Fingerprint Liveness Detection Using Local Ridge Frequencies and Multiresolution Texture Analysis Techniques

摘要: It has been demonstrated that simple and inexpensive techniques are sufficient to spoof fingerprint scanners. Previously, effective use of physiological phenomenon perspiration is shown as a counter-measure against such attacks. These require more than one image for performing the liveness check hence may not be suited on-line processing. In this work, measure based on single developed. The inherent texture density differences between `live' `not live' images exploited. Multiresolution analysis used minimize energy associated with phase orientation maps. Cross ridge frequency performed by means statistical measures weighted mean calculated. different features along reliability or center given inputs fuzzy c-means classifier. proposed algorithm was applied dataset approximately 58 live, 50 28 cadaver images, from three types An error rate 1.4% achieved. provides faster technique doing test which relies only image.