Pattern recognition using versatile hybrid joint transform correlators: some techniques for improving the performance

摘要

This paper reports the outcome of research studies carried out by us in the area of Optical Pattern Recognition. A hybrid JTC architecture has been used to evaluate correlation performance of four different types of JTCs in a non-cooperative situation. A non-zero order JTC has been proposed based on the principle of differential processing of joint power spectrum and its sensitivity to illumination variation has been investigated. A hybrid wavelet transform based JTC has been proposed and demonstrated to achieve high image discrimination. Hartley transform has been introduced in joint-transform correlation. Chirp modulation in HJTC has been demonstrated to extract information on object correlation and absolute object position from correlation output. A discrimination sensitive rotation-invariant JTC has been proposed based on gradient preprocessing and circular harmonic decomposition.