Imaging Chip Works Quickly

Jennifer L. Morey

A group of Stanford University researchers has collaborated with several industry leaders to develop an imaging sensor that produces images faster, easier and cheaper than its predecessors. By moving the analog-to-digital conversion function onto the imaging chip itself, the team stumbled onto a technology that will further bolster the $1.5 billion digital imaging market.
The Stanford chip is made with complementary metal oxide semiconductor (CMOS) technology, which allows engineers to combine the sensor with computer circuitry, thus reducing chip count and cutting production costs. Because the pixels are read in parallel, the chip is also much faster than charge-coupled device arrays, which read out pixels sequentially. Potential advantages include high dynamic range, noise
reduction and pixel-level programmability, which could be used to improve the chip's performance in different environments such as outdoor scenes or text displayed on computer screens.

Commercial interest
Abbas El Gamal, associate professor of electrical engineering, said that his team has tested the second version of its third-generation CMOS chip and the results are promising. The 1.2-million-transistor chip is producing good 8-bit images, he said, but the team has not finished testing characteristics such as dynamic range.
El Gamal is working with other Stanford researchers through the university's Image Systems Engineering program. Electrical engineering professor Joseph Goodman and several of his students are investigating a method of using the chip's programmability to compensate for imperfections caused by optical lenses.
Industry representatives began contacting El Gamal about two years ago with inquiries about the research and then with offers of funding for the $4 million project. Companies such as Hewlett-Packard Co., Rockwell International Corp. and Analog Devices Inc. have subsidized the research in exchange for the use of the image sensor test set.
Other investing companies include Canon Inc., Eastman Kodak Co., Intel Corp. and Interval Research Corp. These partners have assisted in the design and prototyping stages, have fabricated chips for test purposes and likely will be involved with marketing commercial products as soon as next summer.
Once the chip is completed, applications could include printing, digital photography, silicon and video markets. El Gamal also sees great potential in facial recognition systems for automated teller machines.