved ability: to detect the UXOs, to evaluate the related characteristics, and to reduce the number of false alarms from harmless objects and clutter.

ALIS (Advanced Landmine Imaging System), which is a novel landmine detection sensor system combined with a metal detector and GPR, was developed. This is a hand-held equipment, which has a sensor position tracking system, and can visualize the sensor output in real time on a head-mounted PC display. In order to achieve the sensor tracking system, ALIS needs only one CCD camera attached on the sensor handle. The new hand-held system ALIS is a very compact and do not require any additional sensor for sensor position tracking. The acquired signal from the metal detector and GPR is displayed on the PC display on real time, and the sensor trace can be checked by the operator. At the same time, the operator can visually recognize the signal on the same display. The CCD captured image is superimposed with the GPR and metal detector signal, therefore the detection and identification of buried targets is quite easy and reliable. Field evaluation test of ALIS was conducted in Afghanistan, and we demonstrated that it can detect buried antipersonnel landmines, and can also discriminate metal fragments from landmines.

The Autonomous Mine Detection Sensor (AMDS) is a program to develop a suite of advanced sensor technology on a vehicular robot (PackBot). These sensors are characterized as having high detection performance and low false alarm rate. The AMDS program is sponsored by the U.S. Army CERDEC RDES Night Vision and Electronic Sensors Directorate (NVESD). The CyTerra Corporation solution to this problem is the combination of a Ground Penetration Radar (GPR) and a Metal Detector (MD) which comprise the AN/PSS-14. This paper presents the CyTerra Corporation concept and includes performance results from early government sponsored tests.

This study describes the U.S. Army Force Protection Demining System (FPDS); a remotely-operated, multisensor platform developed for reliable detection and neutralization of both anti-tank and anti-personnel landmines. The ongoing development of the prototype multisensor detection subsystem is presented, which integrates an advanced electromagnetic pulsed-induction array and ground penetrating synthetic aperture radar array on a single standoff platform. The FPDS detection subsystem is mounted on a robotic rubber-tracked vehicle and incorporates an accurate and precise navigation/positioning module making it well suited for operation in varied and irregular terrains. Detection sensors are optimally configured to minimize interference without loss in sensitivity or performance. Mine lane test data acquired from the prototype sensors are processed to extract signal- and image-based features for automatic target recognition. Preliminary results using optimal feature and classifier selection indicate the potential of the system to achieve high probabilities of detection while minimizing false alarms. The FPDS detection software system also exploits modern multi-sensor data fusion algorithms to provide real-time detection and discrimination information to the user.

Over the past several years, an enormous amount of airborne imagery consisting of various formats has been collected and will continue into the future to support airborne mine/minefield detection processes, improve algorithm development, and aid in imaging sensor development. The ground-truthing of imagery is a very essential part of the algorithm development process to help validate the detection performance of the sensor and improving algorithm techniques. The GUI (Graphical User Interface) called SemiTruth was developed using Matlab software incorporating signal processing, image processing, and statistics toolboxes to aid in ground-truthing imagery. The semi-automated ground-truthing GUI is made possible with the current data collection method, that is including UTM/GPS (Universal Transverse Mercator/Global Positioning System) coordinate measurements for the mine target and fiducial locations on the given minefield layout to support in identification of the targets on the raw imagery. This semi-automated ground-truthing effort has developed by the US Army RDECOM CERDEC Night Vision and Electronic Sensors Directorate (NVESD), Countermine Division, Airborne Application Branch with some support by the University of Missouri-Rolla.

A significant amount of airborne data has been collected in the past and more is expected to be collected in the future to support airborne landmine detection research and evaluation under various programs. In order to evaluate mine and minefield detection performance for sensor and detection algorithms, it is essential to generate reliable and accurate ground truth for the location of the mine targets and fiducials present in raw imagery. The current ground truthing operation is primarily manual, which makes the ground truthing a time consuming and expensive exercise in the overall data collection effort. In this paper, a semi-automatic ground-truthing technique is presented which reduces the role of the operator to a few high-level input and validation actions. A correspondence is established between the high-contrast targets in the airborne imagery called the image features, and the known GPS locations of the targets on the ground called the map features by imposing various position and geometric constraints. These image and map features may include individual fiducial targets, rows of fiducial targets and triplets of non-collinear fiducials. The targets in the imagery are established using the RX anomaly detector. An affine or linear conformal transformation from map features to image features is calculated based on feature correspondence. This map-to-image transformation is used to generate ground-truth for mine targets. Since accurate and reliable flight-log data is currently not available, one-time specification of a few parameters like flight speed, flight direction, camera resolution and specification of the location of the initial frame on the map is required from the operator. These parameters are updated and corrected for subsequent frames based on the processing of previous frames. Image registration is used to ground-truth images which do not have enough high-contrast fiducials for reliable correspondence. A GUI called SemiAutoGT developed in MATLAB for the ground truthing process is briefly discussed. Results are presented for ground-truthing of the data collected under the Lightweight Airborne Multispectral Minefield Detection (LAMD) program.

