Abstract

There is a large potential of acoustic methods in antipersonnel landmines detection based on materials characterisation. In this paper the resolution capability of one realisation of that principle is described and quantified. Resolution capability is modelled with different requirements imposed. Using appropriate experimental set-up, that quantity is determined in laboratory conditions for a representative, referent set of testing objects. Results obtained in modelling and experimental work show both the usefulness of resolution capability as an indicator of a buried object materials characterisation, and the reached level of buried object materials characterisation using applied approach.

Introduction

Large number of 80 million mines left in more than 80 countries world-wide [1] cannot be efficiently removed using present day humanitarian demining [2, 3]. As a crucial phase in humanitarian demining, which is responsible for a majority of duration and risk, antipersonnel mine detection (APMD) has been recognised [3, 4]. These facts motivated scientists and experts from NDT community and related fields to start research and development of better APMD method and technique [5]. Work conducted brought about necessity to differentiate between the buried object detection and classification of the buried object into mines or other objects. This is especially seen in probing [6, 7, 8, 9, 10], one of APMD techniques, which is regularly performed in humanitarian demining. It is suitable because of relatively low level of logistics required, and applicability in various terrains. Its drawbacks are large duration and large risk caused by physical contact established between the deminer's handprobe and buried mines [11]. The contact is established several times with a particular object in order to establish its shape. As a rule, digging of the object is included, which is again connected with a large risk and time consumption. Overall, as the main problem of the probing technique we extract lack of confirmation capability. Therefore, a combination of probing and confirmation oriented technique brings about improvement in humanitarian demining.

Fig 1: Possible realisations of buried object materials characterisation sensor, a) draft of one realisation, b) reduced cross-section of realisation discussed in text.

There have been several attempts to formulate such a combination of probing and confirmation oriented technique [6, 8, 9]. In these, confirmation was realised as materials acoustic characterisation, either by using induced acoustic emission [9], or by using low [8] and high frequency [6, 12, 13] ultrasonic waves. In a combination of probing and materials acoustic characterisation the former is detection, while the later is confirmation technique. A consequence is safer and faster detection process, hence the probe with characterisation sensor was coined Fast Probe [14]. Such an approach is suitable especially for antipersonnel mines with low metal content, for which other techniques like metal detection are rather inefficient.

Within the OCULAR technique [4], in one of the recent Fast Probe’s formulation, shown on figure 1a, the materials acoustic characterisation is based on acoustic impedance matching. In more detail, ultrasonic impulse emitted from the transducer, mounted near the probe’s tip, is scattered at the interface between the probe and the buried object. Reflected part of the scattered impulse is registered by the transducer and subsequently analysed. In case of matched impedances, the transmission is maximal, i.e. the reflection is minimal. Hence, buried object surface material is recognised as the material causing minimal reflection of the ultrasonic impulse, in case of constant or controlled other parameters in the model. Acoustic impedances are matched so that this extreme case occurs when buried object surface is made of polymer, which in most cases means low metal content antipersonnel mine. Regarding acoustic impedance matching, it has been confirmed that the approach is fruitful in controlled laboratory conditions [6], in reduced geometry shown on figure 1b. Furthermore, in aligning the exploited conditions with realistic ones, the influence of non-characterised contact layer, made of local soil and formed between the probe’s tip and buried object surface, was analysed [12]. It was shown how to include the homogeneous contact layer influence into the materials characterisation results. Additionally, influence of surface roughness was determined [15].

In this paper we define resolving power, a measure of the robustness of the system. This measure determines average difference between transmission factor values for different buried objects’ materials. It should be maximised, as in realistic conditions occur different sources of signal quality degradation, which are not considered here explicitly.

In the second section the underlying model is briefly sketched, and the resolving power defined. Results of the model are presented and discussed in the third section. Conclusions and lines of future development are given in the fourth section.

Resolving Power

Ultrasonic impulse propagation is considered in the linear regime, for perpendicular incidence of longitudinal impulse on mutually planparallel and homogeneous layers. Boundary conditions on interfaces in the model are continuity of total elastic field and stress tensor normal projections. For a monochromatic wave, the transmission coefficient T is calculated using transfer matrix technique. Details of calculations are given elsewhere [6].

