Joint time frequency analysis based liveness fingerprint detection

摘要

Fingerprint identification is most widely used and popular form of biometrics. By use of simple and inexpensive techniques it is possible to spoof fingerprint scanners which has inbuilt liveness detection. This paper deals with an image processing technique in order to detect the liveness of an authenticated person for fingerprint biometric system. The presented system is based on profiling and a wavelet based approach. In this work the difference between a live and a spoof fingerprint image has been exploited. Profiling of fingerprint images gives the results for the same. The wavelet based approach deals with the decomposition of the fingerprint image using Daubechies wavelet till four levels and profiling of the approximated image. By setting proper threshold for both the above methods results are obtained. The proposed approaches are tested for 50 live and 50 silicon spoof fingerprint images and Receiver Operating Characteristics (ROC) is plotted and an equal error rate of 0.078 for profiling and 0.147 for wavelet based approach is obtained. Finally both the approaches are cascaded for liveness detection. The success rate obtained is 100%.