Why we need better mine detection
Lady Diana (figure 1) did the world a great service in publicising the tragedy of anti-personnel land mines to the world. In the years since her death we have seen the signing by some countries of the 1998 "Ottawa treaty" to ban the production of new mines. But not all countries signed the treaty and the decade since then has seen a growth in the number of mines in the ground. Now there are around 100 million buried land mines in 78 countries around the world. They cause around 15,000 casualties each year, mostly civilians including many children. Only around half the casulaties die. The other half live maimed lives. It cost about $10 to make a landmine but about $500 to remove it from the ground. At the present level of funding, and the present methods of location, it will take hundreds of years to remove them.

Methods of detecting buried anti-personnel mines
Sadly mines have become more difficult to detect over the years as the manufacturers seek to make them undetectable by the existing methods. Figure 2 shows a typical PMA2 modern mine, 69 mm in diameter, and made from plastic with only a tiny pin of metal. There are two main methods used today. Metal detectors are much used by the sappers for mine detection. These are much the same as those used by treasure hunters today. Often an audible sound is heard when metal is near. But battlefields are notoriously full of bits of metal. Any sign of a signal and the sapper must lie down and use, say a prodding stick as in figure 3, to help define the cause of the signal. There will be many false alarms, but if it is the right shape the sapper must try to dig the object out. It is dangerous work - around one sapper is killed per 1000 mines recovered.
Many other methods have been tried but no others are widely used. Any new method needs to
(i) have a high probability of detection for any type of mine
(ii) have a low false alarm rate for any other clutter present
(iii) relatively cheap to be affordable in war-ravaged countries
(iv) be useable by local non-expert personnel without too lengthy training
Ground penetrating radar does a reasonable job with the first two criteria above, but conventional pulsed radar equipment is expensive at around $30,000 and this has limited its use. Also the raw pulse echo data do not present an easy interface with an untrained user. The invention of microwave holography offers a chance of completing the second two criteria. It has similar or better performance, it cost only around $5,000 and in particular a manual scan of the microwave head over the surface can give a direct image of the buried objects that can be easily interpreted.

How microwave holography works
The mathemetics is very simple. Suppose an incident wave at the source to the left of the figure y₀ = cos (2pnt + q₀). Here n is the frequency and t the time. q₀ is some unknown phase angle.
The scattered wave will be going in the reverse direction and when it returns to the source will have a phase difference equal to twice (because it goes there and back) the distance d to the obstacle, divided by the wavelength, which is equal to V/2pn, where V is the velocity in the medium: y = cos (2pnt + 4pd/V + q₀).
To define the interference signal it is necessary to multiply the incident and reflected amplitudes at the source point and average over time. The unknown phase angle q₀ disappears and the resulting signal is A(d) = cos (4pnd/V).
A diagram of this interference function as a function of the obstacle distance d, for several different frequencies n is shown in figure 5. The wavelength of the oscillation decreases as the frequency increases.

The RASCAN system for microwave holography
The RASCAN system developed over many years by the Bauman Moscow Technical University in Russia. Figure 6 shows the scanning head being used on a sand bed containing an aluminium plate, as illustrated in figure 7. The plate was inclined so that it dropped by 85±5 mm in 300 mm. This therefore simulates the simple distance scan which we described above. The sand is covered with a plastic sheet ruled with lines 10mm apart and the manual scan consists of moving the head along each of these lines. A small wheel behind the head measures the distance travelled by the head along the scan and records the signal at say 10 mm intervals. It also records the signal with two polarisation directions, parallel to the scan and perpendicular to the scan. This can be very useful in identifying the nature of the reflected object, but will not be discussed here.
The RASCAN images are recorded at five different frequencies. Figure 8 shows a sum of a 4.0 GHz image coloured red, a 3.9 GHz image coloured yellow, a 3.8 GHz image coloured green, 3.7 GHz image coloured blue and a 3.6 GHz image coloured violet. The edge of the plate was about half way down the image, so that the upper half of the image is simply background. However the oscillatory signals from the sheet are clearly seen in the lower half of the figure. The image closely remembles the optical interference pattern as seen by a slightly inclined glass plate. We often call the characteristic pattern from inclined surfaces the "zebra effect". By averaging several of scans together a more precise definition of the depth-dependence of the five frequency scans can be obtained as in figure 9.

A search for simulated mines using microwave holography In order to evaluate microwave holography in the task of searching for mines a test bed was dug in the garden of the Department of Electronic Engineering at the University of Florence, Italy as in figure 10. In the test bed were buried several simulations of mines, as shown in figure 11. The test bed was scanned manually using both microwave holography and also conventional pulsed radar. The holographic radar results are shown in figure 12. The two clear images to the left are in fact tobacco cases, simulating metallic mines. That to the top right is a plastic object simulating a plastic mine, and is also clearly seen but with a distict colour indicating different phase change. The image to the lower right is the PMA2 mine shown in figure 2.