The warfighter analyst in the data processing ground control station plays an integral role in airborne minefield detection system. This warfighter-in-the-loop (WIL) is expected to reduce the minefield false alarm rate by a factor of 5. In order to achieve such a significant false alarm reduction and to facilitate the development of an efficient WIL interface, it is critical to evaluate different aspects of WIL operations for airborne minefield detection. Recently, researchers at the University of Missouri-Rolla have developed a graphical user interface (HILMFgui) application using MATLAB to evaluate minefield detection performance for the operator. We conducted a series of controlled experiments with HILMFgui using ten participants. In these experiments, we video-recorded all the experiments and conducted post-experiment interviews to learn more about the usability of the interface and the cognitive processes involved in minefield detection. The effect of various factors including the availability of automatic target recognition (ATR), availability of zoom and time constraints were considered to evaluate their influence on operator performance. Qualitative results of the factors affecting the warfighter performance in the minefield detection loop are discussed. Through the qualitative data analysis, we observed two different types of participants (classified here as aggressive and cautious). We also identified three primary types of mental models: mine centric, mine-field centric, and logical placement. Those who used a primarily mine focus had a substantially higher false alarm rate than those whose mental models were more consistent with a mine-field centric or logical placement perspective.

A Bayesian network (BN) is a directed acyclic graphical model that encodes probabilistic relationships among variables of interest. BNs not only provide a natural and compact way to represent the domain knowledge and encode joint probability distributions, but also provide a basis for efficient probabilistic inference. We apply BNs to wide area airborne minefield detection (WAAMD) due to their powerful representation ability of encoding the domain knowledge and their flexible structural extendibility for multi-look and multi-sensor data fusion. We first design BN models for both single-look detection and multi-look and multi-sensor data fusion and then refine them via learning from data using a structural expectation-maximization (SEM) algorithm. We evaluate the performance of our landmine detection scheme using data sets collected by three airborne ground penetrating synthetic aperture radars (GPSARs) (Lynx Ku-band, Mirage stepped-frequency (0.3 - 2.8 GHz), and Veridian X-band GPSARs) from various testing sites that have different terrain and vegetation conditions. Experimental results indicate that BNs can help improve the landmine detection performance significantly. The use of BNs for multi-look and multi-sensor data fusion is also shown to provide significant false alarm reductions.

This paper presents semantic risk estimation of suspected minefields using spatial relationships of minefield indicators extracted from multi-level remote sensing. Both satellite image and pyramidal airborne acquisitions from 900m to 30m flying heights with resolutions from 1m to 2cm resolutions are used for identification of minefield indicators. R-Histogram [1] is a quantitative representation of spatial relationship between two objects in an image. Eight spatial relationships can be generated: 1) LEFT OF, 2) RIGHT OF, 3) ABOVE, 4) BELOW, 5) NEAR, 6) FAR, 7) INSIDE, 8) OUTSIDE. R-Histogram semantics are first generated from selected indicators and metrics such as topological proximity and directional relationships are trained for soft classification of risk index (normalized as 0-1). We presented a framework of how semantic metadata generated from remote sensing images are used in risk estimation. The resultant risk index identified seven out of twelve mine accidents occurred at high risk region. More importantly, comparison with ground truth obtained after mine clearance show that three out of the four identified pattern minefields falls into the area estimated at very high risk. A parcel-based per-field risk estimation can also be easily generated to show the usefulness of the risk index.

There has been significant amount of study on the use of Ground-Penetrating Radar (GPR) for forward-looking landmine detection. This paper presents our analysis of GPR data collected at a U.S. Army site using the radar system developed by Planning Systems Inc. (PSI). One property of forward-looking systems is that a target appears in multiple radar images at different distances. To exploit this property, we divide the distance range in the radar images into a number of distance bands. Identification of potential targets is first performed in each distance band independently. Our algorithm then tracks these potential targets through multiple distance bands and computes weighted averages of their geometrical features. The persistence property of the targets is used to further reduce false alarm rates by removing potential targets that only appear spuriously. Results of landmine detection, including performance on blind test lanes, are presented.

Landmine detection using radar is a very challenging problem due to weak signal returns of landmines and extremely complicated surveying environments. In this paper, we present a new landmine detection system using forward-looking ground penetrating radar (FLGPR), which has shown a promising result in a recently conducted blind test. The system uses wavelet packet transform and the sequential feature selection algorithm to extract the most discriminant information distributed in the joint time-frequency domain for detecting landmines. We also propose a cascade training method that allows a WPT based detector to continue learning from the errors made on the unseen environment to improve its detection performance. The effectiveness of the proposed detector is demonstrated through a blind test based on the measured FLGPR data collected over an area of 14400 square meters.

We consider an information-theoretic approach for sensor management that chooses sensors and sensor parameters in order to maximize the expected discrimination gain associated with each new sensor measurement. We analyze the problem of searching for N targets with M multimodal sensors, where each sensor has its own probability of detection, probability of false alarm, and cost of use. Other information, such as the prior distribution of the targets in space and the degree of constraint of the sensor motion, is also utilized in our formulation. Performance of the sensor management algorithm is then compared to the performance of a direct-search procedure in which the sensors blindly search through all cells in a predetermined path. The information-based sensor manager is found to have significant performance gains over the direct-search approach. Algorithm performance is also analyzed using real landmine data taken with three different sensing modalities. Detection performance using the sensor management algorithm is again found to be superior to detection performance using a blind search procedure. The simulation and real-data results also both illuminate the increased performance available through multimodal sensing.

