Evaluation of Onscreen Precompensation Algorithms for Computer Users with Visual Aberrations
Miguel Alonso, Armando Barreto, Julie A. Jacko, Malek Adjouadi · 2007 · Proceedings of the 9th International ACM SIGACCESS Conference on Computers and Accessibility (Assets '07) · doi:10.1145/1296843.1296881
Summary
This paper from Florida International University and Georgia Institute of Technology presents the results of human subject testing of precompensation algorithms designed to improve onscreen image quality for computer users with visual aberrations. The approach works analogously to an audio equaliser: just as an equaliser pre-distorts audio to compensate for acoustic distortions in a listening room, these algorithms pre-modify display images in a way that is opposite to the distortion the user's eye will introduce, so that the image arriving at the retina is closer to the intended original. The process requires measuring each user's specific optical aberrations using a Wavefront Analyser (aberrometer), which produces a mathematical representation called the Point Spread Function (PSF). The inverse of the PSF is then used to generate precompensated images that "cancel out" the user's visual aberration. Twenty subjects participated in the study, including five controls without significant refractive errors, five with myopia (at least -3 Diopters), five with both myopia and astigmatism (at least -3 Diopters sphere and -0.5 Diopters astigmatism), and five with Keratoconus in at least one eye.
Key findings
A factorial experiment with four factors — Group (2 levels), Eye (2 levels), Size (3 levels: 15, 24, or 38mm icons), and Method (2 levels: uncompensated vs. precompensated) — was conducted using a randomised complete block design. ANOVA results showed that both Size and Method had significant main effects (p < 0.002 and p < 0.001 respectively), confirming that the precompensation algorithm significantly improved icon identification accuracy for users with visual aberrations. Several significant interactions were also found at the 5% level: Group*Size (p=0.025), Group*Method (p=0.000), Size*Eye (p=0.025), and Size*Method (p=0.004). These interactions reveal that the precompensation method works better for certain icon sizes and certain groups of users — particularly important for optimising the approach for different types of visual aberrations. The results from human subjects reinforced earlier findings from software simulations and artificial eye testing, validating that the laboratory results translate to real-world benefit.
Relevance
This research represents a fundamentally different approach to visual accessibility than the standard strategies of magnification, contrast enhancement, or simplified layouts. Rather than modifying the interface to work around a user's vision limitations, precompensation modifies the displayed image to work with the specific optical characteristics of each user's eyes, potentially restoring near-normal visual experience. For practitioners, this has implications for personalised accessibility: as eye measurement technology becomes more accessible, per-user display customisation based on optical profiles could become a viable accessibility feature. The approach is particularly relevant for conditions like keratoconus where standard corrective lenses may not fully compensate for the irregular corneal shape. The significant interaction effects found in the study highlight that visual accessibility solutions need to account for the specific type and severity of a user's visual aberration rather than applying one-size-fits-all adjustments.
Tags: low vision · visual aberration · precompensation · image processing · deconvolution · keratoconus · myopia · astigmatism · GUI accessibility · icon recognition