Abstract

In the areas of signal, image, and information processing, more and more emphasis is being placed on parallel processing techniques for solving todays problems in real-time. Optics with its inherent parallelism has succesfully demonstrated its power in performing Fourier transforms, convolutions, and correlations. More recently, interest in performing matrix algebra optically is evident based on the proliferation of articles appearing in print in this field. However, if optical processing is to ever have a real impact on general matrix computation, then it will be necessary to generalise present, concepts to handle floating-point arithmetic as well.

© 1985 Optical Society of America