Optical Security System with Fourier Plane encoding.

We propose a new technique for security verification of personal documents and other forms of personal identifications such as ID cards, passports, or credit cards. In this technique a primary pattern that might be a phase-encoded image is convolved by a random code. The information is phase encoded on the personal document. Therefore the information cannot be reproduced by an intensity detector such as a CCD camera. An optical processor based on the nonlinear joint transform correlator is used to perform the verification and the validation of documents with this technique. By verification of the biometrics information and the random code simultaneously, the proposed optical system determines whether a card is authentic or is being used by an authorized person. We tested the performance of the optical system for security and validation in the presence of input noise and in the presence of distortion of the information on the card. The performance of the proposed method is evaluated by use of a number of metrics. Statistical analysis of the system is performed to investigate the noise tolerance and the discrimination against false inputs for security verification.

Real-time binary-amplitude phase-only filters.

A real-time binary-amplitude phase-only filter (BAPOF) implemented in available phase-only modulators is presented. The BAPOF has an amplitude transmission equal to one only in a region of support, while the transmission is equal to zero in the complementary region. To implement zero transmission in a phase-only modulator we propose to add a linear phase to the region of support. In this way the correlation desired is obtained off axis. Computer simulations and experimental results obtained with this technique are given.

Phase-only filter with improved discrimination.

The optimization of a phase-only filter (POF) in terms of discrimination capability is presented. The notion of a phase-difference histogram and its modification are proposed for selecting the support function of the POF. Some numerical results obtained with the conventional POF and the optimized POF are given. The discrimination capability is increased significantly.

Nonlinearity effects in the pure phase correlation method in multiobject scenes.

The nonlinearity used in the pure phase correlation method can improve pattern recognition, but it causes high-order harmonics in the correlation plane in the case of multiobject scenes. High-order harmonics and the presence of aliasing may cause false alarms. We show that these effects are strongly diminished when the periodicity in the scene is broken (by different objects, random distribution, etc.).