Improvement to human-face recognition in a volume holographic correlator by use of speckle modulation.

We show that a speckle-modulation technique can improve the parallelism and the recognition accuracy of volume holographic correlators. The object patterns are modulated by a speckle pattern generated by a diffuser. These modulated patterns are stored as Fourier holograms by use of angular-fractal multiplexing. With the speckle modulation the sidelobes are completely suppressed, the cross talk is negligible, and the correlation peak becomes a bright sharp spot. Thus higher recognition accuracy is achieved. The angular separation between adjacent patterns in the multiplexing could be much smaller, resulting in larger capacity and higher parallelism of the correlator. Also, this technique can be combined with other methods such as wavelet filtering to achieve a large invariant tolerance range. Theoretical analysis, numerical evaluation, and experimental results are presented to confirm that sidelobes and cross talk are sharply suppressed by the speckle modulation.