Fingerprint recognition systems are vulnerable to spoof attacks, which consist in presenting forged fingerprints to the sensor. Typical anti-spoofing mechanism is fingerprint liveness detection. Existing liveness detection methods are still not robust to spoofing materials, datasets and sensor variations. In particular, the performance of a liveness detection algorithm remarkably drops upon encountering spoof fabrication materials that were not used during the training stage. Likewise, a quintessential liveness detection method needs to be adapted and retrained to new spoofing materials, datasets and each sensor used for acquiring the fingerprints. In this paper, we propose a framework that first performs correlation mapping between live and spoof fingerprints and then uses a discriminative-generative classification scheme for spoof detection. Partial Least Squares (PLS) is utilized to learn the correlations. While, support vector machine (SVM) is combined with three generative classifiers, namely Gaussian Mixture Model, Gaussian Copula, and Quadratic Discriminant Analysis, for final classification. Experiments on the publicly available LivDet2011 and LivDet2013 datasets, show that the proposed method outperforms the existing methods alongside cross-spoof material and cross-sensor techniques
