The aim of the project was to produce a computer simulation of visual impairments. The project included:
The symptoms of the impairments glaucoma, retinitis pigmentosa, macular degeneration, diabetic retinopathy, hemianopia and cataracts were successfully simulated. Myopia, hyperopia and colour blindness were also simulated, but testing these simulations against the requirements highlighted some areas that could be improved. The use of a computer with a visual impairment was simulated by applying the impairments to the display of a simple drawing program and a web browser. The simulations of glaucoma, macular degeneration and cataracts were evaluated using these programs and were shown to have a significant effect on computer use.
The results of the cooperative evaluation described in Section 10 only showed the effect of the simulated impairments and not those of 'real' visual impairments. It is difficult to make a direct comparison between the real and simulated impairments as is not possible to interview a visually impaired person and ask them if their impairment matches the one simulated. The possible exception is cataracts as people are often able to have an operation to remove them. They could be asked post-operation, how the simulation of cataracts compares to their vision before the operation . Alternatively, visually impaired people could be asked to complete the evaluation tasks described in Section 10.2 with their 'normal vision' in order to make a comparison with the simulated impairments. The problem with this approach is comparing the relative sizes of the field of vision and the severity of the impairments.
The simulated impairments were confined to the computer screen. With a real visual impairment, the keyboard, mouse and any other input devices may also be in the impaired field of vision. Visually impaired computer users may have specially adapted input and output devices. Any evaluation of real visual impairments would need to take these into account when designing the evaluation tasks.
The severity of the simulated impairments ranged from 'very low' to 'very high'. Increasing the severity of the simulated impairment increased the area of the screen that was obscured. The severity of visual impairments is often measured using a Snellen Test which involves reading a chart "imprinted with lines of black letters graduating in size from smallest on the bottom to largest on top"  from a measured distance which is usually 20 feet.
It would be useful to estimate a Snellen value for the different severities for each of the simulated impairments and use this when comparing the real and simulated impairments.
With normal vision it is natural for the line of sight to follow the mouse cursor when clicking hyperlinks and buttons in a web browser, or when drawing shapes in a paint program. However, when performing tasks such reading and writing text, the line of sight often has no correlation to the position of the cursor. The 'track to cursor' mode in the simulation accurately reproduced the behaviour for the first set of tasks by redrawing the display whenever the mouse is moved and positioning the centre of vision at the coordinates of the mouse cursor. However, for the reading and writing tasks, the 'track to cursor' mode added a "hidden change"  by requiring the user to position the centre of vision to the required location.
In addition people with macular degeneration or cataracts often develop techniques where they can use their peripheral vision when performing tasks as this part of the visual field is generally less severely unaffected . When completing the reading and writing tasks as part of the evaluation of the macular degeneration simulation, users were observed to move the black patch to one corner of the screen so that they could use the unobscured region of the screen to complete the tasks. This behaviour mimics the techniques used by people with macular degeneration except that users were then able to aim their line of sight to look at the peripheral vision area of the simulation.
Evaluating the simulation against the specified requirements (Section 9) highlighted areas for further development:
The simulations of myopia and hyperopia could be improved by implementing a more efficient blur operation as this is currently a bottleneck in the performance of the simulation.
The impairments of glaucoma, retinitis pigmentosa, macular degeneration, diabetic retinopathy and the cloudy vision variation of cataracts were created by drawing circles to obscure parts of the display. This gave the tunnel in the simulation of glaucoma and retinitis pigmentosa and the black patches in the simulations of macular degeneration and diabetic retinopathy a very regular shape. The simulation of these impairments could be improved by using more complex polygons to give them an irregular appearance
Colour blindness was simulated by producing a greyscale image for all severities. The simulation could be improved to more accurately model the red and green colour deficiencies of protanopia and deuteranopia as described in Section 2.13.
In order to more accurately model the real impairments, the simulation would need to have some way of tracking the user's line of sight and adjusting the centre of vision accordingly. This may be possible with a camera mounted on top of the screen and some additional software.
The simulation of all impairments could be improved by implementing a three-dimensional model using OpenGL. This model could be used to simulate the effects of different focal lengths and would a useful aid to understanding the symptoms of visual impairments.
The system could be improved to simulate the use of further software applications with a visual impairment. Ideally the system should allow the user to interact with the operating system and all applications in the normal manner with the simulated impairments applied to the displayed output.
The cooperative evaluation described in Section 10.2 was carried out using the simulations of glaucoma, macular degeneration and cataracts applied at a very high severity where a large area of the display was obscured. As such, it is not surprising that the results showed the simulated impairments had a big effect on the use of a computer. To gain a better idea of the effects of each of these impairments, the evaluation would need to be repeated at a reduced severity. The evaluation could also be repeated for the simulations of myopia/hyperopia, diabetic retinopathy, hemianopia and colour blindness which were not part of this evaluation, or extended to explore some of the areas highlighted as being difficult with a particular impairment such as reading and writing text with glaucoma.
© Stephen Ratcliffe. 2005
Department of Computer Science
University of York