Mercredi 15 février 2006
Université de Montréal
sous la direction de Jacques Gresset et Pierre Rondeau
Graham Strong, OD, M.Sc.
Centre for Sight Enhancement, University of Waterloo
The first practical and user-friendly closed circuit TV (CCTV) system for people with low vision was unveiled in 1968 by Sam Genensky and Paul Baran at the American Academy of Optometry meeting in California (Genensky, 2001). The earliest CCTV consisted of a close focusing black and white video camera linked and dedicated display monitor equipped with a rudimentary contrast control feature. Magnification was achieved using zoom lens optics. The first commercial CCTV systems were introduced in the early 70's. Vision rehabilitation workers quickly recognized the significant utility of these CCTV systems, but they were invariably perceived as powerful magnifiers that were primarily intended for people with severe vision loss who wanted to read and write. Most low vision clinicians were daunted by their high cost and physical size, invariably viewing them as options of last resort to be utilized when conventional optical magnifiers failed to provide adequate levels of magnification and field of view. Over the ensuing two decades, the essential functionality of CCTV systems changed very little. Most of the changes were attributable to unrelated improvements in the camera and display components ("technology push" innovations).
The post-CCTV evolution of video-based low vision devices began with the introduction of the Low Vision Enhancement System (LVES) in 1994 (Massof, 1992; 1994; 1995; Rohrschneider, 1997). This certainly qualified as a true "paradigm shift," wherein one conceptual world view was replaced by another. Unfortunately, the conceptual novelties of the LVES were poorly understood by many clinicians and service providers whose understanding of video technology was firmly rooted in their CCTV experience. In the end, the LVES system failed because most clinicians really didn't know how to demonstrate it or even who to demonstrate it to. One of the most significant legacies of LVES is that it revealed the potential utility of video-based virtuality as an accessible seeing environment for people with low vision. They no longer were viewing the real world directly. Instead, they were seeing it displayed on a personal television system. This new virtual environment is interactive, adaptive, and reconfigurable. It can be modified to create supernormal viewing conditions. But it is not like the virtual environment in a video game. This environment is more real than virtual, so I prefer to describe it as "video-mediated reality". The user is immersed in a televised view of the world (displays) created from live streaming video (input) from the real world. This creates a true and accurate representation with visibility characteristics that can be selectively enhanced by various electronic manipulations to become much more visible than the real world it replaces.
In the aftermath of LVES, a procession of derivative devices appeared in the marketplace. Many of these subsequent offerings attempted to update and improve the original device, without fully understanding what it was that made it effective. With the historical success of CCTV devices and the consumer and media excitement about the LVES system, it was inevitable that developers would seek to introduce other form factors for video technology. Guided by their limited conception of video devices as electric magnifiers and prompted by the falling cost of the video components, developers hastened to create video device equivalents for each differentiated optical device category, including head and face borne telescopic devices, hand held magnifiers, stand magnifiers, and hand-held monocular telescopes. This expanded device armamentarium poses a significant problem for rehabilitation clinicians who must resolve how to integrate these newer video products into their established assessment protocols.
The developers of these new hand and stand video products clearly sought to replicate the established functionality of conventional CCTV systems, but within smaller and more portable embodiments. The deficiencies of these video devices for some low vision applications accrue from their incorporation of sensor, display, and lighting technologies that were originally developed for other purposes. In spite of these performance deficits, these video devices have been well-received by many low visioned users because they offer several unique advantages over optical magnifiers for some applications.
Video devices use a camera block to capture in-focus images of unseen objects of potential interest to the user. These images are then processed in a way that renders them visible to the user, whereupon the processed images are displayed to the user in this visibility-enhanced format. Accordingly, video-based low vision devices are significantly more multidimensional than their optical counterparts. They defy categorization based on conventional optical metrics such as magnification and range of focus. Consider some of the intrinsic features of this technology platform. Magnification is one of the most obvious and relevant features of video devices for low vision. It can be derived from the camera block in the form of an optical zoom, or it can be derived from the processing block in the form of digital zoom. The variable zoom capability allows users to adjust the level of magnification to suit the unique resolution requirements of many diverse seeing tasks. The optical and electronic features of video devices facilitate much higher levels of magnification than are commonly available with optical low vision aids. Image stabilization, improved focus control, and enhanced localization options (using the zoomable interface) are significant factors in this. One of the most significant differences between video devices and optical devices is the option of contrast enhancement (Peli, Goldstein, Young et al, 1991). Various contrast stretch algorithms can be invoked to generate emboldened virtual images that are much more visible than the actual objects being viewed. Video devices also have the novel ability to independently adjust magnification and contrast features. This introduces a new level of sophistication into low vision operations. The obvious relevance is that spatial frequency and contrast are two of the most significant dimensions of diminished visual performance. Many visualization tasks require people to see objects that are too small and too poorly contrasted for their diminished visual capabilities. Hence it seems logical that magnification and contrast enhancement pose the most effective enhancement options to redress their low vision deficits. Implicit in this conviction is the notion that incremental benefits may accrue when these two image enhancements can be manipulated in tandem (as with video devices). The availability of a contrast enhancement option may lessen the level of magnification required for many sub-optimal seeing tasks. If a device simply enlarges the perceived size of an unseen object until it can be seen, the penalty is a corresponding reduction in the size of the field of view. If a device increases the perceived contrast of an unseen object until it becomes visible, a large field of view is maintained but there is insufficient reserve resolving capacity for sustained viewing tasks. The best solution if often a combination of both options, thereby producing the comfort and facility of a magnification solution in tandem with the field retention benefits of a contrast enhancement solution.
