Could Radar Be a More Cost-Effective Security Screening Alternative to X-Rays?

A key task in the security market is the detection of concealed threats, such as guns, knives and explosives. While explosives can be detected by their chemical constituents the other threats are defined by their shape. A threat detection system must, therefore, be able to produce an image of an object behind an opaque barrier.

X-rays are probably the most commonly known technology for achieving this and they are widely used for both security and medical applications. However, while they produce high-quality images, x-ray machines are not cheap and there are health concerns with their frequent use on or in the vicinity of people.

An alternative to x-rays often used at airports for full-body screening are microwave imaging systems. These allow the detection of concealed objects through clothes though the spatial resolution is relatively low and objects are often indistinguishable (hence the requirement for a manual search). The ability to detect and identify concealed items can, therefore, be improved by using a high-frequency mm-wave (60 GHz) system.

Plextek has investigated this approach through the use of a Texas Instruments IWR6843 60 – 64 GHz mm-wave radar which is a relatively inexpensive consumer component that could be customised to suit many applications. However, a single radar measurement only contains range information and not angle information. It is, therefore, necessary to collect multiple measurements of an object from different viewpoints to form an image. This is achieved through the use of a custom 2D translation stage that enables the radar to be automatically moved to any point in space relative to the target object. In this example, radar data was collected across a regular grid of 2D locations with millimetre spacing between measurements.

This large set of radar measurements can then be processed to form an image. This is achieved by analysing the small variations in the signal caused by the change in viewpoint when the object is measured from different positions. The set of range only measurements is then extended to include azimuth and elevation as well. In effect, this process produces a 3D cube of intensity values defining the radar reflectivity at each point in space. A slice through this cube at a range corresponding to the position of the box allows an image to be formed of an object that is behind an (optically) opaque surface.

In this case, a cardboard box containing a fake gun was used as the target object. Clearly, a visual inspection of this box would not reveal the contents, however, 60 GHz mm-waves can penetrate cardboard and therefore an image of the concealed object can be produced. In this case, the resulting image of the contents of the box clearly shows the shape of the concealed gun.

This example simulates the detection of a gun being sent through the post and automatic image analysis algorithms would presumably be capable of flagging this box for further inspection. This would remove the need for human involvement in the screening process for each parcel.

A more mature sensor system using this approach could be produced that did not require the manual scanning process but used an array of antenna instead. It would also be possible to produce similar custom systems that were optimised for different target sets and applications.

Acknowledgement

This work was performed by Ivan Saunders during his time as a Summer student at Plextek before completing his MPhys at the University of Exeter.