During May and June of 2003, the US Army Night Vision and Electronic Sensors Directorate (NVESD) and the Ohio State University (OSU) measured the thermal behavior of mines in an arid site. Thermistors were placed in contact with both surface-laid mines and native stones and monitored from before sunset until well after sundown. Measurements of local vegetation and measurements of the surrounding soil at 2.5 and 5 cm depths were also performed. A tripod-mounted MWIR sensor was used concurrently to collect high-resolution images to identify and understand the underlying phenomena. Data were collected during both clear, sunlit conditions and during an overcast day, but because of space limitations only data acquired under the (more typical) clear conditions are described here. The results contain a number of findings. First, local soil properties appear to have important implications for the apparent mine contrast. The same type of mine at locations only a few meters apart can show significantly different contrast with the native soil. Second, natural phenomena can be a significant clutter source. The temperature of vegetation can be similar to that of mines, and a small plant will occasionally produce a signature with a shape similar to that of a surface mine. Native stones are also a source of false alarms, but they tend to show somewhat less contrast. Third, at certain times, mines are best viewed with a low-elevation angle sensor. The construction of some mines causes the temperature of the side walls to be significantly different from that of the top surface at those times. Finally, disturbing the surface of desert soil through excavation, vehicle traffic or even repeated pedestrian traffic is often sufficient to produce a strong thermal signature. This fact could be used to advantage to detect buried mines in desert environments.
Polarimetric scattering models are developed to predict the detectability of surface-laid landmines. A specular polarimetric model works well only under the condition that there is either no sunlight or the sun is not close to the specular reflection direction. Moreover, this model does not give insight why certain man-made objects like landmines give a higher polarimetric signature than natural background. By introducing a polarimetric bidirectional reflectance distribution function (BRDF) the specular model is extended. This new model gives a better prediction of the polarimetric signature and gives a close match to the measurements of landmines with different casings as well as the sand background. The model parameters indicate that the landmines have a lower surface roughness and a higher refractive index, which is the reason why these objects are detectable from the background based on their polarimetric signature.
It is widely acknowledged that tree roots and other forms of buried biomass have an adverse effect on the performance of ground-penetrating radars (GPRs). In this work we present experimental and theoretical work that quantifies that effect. Test sites containing extensive root infiltration at Eglin Air Force Base, FL were probed with a GPR. After completing the measurements, the sites were excavated, and the root structure and soil were thoroughly characterized. Supplemental GPR measurements of simple cylindrical objects in a laboratory setting were performed to investigate basic scattering behavior of buried roots. A numerical simulator based on the Discrete Dipole Approximation (DDA), an integral-equation-based method, was developed, validated and subsequently used to compute scattering from root structures modeled by an ensemble of buried cylinders. A comparison of the measurements and numerical calculations is presented that quantifies the potential for false alarms and increased clutter due to buried roots.
Several sensor fusion approaches are in use for detection of land mines. These are based on different techniques and show different results. Meaningful comparisons of fusion algorithm performance are difficult, because the performance achievable may be limited by the data acquired and the sensors used. Especially in comparing such dissimilar situations, it is important to compare measures that are not dependent of sensor peculiarities or layout of test lanes. Nonetheless, a quantitative comparison of algorithms is necessary to identify the most effective fusion techniques. This comparison must sometimes be done when the algorithms are being applied to different data sets. In this paper we describe a methodology to evaluate the performance of sensor-fused mine-detection systems. This process can be used to compare different mine-detecting sensors in well- defined scenarios.
