Fake Finger Detection Based on Thin-Plate Spline Distortion Model

摘要

This paper introduces a novel method based on the elasticity analysis of the finger skin to discriminate fake fingers from real ones. We match the fingerprints before and after special distortion and gained their corresponding minutiae pairs as landmarks. The thin-plate spline (TPS) model is used to globally describe the finger distortion. For an input finger, we compute the bending energy vector by the TPS model and calculate the similarity of the bending energy vector to the bending energy fuzzy feature set. The similarity score is in the range [0, 1], indicating how much the current finger is similar to the real finger. The method realizes fake finger detection based on the normal steps of fingerprint processing without special hardware, so it is easily implemented and efficient. The experimental results on a database of real and fake fingers show that the performance of the method is available.