Over the past decade, we've learned a lot about video technology through our clinical experiences with people who were using the LVES and its derivatives, as well as our own VideoTelscope device. A significant feature of video technology is its versatility. This trait is evidenced by the disparate vision rehabilitation applications described for these devices by others who work extensively in this area (Maino, Lalle, Stelmack et al, 1997; Ballinger, Lalle, Maino et al, 1996; Harper, Culham, Dickinson, 1999). We routinely interview our patients concerning their experiences using their new video devices. We're left with the impression that video devices are somewhat akin to a Swiss Army knife, since they incorporate so many different features that can be brought to bear on so many different seeing tasks under so many different seeing conditions. We also find that most video device users continue to find new applications for their devices long after obtaining them. This contrasts with the experience reported by patients receiving optical devices; who often find that the number of device applications becomes less after obtaining their optical low vision aids. The reported versatility of these devices seems to derive from features that aren't explored very well using standard clinical assessment protocols (Fraser, Cole, Fay et al, 1997; American Academy of Ophthalmology, 2001). This realization caused us to take a closer look at some of the unique and potentially relevant properties of video-based low vision devices.
Individuals with low vision often report being disproportionately affected by adverse seeing environments, such as poor lighting, glare, and contralight viewing conditions. Contemporary video cameras often incorporate unique sensitivity features that were designed to provide improved visibility under several common adverse viewing conditions. Anyone who's taken video pictures in a dim room will see that the video seems to provide new brightness to the material being filmed. Video processing is effective in a number of other viewing environments such as when the natural lighting is poor, variable, or uneven. Similarly, video images can be adjusted to overcome contralight viewing problems similar to those encountered when trying to recognize someone who is standing directly in front of a window. Each of these inherent properties adds to the utility and versatility of video-based low vision devices. Freeze frame is another interesting feature that is a commonplace in video systems. For low vision applications, it allows users to temporarily retain visual information so it can be applied to another one (such as looking up a phone number and then bringing the information to a telephone). It also allows users to share observations of interest with others, such as a teacher or rehab instructor.
Not all low vision tasks and not all low vision patients derive benefit from magnification. In some instances, minification or field compression is the most relevant strategy, especially for some orientation tasks or context visualizations. Minification provides an instant overview or orientation view that makes spotting more efficient and effective. Anyone who's ever visited my house or lab will understand what it is to live and work in a cluttered world. Many real life seeing environments are too complex and confusing to be viewed under constant magnification conditions. The capacity to periodically obtain an overview or reference view provides a very useful coping feature. When magnification and minification are combined in this way with an interactive controller, this can be descried as a zoomable user interface (ZUI). There are two spatial dimensions to many seeing tasks. One important dimension concerns the specific details of each observation and the other concerns the general context in which the details are situated. People with low vision often must attend from one dimension to the other in order to enjoy the same insights as their fully sighted peers. A zoomable user interface enhances the individual's capacity to cope with complex seeing environments by allowing her to zoom in to see details or zoom out to put them in context. This capacity greatly enhances several important visual functions including orientation, visual search, localization, viewing and inspection. This is the same strategy used by people when using a road map. A localization or orientation perspective helps to quickly find an area of interest, while focused concentration (corresponding to a highly magnified view) is used for detailed inspection.
We've conducted some preliminary research to investigate how people use zoom magnification in various seeing environments. In these studies, we used an autofocus digital Canon GL-2 camcorder with an optical zoom magnification system to observe how long it took people with low vision to locate and inspect specific objects within cluttered seeing environments. The task was performed initially using zoom magnification as required. The task was repeated using fixed magnification at the level that was required to make the necessary visualization. This order was used consistently because we expected zoom magnification performance to be better than fixed magnification performance and we wanted the inevitable learning effect to diminish this difference in order to ensure that any time lapse differences were meaningful. One task required participants to read the time on a microwave clock in a cluttered kitchen (See Figure 1). Most subjects required significant levels of magnification to reading the clock.
