17 January 2013—Millimeter-wave scanners are able to reveal hidden guns and knives on a person’s body because that wavelength of light passes through clothing but bounces off metal. But these scanners are expensive, bulky, and, as anyone who has stood in an airport line knows, slow.
Researchers at Duke University have devised a new microwave- and millimeter-wave-imaging technology that could cut the cost, size, and speed of such scanners, opening them up to other applications. For instance, the technology, which relies on metamaterials and computational-imaging techniques, could let self-driving cars see through fog and dust, says John Hunt, a graduate student who developed the new technology. Or it could enable handheld scanners that would let police search a suspect for weapons without laying a hand on him. Hunt, electrical and computer engineering professors David Smith and David Brady, and their colleagues reported the system in the journal Science this week.
Conventional digital cameras contain chips that carry millions of silicon-based detectors. A lens directs light reflected from an entire scene to the detectors. Each detector records the intensity of light hitting it, producing information corresponding to one pixel of the image. Ideally, image sensors for the millimeter and microwave chunk of the electromagnetic spectrum would work the same way. However, traditional silicon sensors are blind to these wavelengths, and the sensors that do see them are much bigger and pricier.
To cut costs somewhat, today’s millimeter-wave scanners have a single detector, or a small array of detectors, that is scanned along an object or scene to create a two-dimensional image. The scanning mechanics add cost, bulk, and complexity, Hunt points out. “We wanted to replace all the moving apparatus with a stationary fixed aperture,” he says.
The Duke team turned to metamaterials, which are engineered materials that interact with light in unnatural ways. The team’s metamaterial is a strip of metal patterned with elements that resonate at a specific frequency to steer radiation. The researchers placed the strip on top of a separate, plastic-covered metal sheet. The plastic sandwiched between the two metal sheets—one of which had the resonators on it—acted as a waveguide, confining the light. The small metal strip would replace the lenses, multipixel detectors, and moving parts in a conventional millimeter- or microwave-imaging system. “We get a 2-D image from this thin metamaterial array,” Hunt says. “This would be a paradigm shift to some degree. Cost and form factors could be so much smaller than conventional systems.”
Here’s how it works: To image a scene, the researchers send microwaves at different frequencies, ranging from 18 to 26 gigahertz, one at a time into one end of the waveguide. As light of a certain frequency travels down the structure, it encounters resonator elements, some of which are designed to resonate at that frequency. So the light radiates out of the metal strip at those resonator spots. The emerging light waves interfere constructively and destructively and create beams that point at a variety of angles. “What propagates away from the structure is a set of beams pointing in different directions,” Hunt says.
The resonance frequency of the metamaterial elements and their layout on the metal strip determines the arrangement and orientation of the interference beams emerging from the waveguide. As each microwave frequency is applied to the strip, different portions of the scene are illuminated by the microwave beams.
A detector collects the light reflected off the scene. The series of measurements generated as the researchers send different frequencies through the waveguide are processed using complicated algorithms developed for the field of compressive imaging. Through some intense computation, compressive imaging allows the construction of an entire image from a relatively small set of measurements of it.
“This is an elegant solution to a lot of problems in microwave, millimeter-wave, and terahertz imaging,” says Eddie Jacobs, an electrical and computer engineering professor at the University of Memphis, in Tennessee. Jacobs and his colleagues have developed a microwave-imaging technique that involves a spinning disk with holes that throws different patterns of light spots from a scene onto a detector and then uses compressive-imaging math to reconstruct the picture. But, says Jacobs, “the new system is completely not mechanical. The novelty is that they use a simple metal strip, and the technology is easy to implement.”
Reconstructing the picture using the algorithms, however, can take longer, Jacobs points out. The Duke researchers were able to capture the complete data set needed to reconstruct a scene containing two or three microwaves scattering objects in 100 milliseconds. Jacobs says that faster compressive-imaging reconstruction algorithms would make the system ideal for practical use.
About the Author
Prachi Patel is a contributing editor to IEEE Spectrum. In the September 2012 issue, she reported on the technology needed to monitor and forecast drought. You can also hear her on Spectrum Radio and on Public Radio International’s “ Living on Earth.”