Cancelable Fingerprint Templates with Delaunay Triangle-Based Local Structures

摘要: Security of biometric templates in a fingerprint authentication system is highly critical because the number fingers person limited and raw fingerprints cannot be reset or replaced. A cancelable template an efficient powerful means to provide protection. However, most templates, many-to-one based non-invertible transformation acts on each single minutia, which may greatly decrease discriminative capability its feature representation, more importantly, minutia sensitive than local structure non-linear distortion present image. In this paper, we propose by utilizing some nice properties Delaunay triangle-based structures, e.g., excellent structural stability, achieve satisfactory performance. The applied rather individual so as mitigate influence retain power data after mapping. Experimental results publicly available databases demonstrate validity proposed scheme.