When attempting this task with a fixed magnification device, the user has available at all times, sufficient magnification to read the display. However, this enabling magnification made it very difficult to actually find the display to be read. With a zoomable user interface, any amount of magnification can be invoked as required. The first order of business is to find the microwave. The next is to find the clock display, and the final requirement is to read the numbers on the clock. All of these task elements require different levels of magnification. If the user gets lost at any time, she can simply zoom out to regain a reference view and then rejoin the hunt. Another experimental task required individuals to locate and observe objects that were more concealed within the visual texture of their environment; a birdhouse located on a tree in a forest. The context view contains few directional clues. In this instance, the fixed magnification search is almost futile. The user begins with a methodical scanning pattern, but this pattern is abandoned once the user gets lost. The final resort is to go from tree to tree starting at the bottom and working her way up to the top. A number of subjects were unsuccessful even after 90 seconds. Most subjects seemed to do much better with a zoomable user interface. They managed to localize probable areas for inspection and then used the zoomable interface to methodically scan these areas with just enough magnification to detect any birdhouse-like objects. The reserve zoom is activated to confirm or reject these initial observations. The zoomable user interface significantly enhances the user's performance on these localization and inspection tasks. These data coincide with the anecdotal reports of users trying to describe how they use their video devices for many distance seeing tasks. Interestingly, we also had people attempt a simplified version of the game called "Where's Waldo?" The object of the game is to discover the location of a Waldo character hidden within each cluttered scene. Perhaps not surprisingly, the zoomable interface provided little tactical advantage for this task. People took about the same amount of time with the zoom feature as without it. In fact, many people quickly abandoned the zoom feature right away so they could systematically scan the scene under high magnification until they found Waldo. Since he was invariably located in the centre of the scene, it made little difference where they started or how they organized this search.
Video-based low vision technology provides an important platform for contemporary assistive devices for people with low vision. The growing range of video devices establishes a viable alternative for most optical sight enhancement systems. The versatility and utility of video-based devices derive from many unique properties beyond their capacity to provide high levels of aberration-free magnification. Significant advantages can be derived from their inherent capacity to allow users to retain satisfactory seeing abilities even under many frequently encountered adverse viewing conditions such as dim illumination, variable lighting, glare, and contralight visualizations. Zoom magnification further enhances the utility of video-based devices for a variety of everyday seeing problems. The concurrent aggregation of separately tunable enhancement options makes it impossible to assess these devices without using task-relevant demonstrations. Optimum utilization occurs when users learn how to tune and adjust the device to compensate for their unique multidimensional impairment attributes relative to the unique multidimensional demands of their desired seeing tasks. The increased availability of new video-based low vision devices poses a significant challenge for contemporary vision rehabilitation service providers, who must now resolve their integration into well-established low vision assessment protocols.
A number of research challenges have been identified to improve the quality and utility of video-based low vision devices. Some of the current limitations of these devices (aside from cost) could be resolved with higher resolution camera sensors, improved image enhancement algorithms, and larger display fields of view. Other opportunities have been identified for marrying video-based devices with other assistive technologies such as face recognition to assist people with multiple disability challenges.
I would like to thank the Ontario Rehabilitation Technology Consortium (ORTC), the Health Technology Exchange (HTX.ca), Canadian Institutes of Health Research (CIHR), the Natural Sciences and Engineering Research Council of Canada (NSERC), and Sight Enhancement Systems for research support. The presenter has a financial interest in Sight Enhancement Systems Inc.
American Academy of Ophthalmology. Vision Rehabilitation Committee. (2001). Vision rehabilitation for adults. San Francisco, CA : American Academy of Ophthalmology. (Preferred Practice Pattern Series).
Ballinger, R., Lalle, P., & Maino, J. (1996). Veterans affairs multicenter low vision enhancement system (LVES) study : Clinical results. Optometry and Vision Science, 73(12S), 210.
Fraser, K.E., Cole, R.G., Faye, E.E., Freeman, P.B., Goodrich, G.L., Stelmack, J.A., et al. (1997). Optometric clinical practice guideline : Care of the patient with low vision. St.Louis, MO : American Optometric Association. (Optometric clinical practice guideline; no. 14).
Genensky, S.M. (2001, March). One man's story : Samuel M. Genensky. The Braille Forum, 39(9). Accessed on-line at www.acb.org/magazine/2001/bf032001.html#bf10
Harper, R., Culham, L., & Dickinson, C. (1999). Head mounted video magnification devices for low vision rehabilitation : A comparison with existing technology. The British journal of ophthalmology, 83, 495-500
Maino, J.H., Lalle, P., & Stelmack, J. (1997). Use of the manual focus Beta 1 LVES by veterans. Optometry and Vision Science, 74(12s), 129.
Massof, R.W., & Rickman, D.L. (1992). Obstacles encountered in the development of the low vision enhancement system. Optometry and Vision Science, 69, 32-41.
Massof, R.W., Rickman, D.L., & Lalle, P.A. (1994). Low vision enhancement system. Johns Hopkins Applied Physics Laboratory Technical Digest, 15, 120-125.
Massof, R.W. (1995). New technology for low vision enhancement. Proceedings of the North American Neuro-Ophthalmological Society Meeting, 151-153.
Peli, E., Goldstein, R.B., Young, G.M., Trempe, C.L., & Buzney, S.M. (1991). Image enhancement for the visually impaired : Simulations and experimental results. Investigative ophthalmology & visual science, 32, 2337-2350.
Rohrschneider, K., Bruder, I., Aust, R., & Blankenagel, A. (1997). [Use of a new optoelectronic vision aid for highly visually handicapped patients] (in German). Klinische Monatsblätter für Augenheilkunde, 210, 105-110.