A Robust Contactless Fingerprint Enhancement Algorithm

Abstract

Compared to contact fingerprint images, contactless fingerprint images have three particular characteristics: (1) contactless fingerprint images have less noise than contact fingerprint images; (2) there are less discontinuities of ridges in contactless fingerprint images; and (3) the ridge-valley pattern of contactless fingerprint is much more unclear than that of contact fingerprint images. These properties increase a great difficulty to the contactless fingerprint enhancement. In this paper, we propose a robust contactless fingerprint enhancement algorithm based on simple sinusoidal-shaped filter kernel to fully take advantage of the properties of contactless fingerprint. First, an effective preprocessing is proposed to preliminarily strengthen the ridge-valley contrast of contactless fingerprint images. Then, simple sinusoidal-shaped filter kernel is proposed to enhance the contactless fingerprint images. Finally, we propose a score-filtering procedure to effectively recover the ridge-valley pattern. Comprehensive experiments were performed to evaluate the proposed method from aspects of image quality, minutiae extraction and fingerprint verification. Experimental results demonstrate the high performance of the proposed algorithm in contactless fingerprint enhancement.