Nonminutiae-Based Decision-Level Fusion for Fingerprint Verification

Abstract

Most of the proposed methods used for fingerprint verification are based on local visible features called minutiae. However, due to problems for extracting minutiae from low-quality fingerprint images, other discriminatory information has been considered. In this paper, the idea of decision-level fusion of orientation, texture, and spectral features of fingerprint image is proposed. At first, a value is assigned to the similarity of block orientation field of two-fingerprint images. This is also performed for texture and spectral features. Each one of the proposed similarity measure does not need core-point existence and detection. Rotation and translation of two fingerprint images are also taken into account in each method and all points of fingerprint image are employed in feature extraction. Then, the similarity of each feature is normalized and used for decision-level fusion of fingerprint information. The experimental results on FVC2000 database demonstrate the effectiveness of the proposed fusion method and its significant accuracy.