Abstract
Visual distortion, known as metamorphopsia, is a serious visual
deficit with no effective clinical treatment and which cannot be
corrected by traditional optical glasses. In this paper, we introduce
a toolkit of approaches for digitally mapping and correcting visual
distortion, that might eventually be incorporated in a low vision
aid VR headset. We describe three different approaches spanning
data-driven and generative designs and leveraging either uniocular
or binocular cues. We present our proposed demonstrator,
our evaluation roadmap, and challenges for the field. Initial tests
with simulated data demonstrate the effectiveness of the approach.
Once clinically validated, we hope these approaches will enable
accurate mapping of visual distortion and eventually lead to the
development of ‘digital glasses’ capable of correcting the effects of
metamorphopsia and restoring healthy vision.