期刊名称：International Journal of Computers and Communications
印刷版ISSN：2074-1294
出版年度：2019
卷号：13
页码：79-87
出版社：University Press
摘要：Iris Recognition methodology is anauthentication mechanism that combines the various methodology. It is used to increase the accuracy of the iris detection method. Iris recognition has inherent weaknesses that can potentially compromise the security of a system. Parodying attacks is one of them and enhanced iris recognition is more vulnerable to parody attack than normal iris recognition method. Parodying is giving duplicate input to the biometric sensor. Parody detection is used to check whether the given input is original or duplicate. The objective is to overcome parodying attacks in iris recognition method. The proposed methodology extracts a set of features from iris using mean, median, variance and local ternary pattern (LTP) techniques respectively and the extracted biometric features are fused and fed to a convolution neural network that employs deep learning to detect parodied features from original features. The proposed method gives better results than the existing liveness detection methods in iris recognition.
