The Argus II system uses a spectacle-mounted camera to feed visual information to electrodes in the eye. "What we are trying to do is take real-time images from a camera and convert them into tiny electrical pulses that would jump-start the otherwise blind eye and allow patients to see," said Professor Mark Humayun, from the University of Southern California.

Retinal implants are able to partially restore the vision of people with particular form of blindness caused by diseases such as Macular Degeneration or Retinitis Pigmentosa. Both diseases cause the retinal cells which process light at the back of the eye to gradually die. The bionic eye works by converting images from a tiny camera mounted on a pair of glasses into a grid of 16 electrical signals that transmit directly to the nerve endings in the retina.

The camera in the glasses sends a signal to a pocket device, the size of a Blackberry, which processes the images in real time into a grid of electrical signals. This is then beamed to the eye implant, which sends it directly to nerve endings in the retina.

In the current devices, the receiver for the signal is implanted under the skin behind the ear with a wire connection to the eye implant. The team has shrunk the device to the extent that the improved version fits under the skin around the eye.

The implants will be most successful in patients who were once fully sighted rather than people who were without sight from birth.

U.S. researchers have been given the go-ahead to implant the prototype device in patients. The implant could be available to patients within two years.

Professor Humayun predicts that future versions of the bionic eye would need at least 1,000 pixels for patients to recognise faces. Future work includes studying the effects the implants have on the brain.