Region growth–based feature extraction method for finger-vein recognition

Abstract

Finger-vein recognition uses the finger-vein pattern extracted from the captured vein images to identify individuals. However, the captured vein images contain irregular shading and noise. To extract effective vein pattern from these unclear images, this paper proposes a novel vein pattern extraction approach by running the region growing operator on the different seeds. As the proposed method emphasizes continuity and symmetry of valleys in the cross-sectional profile, it can extract the robust finger-vein patterns and be against irregular shading and noise. Experimental results show that the proposed method achieves robust vein pattern extraction and provides a lower equal error rate in finger-vein recognition.