Due to the nature of landmine detection, a high detection probability (Pd) is required to avoid casualties and injuries. However, high Pd is often obtained at the price of extremely high false alarm rates. It is widely accepted that no single sensor technology has the ability to achieve the required detection rate while keeping acceptably low false alarm rates for all types of mines in all types of soil and with all types of false targets. Remarkable advances in sensor technology for landmine detection have made multi-sensor fusion an attractive alternative to single sensor detection techniques. Hence, multi-sensor fusion mine detection systems, which use complementary sensor technologies, are proposed. Previously we proposed a new multi-sensor fusion algorithm called Multi-modal Iterative Adaptive Processing (MIAP), which incorporates information from multiple sensors in an adaptive Bayesian decision framework and the identification capabilities of multiple sensors are utilized to modify the statistical models utilized by the mine detector. Simulation results demonstrate the improvement in performance obtained using the MIAP algorithm. In this paper, we assume a hand-held mine detection system utilizing both an electromagnetic induction sensor (EMI) and a ground-penetrating radar (GPR). The hand-held mine detection sensors are designed to have two modes of operations: search mode and discrimination mode. Search mode generates an initial causal detection on the suspected location; and discrimination mode confirms whether there is a mine. The MIAP algorithm is applied in the discrimination mode for hand-held mine detection. The performance of the detector is evaluated on a data set collected by the government, and the performance is compared with the other traditional fusion results.

CyTerra's dual sensor HSTAMIDS system has demonstrated promising landmine detection capabilities in extensive government-run field tests. Further optimization of the successful PentAD algorithm is desirable to maintain the high probability of detection (Pd) while lowering the false alarm rate (FAR). PentAD contains several input parameters, making such optimization using standard Monte-Carlo techniques too computationally intensive. Genetic algorithm techniques, which formerly provided substantial improvement in the detection performance of the metal detector sensor algorithm alone, have been applied to further optimize the numerical values of the dual-sensor algorithm parameters in more practical time frames. Genetic algorithm techniques have also been applied to choose among several sub-models and fusion techniques to potentially train the HSTAMIDS system in new ways. An analysis of genetic algorithm results has indicated that ground type may have a significant impact on the optimal parameter set. In this presentation we discuss the performance of the resulting ground-type based genetic algorithm as applied to field data.

In this paper, we studied the effects of the preprocessing techniques over buried object detection performance. We examined different preprocessing techniques applied before the detection algorithm proposed by Sezgin [1]. It is obtained that used preprocessing techniques decreases false alarm rates in real environment. We used different size of objects and burial depths for both metallic and non-metallic targets.

Identifying unique patterns of energy in ground penetrating radar images plays an important role in landmine/clutter discrimination. Many different geometric features, including size and the distribution of energy values in a radar image, can be exploited in mine detection and discrimination. The granulometry of a random set (image), computed by measuring the integral of a sequence of closings with elements from a family of increasing homothetic shapes, yields a size distribution that can be used for texture analysis or object detection and discrimination. An important complement to granulometries for discrimination is the compactometry, a feature we have identified that is computed by measuring the integral of a sequence of increasing concentric homothetic subsets of a random set. The compactometry yields a characterization of the concentration of the density distribution of the random set at a given point. This paper investigates the properties of compactometry and its derivative, the density spectrum, and demonstrates how they can be used together with granulometry to address the problem of landmine/clutter discrimination using ground-penetrating radar sensors.

We present in this paper the use of frequency domain features deduced from the energy density spectrum to improve the detection of landmines. The energy density spectrum is obtained from the GPR measurements at an alarm location, and a method to estimate the energy density spectrum is proposed. The energy density spectrum is shown to reveal distinct characteristics between landmine and clutter objects and therefore can be explored for their discrimination. The robustness and consistency of the frequency domain features are demonstrated through two different GPRs. The fusion of frequency domain features and time-domain features is also examined. Experimental results at several test sites confirm the advantages of frequency domain features to better discriminate between mine targets and clutter objects, and the effectiveness of fusion in frequency and time domain features to improve detection performance.

In this work we present an application of matched subspace detectors to the problem of target detection and identification using ground penetrating radar data. In particular we apply sets of matched subspace detector filter banks to data containing both anti-personnel and anti-tank targets as well as metallic and non-metallic clutter objects. Current results indicate the potential for robust target detection and identification but further improvements via subspace modeling and signal extraction/enhancement may also improve performance.

The imaging of subsurface targets, such as landmines, using Ground Penetrating Radar (GPR) is becoming an increasingly important area of research. Conventional image formation techniques expend large amounts of computational resources on fully resolving a region, even if there is a large amount of clutter. For example, standard backprojection algorithms require O(N³). However, by using multi-resolution techniques-such as quadtree-potential targets and clutter can be discriminated more efficiently with O(N²log₂N). Because prior work has focused on the imaging of surface targets, quadtree techniques have mostly been developed for 2D imaging. Target depth adds another dimension to the imaging problem; therefore, we have developed a 3D quadtree algorithm. In this case, the mine field is modeled as a volume that is sub-divided at each stage of the quadtree algorithm. From each of these sub-volumes, the energy intensity is calculated. As the algorithm proceeds to finer resolutions, the energy in region containing a potential target increases, while that of background noise decreases. A multi-stage detector applied on intermediate quadtree data uses this change in energy to discriminate between regions of targets and clutter. This is advantageous because only the regions containing likely targets are investigated by additional sensors that are relatively slow in comparison to GPR (e.g. seismic or EMI sensors). This algorithm is tested on synthetic and experimental data collected from a model mine field at Georgia Institute of Technology. Even under near field and small aperture conditions, which hold for the mine detection case, test results show that target location information can be gathered with processing using the 3D quadtree algorithm.

