1. INTRODUCTION

A MEXT (the Ministry of Education Culture, Sports, Science and Technology) Committee of Experts on Humanitarian Demining Technology presented the report："Promoting Research and Development for Humanitarian Demining" on May 27, 2002. The report showed that research and development based on Japanese advanced science and technology should be carried out for supporting safety and effective activities of Humanitarian Demining in mine affected countries and that the technology acquired therefrom is expected to provide a field trial in a country or countries that have suffered from antipersonnel mines around FY2005 and FY2007. According to a sector designated to JST (Japan Science and Technology Agency) from the MEXT on ground of this report, JST set a research area, invited research proposals and started the research and development.

Under this research project, we are developing mine detection systems based on advanced sensing technology and access-and-control technology. As a short-term R&D project, the sensor units use new GPR (ground penetrating radar) with existing MD (metal detector) technology in a single system. The concept, in which the deminers are given a continuous supply of detailed underground imaging data. As a middle-term R&D project, we are developing sensors that can detect explosives in landmines to improve detection efficiency. The research and development have been done by cooperation of universities, industries, and government.

SAR-GPR is a sensor system developed to be equipped on a compact vehicle named “Mine Hunter Vehicle” and the others. This system is composed of a GPR and a metal detector for landmine detection. The GPR employs an array antenna combined with synthetic aperture radar algorithm, and it 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. In order to achieve a compact and fast data acquisition system, we developed a new compact network analyzer. 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 about 30kg, and it can be mounted on a robotic arm on a small unmanned vehicle. In this paper, we summarize the developed component of SAR GPR, which includes a compact vehicle concept, SAR-GPR hardware and software, compact network analyzer, and demonstrate some field test results obtained in a field test in March 2005, in Japan.

2. MINE HUNTER VEHICLE

This compact vehicle under development can negotiate tight turns and rough terrain, and safely access to minefields providing fine underground images. The water and dust-proof sensor system can withstand the severe conditions of minefields, and its remotely-controlled vehicle makes mine detection safe.

There are two interchangeable varieties of the GPR system depending on operational conditions?a high-speed model and a soil-type adaptation model. Metal-collection electromagnets and air blowers can also be attached to the vehicle’s robot arm. To obtain detailed underground data for imaging, accurate positioning of the sensor is necessary. A handheld GPR-MD system developed as a part of the project can also produce imaging data, but inaccuracies due to movement of the user’s hands must be taken into account. However, the highly accurate position control of the main system’s robotic arm makes it possible to acquire extremely clear underground images. The clearer imaging afforded by correct positioning of the sensor is a practical example of Japan’s leadership in advanced robotics.

3. SAR-GPR CONCEPT

3.1 Concept of Development

Metal detectors, using electromagnetic induction theory operating at around 10-100kHz have widely been used by many demining operations in the world. Metal detectors are inexpensive, light weight and easy-to-use tools, but it can detect only the existence of conducting materials. In the real situations of humanitarian demining, the most serious problem is that, although metal detectors can detect buried conducting materials, most of them are not landmines, but fragments of barrel and metal debris. It causes quite high fail alarm rate. On the contrary, Ground Penetrating Radar (GPR) can be used for imaging tool due to its high resolution, therefore it would be quite ideal for identification of buried objects that most metal detectors cannot achieve. However, imaging by GPR is very difficult in strongly inhomogeneous material due to strong clutter. We are proposing to use a synthetic aperture radar (SAR) approach to solve this problem, and have developed a SAR-GPR equipment. SAR-GPR is a combined sensor tool composed of a metal detector and a array-antenna GPR equipments.

When a simple imaging by radar is difficult, we can use a set of scanned radar data for further signal interpretation. This can be achieved by solving inverse scattering problems, but it consumes much time for computation. Another approach is migration signal processing, which are simpler compared to inverse scattering approach, but much simpler and robust. Migration processing, which is often used in geophysical exploration on seismic and GPR is equivalent to synthetic aperture radar (SAR) processing, widely used in remote sensing.

