Abstract: Fingerprint characteristic of every human are unique and stable, therefore widely accepted for personal identification. Accurate and reliable fingerprint recognition is a challenging task, which heavily depends on the quality of the fingerprint images. It is well-known that the fingerprint recognition systems are very sensitive to noise and other image transformations such as rotation. In this paper an enhanced algorithm is introduced to make the matching process rotational invariant. The proposed algorithm is robust to match two identical fingerprint images which are spatially aligned at different rotational angles and gives better FMR and FNMR ratio for images rotated at different angles. A Biometric Fingerprint authentication algorithm using firefly algorithm (FA) based feature selection approach which we named firefly optimization algorithm (FOA) is proposed to address the above issues.The search in (FOA) is iteratively guided by a fitness function defined to maximize class separation to identify new features instead of the traditional Minutiae (Termination and Bifurcation features). The first contribution is the formulation of a new feature selection algorithm for fingerprint recognition based on the DWT algorithm, which solves the localization problem by applying (FOA) separately to four sub-bands to increase the recognition rate and to speed up feature selection. The second contribution is an invariant moment matching algorithm which is proposed as a matching algorithm to address some misclassified features and to increase the matching accuracy. The proposed algorithm was found to generate excellent recognition results which admit 100% accuracy when applied on the FCV2002, 2004 and 2006 dataset. Also, it admits accuracy ranged 96.35-100% when applied to fingerprints with a rotation of 0-360.

Keywords: Authentication system, Swarm Intelligence, PSO, Invariant moment matching algorithm.