抄録
A study was carried to examine the feasibility of an artificial retina using the charge simulation method algorithm to recognize images. An artificial retina machine vision system was developed for this purpose. The retina consisted of photo sensors which detected image features for solid pyramids and wedges that were located in the retina's vision field. The charge simulation method was utilized in compressing the image features. Neural networks that were trained with the compressed image features were then used and resulted in overall shape classification rates ranging from 78.8 to 95% when different object sizes were considered. Overall size classification rates of 90% were obtained for different shapes. The results show that it is feasible for the artificial retina using the charge simulation method to recognize different 3-D objects.
