From The Morning Call -- August 2, 2003

Lehigh professor's insight could give military new vision
Computer system may help detect suicide bombers, hidden guns.

By Christina Gostomski
Of The Morning Call
(original report)

The next high-tech tool that could prevent terrorism, and the brains behind it, reside in a nondescript laboratory at Lehigh University.

The equipment looks innocuous ! two cameras connected to a computer.

But thanks to the mathematical equations of university professor Rick Blum, the system can do something even Superman can't ! project pictures to law enforcement or military personnel that show weapons hidden behind walls or concealed on humans under clothing.

Blum's ingenuity could have far-reaching benefits for law enforcement officials and the U.S. military, which is underwriting the project. It could help them identify from a distance potential attackers such as armed suspects or suicide bombers.

''Certainly the application of detecting concealed weapons is one that would be highly useful for our forces in Iraq today,'' said Bill Sander, associate director of computing and information science for the Army Research Office in Durham, N.C. ''To be able to tell whether approaching Iraqi people are carrying concealed weapons or explosives could save the lives of soldiers.''

The key to the system rests in intricate software developed around a series of complicated algorithms created by Blum, whom the military sought for his expertise in image fusion.

The 43-year-old Quakertown man started working on the project eight years ago after learning about cameras using millimeter-wave sensors. The sensors emit high-frequency radio waves that penetrate walls and apparel to detect metal objects such as weapons. The cameras then show images of the weapons that look similar to X-rays.

The problem with the images is that they show humans as essentially black-and-white skeletal frames. Military and law enforcement officials complained that didn't work for their purposes. They wanted to see human features and clothing ! items that could help them quickly identify suspects in a crowd.

A normal digital camera shows those details but doesn't show hidden weapons.

So Blum began searching for a way to marry the two images.

He found the answer in math.

The pictures taken by millimeter-wave cameras and digital cameras consist of pixels represented by different voltage levels. When connected to a computer, the cameras send signals to the computer with the voltage information for each pixel.

The computer converts the voltages into numbers, through a software program Blum designed.

The software then runs Blum's algorithms, which pick out the most important voltage levels from each picture, and put them together to generate a new picture.

In other words, the algorithms take the images of metal objects, which appear in the millimeter-wave pictures, and the detailed images of humans, which appear in digital pictures, and merge them into one photograph.

''It's a lot of number crunching,'' said Blum, who came to Lehigh in 1991. In 2001, he received an Army grant of approximately $300,000 to fund his work and over the years he, with the help of graduate students, fine-tuned the algorithms to run more efficiently and create pictures in full color. Today, his algorithms are considered among the best in image fusion.

''Because of his track record, he demonstrated to us that he had the knowledge and ability to uncover new ideas in this regime and try to find new algorithms that hadn't been developed,'' said Sander, who helped make the decision to award Blum the federal grant and is his project manager.

Sander said Blum's image fusion algorithms are considered among the best because of the way they align points from the different images and incorporate color.

Before the military and law enforcement start using the system, some glitches ! namely cost ! must be worked out.

The millimeter-wave cameras, based on millimeter-wave radar developed by the military, are extraordinarily expensive and make the system too costly.

''It's going to take somebody to spend some money to make a millimeter-wave sensor inexpensive and then have it mass produced to bring the price down,'' Blum said.

The system would also need to be lightweight and portable for police and soldiers on the move. Such a system will require shorter, faster algorithms, something Blum is already working on for a similar system that connects night vision and digital images.

Sander said it would take at least a year to create a prototype unit for the millimeter-wave and digital fusion system and longer to begin production and deployment. The Army has no current plans to begin such a costly production, he said, adding that the ideal situation would be for a commercial industry to develop and mass produce the unit.

Lehigh University officials have not decided whether to patent the project, but Blum is writing what is expected to be the first published book on the topic.

Copyright © 2003, The Morning Call