Joint transform correlation in security applications

The Joint Transform Correlation (JTC) technique is the one of the most frequently applied methods in the field of optical classification and identification systems. Nowadays there are a lot of different modifications that have been tested and verified. Because of extremely high computational throughput the JTC has been implemented in many special purpose (security) applications. This paper deals with an implementation of the JTC for two basic classes of objects-fingerprints and faces. In the first part the sensitivity to rotation and zooming is studied. Consequently in the second part two types of thresholding in the spectral domain are tested in order to improve performance of the system. All procedures are simulated on a computer first and then the real optical FT setup is employed. The results are compared and discussed.