OPTONEURAL SYSTEM FOR INVARIANT PATTERN-RECOGNITION

An optoneural system is developed for invariant pattern recognition. The system consists of an optical correlator and a neural network. The correlator uses Fourier-Mellin spatial filters (FMF) for feature extraction. The FMF yields an unique output pattern for an input object. The present method works only with one object present in the input scene. The optical features extracted from the output pattern are shift, scale, and rotation invariant and are used as input to the neural network. The neural network is a multilayer feedforward net with back-propagation learning rule. Because of substantial reduction of the dimension of feature vectors provided by optical FMF. the small neural network is simply simulated in a personal computer. Optical experimental results are shown.