Adaptive methods are currently in use for GPR to detect shallow-buried targets in the presence of ground-bounce that may be orders of magnitude greater than the target response. One such method that has been used with noteworthy success to reduce the soil background has been called the Principal Components Algorithm (PCA) that applies the Mahalanobis distance measure, a quadratic form, to distinguish the energy in a target from the background energy. We discuss an alternative vector subspace discrimination method borrowed from adaptive spatial filtering that has been used successfully for suppressing Doppler-spread ionospheric clutter in a high frequency radar application [4,5]. This method first estimates the background vector subspace containing the clutter during a training phase and then projects the current response on the orthogonal complement of the estimated subspace to extract the desired signal during the search phase. Discrete clutter in the output response is reduced without attenuating the target signal by applying constraints derived from the estimated target eigenstructure, either measured or modeled. Mathematical details of the algorithm are presented in an appendix. Comparison of results of the PCA and vector subspace projection methods applied to a stepped-frequency GPR have shown the performance of the proposed algorithm to be better in terms of clutter suppression and principal component economy while also retaining the phase and amplitude of the signal samples to facilitate array processing useful for forward looking ground penetrating radar.

We are developing a new landmine detection system, called advanced landmine imaging system (ALIS), which is equipped with metal detector (MD) and ground penetrating radar (GPR). Although this is a hand-held system, we can record the MD and GPR signal with the sensor position information acquired by CCD camera. Therefore, 2D MD image and 3D GPR image are possible after signal processing. But because ALIS is a hand-held system, the sensor position is random when it is operated in the field. So interpolation processing is used to deal with the problem and offer grid data set for both MD and GPR. Good MD image can be achieved after interpolation. Also, interpolation can prepare good data set for migration to get good horizontal slice image. After interpolation, 3D diffraction stacking migration with migration aperture is used to refocus the scattered signals and enhance the signal-clutter ratio for reconstructed good GPR image. The ALIS was tested in Afghanistan in December 2004 and could achieve good landmine image. Especially, GPR could obtain good image of anti-person (AP) mine buried at more than 20cm depth. Also MD image and GPR image could combine to distinguish mine from metal fragment.

We investigated the application of two state-of-the-art feature selection algorithms for subsurface target discrimination. One is called joint classification and feature optimization (JCFO), which imposes a sparse prior on the features, and optimizes the classifier and its predictors simultaneously via an expectation maximization (EM) algorithm. The other selects features by directly maximizing the hypothesis margin between targets and clutter. The results of feature selection and target discrimination are demonstrated using wideband electromagnetic induction data measured at data collected at the Aberdeen Proving Ground Standardized Test Site for UXO discrimination. It is shown that the classification performance is significantly improved by only including a compact set of relevant features.

Landmine data for electromagnetic induction (EMI) and ground penetrating radar (GPR) sensors has been collected in two background environments. The first environment is clay and the second is gravel. A multi-modal detection algorithm that utilizes a Maximum A Posteriori (MAP) approach is applied to the clay background data and compared to a pair of similar MAP detectors that utilize only the single sensors. It is shown that the multi-modal detector is more powerful than both single mode detectors regardless of landmine type. The detectors are then applied to the data from the gravel background. It is shown that a more powerful performance is achieved if the MAP detector adapts to the statistics of the new background rather than training it a priori with broader statistics that encompass both environmental conditions.

A method known as active sensing is applied to the problem of landmine detection. The platform utilizes two scanning sensor arrays composed of ground penetrating radar (GPR) and electromagnetic induction (EMI) metal detectors. Six simulated confirmation sensors are then dynamically deployed according to their ability to enhance information gain. Objects of interest are divided into ten class types: three classes are for metal landmines, three classes for plastic landmines, three classes for clutter objects, and one final class for background clutter. During the initial scan mode, a uniform probability is assumed for the ten classes. The scanning measurement assigns an updated probability based on the observations of the scanning sensors. At this point a confirmation sensor is chosen to re-interrogate the object. The confirmation sensor used is the one expected to produce the maximum information gain. A measure of entropy called the Renyi divergence is applied to the class probabilities to predict the information gain for each sensor. A time monitoring extension to the approach keeps track of time, and chooses the confirmation sensor based on a combination of maximum information gain and fastest processing time. Confusion matrices are presented for the scanning sensors showing the initial classification capability. Subsequent confusion matrices show the classification performance after applying active sensing myopically and with the time monitoring extension.

The Handheld Standoff Mine Detection System (HSTAMIDS system) has achieved outstanding performance in government-run field tests due to its use of anomaly detection using principal component analysis (PCA) on the return of ground penetrating radar (GPR) coupled with metal detection. Indications of nonlinearities and asymmetries in Humanitarian Demining (HD) data point to modifications to the current PCA algorithm that might prove beneficial. Asymmetries in the distribution of PCA projections of field data have been quantified in Humanitarian Demining (HD) data. The data suggest a logarithmic correction to the data. Such a correction has been applied and has improved the FAR on this data set. The increase in performance is comparable to the increase shown using the simpler asymmetric rescaling method.

