There's no getting away from a satellite that can see through camouflage. - Popular Science
by Dawn Stover
It's said the Pentagon's spy satellites can read the license plate numbers on cars. But what about a car -- or an enemy tank -- that's parked under a tree and covered with camouflage? It's virtually invisible to the prying lens of a conventional camera, no matter how powerful.
But that doesn't mean it can't be seen. The military can spot an enemy hiding in the woods thanks to a new way of looking at things from space called hyperspectral imaging. Instead of merely viewing objects in the colors that the human eye can see, hyperspectral cameras also detect colors that are invisible to the naked eye-including the infrared portion of the electromagnetic spectrum. To a hyperspectral camera, the leaves on a tree don't look anything like camouflage netting or paint. And a decoy doesn't look anything like a real vehicle, because they are made from different materials. In fact, hyperspectral imaging can even detect the subtle differences between two types of similarly colored paint-one applied to a U.S. military vehicle and the other used by an enemy. That very well could spell the end of camouflage as we know it.
By the time you read this, the military may have already begun testing an experimental hyperspectral imaging system known as Warfighter-1. In addition to spotting hidden enemies, hyperspectral images also provide clues about the conditions ground troops are likely to encounter: The soil may be too muddy for heavy armored vehicles, for example.
Hyperspectral imagery isn't only for the military. It can also be used for civilian applications-such as detecting crop blights, mapping property damage, and nabbing marijuana growers in remote areas. The military payoff, however, is expected to be huge. Warfighter-1 is piggybacking on the OrbView-4 satellite, which is scheduled to be launched by the Orbital Imaging Corp. (OrbImage) this summer. OrbView-4 will not only provide the Air Force Research Laboratory with hyperspectral images of military vehicles and other test targets at seven U.S. military bases, it will also be the world's first commercial hyperspectral imaging system in space.
OrbView-4 is a huge flying cam-era with a half-meter aperture that admits sunlight reflected by objects or surfaces on the ground. At the focal point behind the camera are two arrays of detectors. One array consists of panchromatic sensors that detect light in a single spectral band -the portion of the electromagnetic spectrum visible to the human eye. Information from these sensors is combined to form a black-and-white image with a resolution of 1 meter -equivalent to the high-resolution images produced by Space Imaging's Ikonos satellite, launched in October 1999. (A 1-meter resolution means these satellites can pick out any objects on the ground that are larger than a meter in size.)
OrbView-4's second detector array consists of multispectral sensors with a 4-meter resolution. These sensors detect light in the red, green, blue, and very-near-infrared regions. By adding multispectral data to a panchromatic image, you can create a color image that looks as though it were shot from the cockpit of an airplane.
But even in a color picture, some objects are difficult to see. The human eye cannot distinguish between vegetation and camouflage, for example. That's where hyperspectral imaging comes in-it allows us to "see" objects in colors that are normally beyond our range of vision.
Some of the reflected light entering the OrbView-4 camera is diverted to a third set of detectors comprising the Warfighter-1 imaging system. Here light is sorted into 200 different colors, or spectral bands, ranging from visible light to shortwave infrared. Each band is only about 10 nanometers wide.
"You can think of a hyperspectral image as being a stack of 200 pictures, each of a separate spectral band," says Randy Kindsfather, program manager for the Warfighter program at Northrop Grumman, which built the imaging system. When you combine all of these pictures, you get a "spectral signature" of a scene.
It's difficult to analyze such an image simply by looking at it. Because the identification of objects in the image relies on their spectral signatures rather than literal interpretation, analysts use computers to help them process the images and extract useful information. One way to process a hyperspectral image is to look at individual color bands. "Depending on what colors you select, you might highlight different things," says Joe Dodd, vice president of national programs for OrbImage, developer of OrbView-4.
Camouflage paint won't keep an enemy hidden, either. In the shortwave infrared portion of the spectrum, many paints are transparent. So you can use these bands to look at materials beneath the paint.
Another way to process a hyperspectral image is to look at each element of the image and compare it with the elements around it. "You can look at the background, look at something you suspect might not be background, and ask 'Are they really different?'" says Kindsfather. You can also compare the spectral signature of an unidentified object with a library of known signatures.
A typical hyperspectral image, after processing, looks like an ordinary picture except that you can add special colors to identify objects of interest: For example, you can make all the oak trees in a forest look red. With a resolution of 8 meters, Warfighter-1 will be able to spot large targets of military interest. Future hyperspectral systems with higher resolution may even be able to find recently buried land mines. "The disturbed soil is spectrally different than undisturbed soil," explains Rich Madonna, senior adviser for electro-optical systems at Northrop Grumman.
Although commercial customers will be able to purchase hyperspectral images from OrbImage, the resolution won't be as high as for military images-only 20 meters. The cost for each image will probably be around $20,000, says Dodd of OrbImage.
If that sounds high, consider the potential value of such an image to a vineyard owner, for example. "If we can produce imagery that allows early detection of insect blight, and early intervention, that information may be nearly priceless," says Dodd.
Hyperspectral images could also be a valuable tool for mineral exploration companies searching for rare ores in remote areas. "We can take images of parts of the world where you can't even get people to, or it's extraordinarily expensive to do any exploration," says Dodd.
Although hyperspectral imaging satellites offer global coverage, they do have some limitations. They can't see at night or through heavy cloud cover, and enemies may know exactly when the imaging system will be passing overhead because orbits are predictable. Still, it won't be easy to hide from these all-seeing eyes in space. The only camouflage that can fool a hyperspectral imager is a material with the same spectral signature as the real thing.
Hyperspectral "tags" on vehicles may also save soldiers' lives by reducing the incidence of friendly fire casualties in war. "You could basically put a coating on your assets so you could identify friend from foe," says Madonna. "The foe must have a hyperspectral sensor to see the marker, and must know that it is a marker."
For now, Warfighter-1 is experimental and not ready for combat. "It's in its infancy," says Alan Murdoch, director of commercial and civil space programs at Northrop Grumman. "But it's going to grow, not only in the military uses but also in the commercial arena." |
