Digital image processing for visual prosthesis: Filtering implications

Investigators around the world are working on retinal neurostimulation as it may restore functional vision to the blind. The image is captured by a camera and after being processed, a series of electrical stimuli are applied to the surviving ganglion cells of the retina. This visual perception is expected to have low resolution. Therefore, there is a need of new algorithms that present the information contained in a visual scene understandable to humans. This study presents a novel multi-resolution algorithm based on wavelet analysis to extract the useful features of an image. Participants in this experiment were able to configure a filter bank to complete a set of everyday tasks. This study shows that wavelet-based algorithms may facilitate improved functional performance in prosthetic vision.