Abstract

A synthetic discriminant function-based nonlinear image correlator is presented. This pattern recognition system is capable of storing a large number of reference patterns in the spatial domain. The synthetic discriminant function is synthesized for a training set of images and is displayed at one side of the input plane. The input scene is displayed at the other side of the input plane. Thus the correlation between the input image and the reference patterns stored by the synthetic discriminant function is produced by nonlinear joint transform optical correlation.

© 1988 Optical Society of America