Generalized Vander Lugt correlator as an optical pattern classifier and its optimal learning rate

Abstract

Generalized Vander Lugt correlator (GVLC) that employs fractional Fourier transforms is an extention of the conventional Vander Lugt correlator. Like neural networks, the error backpropagation algorithm provides the learning rule by which the filter values are changed iteratively to minimize the given error function. We apply the GVLC to pattern classification and develop the optimal learning rate in order to improve the learning convergence and the classification performance.