Jennifer L. Morey
Hoping to overcome the limitations previously associated with automated target recognition, Litton Data Systems has developed a miniature optical correlator that could boost the processing speed of these systems tenfold.
Although the project has been in the works since the early 1990s, it has been hampered by limited funding from outside sources.
"Automated target recognition has a less-than-stellar reputation among government people," said Gary Mallaley, director of recognition and advanced systems. "The problem is, billions [of dollars] have been spent without much return."
Mallaley attributes this discrepancy to the fact that developers have focused too much on feature extraction at the expense of correlation methods. Therefore, he said, problems occur because of the limited computational capabilities: Electronic processors operate too slowly to be effective in real-time situations.
Unlike traditional electronic systems, the Litton correlator relies on optical computation and can make eight to 10 comparisons per frame. It is built in a rugged integrated block and employs Litton's ring laser gyro technology.
Because the device exhibits a low coefficient of thermal expansion, it is suited for a wide variety of environments and temperature ranges.
The correlator's characteristics have boosted its potential for use in military applications. In military testing, it was estimated that the optical correlator could correctly identify an unobstructed target 98 percent of the time. The false detection rate was less than one in a million.
Nonmilitary applications
Mallaley sees other uses of the optical correlator in fingerprint identification and in medical applications from tumor recognition to ophthalmology.
Litton investigators hope that they will achieve the milestone of 1000 comparisons per second in gray scale this year.
Meanwhile, the government has shown more interest in funding the project. Mallaley said Congress has allocated $5 million toward additional developments in 1999.