Ateb-Gabor Filtering Simulation for Biometric Protection Systems

Personal authentication by fingerprint recognition depends on the correct identification of characteristic points of biometric images. This paper presents a scheme for identifying characteristic points. However, poor fingerprint input quality is generally observed due to unstructured patterns, unclear spine structures, and various background noises that have resulted in poor fingerprint recognition. Therefore, improving the input image is a crucial step for more accurate recognition. This paper proposes a new method of image filtering by filtering by non-periodic Ateb-functions. The functions of hyperbolic sine, cosine, tangent, cotangent are considered. The method of calculation of nonperiodic Ateb-functions is shown. To identify the characteristic points, a set of bifurcation patterns was constructed, oriented along with different directions. The proposed method is implemented and tested on fingerprints The reliability results were tested based on NIST Special Database 302. A data set for estimating the parameters that verify fingerprints obtained from 162 samples of different quality. Experimental results show the effectiveness and accuracy of the method.