Berkeley — What if computer screens had glasses instead of the people staring at the monitors? That concept is not too far afield from technology being developed by UC Berkeley computer and vision scientists.
The researchers are developing computer algorithms to compensate for an individual’s visual impairment, and creating vision-correcting displays that enable users to see text and images clearly without wearing eyeglasses or contact lenses. The technology could potentially help hundreds of millions of people who currently need corrective lenses to use their smartphones, tablets and computers. One common problem, for example, is presbyopia, a type of farsightedness in which the ability to focus on nearby objects is gradually diminished as the aging eyes’ lenses lose elasticity.
More importantly, the displays could one day aid people with more complex visual problems, known as high order aberrations, which cannot be corrected by eyeglasses, said Brian Barsky, UC Berkeley professor of computer science and vision science, and affiliate professor of optometry.
“We now live in a world where displays are ubiquitous, and being able to interact with displays is taken for granted,” said Barsky, who is leading this project. “People with higher order aberrations often have irregularities in the shape of the cornea, and this irregular shape makes it very difficult to have a contact lens that will fit. In some cases, this can be a barrier to holding certain jobs because many workers need to look at a screen as part of their work. This research could transform their lives, and I am passionate about that potential.”
Using Computation to Correct Vision
The UC Berkeley researchers teamed up with Gordon Wetzstein and Ramesh Raskar, colleagues at the Massachusetts Institute of Technology, to develop their latest prototype of a vision-correcting display. The setup adds a printed pinhole screen sandwiched between two layers of clear plastic to an iPod display to enhance image sharpness. The tiny pinholes are 75 micrometers each and spaced 390 micrometers apart.
The research team will present this computational light field display on Aug. 12 at the International Conference and Exhibition on Computer Graphics and Interactive Techniques, or SIGGRAPH, in Vancouver, Canada.
“The significance of this project is that, instead of relying on optics to correct your vision, we use computation,” said lead author Fu-Chung Huang, who worked on this project as part of his computer science Ph.D. dissertation at UC Berkeley. “This is a very different class of correction, and it is non-intrusive.”
The algorithm, which was developed at UC Berkeley, works by adjusting the intensity of each direction of light that emanates from a single pixel in an image based upon a user’s specific visual impairment. In a process called deconvolution, the light passes through the pinhole array in such a way that the user will perceive a sharp image.
“Our technique distorts the image such that, when the intended user looks at the screen, the image will appear sharp to that particular viewer,” said Barsky. “But if someone else were to look at the image, it would look bad.”
In the experiment, the researchers displayed images that appeared blurred to a camera, which was set to simulate a person who is farsighted. When using the new prototype display, the blurred images appeared sharp through the camera lens.
This latest approach improves upon earlier versions of vision-correcting displays that resulted in low-contrast images. The new display combines light field display optics with novel algorithms.
Huang, now a software engineer at Microsoft Corp. in Seattle, noted that the research prototype could easily be developed into a thin screen protector, and that continued improvements in eye-tracking technology would make it easier for the displays to adapt to the position of the user’s head position.
“In the future, we also hope to extend this application to multi-way correction on a shared display, so users with different visual problems can view the same screen and see a sharp image,” said Huang.
For more information, visit www.newscenter.berkeley.edu.
Filed Under: M2M (machine to machine)
