Source: The Economist

        A new type of camera that could detect explosives and other hidden items

        SALESMEN flogging digital cameras boast of the number of pixels in the images captured as an indicator of quality. The more pixels there are?and they are usually counted in the millions?the better the picture.

        That may cease to be the case. Scientists have devised a way to get the same quality using a single pixel. And it gets better?the new camera will need less memory and power, and may be able to capture images at wavelengths that could prove useful for airport security.

        The trick is to manipulate the way in which the camera captures and stores information. Doing this involves a whole new branch of information theory developed just a couple of years ago, known as compressed sensing.

        When a picture is taken with a digital camera that has, say, 1m pixels, the device momentarily captures 1m pieces of information. A computer chip inside the camera then immediately compresses this data, using a piece of mathematical wizardry called a Fourier transform. The information about the image is converted from 1m to about 10,000 numbers that correspond to the most important items of the mathematical transformation. It is these numbers that are stored to recreate the image; the rest of the data is ditched.

        Richard Baraniuk and Kevin Kelly of Rice University in Houston, Texas, used compressed sensing to eliminate this waste. As the name suggests, it cuts the quantity of data required by using a new way to sense an object. The pair built a camera with just one pixel that captures many images over time and uses information from them to create a single picture. Instead of looking once at the object using a million pixels, it looks 10,000 times using one pixel.

        The researchers, who reported their results at the Optical Society of America¡¯s annual meeting earlier this month in Rochester, New York, used a digital micromirror device, a piece of kit invented 20 years ago that is now commonly used in digital televisions and projectors. It contains several hundred thousand microscopic mirrors arranged in a rectangular array. The position of these mirrors can be altered millions of times per second. The mirrors appear either bright or dark, depending on the angle at which they are set. By changing these angles at random between their 10,000 closely spaced shots, the researchers captured enough information to create a high quality image.

        Standard cameras process the information they capture to turn it into a picture. With the single-pixel camera, by contrast, a separate computer recombines information about the light levels with information about the patterns of the micromirrors to create an image. A commercial version of the camera would therefore need far less power than cameras that do their own computing, and its batteries would last far longer.

        The new camera may have another big advantage over standard devices?the ability to capture images at terahertz wavelengths, which sit between the infra-red and microwave regions of the electromagnetic spectrum. These wavelengths can penetrate clothes, paper and plastic, so are useful for detecting hidden items. Handily, explosives and drugs have characteristic fingerprints in the terahertz range.

        There is some way to go before the single-pixel camera appears in the shops. The prototype is as big as a coffee table and takes five minutes to capture enough images to generate a high quality picture. But the technology¡¯s potential uses may make bringing it to market worthwhile.