Evaluation of Dynamic Image Pre-Compensation for Computer Users with Severe Refractive Error
Jian Huang, Armando Barreto, Malek Adjouadi · 2012 · Proceedings of the 14th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2012) · doi:10.1145/2384916.2384947
Summary
This paper presents a dynamic image pre-compensation method designed to improve computer accessibility for users with severe refractive errors such as high-degree myopia, hyperopia, and astigmatism. Refractive error affects over 30.5 million Americans aged 40 and older, and while correctable with glasses, contact lenses, or surgery, some individuals cannot fully correct their vision through these means. The core idea is to preprocess images displayed on screen so that when viewed through the user's imperfect optics, they appear clearer. The method works by measuring each user's specific ocular aberrations using a Shack-Hartmann aberrometer, then applying Inverse Wiener Filtering to generate pre-compensated versions of on-screen images. The key innovation over the authors' previous static pre-compensation approach is making the process dynamic: an eye tracker (Tobii T60) monitors the user's pupil size in real-time at 60 Hz, and the aberration model is continuously updated to account for pupil size changes caused by varying illumination conditions. This matters because ocular aberrations are pupil-size-dependent — the Zernike coefficients that describe aberrations change as the pupil dilates or constricts, so a static model based on one-time measurements becomes inaccurate during actual use.
Key findings
An empirical study with 20 participants (40 eyes tested, spherical errors ranging from -3.24D to -10.34D) evaluated the method through an icon recognition task using 8 common computer icons (Copy, Document, Folder, Email, Picture, Printer, Print, Save, Delete) at two sizes (48px and 72px). Recognition accuracy improved significantly with dynamic pre-compensation (DPC): mean correct recognitions increased from 3.98 out of 8 for original icons (49.8% accuracy) to 5.58 for DPC icons (69.8%), a statistically significant improvement (p<0.01). Crucially, DPC also significantly outperformed static pre-compensation (SPC), which actually performed worse than original icons (3.20 correct, 40%), likely because pupil sizes during testing were markedly smaller than the base measurement pupil size, causing the static model to introduce worse distortion than the original. Icon size had a significant effect (p<0.01) but did not interact with pre-compensation method, meaning the dynamic approach helped equally for small and large icons. Participants identified low contrast (55%) and ringing artifacts (25%) as the main remaining limitations of the pre-compensated images.
Relevance
This research addresses an underexplored area of computer accessibility: supporting users whose visual impairments stem from refractive error rather than eye diseases. While most assistive technology for low vision focuses on magnification or contrast enhancement, this approach takes a fundamentally different path by modifying the image itself to counteract the specific optical characteristics of an individual user's eyes. For accessibility practitioners, the work highlights that visual accessibility is not one-size-fits-all — personalization based on individual vision profiles can yield significant improvements. The concept of real-time adaptive display based on physiological monitoring (pupil tracking) foreshadows modern approaches to personalized accessibility. Although the method has practical limitations (requiring aberrometer measurement and eye tracking hardware), it demonstrates the potential for display systems that actively adapt to users' visual capabilities rather than relying solely on user-configured settings like zoom or high contrast modes.
Tags: low vision · refractive error · image enhancement · visual impairment · eye tracking · computer accessibility · image pre-compensation · ocular aberration