Pittsburgh, PA
January 29, 2023
    News           Sports           Lifestyle           Classifieds           About Us
Health & Science
Place an Ad
Running Calendar
Travel Getaways
Headlines by E-mail
Home >  Health & Science >  Science Printer-friendly versionE-mail this story
Carver Mead's career is based on practical problem-solving

A better digital camera color sensor is Caltech professor's latest venture

Monday, March 18, 2002

By Byron Spice, Science Editor, Post-Gazette

As a boy growing up in rural California, Carver Mead knew he wanted a career in science and mathematics. But it would have to be anchored in the physical world and certainly not in the realm of theoreticians spinning ideas about fundamental forces and extra dimensions.

Carver Mead holds a color chart in the Foveon photo gallery in Santa Clara, Calif. (Paul Sakuma/Associated Press)

"I'm not satisfied with just making up stories," said Mead, an electrical engineer and a professor emeritus at the California Institute of Technology. His work, he explained, "needs to have a real tangible form."

Few people have had a more tangible impact on the Information Age. Mead, 67, may not have invented the computer chip, but he pioneered Very Large Scale Integration, or VLSI, design, which allowed millions of transistors to be packed onto a single silicon chip. That breakthrough made Moore's law -- the notion that computer power would double every 18 months -- a reality and in turn helped make the computer a ubiquitous presence.

He also created a high frequency transistor, called the HEMT, that is less well known, but nevertheless is a crucial amplifying device for microwave communications systems; consumers rely on them every day when they make phone calls or dial up the Internet. One of his start-up companies, Synaptics, developed the touchpad that replaces the mouse on some laptop computers; another, Sonic Innovations, makes digital hearing aids.

  Awards day

Carver Mead will receive the Dickson Prize in Science during a brief award ceremony at 4:30 p.m. tomorrow in McConomy Auditorium, inside CMU's University Center. He will then deliver a lecture, "The Coming Revolution in Photography," that is free and open to the public.


"In many different ways, he's used the fundamental properties of materials to create new computer hardware," said Dana Scott, a professor of computer science, philosophy and mathematical logic at Carnegie Mellon University. "He's a real hands-on guy."

Even his most abstract work -- a 2000 book, Collective Electrodynamics, which argues that the counterintuitive world of quantum physics can be explained in terms of electromagnetic waves -- has applications in measuring time and voltages to high precision, Mead said.

That's why he has been called "the most important practical scientist of the late 20th century," why the $500,000 Lemelson-MIT Prize for Invention and Innovation was presented to Mead in 1999 and why Scott nominated Mead for Carnegie Mellon's $47,000 Dickson Prize in Science, an award Mead will accept on campus tomorrow.

His penchant for doing the impossible was evident again last month, when his start-up company, Foveon, announced that it has developed a new sensor that could revolutionize the way digital cameras record color images.

"It's done for electronic photography what Kodachrome did for film photography," Mead said. The images produced by the sensor, dubbed the X3, equal or surpass those possible with photographic film, he said. "There's no longer a reason to use film just to get high quality," he added.

That doesn't mean that film photography is in immediate danger; the advent of computer word processing, for instance, didn't eliminate mechanical typewriters. But, then again, it's hard to find a typewriter these days.

The X3 sensor still must clear some competitive hurdles, said Alexis Gerard, publisher of the Future Image Report, a newsletter that monitors the digital photography industry. But the chip delivers what techies call "fully measured color" and promises to be better -- and ultimately cheaper -- than the sensors now used in digital cameras.

"This is something that imaging engineers have been trying to accomplish for years," Gerard said. The Foveon researchers "are the only ones who persevered."

Mead was born in Bakersfield, Calif., but grew up in the Sierra Nevada mountains, where the local grade school had only 20 students and one or two teachers. A trigonometry textbook, given to him by one of his teachers, and his own tinkering with war-surplus electronics gear pointed him toward a career in math or science.

He enrolled in 1952 at Caltech, where he earned his doctorate in electrical engineering.

Mead, who lives in Woodside, Calif., said his career has followed a 13-year cycle -- he becomes intrigued by, then immersed in, a new research project and then, burned out, shifts the focus of his research.

"I'm not at all sure I could describe how that happens," he said of choosing new research projects. "Eventually they sort of find me, I think."

The Foveon work, which began 4 1/2 years ago, was an outgrowth of Mead's interest in "neuromorphic" circuitry -- electronic systems that mimic the brain and nervous system -- and of his design for an electronic retina.

As Mead became acquainted with the sensors used for digital cameras, "it became clear to me that the way people perform digital image capture is just brain-dead," he said.

The problem facing sensor designers is the same one that faced the photographers who made the first color photographic films -- How do you register all three primary colors at every point in a photographic image?

A digital camera sensor divides each image into millions of tiny "pixels" arranged in a grid pattern. A photodetector records the amount of light striking each pixel and this information is then used to assemble an image. It works well for black-and-white photography, but the inability to record more than one color at each pixel has been a continuing bugaboo.

The cheapest and most popular method of solving the color problem -- and the one that drove Mead to distraction -- has been to devise a "mosaic" sensor. Each pixel detects just one of the three primary colors -- blue, red or green. The pixels are arranged in a mosaic pattern, with every other pixel detecting green light and the pixels in between detecting either blue or red.

Not only does this make the sensor less sensitive to light -- two-thirds of the light is reflected away -- but it forces the camera to guess about two of the three primary colors at each pixel. So these digital cameras must do considerable computing to fill in the blanks and produce an image.

This and other schemes, such as using a prism to direct the different colors onto three separate silicon sensors, seemed awkward and inefficient to Mead. Instead, he wondered, why not take advantage of a fundamental property of silicon -- that light of different colors penetrates silicon to different depths? The solution could be as simple as creating separate photodetectors at different depths in the silicon chip.

"The idea has been around for a while," Mead acknowledged. "It's just that nobody was able to make it work."

In fact, the concept was quite similar to that of Kodachrome. That film is a sandwich, with different layers treated to respond to different primary colors.

Making the color digital sensor work required innovations in materials science, in semiconductor processing and in computer software, said Mead, who emphasized that he primarily served to identify the best solutions. The chip itself was invented by Foveon co-founder Dick Merrill, while chief scientist Richard Lyon worked out the method for converting the sensor's signals into images.

Foveon says the images produced by the sensor are sharper and have greater color clarity than those produced by mosaic sensors; the sensor also uses less power and promises to be cheaper to produce. The Santa Clara, Calif., company says the sensor is the first with a variable pixel size -- small for high-resolution photography, large for use in video cameras that need to capture images faster and operate in low-light conditions.

One company, Sigma Corp., already has announced it will begin selling a $3,000 camera using the X3 chip. Lower cost cameras reportedly are in the pipeline.

It's a magnificent technical achievement, Future Image's Gerard said, though its commercial success is not certain. Other camera makers will not be in a rush to replace their own sensors and most digital camera users have not been complaining about poor color.

Users might settle for the images now available on digital cameras, Mead acknowledged, just as television watchers settled for scan lines on their TV screens. But he said he's been encouraged by the response thus far to the Foveon announcement, both from camera manufacturers and from the photographic public.

"It's just overwhelming," Mead said. He remains confident that the sensor will eventually dominate the market. "It's really just a matter of time."

Search | Contact Us |  Site Map | Terms of Use |  Privacy Policy |  Advertise | Help |  Corrections