New Mine-Clearing Technology Performs Well in Early Tests
ARLINGTON, VA., April 1, 1999 -- The Office of Naval Research and the Defense Advanced Research Projects Agency are sponsoring development of a new mine-detection technology that promises to significantly improve the Navy's ability to detect and clear underwater mines.
"In early but pretty realistic tests we've reduced the number of undetected mines by half and decreased the number of false alarms by a factor of five," said Dr. Richard Lau of ONR's Mathematical, Computer and Information Sciences Division.
The advanced technology is part of the Navy's stepped-up effort to develop an "organic" mine-hunting capability for the Fleet's surface combatants, amphibious ships and submarines. Such a capability would give a ship the ability to identify and avoid or deactivate mines as it encounters them instead of relying on specialized minesweeping ships which may not be available when needed.
The price of a state-of-the-art U.S. Navy Aegis cruiser is about $1 billion. The price of a low-tech underwater mine capable of crippling an Aegis cruiser is as low as $2,000. Protecting ships and sailors from these cheap, easy-to-deploy weapons -- which nearly sank the Aegis cruiser U.S.S. Princeton and the amphibious ship U.S.S. Tripoli during the Gulf War -- is a Navy priority. Defense Secretary William Cohen recently directed an additional $315 million to the Navy's mine warfare budget for fiscal years 2000 to 2005 "to translate the 'organic vision' to executable programs and deliver this capability to the fleet."
The technology uses a new wavelet technique for computer noise removal from and compression of side-scan sonar images, and a statistical classification method to make very subtle judgements about whether or not an object really is a mine. Researchers working on the project report that the wavelet technique decreases the number of undetected mines by at least 50 percent. Statistical classification, combined with wavelet-based filtering dramatically decreases the number of false alarms.
"These are terrific results so early in the project," said Dr. Dennis Healey, DARPA project manager. "We expect to see further improvement once the technologies are integrated into a single system."
Funded jointly by ONR and DARPA, the research is coordinated among Boston, Brown and Yale universities, the Naval Coastal Systems Center, Panama City, Florida, the Ecole Nationale Superiore in France and Cognitech, Inc. in Pasadena, CA.
In side scan sonar images, mines appear as tiny bright objects with telltale dark shadows. The mines are difficult to distinguish from the rest of the image because they blend with background features in the images; for example, the mine's shadow may be obscured by pebbles on the ocean floor. Wavelet analysis cleans up the images by isolating and discarding the small features that may interfere with seeing mines. Wavelets are also good at representing images parsimoniously, with very few bits of information. The technology can reduce some megabyte images to 1,000 bytes, so there is much less data to look through to find the mines. Next, statistical analysis of the data makes very sophisticated judgments that reveal the mines and pinpoints them on the electronic images.
"In early tests, the system missed just three mines out of 100 with well less than one false alarm per image," Dr. Lau said. Reducing the number of false alarms is a tremendous time saver given the current slow methods of deactivating mines. Eventually, advances in robotics technology will make autonomous remote deactivation possible, Dr. Lau predicted. "The new wavelet/statistical technology is very well positioned to become the signal processing 'brains' of these robot crawlers."
Healey and Lau noted that while integrating the system's components should return even better results, 100 percent detection is practically impossible because mines buried in sand are indistinguishable from background noise.
The Office of Naval Research pursues an integrated science and technology program from basic research through manufacturing technologies. Research areas include oceanography; advanced materials; sensors; electronics; weapons; surveillance mine countermeasures; and surface ship, submarine and aircraft technologies. For more information about ONR programs, refer to the ONR home page at http://www.onr.navy.mil on the World Wide Web.