| Biometric identification technology is a new field and highly expected to authenticate identity.Now it has been basically completed the initial assumption.As an important branch of biometric research facial feature is widely noted by the advantages of low cost,friendly and easy to be accepted by people.Currently the face feature authentication system needs to put the biometric features of users into database in the form of template,but the study found that this approach is not safe.This makes the development of biometric research has been greatly hindered.Therefore,the protection of biometric template has become an important issue in the direction of biometric authentication system application.This paper proposed a targeted template protection algorithm which fuses the Chaos theory and RSA according to the characteristics of statistical properties of face template.First,this paper adopts the feature extraction algorithm which can highly reduce the dimension of face image to reduce the dimension and get the face template.Next,according the matrix characteristics of face template the proposed algorithm choose the two-dimensional Logistic mapping to produce two chaotic sequences.One is used to generate scrambling matrix,the other is to produce the encryption parameters of RSA.Then,In order to meet the plaintext requirements of RSA,proposed algorithm did the pretreatments of element interception,deformation and integration for feature template.Finally,complete the encryption process of face feature template after the secondary encryption of RSA.The proposed algorithm makes the matching verification and safety analysis under two different feature extraction algorithms of(2D)2LDA and PCA by using ORL database.The classification method is the nearest neighbor classifier,and the criterion is Euclidean distance.Experimental results show that the proposed fusion algorithm not only has good security,but also protected the integrity of the matching rate of the original system. |
PDF Full Text Request