Ir thermography for the detection of buried objects: a short review

IR Thermography for the Detection of Buried Objects:
A Short Review
ABSTRACT

Sub-surface buried objects, such landmines and archaeological artefacts, and the surrounding environment constitute a
complex system with variable characteristics. As a consequence, the detection and recognition of these objects may be
extremely difficult.
IR thermography, which is widely employed in the detection of discontinuities in materials and structures, would be in
principle suitable also for this kind of application. The issue in this case appears to be the presence of excessive levels
of background noise, whose modelling is difficult, in that it results from a number of factors e.g., moisture content,
presence of vegetation, and variation of solar radiation at topsoil level. In recent years, a number of studies have tried to
overcome these limitations and improve the reliability of this method, using filtering and automatic pattern recognition
techniques, specific for the detection of buried objects.
This work is aimed at revising and commenting the most recent experiences in this application of IR thermography. In
particular, the possibility of combined use of IR thermography with other techniques, in particular ground probing radar
(GPR), and high frequency electromagnetic techniques are discussed, in order to improve the reliability of buried
objects detection by fusion of data obtained from different sensors.
Keywords: IR thermography, data fusion, soil conditions, buried objects detection
1. Introduction

In recent years, the detection of buried objects using thermographic techniques has been widely
studied. A large number of scientific papers are available, a list of which, including considerations
on the specificities and the novelty of each work, is presented in Table 1. In general, this application
can be interpreted as a non-destructive evaluation system, whose performance depends on three
elements: the intrinsic detection capability of the method (or combination of methods) employed,
application-related factors, in particular the state of environment surrounding the buried object, and
the human factor, connected with the skills of NDE operators. The NDE nature of this issue can be
clarified e.g., by considering the electromagnetic systems for the detection of buried metallic
objects, in which case the source and the sensor are arrays of loop antennas. These systems are
based on the measurement of the EM field scattered by the target [1]. Discrimination is generally
pursued through the empirical comparison of measured data obtained by known targets with real-
field measurements. The typical working frequency band is between 30 Hz and 30 kHz (low
frequency) when the target is metallic, or AM radio frequencies, when the target is mainly poor
conductor (dielectric). In the latter case, soil characteristics affect the response of buried objects and
increase the difficulty of discrimination [2]. In the more general case of multi-technique detection
of buried objects, systems for data fusion or sensor fusion are needed, which are not always
synergic in the sense of increasing the probability of detection (POD) and decreasing the probability
of false indications (PFI) [3]. In this case, statistical methods to account for some aspects of the
environment, such as the cooperative/uncooperative or homogeneous/inhomogeneous soil, have
also been proposed [4].
Main results of the work
Evaluation of the effect of soil anisotropy and sunlight on parameters for thermal transport for objects detection Adoption of a detection system based on a concentrated heat source and a non-contact thermometer mounted on a suspended transportation system exploring the soil surface Effect and parametrisation of vegetation in Thermographic detection of buried objects Thermographic detection of objects buried in sandy soil at different depth and with different humidity content Study of the effect of object orientation on the signal obtained on the thermograms Comparison of thermographic detection results with a tri-dimensional model of heat flux transport In-lab reproduction of thermographic tests for buried objects detection with parametric evaluation of scale-effect Table 1 Some recent works on the application of IR thermography to the detection of buried objects This paper concentrates on the possibility that some results obtained in the case of the detection of anti-personnel landmines (APL), are applicable to the more general detection of buried objects (e.g., archaeological artefacts). In this regard IR thermography is competing with a number of other methods, which are listed and detailed in Table 2. Advantages
Drawbacks
