Recognition Of Moving Objects Under Simulated Prosthetic Vision Using Motion Detection

Visual prosthesis repairs the visual experience by applying electrical stimulation to different part of the visual pathway (retina, optic nerve or primary visual cortex) and eliciting phosphenes with different sizes, brightnesses and shapes. Phosphene is the basic component unit of artificial vision. However, due to the limited number of electrode arrays and existing techniques, it will be difficult to increase the pixel number of the actual visual experience. Therefore, it becomes the focus of prosthesis designers and physicists that how to optimize the information content of a limited visual display to provide implant users with the best visual perception. This study focused on the image processing strategy, based on simulated prosthetic vision using motion detection.The study consisted of three parts, including the design of the psychophysical experiment platform based on simulated prosthetic vision, the choice and processing of video-materials, and the design of the experiments as well as the results and discussion. The results showed that pixel number had significant impact on the moving objects' recognition. In the four image processing strategy, the BS, BR, and BRFE performed much better than the BPLR strategy. BR gave the best performance with both high recognition accuracy and low distortion. Thus the BR strategy could provide the subjects with the most useful information and the best visual perception.