Superior radar images of the moon, inner planets and asteroids are possible with a "polar-format" radar-processing algorithm developed at the University of Illinois. The algorithm, from spotlight-mode synthetic aperture radar, provides improved image quality over conventional processing, without an excessive increase in computational requirements or complexity.
"In astronomy, radar reflectivity data are sometimes used to supplement
other types of observations," said David Munson, a U. of I. professor of
electrical and computer engineering. "Radar measurements, for example, have
proven particularly useful for imaging Venus, where a thick layer of clouds
perpetually obscures the planet's surface. Oftentimes, however, the radar
images produced from Earth-based measurements are of poor quality."
Conventional radar-imaging systems are based on range-Doppler techniques,
Munson said. "But in the time it takes to collect astronomical data —
typically 10 to 20 minutes — the object's rotation causes both the range
and the radial velocity of reflectors to change with time. Because
conventional range-Doppler processing makes no allowance for this relative
motion, the resulting image is blurred."
The polar-format, spotlight-mode synthetic aperture radar approach avoids
this problem by affixing a spatial-domain coordinate system to the target.
As the object rotates with respect to the radar, the coordinate system
rotates with it, thus avoiding smearing in the image.
Munson, former graduate student Jennifer Webb (now a researcher at Texas
Instruments) and Nick Stacy, a researcher with the Microwave Radar Division
of the Defense Science and Technology Organization in Australia, recently
applied the polar-format radar-processing algorithm to lunar reflectivity
data collected at Arecibo Observatory in Puerto Rico.
"Although the moon constantly presents the same face toward Earth, we get
to see it from different angles as it moves," Munson said. "And in our
mathematical model, that's all that's required. The principle employed is
nearly identical to that used in computer tomography in medical imaging."
The high-resolution lunar images produced by the researchers were far
superior to what had been obtained in the past with conventional
radar-processing techniques. "The amount of computational work was only
three times what was formerly required," Munson said. "So, with a small
increase in computational effort, we can get vastly improved imaging."
As an additional benchmark, the researchers compared their images with
those obtained with another approach developed by Stacy, called focused
range-Doppler processing. This latter technique "produces the best known
results, but is much more expensive, computationally," Munson said. "Our
polar-format processing algorithm performed nearly as well, with
considerably less effort."
The researchers describe their algorithm and present their results in the
November issue of IEEE Transactions on Image Processing.