contact, Reflections of both soil surface and the antenna possible selection of most adapted frequency used require a very broad detection band to have band a resolution lower than 10 mm, and often data filtering methods are needed Ease of detection for known types of buried Strong dependence on geometry and orientation objects, e facility in obtaining a uniform of the buried object, usually avoided by the use of magnetic field with remote sensing Electrical impedance Possibility and relative ease of measuring Problems with electrical contact can be revealed tomography (EIT) conductivity perturbations and simulate in case the soil is very dry. Strong dependence of realistic conditions per la presence of the detection reliability from object geometry object. Advantageous for the detection on humid soil and underwater Easy detection of small quantity of Performance limited from presence of humidity explosive substances also at significant and strongly dependant on objects scale. depths (up to 300 mm) Need to evaluate neutron distribution to reduce false alarms. Virtual impossibility of detecting buried objects other than landmines Portability and auto-feeding. Inspection at Dependence on density, not necessarily related to evaluations to reduce levels of error. Resonance effects due to the substances present in the object (e.g., explosives, dust), and therefore need to know in advance its nature and composition. Capable of ensuring detection by comparing Requires acoustic excitation, in which case the ratio of the velocity magnitude of the detection reliability is strongly dependant on the ground surface over and away from the acoustic relaxation time of the soil surface target, and the presence of wave-like or scattering phenomena. The dimension of the buried object can be measured, with suitable excitation wavelength Table 2 Advantages and drawbacks of some techniques for buried objects detection To optimise the rate of detection, it can also be advisable to use sensors for different methods (e.g., geo-radar and electromagnetic induction), sometimes installed on mobile stations, and operate with data fusion techniques, according to a procedure outlined in Figure 1, aimed at the comparative analysis of results with probabilistic factors. In this case, it may be useful training an artificial neural network (ANN) system in order to automatically recognise the buried object. However, the acquisition of an optimised and unique method for this purpose appears still a very ambitious objective. Figure 1 Structure of a typical system for data fusion aimed at buried objects detection The use of IR thermography enables a fast no-contact detection of the artefact: in particular, the latter characteristics can be desirable both for antipersonnel landmines (APL) than in the case of archaeological artefacts, where the intervention of specialised operators follows detection [5]. In reality, the depth of buried object appears still a limiting factor: error-free detection is hardly ever achieved at depths exceeding 10 mm [6]. Moreover, the issue of statistical reliability in buried objects detection using IR thermography has not been addressed yet. The background noise present in thermograms is still quite high, so that it is not easy to optimise the severity of control, whose incorrect setting may result in false alarms or else in missed detection. In practice POD and PFI are influenced by a large number of factors, which are schematically presented, divided in categories, in Figure 2, assuming for simplicity that a last generation thermographic system is used, and that human factor is optimised. Figure 2 Principal factors to be considered in the detection of buried objects using IR thermography 2. Characteristics of Thermographic Investigation

The detection of buried objects using passive thermography technique is based on the difference in
thermal conductivity between the object and the surrounding environment (soil), which ideally
allows the object to be individuated from the presence of a recognisable mode of emission, the so-
called "thermal signature" of the artefact.
A possibility is offered by increasing image by soil heating using microwaves [7]: more in general,
the use of microwaves for sample heating has been sometimes applied in IR thermography studies,
although at a larger scale obtaining a uniform heating on the sample may not be easy [8]. Also pulse
thermography has been attempted, based on the assumption that distribution of temperatures around
the surface is influenced by the presence of buried object, resulting in a "hot spot" on the surface,
which is recognisable for a transient time, depending on the heating method used, but possibly
extending also to several minutes [9]. The main limitation appears in this case the depth of the
buried object, which does not exceed 20-40 mm: in a typical application of pulse thermography,
temperature profiles include a faster heating phase, followed by a slower cooling phase, with time
intervals growing with buried object depth [10].
3. Experimental Studies and Thermal Modelling of Buried Objects Detection

From the theoretical point of view, modelling the temperature response of a buried object requires
the knowledge of equation for turbulent heat flux in a soil assumed dry (zero moisture content) and
homogeneous (granulometry of soil particles constant), transport which is modified from the
presence of the buried object [10]. Heat transport phenomena can be modelled as bi-dimensional,
although, with growing distance from buried object, also mono-dimensional models become
progressively more reliable, especially if the principal objective is the determination of the
maximum temperature gradient which may develop at soil surface, in correspondence with the
buried object [11].
