指纹和虹膜特征融合的可撤销模板保护方法  

Cancelable biometric template protection method based on feature fusion of fingerprint and iris features

作　　者：张雪锋[1] 常振会 张俊杰 王超飞 ZHANG Xuefeng;CHANG Zhenhui;ZHANG Junjie;WANG Chaofei(School of Cyberspace Security,Xi’an University of Posts and Telecommunications,Xi’an 710121,China;School of Communications and Information Engineering,Xi’an University of Posts and Telecommunications,Xi’an 710121,China)

机构地区：[1]西安邮电大学网络空间安全学院,陕西西安710121 [2]西安邮电大学通信与信息工程学院,陕西西安710121

出　　处：《西安邮电大学学报》2023年第4期51-60,共10页Journal of Xi’an University of Posts and Telecommunications

基　　金：陕西省自然科学基础研究计划项目(2021JQ-722)。

摘　　要：针对单模态生物特征模板保护识别率低和安全性差等问题,提出一种基于指纹和虹膜特征融合的可撤销模板保护方法。首先,对指纹和虹膜图像进行预处理和特征提取,采用t分布随机近邻嵌入(t-distributed Stochastic Neighbor Embedding,t-SNE)方法对提取的特征数据进行降维和串联融合。其次,对融合后的特征向量进行随机索引置乱、小波变换和离散傅里叶变换(Discrete Fourier Transform,DFT)运算,以进一步提取生物信息,增强模板保护方法的安全性。最后,通过部分Hadamard变换生成可撤销生物特征模板。理论分析和实验结果表明,该方法生成的模板满足不可逆性和可撤销性,具有较高的识别率和安全性。In view of the problems that the uni-biometric template protection has poor recognition accuracy and lower security,a biometric template protection method based on feature fusion of fingerprint and iris features is proposed.Concatenation fusion and t-distributed stochastic neighbor embedding(t-SNE)were performed on the features that extracted from the preprocessed fingerprint and iris image.Subsequently,random index scrambling,wavelet transform and discrete Fourier transform(DFT)were adopted on the fused feature vectors to further extract biological information and enhance the security of the proposed template protection method.Finally,the revocable biometric template was created through the partial Hadamard transform.Theoretical analysis and experimental simulation results show that the template generated by this method meets the irreversibility and revocability,and has better recognition rate and security.

关 键 词：特征融合 可撤销模板 t-SNE 小波变换 HADAMARD变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]