Affiliation:
1. University of Leeds
2. University College London
3. UNC Chapel Hill, Princeton University
Abstract
The enthusiasm for Virtual Reality (VR) brings forward the question of accessibility. Current hardware is not suitable for usage with prescription correction glasses despite the prevalence of refractive eye problems, and algorithmic solutions are not optimal. We tackle this issue by proposing a differentiable visual perception model that derives key insights from the human visual system, the target display, and the target user's eye refractive errors. Using our algorithmic approach, we optimize the rendered images using stochastic gradient‐descent solvers to provide contrast enhanced images at interactive rates for a visually impaired user.