Fingerprint Identification - New Directions

Abstract

In most of the algorithms that have been suggested in this report, the fingerprint image is reduced to a relatively short sequence of integers. This reduces the memory size required by the database. Each algorithm is intended to exploit specific properties and features of the fingerprint that vary from finger to finger, and that can be localized relatively fast using digital techniques, thus also reducing the computational time requirements to a minimum. In each case, the sensitivity of the algorithm to small variations in the image was also discussed, with the aim of reducing the False Rejection Rate, and of increasing the general robustness of the algorithm.