James Ferwerda, a research associate in the Cornell Program of Computer Graphics, is developing computer simulations of the ways in which people with several kinds of low vision see the world.
Working backward from these computer models, he plans to process images of the real world into forms that low-vision sufferers can more easily comprehend. The work is funded by a three-year, $450,000 grant from the National Science Foundation’s (NSF) Information Technology Research program.
By the end of the project, Ferwerda hopes to create small hand-held devices that would help visually impaired people read and move around.
Ferwerda, whose background is in both experimental psychology and computer science, says he undertook the project because ‘it offers an opportunity to use computer graphics technology to make a real difference in people’s lives.’ He will collaborate with Gordon Legge, the Distinguished McKnight University Professor of Psychology at the University of Minnesota and director of the Minnesota Laboratory for Low-Vision Research, who will test the new techniques with subjects with a variety of visual impairments.
Common low-vision disorders include glaucoma, cataracts, macular degeneration, diabetic retinopathy and retinitis pigmentosa, as well as the overall loss in visual ability that comes with ageing. More than 10 million people in the United States have some form of low vision.
Each type of disorder presents the sufferer with different problems. Glaucoma, for example, results in a loss of peripheral vision, while macular degeneration causes a loss of fine detail in the centre of the visual field. Other disorders, such as cataracts, lead to a general loss of contrast over the entire field. Textbooks often include illustrations prepared by artists to show how the world appears to low-vision sufferers.
‘The trouble is, most of these illustrations are completely wrong,’ Ferwerda says. For example, an illustration of the effect of macular degeneration is usually a picture with a hole in the middle. In fact, Ferwerda says, while the retina might not gather information about the centre of the field, the brain is very good at filling in the blanks, and a person sees a poorly detailed image, rather than one with a hole. Glaucoma patients with ‘tunnel vision’ don’t see the edges of their visual field as dark; they just have trouble orienting themselves in space.
Rather than manipulating images optically, Ferwerda will work from computer models of human visual processing. Experimental psychologists have broken down visual processing into a series of steps, beginning with the absorption of light by rod and cone photoreceptors, moving through preliminary processing in nerve tissue in the retina, proceeding to several steps in the brain. Each step can be represented mathematically and modelled in a computer program. Starting with a model of normal vision, Ferwerda can introduce changes that correspond to various defects in the system.
A reliable computer model of a visual defect, Ferwerda says, should make it possible to process images in a way that compensates for the defect. For example, one way to aid people with macular degeneration might be to shift the central portion of the visual field to an undamaged part of the retina, then modify the contrast of edges in the image to make up for the fact that the off-centre parts of the retina deliver less detail. For glaucoma sufferers with restricted peripheral vision, a very simple outline of the larger visual field might be overlaid on the central image to help the person stay oriented. (This approach, called vision multiplexing, already is under development elsewhere.) Image enhancement techniques also might be used to counter the glare effects produced by cataracts or to compensate for the losses in visual sensitivity that accompany the ageing process.
Eventually these ideas might be built into a lightweight pair of glasses, but current technology is not that advanced, Ferwerda says. Head-mounted devices, looking something like virtual-reality headsets, take over the whole visual field and often are rejected by people with low vision.
Instead, Ferwerda’s idea is to incorporate new technology into a small, hand-held device that could be held up and looked into as needed. Such a device would take advantage of microdisplay technology currently in development that creates very high-resolution images on a very small screen. The same technology can be incorporated into web browsers to give low-vision users better access to graphical content on the Internet, Ferwerda added.
In addition to the NSF, the project is supported by the Cornell Program of Computer Graphics, founded and directed by Donald Greenberg, the Jacob Gould Schurman Professor of Computer Graphics at Cornell.