3.2 Array antenna and data acquisition

Landmines are targets which are buried in shallow subsurface, normally less than 10cm. A GPR antenna has to have a good electromagnetic coupling to subsurface material, so it has to be set very closely to the ground surface. Hence, the distance to the buried target from the antenna is less than 20cm. We are designing our GPR for the use in a very dry conditions such as Afghanistan, therefore, we can use relatively high frequency compared to conventional GPR applications, which normally use a frequency up to 1GHz. We adopted an antipodal Vivaldi antenna for our GPR, because it has relatively sharp radiation pattern in very broad frequency range, and it requires no balance-unbalance transformation. At the same time, this type of antenna can be fabricated easily, and suitable for an array antenna. We used FDTD for designing the antenna, and the prototype antenna showed a good operation in the frequency range of 1GHz and 4GHz.

The antennas will be scanned mechanically near the ground surface and acquire the radar data. The data will be processed afterwards for subsurface imaging. At the same time, we use an array antenna having 6 elements for data acquisition, in order to suppress the ground clutter. We adopt a Common Midpoit (CMP) technique for gathering a data sets acquired at one position by the array antennas. Fig.2 shows the laboratory test of an array antenna for SAR-GPR, which uses three normal vector network analyzers for data acquisition. In the CMP technique, we acquire three sets of radar data using the three pairs of antennas at one position. Fig.3 shows the antenna unit of SAR-GPR to be mounted on Mine Hunter Vehicle.

Fig.3 shows the 6 Vivaldi antennas packed in a plastic box, and Fig.4 shows the its test operation.

3.3 Compact vector network analyzer

In order to achieve the optimum SAR-GPR performance, selection of adaptive operational frequency is quite important. Also, antenna mismatching causes serious problems in GPR. Most of the conventional GPR systems employed impulse radar, because it is compact and data acquisition is fast. However, most of the impulse radar system had disadvantages such as instability in signal, especially time drift and jitter, strong impedance mismatching to a coaxial cable, which causes serious ringing, and fixed frequency range. Vector network analyzer is a synchronized transmitter- receiver measurement equipment. It is composed of a synthesizer and coherent receiver. It enables quite flexible selection of operation frequencies, and stable data acquisition. In addition, commercial vector network analyzers are equipped with a calibration function, which masks impedance mismatching caused by RF hardware. Impedance matching of antennas to coaxial cables in GPR is quite difficult in all the frequency range of operation. Therefore refection caused by impedance mismatching returns to a generator, and signal wave deforms. In order to avoid these effects, many GPR antennas adopt strong damping by impedance loading, which decreases antenna efficiency. If we use a vector network analyzer, reflection from antennas can be perfectly absorbed by the vector network analyzer, therefore, we can operate antennas without heavily impedance loading.

Table 1 Comparison of commercial and developed vector network analyzers (VNA)

However, due to the large size and weight, conventional network analyzer cannot be mounted in SAR-GPR. A compact vector network analyzer has been available, which can be used field measurement. However, the existing compact vector network analyzer had limited frequency range, which cannot be adopted in GPR for landmine detection, and data acquisition speed was too slow for practical use. Therefore, we develop a new compact vector network analyzer which fits to our requirements. Table 1 show the comparison of the specifications of commercial and new vector network analyzers. The prototype of the compact network analyzer was completed, and was installed in the SAR-GPR system. In order to achieve fast data acquisition, three vector network analyzers were installed, and they synchronize and acquire data works simultaneously.

