Rotation-invariant volume holographic correlator by combining the synthetic discriminant function and wavelet filtering

In this paper, it is proposed to record the matched filter in the photorefractive crystal by using the synthetic discriminant function as the input. The synthetic discriminant function is generated by training a set of rotated object images. Multiple matched filters are stored by angle-multiplexing in the same volume. Due to the large difference among the optical throughouts of the images, a wavelet transform filter is directly introduced into the experimental setup to improve the object discrimination largely. Computer simulation and experimental results demonstrate that it is possible to achieve the rotation-invariant pattern recognition with the proposed method