Optical spatial frequency correlation system for biometric authentication

An optical information processing system for biometric authentication using the spatial-frequency correlation of input and registered biometric identifiers is proposed. The merit of our system is to speed up the authentication because it composed of all optical elements. In this study, we adopted fingerprints as an example of biometric identifiers and analyzed the basic properties of our system by regarding fingerprints as the 1D finite rectangular wave with a period of 0.5mm and the whole width of 15mm. Concretely, the behavior of the second maximum value of the intensity distribution of the spatial-frequency correlation function of input and registered fingerprint images was numerically analyzed by considering the ending points in the input fingerprint. As a result, the threshold level for the fingerprint recognition in our system was determined. In addition, the effects of the random noise added to the input fingerprint on the second maximum value were statistically analyzed. Furthermore, the behavior of the second maximum value was numerically analyzed to investigate the effects of various kinds of shifts of the input fingerprint on the recognition. As a result, the false reject rate (FRR) was the highest for the transverse shift in comparison with the longitudinal and rotational shifts.