In case when the analysis is performed in the frequency domain of the registered signal, the quantity that measures dependence of T on buried object acoustic impedance Z₄ is the following

(1)

where t_i = d_i/v_i Index 2 (3) refers to a layer which is adjacent to piezoelectric slab (buried object). Quantity Z_Max is taken as the upper limit of acoustic impedances which buried objects’ surfaces achieve. Meaning of quantity s_T is the following: in case of impedance matching there is pronounced maximum in T, and s_T obtains relatively large value, curve A on figure 2. Quite the opposite characterises a case when impedances are not matched. Then T achieves values in relatively narrow interval, hence s_T obtains relatively small value, curve B on figure 2. Therefore, maxima of s_T point to set of values t_2,3 and Z_2,3 for which T spans rather large interval, what implies that there is clear resolution of different materials possible. Because larger s points to better resolution of different materials in contact with probe’s tip, it is called resolving power. This quantity is a particular realisation of one-parameter based classification of buried objects. More general situations require larger parameter set for classifications [16].

Fig 2: Transmission coefficient as function of the buried object acoustic impedance. Z4. A – matched impedances for Z4 = 3,2 MRayl, – unmatched impedances. Values of resolving power are 0,05 and 0,005 for curves A and B, respectively.

Results and Discussion

Frequency and parameters of the two-material slab are free parameters of the model. Because of the convenience frequency is fixed to 2 MHz in further considerations. Mechanical requirements, as discussed elsewhere, limit Z₃ to few possible values. Here Z₃ = 45 MRayl and Z_Max = 50 MRayl were taken.

Material	Piezoelectric slab	Layer 2	Layer 3	Buried object
Z, MRayl	7,2	3,2;6,8 and 18	45	3,2
Table 1: Acoustic impedances of layers as in figure 1b, which were used in calculations shown in this paper.

Resolving power as a function of t₂ and t₃, for several Z₂ is shown on Figures 3 and 4. Phase change for a wave of frequency 2 MHz between i-th layer boundaries is 2p rad for t₁= 1/2. In graphs, the slightly smaller intervals of t^’s are shown.

From figure 3 it can be inferred that local maxima caused by changes in t₃ are higher and narrower than local maxima caused by changes in t₂. This can be related to corresponding acoustic impedance values. Furthermore, from intersection of the lines of local maxima it is seen that there does not occur constructive effect resulting in the brightest point in the graph, but that both peaks are averaged in a way. Graphs on figure 4 show similar trends. Marginal changes of one series of local maxima by variations in the parameters of the other layer show that in the exploited parameter region there is strong separation of impedance matching effects into one layer of two-material slab. If the acoustic impedances of these two layers were similar, then their effect would be that of a single matching layer. Graphically, a form of local maxima curves would be diagonal in s vs. t_2,3 representation.

Fig 3: Resolving power for Z2 = 3,2 MRayl, a) 3D plot of one local maxima, b) density plot showing several local maxima, represented as brighter regions. Lines of intersecting local maxima repeat periodically.

Fig 4: Resolving power for a) f = 2 MHz and Z2 = 6,8 MRayl, b) f = 2,2 MHz and Z2 = 3,2 MRayl.

While the aforementioned refers to matching of impedances in case when the polymer surface of buried objects is to be recognised, the contrary approach could be formulated. In it, the surfaces of buried objects not made of polymers should be differentiated. In that case, one obtains a false alarm reductor, which differentiates between metals, rocks and other similar objects on the one side and polymer objects on the other side. One example of parameter set bringing which brings about such a configuration is shown on figure 5. The effects are caused by the fact that for a given set of layer 2 parameters, layer 3 figures as a diffuse boundary of layer 4, which is the buried object. In that case, effectively there is one –material layer, hence impedance matching conditions are known in analytic form

(2)

In case of Z₁ = 7.2 MRayl and Z₃ = 45 MRayl this gives Z₂ = 18 MRayl. Therefore, on figure 5 is shown the case of one-layer impedance matching between piezoelectric ceramics and external protection layer. An additional consequence is that results for T and s depend negligibly on t₃ , what has implications in production simplification of the accompanied device. In case of large enough values of buried objects acoustic impedances, these are similar to acoustic impedance of steel, what brings about impossibility of resolving these materials from steel. However, in the configuration of false alarm reducer this property is its advantage.