A specular model has been used to predict the passive polarimetric infrared (IR) signature of surface-laid landmines. The signature depends on the temperature of the landmine and the sky radiance. The temperature of the landmine is measured using a thermocouple. The signature itself is measured using a polarimetric IR camera setup. The predictions are fit to the measurements using the refractive index as an optimization parameter. The obtained refractive indices of each landmine type are consistent, but for the PMN landmine much lower than determined in a previous indoor experiment. Throughout the measurement day, the average landmine polarimetric signature was higher than the average background signature. Moreover the polarimetric signature appears to be a more robust indicator of the shape of the landmine's top surface than the normal IR signature. A simulator of passive polarimetric imagery is also being developed. That work is based on a physical model for both the thermal and radiometric processes, and it includes a finite-element solution for the heat transfer problem, ray tracing to describe the incident sunlight and the effects of shadowing, and analytical models for the Mueller matrices of rough dielectric surfaces. Preliminary results from that model show substantial qualitative agreement with measured images.
A numerical model for polarimetric signatures of surface-laid land mines is presented. The model simulates high-resolution images formed from Stokes parameters that describe thermal emission and reflection of sunlight and skylight. The temperature of the mine and its surroundings is computed as a function of time using a finite element solution of the heat transport equation. The effects of surface roughness are included via a two-scale model, in which the gross shape of the mine is represented by triangular facets (the surface facets of the finite element tetrahedra). Theoretical solutions for rough surface scattering are used to describe small-scale roughness on a facet. Multiply reflected contributions are neglected in the current implementation. An example is presented in which the role of each component is described and related to the observed image.
The potential for performance improvements through sensor fusion is explored for two electro-optical (EO) imaging sensors: a passive thermal IR camera and an active polarimetric system. Tests of decision-level fusion using a small data set (roughly 60 mine signatures) suggest that a significant performance improvement can be obtained by using an AND fusion approach. The source of this improvement derives from correlation among the sensors. Specifically, the sensors exhibit a strong positive correlation when a mine is present, and a negligible correlation when viewing clutter. The observed improvement is independent of the local ground clutter, but it depends strongly on the decision thresholds used for the individual sensors.
Thermal IR signatures of buried land mines are affected by various environmental conditions as well as the mine's composition, size and burial geometry. In this work we present quantitative relations for the effect of those factors on the signature's peak contrast and apparent diameter. We begin with a review of the relevant phenomena and the underlying physics. A three-dimensional simulation tool developed by the authors is used to simulate signatures for the case of a static water distribution. We discuss efforts to validate the model using experimental data collected at Fort A.P. Hill, VA. Using this simulation tool a variety of factors are considered, including soil water content, soil sand content, wind speed, mine diameter and mine burial depth.
Predicting the thermal signature of a buried land mine requires modeling the complicated inhomogeneous environment and the structurally complex mine. It is useful, both in checking such models and in making rough calculations of expected signatures, to have an accurate, easily computed solution for a relatively simple geometry. In this paper, a reference solution is presented for the integral equation that governs the temperature distribution. Our solution procedure uses the method of weighted residuals. The problem comprises a homogeneous cylindrical body (the mine model) buried in an infinite homogeneous half space (the soil model) with a planar interface. Using periodic boundary conditions in time at the planar interface, the temperature distribution in the lower-half space is expanded in a Fourier series. A volume integral equation for the Fourier series coefficients is obtained via Green's second identity. The Green's function for the Fourier coefficients is derived and reduced to a computationally efficient form. The integral equation is reduced to a matrix equation, which is then solved for the unknown temperature distribution. The integral equation solution is compared with a finite element model.
Land mines are a major problem in many areas of the world. In spite of the fact that many different types of land mines sensors have been developed, the detection of non-metallic land mines remains very difficult. Most landmine detection sensors are affected by soil properties such as water content, temperature, electrical conductivity and dielectric constant. The most important of these is water content since it directly influences the three other properties. In this study, the ground penetrating radar and thermal IR sensors were used to identify non-metallic landmines in different soil and water content conditions.
