Optical correlator-neural networks hybrid system for automatic angle measurement of two-dimensional objects

Abstract

A new interpolation algorithm is proposed and demonstrated to perform automatic angle measurement of two-dimensional (2D) objects. The proposed algorithm works in conjunction with optical correlator neural network hybrid architecture (OCNN). The OCNN is trained with a combined algorithm of direct binary search and error back propagation. Input of the OCNN is presented with an image whose angle of rotation is to be measured, and output from the OCNN is fed into the proposed interpolation algorithm, which finally produces the rotation angle of the input image. Results of both computer simulation and experimental set-up are presented for an English alphabetic character as a 2D object. The experimental set-up consists of a real optical correlator using two spatial light modulators for both input and frequency plane representations and a PC based model of a single layer neural network. We obtained very low experimental mean absolute error of 3.18 deg with standard deviation of 2.9 deg.