Objective fingerprint image quality assessment using gabor spectrum approach

Fingerprint image quality assessment is crucial for many fingerprint applications. It affects the performance and interoperability of fingerprint identification, authentication, and built on based crypto systems. In this paper we present a novel approach for objective fingerprint image quality measure derived from power-spectra of two dimensional Gabor features representation. We also classify existing assessment algorithms into two classes: structural and intelligent representation approaches. Approaches were implemented for accuracy and reliability tests using TIMA Database including good and faulty fingerprint images and compared with Gabor spectrum approach. The goal of our quality assessment approach is to automatically assess the quality of fingerprint image in agreement with correlation of human visual system judgment. The proposed approach shows improvement performance in the quality classification of finger print images than existing classified approaches. Experimental verification demonstrates very good correlation of Gabor spectrum (GS) with subjective quality assessment. 3% improvement of power spectrum in false rejection rate, and 14% improvement of Gabor features in true acceptance rate were achieved.