Modeling fixational eye movement for the vision prosthesis

Abstract

Spatiotemporal image pixelization is a technique useful to improve characters recognition to visual impaired subjects through image projection, using a prosthetic vision device. Subjects gifted with the most spread on-market devices, which exploit a camera to acquired images from the surrounding environment and electrically stimulate the visual pathway to elicit vision, don't share one of the characteristic eye behaviours along a visual task: fixational eye movement. Emulating the missed phenomenon using biological inspired models may provide a tool helpful to develop a spatiotemporal image sampling which may improve character recognition, furthermore replacing a physiological feature in the human eye system. In this study a model which mimic fixational eye movement has been developed, jointly investigating physiological features and feasible implementation on a real device, through simulated prosthetic vision.