3.4 Image reconstruction algorithm

Two-stage signal processing was carried out after the acquired data was transformed into time-domain data by inverse Fourier transformation. At first, the CMP stacking was carried out for suppression of ground clutter. In this processing, 5 data sets acquired at one position is stacked by calculating the time delay differences due to different propagation length between antenna sets. Clutter from the ground surface and homogeneous gravel can be suppressed by this CMP processing. The stacked signal is then processed by the diffraction stacking algorithm and a 3D image is reconstructed. The diffraction stacking is one of the standard migration algorithms used for GPR and seismic measurement. The reconstructed image  by the diffraction stacking is given by

        (1)

where is the measured signal at the antenna position  at time where

                    (2)

is the travel time from the antenna position to the focusing point, and is the assumed velocity of wave in the material.

At the same time, we are developing more advanced signal processing which can be used together with SAR-GPR. We are planning to use the GPR data set for estimation of ground surface topography, and estimation of velocity model, which can be implemented in the image reconstruction algorithm

3.5 Metal detector

We believe that detection of buried landmine from GPR image is too difficult in practical situation, because the GPR mages is always suffered artifacts caused by strong clutter, even in SAR-GPR signal processing was adopted. Therefore, all the judgment of data is carried out by combination of metal detector information. CEIA MIL-D1 sensor was equipped on the SAR-GPR system, and the metal detector signal is visualized at the same time.

3.6 Laboratory evaluation test

Test measurement was carried out using a sand box in laboratory condition, which simulates very rough ground conditions. A sand box was filled with gravels having a diameter about 50mm and crashed rocks having a diameter about 5mm. The averaged dielectric constant of the mixed material is about 3.8, which was determined from the radar measurement. The correlation length of the ground surface is 20mm and the RMS height is 15mm. The radar target is a model of a plastic landmine Type-72. It contains very small amount of metal part, and it is mostly filled with dielectric material. This model has a cylindrical structure having a diameter of 78mm and the thickness is 40mm. The antenna array was moved by a horizontal X-Y stage and acquired the radar data at 1cm interval. The antenna height from the mean ground surface was about 50mm. Figure 6 and 7 show the GPR images of CMP stacked data, before and after SAR processing. Figures 8 and 9 show the 3-D GPR reconstructed images. The effect of SAR processing is very clear and only after SAR processing, buried landmine image can be clearly found in soil. Now we can distinguish the reflections from the buried landmine model and the ground surface separately.

We could find a good improvement of GPR image by using the CMP and migration. We tested this technique for different conditions. We also tested the same material by changing the roughness condition, because reconstructed image can be changed according to the each shape of the ground surface. However, we could find that almost same quality of image can be obtained if we use the same material as shown in Figs.6 and 7.

Figs. 8 and 9 compare the improvement of the image quality by signal processing. Fig.10 Shows the SAR-GPR image acquired in the evaluation test held in Japan in March 2005.

4. CONCLUSION

SAR-GPR was designed and prototype was produced. It is equipped with three sets of GPR systems having Vivaldi antenna and a compact vector network analyzer. CMP and Kirchhoff type migration algorithm was adopted, and we could show that the signal processing can reduce the ground clutter quite well, and SAR-GPR can image buried landmines even in very high inhomogeneous material case.

Acknowledgements

This work was supported by JST (Japan Science and Technology Agency). A part of this work is also supported by JSPS Grant-in-Aid for Scientific Research (S)14102024. We thank Mr. Tatsuki Yagi for his contribution in the fabrication and tuning of the system.

REFERENCES

[1]       M.Sato, Y.Hamada, X. Feng, F.Kong,Z.Zeng, G. Fang,”GPR using an array antenna for landmine detection,” Near Surface Geophysics, 2, pp3-9,2004.

[2]       X. Feng and M. Sato,  “Pre-stack migration applied to GPR for landmine detection,” Inverse problems, 20, pp1-17,  2004.

[3]       X. Feng,, Z. Zhou., T. Kobayashi, T. Savelyev, J. Fujiwara and M. Sato, Estimation of ground surface topography and velocity models by SAR-GPR and its application to landmine detection, Proc. Detection and remediation technologies for mines and minelike targets X, March 2005.

[4]       http://www.jst.go.jp/kisoken/jirai/EN/index-e.html