There has been significant amount of study on the use of Ground-Penetrating Radar (GPR) for forward-looking landmine detection. This paper presents our analysis of GPR data collected at a U.S. Army site using the Synthetic Aperture Radar system developed by Stanford Research Institute (SRI). Various types of features are extracted from the GPR data and investigated for their abilities to distinguish buried landmines and false alarms; the list include intensity and local-contrast features, fuzzy geometrical image features, ratio between co-polarization and cross-polarization signals, and features obtained using two different approaches of polarimetric decomposition. We also describe the feature selection procedures employed to find subsets of features that improve detection performance when combined. In addition, our analysis indicates that images formed with different frequency bands have different qualities, and that the selection of proper frequency bands can significantly improve the detection performance. Results of landmine detection, including performance on blind test lanes, are presented.

Landmines have become a problem and a daily risk in approximately 70 countries. There exists a broad variety of types of mines in which trinitrotoluene (TNT) is mostly used as the main explosive charge. TNT has a number of specific properties that make it a nearly ideal explosive for military applications. There are several methods currently used to detect buried landmines. The goal of this work is develop new methods for detecting TNT in contact with soil and sand. Raman microscopy is used to provide information about identity, composition, molecular orientation or crystal formation in very small samples or small heterogeneities in large samples. The possible interactions of the energetic material with sand particles have been studied by quantitative vibrational spectroscopy. Ambient conditions that may affect the spectroscopic signature of the explosive in contact with soil were also studied. Among the parameters investigated were: Sand-TNT mass ratio, temperature, pH of soil, incidence of UV light and water content. The characteristic bands of TNT are not significantly shifted, but rather appear constant with respect of the characteristic band of Si-O in sand (~464 cm^-1).

The detection of explosive materials is not only important as an issue in landmines but also for global security reasons, unexploded ordnance, and Improvised Explosive Devices detection. In such areas, explosives detection has played a central role in ensuring the safety of the lives of citizens in many countries. Raman Spectroscopy is a well established tool for vibrational spectroscopic analysis and can be applied to the field of explosives identification and detection. The analysis of PETN is important because it is used in laminar form or mixed with RDX to manufacture Semtex plastic explosive and in the fabrication of Improvised Explosive Devices (IEDs). Our investigation is focused on the study of spectroscopic signatures of PETN in contact with soil. Ottawa sand mixed in different proportions with PETN together with the study of the influence of pH, temperature, humidity, and UV light on the vibrational signatures of the mixtures constitute the core of the investigation. The results reveal that the characteristic bands of PETN are not significantly shifted but rather appear constant with respect of the ubiquitous band of sand (~463 cm^-1). These results will make possible the development of highly sensitive sensors for detection of explosives materials and IDEs.

The detection of hidden explosives using vapors emanating from explosives has been considered an area in explosives technology that requires high sensitivity and selectivity. In this work is reported the results of two methods for vapor explosive detection, GC-μECD and GC/MS coupled to a Tunable Electron Energy Monochromator (TEEM-GC/MS). Both used Solid Phase Microextraction (SPME) in Headspace (HS) mode to collect vapors above the samples. Optimum parameters for SPME were determined with the purpose of obtaining a high-quality extraction. The parameters were: type of SPME fiber, exposure time and desorption time at the injection port of the GC. Headspace SPME procedure was carried out in samples with crystals of TNT buried in soil. These samples were analyzed under important environmental conditions such as temperature and water content. Analyses at contact times after the TNT-soil mix preparation were carried out during 1 month. A comparison of results from both techniques was performed. Vapors of TNT and 2,4-DNT were found predominantly in the samples. HS-SPME coupled with GC-μ ECD and TEEM GC/MS exhibited excellent selectivity and sensitivity.

The detection of trace amount of explosives is of utmost importance in many day-to-day military operations. Moreover, the detection of landmines is a complex and urgent worldwide problem, which needs specific, rapid and cost effective solutions. The most commonly used explosive in landmines is 2,4,6-trinitrotoluene (TNT). Almost 80% of the types of mines manufactured worldwide contain TNT. This contribution describes the use of Immersion Mode Solid-Phase Microextraction (I-SPME) for extraction of TNT and their degradation products from surface soil samples for subsequent analysis by either GC with ⁶³Ni micro cell Electron Capture Detector or gas chromatograph-mass spectrometer coupled to a Tunable Electron Energy Monochromator. A pretreatment step was introduced for the soil samples which extracted the target compounds into an aqueous phase. The experimental results demonstrated the effects of controllable variables. Parameters studied include the chemical properties of the fiber coating, extraction and desorption times, fiber extraction and matrix effect. Surface soil samples containing TNT were evaluated to study the detection of the nitroaromatic explosives and its degradations products using different environmental conditions such as sample temperature, sample contact time and water content.