We investigate the potential for improving land mine detection by fusing data from ground penetrating radars (GPRs) and sensors of acoustically induced soil motion. We present a case study involving data from the SRI forward-looking radar and a laser Doppler vibrometer (LDV) system developed by the University of Mississippi. The LDV sensor detects acoustically induced soil vibrations, while the GPR detects scattering from dielectric discontinuities or metal objects in the soil. Since these sensors exploit different target physical properties, it is reasonable to expect a benefit in fusion. Although the sensors are relatively new, the LDV and GPR data exhibit evidence for complementarity, in that the GPR is significantly better at detecting metal mines, while the LDV is somewhat better at detecting plastic mines. Decision-level fusion is shown to improve performance. A simple OR fusion approach is found to perform similarly to an optimum hard decision fusion algorithm.
Many aspects of a buried mine's thermal IR signature can be predicted through physical models, and insight provided by such models can lead to better detection. Several techniques for exploiting this information are described. The first approach involves ML estimation of model parameters and followed by classification of those parameters. We show that this approach is related to an approximate evaluation of an integral over the parameters that arises in a Bayesian formulation. This technique is compared with a generalized likelihood ratio test (GLRT) and with computationally efficient, model-free approaches, in which soil temperature data are classified directly. The benefit of using the temporal information is also investigated. Algorithm performance is illustrated using broadband IR imagery of buried mines acquired over a 24 hour period. It is found that the detection performance at a suitably selected time is comparable to the performance achieved by processing all times. The performance of the GLRT, for which detection is based only on the residual error, is inferior to a classifier using the parameters.
Simple theoretical models can be constructed to study the behavior of sensor-fused systems using idealized sensor suites. Models are available for feature-level and decision-level fusion, both of which are now being used with demining sensors. These models are attractive as design tools and for estimating the expected performance of new sensor suites, since their performance can be evaluated with relatively little effort. In this paper we review some simple idealized models and their predictions for fused system performance. The data produced by demining sensors are often correlated, and the effect of correlation is explored for both feature-level and decision-level fusion.
Algorithms are presented for detecting surface mines using multi-spectral data. The algorithms are demonstrated using visible and MWIR imagery collected at Fort A.P. Hill, VA under a variety of conditions. For imagery with a resolution of a few centimeters there is significant correlation in the clutter. Using a first-order Gauss Markov random field model for the clutter, an efficient pre-whitening filter is proposed. A significant improvement in detection is demonstrated as a result of this whitening. Further improvement in the detection of specific mine types is demonstrated by using a random signal model with a known covariance matrix. That approach leads to an estimator-correlator formulation, in which the random signature estimate is the output of a Wiener filter. It is suggested that by fusing the output of a bank of such filters one could improve detection of all mine types.
It has long been recognized that surface-laid land mines and other man-made objects tend to have different polarization characteristics than natural materials. This fact has been used to advantage in a number of mine detecting sensors developed over the last two decades. In this work we present the theoretical basis for this polarization dependence. The theory of scattering from randomly rough surfaces is employed to develop a model for scattering and emission from mines and natural surfaces. The emissivity seen by both polarized and unpolarized sensors is studied for smooth and rough surfaces. The polarized and unpolarized emissivities of rough surfaces are modeled using the solution of the reciprocal active scattering problem via the second order small perturbation method/small slope approximation(SPM/SSA). The theory is used to determine the most suitable angle for passive polarimetric IR detection of surface mines.
3D thermal and radiometric models have been developed to study the passive IR signature of a land mine buried under a rough soil surface. A finite element model is used to describe the thermal phenomena, including temporal variations, the spatial structure of the signature, and environmental effects. The Crank-Nicholson algorithm is used for time-stepping the simulation. The mine and the surroundings are approximated by pentahedral elements having linear interpolation functions. The FEM grid for the soil includes a random rough surface having a normal probability density and specified covariance function. The mine is modeled as a homogeneous body of deterministic shape having the thermal properties of TNT. Natural solar insolation and the effects of convective heat transfer are represented by linearized boundary conditions. The behavior over a periodic diurnal cycle is studied by running the simulation to steady state. Finite element solutions for the thermal emissions are combined with reflected radiometric components to predict the signatures seen by an IR camera. Numerical simulations are presented for a representative target, a 25 cm anti-tank mine simulant developed by the US Army. The temporal evolution of the temperature distribution and IR signature are presented for both smooth and rough surfaces.
