LiveNet: Improving Features Generalization for Face Liveness Detection using Convolution Neural NetworksDate:2021-09-17
Biometrics Find Support from an Unlikely Demographic: Seniors.
Discusses the use of biometrics in financial institutions in the U.S. Factors affecting the decision of U.S. banks to adopt biometric options; Reason for the faster growth of the cell phone revolution in Europe than in the U.S.; Examples of European countries that instituted national identification cards with fingerpints.
It has been demonstrated that fingerprint recognition systems are susceptible to spoofing by presenting a well-duplicated synthetic such as a gummy finger. This paper proposes a novel software-based liveness detection approach using multiple static features. Given a fingerprint image, the static features, including fingerprint coarseness, first-order statistics and intensity-based features, are extracted. Unlike previous methods, the fingerprint coarseness is modeled as multiplicative noise rather than additive noise and is extracted by cepstral analysis. A random forest classifier is employed to select effective features among the extracted features and to differentiate fake from live fingerprints. The proposed method has been evaluated on the standard database provided in the Fingerprint Liveness Detection Competition 2009 (LivDet2009). Compared with other state-of-the-art methods, the proposed method reduces the average classification error rate by more than 20%.
|0||Digital Identity Verification||Default|