Landmines have been a part of war technology for many years. As a result of the continued and indiscriminate use in approximately 90 countries landmines pose a severe and ever growing problem and a daily risk. Raman Spectroscopy is capable of providing rich information about the molecular structure of the sample and pinpoint detection of many chemicals, both of organic and inorganic nature. The presence of landmines in soils can be detected by Raman Spectroscopy sensing in a Point Detection modality, using characteristic vibrational signals of each explosive present in landmines. Detection of 2,4-DNT in sand and studies on how the vibrational signatures of 2,4-DNT is modified by interacting with soil particles and environmental conditions is reported. Raman Microspectrometers equipped with 514 nm and 785 nm laser excitation lines were used. The work focused in how the spectroscopic signatures of DNT in contact with Ottawa Sand are affected by the presence of humidity, pH, temperature, UV light and reaction times. Samples of mixtures of sand/2,4-DNT were analyzed by Raman Spectroscopy at 10, 50 and 100% water content and temperatures in range of 40-80 °C. Mixtures were also analyzed at different pH: 4, 7 and 10 and under ultraviolet light at 254 nm. Raman spectra were taken as a function of time in an interval from 24 to 336 hours (two weeks). Characteristic signals of 2,4-DNT were analyzed in different ranges 100-3800 cm^-1, 600-1200 cm^-1, 300-1700 cm^-1 and 2800-3500 cm^-1. The effect of these variables was measured during 45 consecutive days. It was confirmed that the decrease of characteristic vibrational signatures of 2,4-DNT can be attributed to increase of the degradation of 2,4-DNT by the simulated environmental conditions. Spectroscopic characterization of degradation products, both in contact with sand as well as airborne is under way. These results will make possible the development of highly sensitive sensors for detection of explosives materials and correlated with their degradation products in landmines.

Experimental studies have shown that a key factor affecting the bioavailability and biodegradability of nitroaromatic compounds (NAC's) in subsurface environments is their sorption onto clay minerals. This study present the recent ab initio quantum mechanical calculations on the interaction of 2,4-DNT (DNT) with the basal siloxane site surface of kaolinite, a clay mineral. Theoretical calculations of the low energy conformation of DNT interacting with the siloxane site surface of clay minerals were performed in order to obtain their properties adsorbed on soil environments as well as the structure of the adsorbed molecule. The calculations also yielded the way of orientation and the effect of the adsorption. This study was performed using DFT//HF and MP2//HF methods taking into account the contribution of the Coulombic (CEb) and dispersion (DEb) energies, to obtain the binding energies between DNT and siloxane surface. A comparison of the CEb and DEb energies shows that the stabilization of DNT at the siloxane sites, using a small molecular model (single tetrahedra), is mainly provided by dispersion interaction energy. Considering the accuracy and cost of the computation methods the 6-31+G* basis set produced the best representation of the interaction energy (42 kJ/mol) using the MP2//HF level of theory for the DNT-Siloxane surface. These theoretical calculations give a good prediction of the interaction between the 2,4-DNT molecule with soil clay minerals. The computational results are compared with the experimental results obtained with the FT-IR microscopic technique.

Computational algorithms have been very useful to study molecular interactions between explosives and different types of soils. In this work ab initio molecular orbital calculations were employed to study the interaction of 2,4,6-trinitrotoluene (TNT) with the basal siloxane surface of clay minerals. The intermolecular interaction energy, the vibration frequencies and efficient computational algorithms have been tested for the complex of TNT with the siloxane surface site of clay minerals. Two cluster models have been developed to represent the TNT on the siloxane surface of clay minerals. They have been employed in order to determine the changes in the spectroscopic signature of TNT. The results obtained provide information about the interaction energy of TNT on clays. The binding energy between the TNT and the basal siloxane surface was -38 kJ/mol, obtained with MP2//HF/6-31+G(d) level of theory and basis set, respectively. The calculated interaction has their minimal at separation between the two molecules of 3.5 Å. The theoretical IR spectra of the interaction was obtained with DFT//HF methods and the 6-31+G(d) basis set. The calculation predicted a shifting effect in NO₂ bands, due to the interaction. The results are in excellent agreement with available experimental data. Further, result of such theoretical studies could contribute to an understanding of the interaction energy of the other kinds of explosives that may be occurring in other environments.

The JHU/APL Path Planning team has developed path planning techniques to look for paths that balance the utility and risk associated with different routes through a minefield. Extending on previous years' efforts, we investigated real-world Naval mine avoidance requirements and developed a tactical decision aid (TDA) that satisfies those requirements. APL has developed new mine path planning techniques using graph based and genetic algorithms which quickly produce near-minimum risk paths for complicated fitness functions incorporating risk, path length, ship kinematics, and naval doctrine. The TDA user interface, a Java Swing application that obtains data via Corba interfaces to path planning databases, allows the operator to explore a fusion of historic and in situ mine field data, control the path planner, and display the planning results. To provide a context for the minefield data, the user interface also renders data from the Digital Nautical Chart database, a database created by the National Geospatial-Intelligence Agency containing charts of the world's ports and coastal regions. This TDA has been developed in conjunction with the COMID (Cooperative Organic Mine Defense) system. This paper presents a description of the algorithms, architecture, and application produced.

It is estimated that 10% of war-time bombs did not explode and can be found at the ground surface or buried at a depth of up to 8 meters depending on the formation of the soil. These unexploded bombs or ordnance (UXO) pose a real danger to construction workers and properties. Ground surface based methods become ineffective for objects sinking into deep places due to rapidly diminishing anomalous field and interfering metal debris distributed over ground surface. To overcome the difficulties, a unique inversion algorithm is proposed in this work with advantages of fast convergence and maximization of information extracted from individual hole measurement. It is more reliable than traditional methods by examining the possibilities within a number of estimations. The information from individual hole measurement is fully interpreted hence suggestion can be made for the positioning of next drilling in order to minimize the number of holes required for clearance. Based upon the recovered information, a comparison method is proposed for the identification and discrimination of UXO items from other objects that may be found in the environment, such as steel pipes and steel barrels. It is not sensitive to the interference in the data once the dipole moment is recovered. The results from a test site demonstrates its supreme capability to deal with real-world inversion problems having small number of available data points.