A method is described to improve the performance of sensor fusion algorithms. Data sets available for training fusion algorithms are often smaller than described, since the sensor suite used for data acquisition is always limited by the slowest, least reliable sensor. In addition, the fusion process expands the dimension of the data, which increases the requirement for training data. By using structural risk minimization, a technique of statistical learning theory, a classifier of optimal complexity can be obtained, leading to improved performance. A technique for jointly optimizing the local decision thresholds is also described for hard- decision fusion. The procedure is demonstrated for EMI, GPR and MWIR data acquired at the US Army mine lanes at Fort AP Hill, VA, Site 71A. It is shown that fusion of features, soft decisions, and hard decisions each yield improved performance with respect to the individual sensors. Fusion decreases the overall error rate from roughly 20 percent for the best single sensor to roughly 10 percent for the best fused result.
Ground-reflected clutter is often a performance-limiting factor in ground-penetrating radar detection of near-surface targets including anti-personnel mines. When a down-looking antenna is scanned across the surface this reflection produces a strong band in the image, which obscures shallow targets. Imperfections in the system impulse response can produce similar bands. Radar images of buried targets can be degraded by these forms of clutter.
An image processing technique is described for the detection of miens in RI imagery. The proposed technique is based on a third-order neural network, which processes the output of a wavelet packet transform. The technique is inherently invariant to changes in signature position, rotation and scaling. The well-known memory limitations that arise with higher-order neural networks are addressed by (1) the data compression capabilities of wavelet packets, (2) protections of the image data into a space of similar triangles, and (3) quantization of that 'triangle space'. Using these techniques, image chips of size 28 by 28, which would require 0(109) neural net weights, are processed by a network having 0(102) weights. ROC curves are presented for mine detection in real and simulated imagery.
We present and compare methods for pre-detection and post- detection fusion of multi-sensor data. This study emphasis methods suitable for data that are non-commensurate and sampled at non-coincident points. Decision-level fusion is most convenient for such data, but this approach is sub- optimal in principle, since targets not detected by all sensor will not achieve the maximum benefits of fusion. A novel feature-level fusion algorithm for these conditions is described. The optimal forms of both decision-level and feature-level fusion are described, and some approximations are reviewed. Preliminary result for these two fusion techniques are presented for experimental data acquired by a metal detector, a ground-penetrating radar, and an IR camera.
We describe sensor-based and signal-processing-based techniques for improving the detection of buried land mines in thermal IR imagery. Results of experimental studies using MWIR and LWIR imaging systems are reported. Thermal clutter due to surface reflected sunlight and skylight are investigated and shown to be the dominant clutter component for both MWIR and LWIR imagery collected during daylight hours. A sensor-based clutter reduction technique, spectral differencing, was considered and found to provide some benefit. The temporal evolution of thermal signatures was investigated. The imagery are found to have near-Gaussian statistics, and therefore the deflection coefficient is a valid measure of detectability. The deflection coefficient for some buried mines was found to improve with time after sunset. In addition, the LWIR band appears to offer some advantages in detection. Clutter mitigation via signal processing is also explored using an 'estimator-classifier' technique in which target-related parameters are estimated from the data and detected with a classifier. The theoretical basis of the method is discussed. MWIR and LWIR imagery are used to illustrate both the sensor-based and signal-processing-based techniques.
A sensor-fused system is being developed for detection of buried land mines. The system uses a ground-penetrating radar, an infrared camera, and an electromagnetic induction sensor. In the current implementation each sensor is used independently, and fusion is performed during post-processing. We briefly describe the sensors and a data collection involving buried mine surrogates. Algorithms for preprocessing and feature extraction are reviewed. To deal with non- coincident sampling we have developed a new feature-level fusion algorithm, which does not require detection and subsequent association of putative targets. Results are presented for fusion of simulated data.