In practical NDE, thermographic modelling of buried object geometry requires evaluating thermal
inertia in the soil above the object. Thermal inertia I is defined as: I = (kρC)½, where k is the bulk
thermal conductivity, ρ is the bulk density and C the specific heat capacity. The practical evaluation
of thermal inertia would require also the measurement of the content in moisture and chemicals in
the soil, which have an effect on the value of ρ in the same way a buried object would. A possible
calculation of thermal inertia can be carried our by heating soil samples with different moisture
content by using infrared thermometers. The measurement of soil heat flux and infrared radiation
temperature allow evaluating thermal inertia [12].
In a model more respondent to reality, the presence of vegetation needs also accounting for, as a
parameter affecting soil heat capacity [13]. For as regards buried objects, modelling their geometry
requires also knowing their orientation with respect to soil surface, which may be expressed as a
function of two angles, a θ angle between Y-axis and the symmetry axis of the object, and a ψ angle
between Y-axis and a line perpendicular to the soil surface. These two angles are then statistically
plotted against δ, which is the variation of object depth consequent to orientation [4].
An essential aspect of experimental studies performed using IR thermography on the detection of
buried object is the possibility of obtaining reproducible results also in absence of sunlight. Studies
in this field have clarified that thermal images showing buried object profile can be acquired, also in
case thermal contrast owed to solar irradiation is negligible. However, it has not been possible to
obtain a clear trend for probability of detection (POD) of buried objects with depth, which would
have allowed establishing limits of reliability for the technique [14]. In this regard, the use of data
fusion techniques with other non destructive techniques, such as geo-radar, has been suggested. On
the different techniques for pixel-to-pixel data fusion it is possible to refer to information given in
[15], where possibilities are suggested for almost totally excluding subjective evaluations from data
fusion.
Insufficient detection reliability has been revealed also in tests for buried objects detection in sandy
soil with different moisture content (0, 2.5, 5, 7.5 and 10%), heated using IR lamps [11]. In this
case, the transient enabling the measurement of temperature difference following soil heating for 8
minutes, lasts for about 30 minutes, up to a level of about 40 mm below the surface. In general,
higher moisture content facilitates buried object detection: however, detection at higher depth
requires a longer cooling phase before the thermogram is able to produce an object signature.
In fact, the presence of an interposed air cushion in sand layer above the object modifies the
conductive properties of the soil as well, since it acts as an insulating layer and can as a whole
increase the POD of the buried object. As a whole, however, transient thermography following
heating with IR lamps, appears to be preferable, especially in the case it is used as a stand-alone
technique.
4. Conclusions
Trying to sum up the possibilities of using IR thermography for an application aimed at buried objects detection, it is noteworthy, as specified in Figure 2, that the reliability of such detection is dependant on a large number of factors, relative both to the environment and to the limitation of non-destructive technique used, obviously taking into account also the human factor. The majority of studies demonstrate that a simple mono-dimensional modelling yields results, measured as temperature gradients on the buried object, very close to the experimental ones. Thermal transient may be sufficiently high to possibly allow a high precision in detection and reduce the incidence of false alarms. In addition, variations of soil characteristics do not necessarily appear critical for object recognition. As a consequence, the real limiting factor appears to be represented by the shallow depth at which IR thermography is capable of offering indications on the buried object. It is likely that the use of thermographic systems with higher accuracy in the measurement of temperature (up to differences of less than 100 mK) would allow making sense also of shorter thermal transient (possibly in the order of the second or less) for buried object detection at “realistic” depths (e.g., between 100 and 300 mm), where other techniques, such as e.g., neutron backscattering or neutron activation [16]. This publication has been financed by the Programme "Incentivazione alla mobilità di studiosi stranieri e italiani residenti all'estero" 1. Bell TH, Barrow BJ, Miller JT “Subsurface discrimination using electromagnetic induction sensors,” IEEE Transactions Geoscience Remote Sensing 39, 2001, pp. 1286–1293.
