Ionic Vision System Mimics the Human Eye for Software-Free Image Processing — and Optical Illusions

Researchers from Japan's National Institute for Materials Science (NIMS) have created an artificial vision system inspired by the operation of human vision — and potentially leading to compact, energy-efficient image sensing and processing system for analog signals.

The NIMS team set about finding a solution to one of the biggest problems of artificial intelligence systems attempting to replicate the functionality of the human sensory system: They're bulky, complex, and use a lot of power. The solution, for at least one domain: An "ionic artificial vision device," which simulates the way human retinal neurons work.

The device is constructed from mixed conductor channels placed on a solid electrolyte at regular intervals, whereby input voltages cause ions within the electrolyte the migrate across the channel. The result: A system which can process input image signals and provide an output image with enhanced edge contrast between lighter and darker areas, entirely independent of any controlling computer or software.

Interestingly, the mimicry of the human vision system may go a step further: The team proposes that the ionic vision system could be used to reproduce optical illusions to do with tilt angle, size, color, and movement, and could potentially be built into a visual sensing system capable of performing all the functions of the human retina.

"This scheme," the researchers note in the paper's abstract, "enables computer vision tasks with simple and effective operations, owing to the intrinsic properties of the materials employed."

The team's work has been published under closed-access terms in the journal Nano Letters.