Knowledge of the chemical state of explosive materials on soil surfaces is central to our understanding of the environmental effects of landmines as well as to employ existing technologies for landmine sensing. We used X ray photoelectron spectroscopy to study the surface chemistry of TNT and RDX on soil substrate surfaces as a function of soil pH. These explosives exists in many forms in the solid state. At least two forms have been reported for RDX and TNT in the solid state. Different forms are observed for both TNT and RDX adsorbed on soil substrate surfaces. These forms have a markedly different morphology than the one observed on flat surfaces and are found to be pH sensitive. The N 1s binding energy is found to be markedly different for TNT and RDX adsorbed on soil particles as compared to the one measured for these chemicals on other surfaces. The results are consistent with a change in the chemical state of the explosive upon its interaction with soil particles.

We report on the microscopic mass transfer of TNT on soil particles using scanning electron microscopy (SEM) and energy dispersed X ray fluorescence spectroscopy (EDAX). A change in the morphology and spatial distribution of TNT on soil is observed in SEM measurements a few hours after sample preparation. The change in morphology and spatial distribution is found to depend on the soil pH. Evidence is also found for the disappearance of TNT from soil as reflected by a decrease in the N fluorescence signal intensity with time after the preparation of TNT/soil mixtures. The results are modeled with the lattice model that describes the diffusivity solid-solid and provide the transference area and the diffusivity to which TNT transports in soil.

Atomic force microscopy (AFM) and near field scanning optical microscopy (NSOM) are promising analytical techniques for the determination of trace amounts of explosive materials on a variety of surfaces. Information regarding the forces of interaction and explosive spectroscopic signatures, in addition to explosive morphology, can be obtained with AFM and NSOM techniques. Basic work toward the development of methodology that can enable the employment of these techniques is needed to facilitate their employment in the field and real life scenarios. In this work, we report on the use of AFM and NSOM for the determination of the morphology and spectroscopic signature of TNT particles on optically transparent substrates open to air environments. The TNT particles are about 1 mm in diameter. Transmission NSOM on the particles following 265 nm excitation reveals that the fluorescence peak is centered at 255, about 10 cm-1 lower than the excitation wavelength. The fluorescence yield is found to increase non-linearly with incident laser power, consistent with a multi photon absorption process. The results encourage future work in the area, in particular, with the use of multi-cantilevers that can increase the surface area examined in real life scenarios.

2,4,6-Trinitrotoluene is a high explosive used in manufacturing landmine, bombs, and other explosive devices. It has been the main source of contamination in groundwater, soil as a result of intentional or accidental releases at many places around the world. Crystallization of TNT in soil from TNT/methanol solutions was carried out and characterized using vibrational spectroscopy. TNT exhibits a series of characteristic bands that allow its detection when in soil. The spectroscopic signatures of neat TNT and TNT in soil samples were determined with Raman Microspectroscopy and Fourier Transform Infrared (FTIR) Microscopy. The spectroscopic signature of neat TNT is dominated by strong bands about 1380 and 2970 cm^-1. The intensity and position of these bands are found remarkably different in soil samples spiked with TNT. The 1380 cm^-1 band is split into a number of bands in that region. The 2970 cm^-1 is reduced in intensity and new bands are observed at about 2880 cm^-1. The results are consistent with a different chemical environment for TNT in soil as compared to neat TNT. Further measurements are required to fully understand the spectroscopic signature of TNT in soil samples. Its detection in soil is essential in landmine detection technology, and could address the improvement of the devices in the mentioned technology.

We present the general concept and results of a pilot study on land mine detection based on the application of Time Reverse Acoustics (TRA). Applying TRA is extremely effective at focusing seismic waves in time and space, significantly improving detection capabilities using both linear and nonlinear wave methods. The feasibility of the system was explored in the laboratory and in small scale field experiments. The system included a multi-channel TRA electronic unit developed at Artann, five speakers for seismic-wave excitation and noncontact (laser vibrometer) or contact (accelerometer) devices for measurements of the surface vibration. Experiments demonstrated the high focusing ability of the TRA system. We observed excitation of highly focused seismic waves in an area with dimensions of the order of one wavelength. In the presence of a buried mock mine, the method led to an increase in the surface vibration amplitude and to significant nonlinear distortion of the TRA focused signal. Localization via TRA depends on the frequency of excitation, the depth of the buried mine, and the form and size of a mine mock. The nonlinear acoustic effect-higher harmonic generation-provides higher contrast for the mock-mine signal-response than for the surrounding medium. We also successfully tested an inversion method of the nonlinear TRA measurements earlier developed for medical ultrasound applications.

