3/12/02

Chria Faile, 334-844-5964

-- Peaks of Excellence --

AU RESEARCHERS DEVELOP USER-PROGRAMMABLE HANDWRITING RECOGNITION

AUBURN -- Auburn University researchers have made it possible for users of personal data assistants and other handheld computer devices to soon create their own alphabet -- an advance that could be useful to sufferers of Multiple Sclerosis or Parkinson's Disease.

Richard Chapman, an associate professor in the Department of Computer Science and Software Engineering, and William Confer, a doctoral student, developed the technology.

The innovation allows handheld computer users to create their own alphabet, or symbol library, based on the way they already write. Unlike other handwriting recognition programs, the AU technology does not come loaded with a predetermined alphabet, according to Chapman.

"I believe this is one of the major advantages of our program," he says "Rather than forcing users to learn a new written alphabet, they can teach the program to learn the alphabet they already use."

Since it has no predetermined language, the program offers advantages to people who have difficulty using a Roman-like alphabet.

"The program is ideal for people suffering from a motor condition, such as Multiple Sclerosis or Parkinson's Disease," Confer said. "Using our system, they would be able to use handheld devices to communicate with others, an option that is not presently available to them."

Unlike other technologies that use a predetermined alphabet, the AU development has an extremely high rate of character recognition.

"There are other programs on the market that allow you to change letters, but only to a minimal degree," Chapman said. "The fact that you are able to change and create new letters and symbols, is one of the major advantages that sets our program apart.

"What otherwise might be viewed as noise or a scribbled mess by other handwriting recognition programs, is easily recognized by this one. The letters created by the user don't even have to look like Roman characters."

Another advantage of the AU program is its simple algorithmic code. The simplicity of the code allows it to be used on machines with very low memory capability.

"The program could even be used on a pager, as long as you had the correct writing pad," Chapman said.

Confer said, in developing the program, he tried to envision how he could tell the difference between written letters when he had his eyes closed.

"I wanted to be able to write more like I do on paper," he said. "So I asked myself, when I write letters with my eyes shut, how do I know the difference? That's when I began writing the algorithm."

Work actually began when the university designated Information Technology as a Peak of Excellence. The Department of Computer Science and Software Engineering began exploring the development of a handwriting recognition program that was less complex to calculate than others on the market.

Confer wrote the algorithm as a senior class project, with Chapman serving as his advisor.

Other students also are working on the project, and Chapman said it is his hope that they will be able to demonstrate its capabilities to high-tech companies looking to purchase the application.

The researchers have filed a provisional patent application on the technology. They say it will be available to the public as soon as a commercial developer can be secured.

# # #

mar02:AU-handwriting

CONTACT: Chapman, 334/844-6314.