A computer chip with the power of a supercomputer, but small enough to fit in your eye, may someday enable the blind to see and the sighted to see better.
This possibility is foreseen by researchers at the University of California at Berkeley and their colleagues at the Hungarian Institute of Sciences in Budapest. They have developed an artificial visual system known as the Cellular Neural Network Universal Machine, which can mimic all the functions of the human eye, from motion detection to color processing. It is also more sensitive to the nuances of motion and light than the human eye, giving it visual capabilities found in many animal species, such as cats or bats.
The new device, which is designed specifically to analyze and process visual images, can do in a fraction of a second what a refrigerator-sized image-processing computer requires hours or days to do.
"While our retina takes a fifth of a second to process all the information from one image, the [planned] chip will do it in a few millionths of a second," says neurobiologist Frank Werblin of Berkeley.
Based on a neural network design, the system employs an analog-type system that gives it greater flexibility than the "on-and-off" structure of digital systems. …