A computerised eye developed by scientists in California can spot a chunk of mineral, camouflaged tanks, or campers lost in the woods, from thousands of metres up in the air.
The system, which scans vast areas for trace spectral signatures in light reflected from the ground, will immeasurably aid the human eye in commercial projects such as exploring for minerals. And it can peer into dense foliage and find objects that even radar can't uncover.
At the heart of the device, called Hydice, or Hyperspectral Digital Imagery Collection Experiment, is a sensor made of indium antimonide. It is capable of detecting a broad range of wavelengths simultaneously, from the visual part of the spectrum to the shortwave infrared, says Glen Healey of the University of California at Irvine.
Images collected by the sensor are divided into thousands of pixels like those on a computer screen. The system splits each pixel into 210 different wavelengths and analyses each for known characteristics of the object being sought, whether this is a particular metal or a colour of clothing. "You can cover huge areas of ground very quickly from 20 000 feet," says Healey.
However, since its development nearly a decade ago, Hydice has found relatively few practical uses for the device because of a trade-off between spatial resolution and spectral resolution, says Jian Guo Liu, a remote-sensing and image-processing expert at Imperial College, London. "If you're looking for minerals it's potentially a very powerful tool," says Liu. But this sort of work requires high resolution-in other words putting a small area into a single pixel. Because a smaller area means less light is available to analyse, high-resolution images have too little spectral detail to be useful.
Healey and his colleagues believe they have overcome this with an algorithm that allows them to identify pixels containing minerals, even if that mineral makes up only 5 per cent of the pixel. In theory, Healey could scan at a resolution of half a square metre per pixel when flying a thousand metres up and spot a fist-sized lump of sulphur.
Healey devises a separate algorithm for each target material by first determining its spectral features under all possible light conditions. During a mission, the system continuously compares the background as a whole with the signal in each pixel. If the signature of the material appears above the background, the pixel is ranked as positive.
The trick, says Healey, lies in looking for "invariants"-unique spectral characteristics that remain relatively constant despite changes in weather or daylight. "What the material looks like can change under different atmospheric conditions," he explains. But their system will deliver the same results, in rain or shine.
Author: Duncan Graham-Rowe
New Scientist issue 14th August 1999
PLEASE MENTION NEW SCIENTIST IF YOU USE THIS ARTICLE - THANK YOU