2. Araneo R, Celozzi S, “Numerical Analysis of Subsurface Object Discrimination Systems”, IEEE Transactions on Magnetics 39, 2003, pp. 1219-1222.
3. Müller C, Scharmach M, Gaal M, Guelle D, Lewis AM, Sieber AJ “Performance demonstration for humanitarian demining” Materialprüfung 45, 2003, pp. 504-512.
4. Stepanic J, Malivovec M, Svaic S, Krstelj V, "Parametrisation of non-homogeneities in buried object detection by means of thermography", Infrared Physics & Technology 45,
2004, pp. 201-208.
5. Boras I, Malinovec M, Stepanic J, Svaic S, "Modelling of buried object detection using thermography", QIRT Conference Proceedings, Reims, 2002, pp. 176–181. 6. Di Marzio CA, Rappaport CM, Wen L, Sauermann GO, "Microwave-Enhanced Infrared Thermography", Proceedings of SPIE, vol. 3710, 1999, pp.173-179. 7. Santulli C, Jeronimidis G, "Measurement of surface void content on balsa wood using IR thermography", E-Journal of Non-Destructive Testing 11 n.6, June 2006.
8. Deans J, Gerhard J, Carter LJ, "Fehler! Hyperlink-Referenz ungültig.Infrared Physics &
Technology 48, 2006, pp.202-216.
9. Boras I, Malinovec M, Stepanic J, Svaic S, "Detection of underground objects using thermography", 15th WCNDT proceedings, October 2000, Rome. 10. Moukalled F, Ghaddar N, Kabbani H, Khalid N, Fawaz Z, "Numerical and experimental investigation of thermal signatures of buried landmines in dry soil", Journal of Heat Transfer
128, 2006, pp.484-494.
11. Khanafer K, Vafai K, "Thermal analysis of buried land mines over a diurnal cycle", IEEE Transactions on Geoscience and Remote Sensing 40, 2002, pp.461-473.
12. Sun XM, Zhu ZL, Tang XZ, Su HB; Zhang RH, "A new measuring technique of soil thermal inertia", Science in China Series E - Technological Sciences 43, 2000, pp.S62-69.
13. Agassi E, Ben Yosef N, "Parameterization of sparse vegetation in thermal images of natural ground landscapes", Optical Engineering 36, 1997, pp.2910-2917.
14. Carter LJ, Kokonozi A, Hosgood B, Coutsomitros C, Sieber A, "Landmine detection using stimulated infrared imaging", Geoscience and Remote Sensing Symposium, 2001, IEEE, vol. 3 pp. 1110-1112. 15. Grosse XE, Bousigue J, Takahashi H, "NDT data fusion at pixel level", NDT&E International 32, 1999, pp. 283-292.
16. Datema CP, Bom VR, van Eijk CWE, “Landmine detection with the neutron backscattering method”, IEEE Transactions on Nuclear Science 48, 2001, pp.1087-1091.
17. Hadas Z, Wilner K, Ben-Yosef N, "Introducing anisotropy in the autocorrelation function of natural terrain infrared images", Optical Engineering 42, 2003, pp.1683-1689.
18. Balsi M, Corcione M, "Thermal detection of buried landmines by local heating", International Journal of Systems Science 36, 2005, pp.589-604.
19. Lester J, Bernold LE, "Innovative process to characterize buried utilities using ground penetrating radar", Automation in Construction 16, 2007, pp. 546-555.
20. Muscio A, Corticelli MA, "Experiments of thermographic landmine detection with reduced size and compressed time", Infrared Physics & Technology 46, 2004, pp. 101-107.

Source: http://www.ultrasonic.de/article/v12n12/santulli.pdf