Stanford Out to Shrink the Camera

Eat your heart out, Dick Tracy.

Researchers at Stanford University are developing revolutionary new technologies that could lead to digital cameras so small and efficient they could fit on a watch band. That’s one up on the famous detective’s wristwatch-telephone.

Preliminary results have been so promising that Stanford danced around the primary hurdle confronting most cutting-edge research: funding. Five major companies--Canon Inc., Eastman Kodak Co., Hewlett-Packard Co., Intel Corp. and Interval Research Corp.--are putting up $4 million for a three-year research project.

“These are the only companies we talked to, and they all joined,” said Abbas El Gamal, associate professor of electrical engineering at Stanford and the project’s principal investigator. “It’s not like we went out to 50 companies and got four or five to join. It was one of those interesting funding situations where we didn’t go around shopping.”

Jim Duley, strategic analysis manager for Hewlett-Packard, one of the corporate sponsors, sees the Stanford work as “potentially revolutionary.”

At the heart of the project is a radical approach to digital imaging that some had thought was impossible.

The current generation of digital cameras, ranging from space probes to camcorders, use charged coupled devices--or CCDs--to collect light and send it on to a series of processors that convert the light energy to digital signals, which can then be processed and compressed into images.

“Lots and lots of processes happen,” El Gamal said, “so you end up with many chips, which means high cost and high power, which is not good for cameras and portables.”

For several years, a number of companies have been developing standard computer chips that could ultimately replace CCDs. These chips capture the signal and digitize it on the same chip.

About five years ago, El Gamal and a graduate student, Boyd Fowler, took a hard look at that research.

“We said, ‘Wait a minute. If you are going to integrate processing and capture on the same chip, you should really rethink the whole thing,’ ” El Gamal said.

“Instead of having a block on a chip that does the sensing, and the output from that block goes to another block on the chip that does analog conversion [to digital], and then it goes to another block for signal processing . . . instead of that, can you integrate these things more deeply.”

They wanted to know, for example, if it might be possible to digitize the signal on the pixel level. Pixels are tiny parts of an image--packets of light, like grains of silver in photographic emulsions--that make up the picture.

If individual pixels could be digitized at the time of their capture, El Gamal reasoned, a major step in the process would be eliminated, along with a lot of the background noise that degrades the electronic image.

“At first it seemed like a crazy idea because ADC (analog-to-digital conversion) chips are very complex and require a large number of transistors,” El Gamal said.

But the researchers persisted, and Stanford now has four patents covering pixel-level digitizing.

Five chips have been produced in the Stanford lab, and El Gamal said they prove that the concept has tremendous advantages over CCD technology.

It is possible, for example, to program the camera on a pixel-by-pixel basis. That greatly increases its dynamic range, or its ability to capture features in bright sunlight and dark shade in the same image.

With film, by contrast, the entire roll is the same “speed,” so every photo on the roll has the same light sensitivity. The Stanford camera will be able to measure the light from every area of the photograph, and adjust sensitivity to the specific conditions, thus avoiding washing out brightly lighted areas and losing detail in the shadows.

El Gamal is a little coy about the results so far because he is waiting to publish in a professional journal. But he insists the “dynamic range is going to be pretty incredible.”

Currently, however, the lab does not have the kind of equipment it needs to reach the next plateau. The funding from corporate sponsors will be used for better optics, for example, which should enhance the resolution of the images considerably.

Since fewer chips and processors are needed, the technology should lead to improved consumer products that are smaller and require less energy to operate. And they could be ubiquitous.

“Digital cameras are going to be enormous markets, enormous,” El Gamal said. He sees them eventually supplanting film in most cases, although not necessarily at the high end where film will probably hold an advantage for years to come.

In time, he said, every vehicle will be equipped with seven or eight digital cameras that will monitor everything from road conditions to nearby traffic.

El Gamal’s team includes specialists from a wide range of disciplines. Stanford professor Joseph Goodman, an expert on optics, is looking into ways of using the chips’ programmability to compensate for flaws in the optical system, similar to removing atmospheric distortion from astronomical images.

Psychologist Brian Wandell is studying possible applications for the new technology. He sees school kids with cameras on their lapels, capturing images that will record their daily adventures. He also sees all sorts of surveillance and personal identification applications, like at ATM machines.

So, Mr. Tracy, when you finish eating your heart out, look over your shoulder. Someone may be watching you.

*

Lee Dye can be reached via e-mail at leedye@compuserve.com.