Wavelet-Markov local descriptor for detecting fake fingerprints

The use of a wavelet-Markov local descriptor, which exploits joint dependencies among wavelet coefficients, for fingerprint liveness detection, is proposed. On the LivDet 2009 datasets, a properly trained support vector machine classifier based on this descriptor guarantees an average error below 3%, as opposed to the 8% average error of the best conventional techniques. © The Institution of Engineering and Technology 2014.