Abstract

The extremely large storage density of volume holographic memories is applicable to numeric and neural network optical computing. Such photorefractive memories can be used to provide optical storage of binary bit-plane or gray scale images as well as weighted interconnects for optical digital and neural network computing. However, success in applying optical memories to computing will strongly depend not only on the memory's large capacity and short random access time, but also on the memory's dynamic characteristics.

© 1991 Optical Society of America