This is a proposal to test the feasibility of a """"""""Smart Telescope"""""""" for use to improve the ability of visually impaired persons in tasks such as travel, navigation and social interactions. Advances in low-power high-speed portable computers combined with novel computer vision algorithms enable us to build an affordable, portable, and cosmetically acceptable digital telescope that can enable visually impaired persons to perform these tasks with greater ease than with current telescopes. The computer vision algorithms first detect regions of interest in an image where targets are likely to be, even if these targets occupy only a small portion of the visual field, obviating the need for a user to scan or search a scene as would be necessary with an ordinary telescope. Next, novel object-specific super-resolution enhancement algorithms use target-specific knowledge to magnify and enhance these regions so that users can interpret them, similar to pointing a telescope at those regions. Algorithms can then track the targets as the observer moves, and indicate their relative locations. Finally, like today's digital telescopes for the low vision community, the Smart Telescope will output either to a monocular viewfinder display or to a bioptic display. The project process involves: (1) Data collection and analysis (Iow-vision persons will be used to collect image data); (2) Detection, enhancement and tracking algorithm development; (3) Integration of algorithms with user interface, and (4) Testing human factors. Our field prototypes will range in cost from approximately $3,000 to $5,000 each, while the target cost of a commercial version is under $1,000. Our price estimates are conservative, and we anticipate that the rapid development of computer technology will lower these costs substantially in the next few years.
