Fixed-length asymmetric binary hashing for fingerprint verification through GMM-SVM based representations
作者：
Highlights：
• This study is a step towards a complete secure fingerprint authentication system in which fixed-length binary minutiae templates generated by our framework can be encrypted and matched via homomorphic encryption methods in an encrypted domain.
• A method for converting unordered and varying number of minutiae into a fixed-length binary string is proposed.
• The framework is based on GMM-SVM fingerprint verification scheme. Novel features are proposed for GMM-SVM since it is not clear what would be the best features to use for fingerprint.
• GMM-SVM feature vectors are binarized using asymmetric locality sensitive hashing.
• Higher recognition accuracies are obtained compared to the state-of-the-art methods.
摘要
•This study is a step towards a complete secure fingerprint authentication system in which fixed-length binary minutiae templates generated by our framework can be encrypted and matched via homomorphic encryption methods in an encrypted domain.•A method for converting unordered and varying number of minutiae into a fixed-length binary string is proposed.•The framework is based on GMM-SVM fingerprint verification scheme. Novel features are proposed for GMM-SVM since it is not clear what would be the best features to use for fingerprint.•GMM-SVM feature vectors are binarized using asymmetric locality sensitive hashing.•Higher recognition accuracies are obtained compared to the state-of-the-art methods.
论文关键词：Fingerprint verification,Biometric template protection,Minutiae protection,Fixed-length minutiae representation
论文评审过程：Received 17 April 2018, Revised 17 October 2018, Accepted 27 November 2018, Available online 28 November 2018, Version of Record 6 December 2018.
论文官网地址：https://doi.org/10.1016/j.patcog.2018.11.029
