Computer Vision Applications for the Blind and Visually Impaired:
Abstracts of Talks
Dr. Paul J. Beckmann, Research Associate, Minnesota Laboratory for Low Vision Research, Univ. Minn., "Behavioral Evaluation of the Digital Sign System (DSS)"
http://www.senderogroup.com/wayfinding/about.htm#uofm
Localization within a building is a difficult task for those with impaired vision. Outdoor solutions based on GPS technology are gaining a foothold but GPS signals do not penetrate buildings. While some solutions such as Talking Lights and Talking Signs have been introduced, these require expensive installation and maintenance and generally provide a fixed message to the wayfarer. We have evaluated a different approach to localization within a building that we call the Digital Sign System (DSS). A critical component of this system is a small hand-held camera ("Magic Flashlight") and associated computer-vision software that allows the user to detect, localize, and read (with synthetic speech) inexpensive credit-card sized signs within a hallway environment in real time. In addition to information directly associated with a sign, a user can also request information about building attributes in the immediate and extended vicinity of the sign from a database system called the Building Navigator.
Three groups of subjects participated in the study: those who were visually-impaired but used vision to avoid walls while walking in buildings ("low vision"), those who did not use vision while walking in buildings ("blind"), and normally sighted but blindfolded subjects. In the first study, we placed signs one at a time on walls within a simple environment. We compared the time for the subjects to find a sign using their well-practiced common mobility skills with and without the DSS Magic Flashlight. In the second study we measured how well subjects navigated from one location to another in a building layout, comparing performance using typical tactile signage and the Digital Sign System. In the first study, the mean times for visually-impaired subjects to localize a sign with the Magic Flashlight compared with direct physical localization were 25% shorter even though the subjects had no previous training with the device.
In collaboration with: Gordon E. Legge, Advait Ogale (University of Minnesota); Bosco S. Tjan (University of Southern California); and Kevin Kramer (Advanced Medical Electronics Corporation)
Prof. Serge Belongie, UCSD, "Project GroZi: Assistive Navigational Technology for the Visually Impaired"
http://www-cse.ucsd.edu/~sjb/
The contemporary urban environment is brimming with rich visual cues that provide valuable directional and informational content to sighted individuals. The goal of the GroZi project at the UCSD division of Calit2 is to make significant advances toward making these visual cues universally accessible in a problem domain that is constrained but at the same time challenging and representative of a larger class of related real-world problems. In particular, our anticipated project outcome is to develop computer vision technology combined with a haptic/auditory feedback interface to help blind and visually impaired individuals execute the task of grocery shopping in a safe and independent manner. By enabling users with a device that can exploit existing visual cues, we avoid the overhead of requiring expensive and time consuming retagging of the urban environment. In this talk I will present our progress and outline the challenges that lie ahead on the problems of detection and recognition of groceries and aisle signs.
Charles LaPierre, CTO, Sendero Group, "Current and future accessible wayfinding for the blind: From GPS systems to indoor navigation"
http://www.senderogroup.com/
The two fundamental challenges in compensating for lack of vision are reading and getting from one place to another. Reading of books has been mitigated by OCR (optical character recognition), and screen readers have provided access to computers and the internet. Getting around is a different story, with two separate components, obstacle detection and accessing location information.
I have been focusing on accessible location information with GPS since 1993 and more recently working on indoor navigation. Four accessible GPS systems are now on the market; however, indoor options are primarily still in the research stage.
The Miniguide, Ultra Cane, Laser Cane and Kay-Sonar are the latest in obstacle detection devices. The blind community maintains that the dog and cane provide adequate guidance and that there is no need for obstacle detection unless it enables access to driving or cycling. The future holds the possibility of seamless indoor/outdoor navigation and integration with obstacle detection on a Segway or perhaps even in a semi-autonomous vehicle.
Prof. Roberto Manduchi, UCSC, "Sensors and Sensibility: Is Computer Vision Appropriate for Assistive Technology?"
http://www.soe.ucsc.edu/~manduchi/
Computer vision is the modality that most frequently comes to mind when talking about Assistive Technology (AT) for the visually impaired. If a camera "sees" what the eye can see, then a blind person should be able to access visual information, suitably digested by a computer program. Factual evidence, however, points to the contrary. Only a handful of AT systems based on computer vision have been marketed successfully, and the functions they perform are highly specialized (OCR, color tellers, bank note readers).
In this talk I will ponder the pros and cons of computer vision as a sensing modality for AT. The discussion will focus on issues such as reliability, hardware needs, and user interface. Case studies (both failures and successes) will be presented. Based on this critique, I will attempt a functional characterization of computer vision as a sensor for AT, and will highlight a number of application areas with realistic opportunities for this technology.
Prof. Eli Peli, Harvard Medical School, "Electro-Optical Vision Multiplexing Devices for Vision Impairments"
http://www.eri.harvard.edu/faculty/peli/index.html
A novel concept of vision multiplexing aims to combine the benefits of a wide field-of-view with high resolution to enable people with impaired vision regain the lost function without losing the surviving one. In a number of electro-optical devices we implemented spatial multiplexing by superimposing contour images over the natural view, or over images presented on a TV. For patients with peripheral field loss (tunnel vision) we present minified contours in an optical see-through head-mounted display, while for acuity loss registered contours provide edge enhancement. We have also implemented the concept in a video player with dynamic magnification for patients with central vision loss. Using the device patients with tunnel vision can search for targets more directly and find them more rapidly. In a walking simulator we found that using only the minified contours subjects can detect impeding collision almost as well as with natural vision. Due to inattentional blindness users may detect objects or event of interest occurring out side their field at about 50% of the cases. The use of contours did not change this rate. We have shown that the effect is due to allocation of attention and not due to interference from the superimposed images.
Last updated Aug. 2007.