Rotationally invariant pattern recognition in a photorefractive joint transform correlator using circular harmonic filters and the wavelet transform

摘要

A photorefractive joint transform correlator using circular harmonic filters and the wavelet transform for high discriminant and rotationally invariant pattern recognition is reported. In the experimental implementation, a photorefractive crystal, BSO, is used as the dynamic holographic medium and a bank of wavelet functions, derived from a mother Mexican-hat wavelet have been selected to achieve a band-pass operation in the Fourier domain. The zero-order and second-order circular harmonic components are used in the investigation. It is shown that, with the proper dilation factors, there is a trade-off between peak correlation intensity and signal-to-noise ratio (SNR) with the wavelet transform filters. With the second-order circular harmonic, an improvement of 63% in the SNR is achieved with only a 4% loss in the peak correlation intensity.