Dynamic, accurate near-real time environmental data is critical to the success of the mine countermeasures operations. Bathymetric data acquired from the AQS-20 mine hunting sensor should be adjusted for local tide variations related to the specific geographic area and time interval. This problem can be overcome by a spatio-temporal estimate of tide corrections provided for the area and time of interest by the Naval Research Laboratory tide prediction code PCTides. For each geographic position of the AQS-20 sonar, a tide height relative to mean sea level is computed by interpolating the tidal information from the K - nearest neighbored stations for the corresponding time. The value is used to correct the measured depth generated by the AQS-20 sonar in that location to mean sea level for fusion with other bathymetric data products. It is argued that this paper provides a useful tool to the MCM decision factors during Mine Warfare operations.

The Humanitarian Demining Research and Development Program of Night Vision and Electronic Sensors Directorate (NVESD), under the direction of the Office of Assistant Secretary of Defense for Special Operations and Low-Intensity Conflict (OASD/SOLIC) and with participation from the International Test and Evaluation Project (ITEP) for Humanitarian Demining, conducted an in-country field evaluation of HSTAMIDS in the region of Humanitarian Demining Unit #1 (HMAU1) in Thailand. Participants included the US Humanitarian Demining Team of NVESD, ITEP personnel, Thailand Mine Action Center (TMAC), HALO Trust organization from Cambodia, and CyTerra Corporation. The primary objectives were to demonstrate the performance of the U.S. Army's latest handheld multisensor mine detector, the AN/PSS-14, in a demining environment in comparison to the performance of the metal detector being used by the local deminers and also to assess the performance of the trained deminers after limited experience and training with the HSTAMIDS.

Detection of buried and flush buried landmines has been dangerous and time consuming for both military and humanitarian de-mining personnel throughout the world. In an effort to make the process safer, faster, and more reliable, scientists have successfully employed Ground Penetrating Radar (GPR) systems in nadir and near nadir viewing angles. Leveraging this successful technology, Forward-Looking Ground Penetrating Radar (FLGPR) technology, using low grazing angles, is being developed which promises to provide an increase in detection stand-off distance thereby increasing safety of personnel during land-based mine detection efforts. However, the application of GPR for the detection of buried plastic mines has been problematic, research has begun to exploit the comination of broadband and hyper-spectral passive electro-optical technologies with GPR technologies. One such embodiment is to use Forward Looking InfraRed (FLIR) technology with the intention to augment the capability of, and overcome limitations inherent to, current FLGPR technology. The emphasis in using FLIR is to understand and exploit specific spectral features which are complementary fo FLGPR and exhibited by buried metal and plastic mine targets. One spectral feature being investigated is the resstrahlen emission which results when soil is excavated or disturbed. This paper is a preliminary investigation of the performance of a vehicle based FLIR camera system for detecting resstrahlen emissions from disturbed soils. Specifically, this paper will examine the robustness of the resstrahlen feature in a forward-looking low grazing angle application. The data presented in this paper was collected at an eastern US Army testing site over targets deployed in soils which had been disturbed from one day before the start of the collection.

The Rapid Route Area and Mine Neutralisation System (RRAMNS) Capability Technology Demonstrator (CTD) is a countermine detection project undertaken by DSTO and supported by the Australian Defence Force (ADF). The limited time and budget for this CTD resulted in some difficult strategic decisions with regard to hardware selection and system architecture. Although the delivered system has certain limitations arising from its experimental status, many lessons have been learned which illustrate a pragmatic path for future development. RRAMNS a similar sensor suite to other systems, in that three complementary sensors are included. These are Ground Probing Radar, Metal Detector Array, and multi-band electro-optic sensors. However, RRAMNS uses a unique imaging system and a network based real-time control and sensor fusion architecture. The relatively simple integration of each of these components could be the basis for a robust and cost-effective operational system. The RRAMNS imaging system consists of three cameras which cover the visible spectrum, the mid-wave and long-wave infrared region. This subsystem can be used separately as a scouting sensor. This paper describes the system at its mid-2004 status, when full integration of all detection components was achieved.

This paper deals with a three-dimensional thermal model for landmine detection problems and an inverse problem for reconstructing the physical parameters of buried objects. Moreover, solutions are given for the estimation of the soil thermal diffusivity and meteorological parameters, needed for solving the inverse problem. The paper describes the main fundamental principles of thermal modelling for buried object identification and illustrates the results on data acquired from a real minefield, together with qualitative and quantitative results illustrating the validity of the model.

A detailed description of the CLEARFAST system for thermal IR stand-off minefield survey is given. The system allows (i) a stand-off diurnal observation of hazardous area, (ii) detecting anomalies, i.e. locating and searching for targets which are thermally and spectrally distinct from their surroundings, (iii) estimating the physical parameters, i.e. depth and thermal diffusivity, of the detected anomalies, and (iv) providing panoramic (mosaic) images indicating the locations of suspect objects and known markers. The CLEARFAST demonstrator has been successfully deployed and operated, in November 2004, in a real minefield within the United Nations Buffer Zone in Cyprus. The paper describes the main principles of the system and illustrates the processing chain on a set of real minefield images, together with qualitative and quantitative results.

A method is presented for identifying buried objects using electromagnetic induction metal detectors. The method uses a physics based model for identifying two basis functions that fundamentally compose metal detector signals. These bases form a signal subspace that contains the signals from all objects at the same depth regardless of their shape, size, or metal content. First, an algorithm for determining this subspace is presented. Then utilizing the proper signal subspace, the shape of the object is determined by estimating the object's directional polarizablity.