基于重加密的随机映射指纹模板保护方案

摘要

In random projection (RP) based biometric template protection methods, the generated template and key are vulnerable to attacks, which may cause the leakage of users' privacy. To solve this problem, an improved RP-based fin-gerprint template protection method was proposed. First, based on the RP result, the proposed method divided the projec-tion domain into fingerprint matching domain and noise adding domain that were mutually independent, then fused them with two sub-matrices of the random projection matrix and saved the result as template. In addition, re-encryption mech-anism was introduced to realize secure storage and transmission of the key (RP matrix). Experimental results show that the proposed method can achieve stronger ability to resist different attacks than existing RP-based biometric template protection methods, and also guarantee high matching accuracy and revocation.%针对基于随机映射(RP, random projection)的生物特征模板保护算法在模板生成和密钥管理中面临易被攻击而泄露用户隐私的问题,提出一种改进RP算法的指纹模板保护方案.首先,在随机映射的基础上,将变换域划分为相互独立的指纹特征匹配域和加噪干扰域;在子域内加噪后利用子随机映射矩阵交叉融合生成模板.同时,引入重加密机制实现对变换密钥(RP矩阵)的安全存储和传输.实验结果和分析表明,与现有的RP模板保护方法相比,所提方案具有更高的抵抗攻击能力并能保持RP算法的匹配性能和模板可撤销性.