A method for fingerprint image identification based on Gabor filter and power spectrum

Abstract

This work presents a fingerprint identification scheme using a coefficient map based on Gabor filter and power spectrum. This process uses Poincaré index in order to locate fingerprint core and delta points and the images are aligned through these singular points before matching. From the core point, this technique divides an input image into quadrants, where information about texture orientation on three Gabor-filter directions is extracted. In the frequency domain, the analysis is based on the power spectrum, aiming towards reducing computational time. Results illustrate the method efficiency and accuracy.