Fingerprint Identification Based on Frequency Texture Analysis

Nowadays the AFIS (automatic fingerprint identification system) plays more and more important roles in various applications such as access control, ATM card verification and criminal identification. The most widely used fingerprint identification algorithm today is based on spatial minutia detection and matching, which needs a time consuming revision during the matching phase. The proposed frequency-based algorithm uses Fourier transform to code an enhanced fingerprint image into a series of float numbers namely feature sector, where the minutia and the ridge information are fused together in the fingerprint enhancing phase. The matching is simply to calculate the absolute distance between the two feature sectors. Because the use of FFT and the concentric circle sampling method, our algorithm achieves the geometric distortion invariant ability and a high matching speed. The experiment results prove the feasibility of our algorithm, and our future research work to improve the performance is presented finally