Fig 5: a) Transmission factor and b) resolving power. In both graphs f = 2 MHz and Z2 = 18 MRayl. In a) t2 = 1/8 while results depend negligibly on t3.

Conclusions and Lines of Future Development

The resolving power was defined for the approach to buried object materials characterisation based on acoustic impedance matching. This quantity is related to a pronounced structure in transmission factor taken as function of buried object acoustic impedance. The pronounced structure is related to existence of acoustic impedance matching with one type of materials of the buried objects. Generally, larger resolving power means that there is larger difference in transmission coefficient for a collection of different materials. If set of objects that are of interest has several distinct values of acoustic impedances, then it is appropriate to consider reduced version of resolving power in which integration is exchanged for summation over set of relevant acoustic impedances.

Acknowledgments

This work is part of the author’s Ph.D. thesis.

The work was financed through the project CRO MoST 120 098.

References

1. International Campaign to Ban Landmines, Landmines Monitor Report 2001, Landmine Monitor Core Group, USA, 2001,
2. N.Kingsbury, “Land Mine Detection DOD’s Research Program Needs a Comprehensive Evaluation Strategy”, US GAO Report, GAO-01 239, 2001,
3. V.Krstelj, J. Stepanic, Jr., “Non-destructive Testing in Antipersonnel Landmines Detection”, Int. Conference MATEST’99, Cavtat, 1999, Proceedings, pp. 109-
4. V.Krstelj, J. Stepanic, Jr., “Humanitarian de-mining detection equipment and working group for antipersonnel landmines detection”, Insight, Vol.42, No.3, 2000, pp.187-190,
5. EFNDT WG5 – APMD Programme, Doc1, 1999,
6. J.Stepanic Jr., H.Wüstenberg, V.Krstelj, H.Mrasek, “Contribution to Classification of Buried Objects Based on Acoustic Impedance Matching”, submitted to Ultrasonics,
7. K.M.Dawson-Howe, T.G.Williams, “Automating the probing process”, SusDem’97 – International Workshop on Sustainable Demining, Zagreb, 1997, Proceedings, pp.4.24 –4.29, see also: www.cs.tcd.ie/Kenneth.Dawson- owe/Landmines/#PUBS,
8. M.A.Borza, F.G.De Witt, “Acoustic landmine prodding instrument with force feedback”, US Patent #6 109 112, 2000,
9. D.Antonic, I.Ratkovic, “Ground probing sensor for automated mine detection”, KoREMA ’96 – 41^st Annual Conference, Opatija, Croatia, 1996, Proceedings, pp.137-140,
10. P.Horowitz, http://seti.harvard.edu/mines/jason.htm,
11. R.Gasser, “Feedback Prodders; A Training Tool to Improve Deminer Safety”, Journal of Mine Action, Vol.5, No.2, 2002, see also: http://maic.jmu.edu/journal/5.2/features/prodders.html,
12. J.Stepanic Jr., D.Markucic, “Contact Region in Ultrasonic Prodder Technique”, Int. Conference MATEST'01, Cavtat-Dubrovnik 2001, Proceedings, pp.56-61,
13. J.Stepanic Jr., “Material Characterization by Mechanical Point Contact Impact Emitted Ultrasound”, 15^th WCNDT, Rome, Proceedings, 2000, see also http://www.ndt.net/article/wcndt00/papers/idn074/idn074.htm,
14. V.Krstelj, “Buried Landmines Detection using Ultrasonic Materials Characterisation”, International Patent ApplicationPCT/HR/0036,
15. D.Markucic, “Possibilities of Material Classification by Means of Ultrasound”, 15^th WCNDT, Rome, Proceedings, 2000, see also: http://www.ndt.net/article/wcndt00/papers/idn054/idn054.htm,
16. D.Antonic, M.Žagar, “Method for determining Classification Significant Features from Acoustic Signature of Mine-like buried objects”, 15^th WCNDT, Rome Proceedings, 2000, see also: http://www.ndt.net/article/wcndt00/papers/ idn085/idn085.htm.

© NDT.net - info@ndt.net

|Top|