基于低质量虹膜与热成像人脸的多模态模板保护算法  

The Multibiometrics Template Protection Algorithm for Lower Quality Iris and Thermal Face Image based on Fuzzy Commitment

作　　者：王宁[1] 李琼[1] 牛夏牧[1] 

机构地区：[1]哈尔滨工业大学计算机科学与技术学院,哈尔滨150080

出　　处：《智能计算机与应用》2013年第3期69-71,74,共4页Intelligent Computer and Applications

基　　金：国家自然科学基金(61100187);哈尔滨工业大学科研创新基金(HIT.NSRIF.2013061)

摘　　要：基于多模态融合的生物身份认证能够较大程度上克服单模态生物特征的存在的问题,基于低质量虹膜和热成像人脸图像特征的融合具有较大的研究价值。多模态生物模板一旦泄露会造成更大的危害,有必要研究针对多模态的模板保护算法。首先将复数热成像人脸特征转换为二进制比特特征,在模糊承诺架构下,利用热成像人脸图像特征产生安全略图和生物密钥,用此生物密钥对低质量虹膜特征进行加密存储,达到安全保护模板的目的。实验表明,该多模态模板保护方法能够解决生物特征类内误差较大,不利于容错的问题,保障了认证效果又具有一定的安全性。分析了在部分生物特征泄露下,交叉匹配带来的安全性,得出在有生物特征泄露的情况时,需要降低容错能力牺牲认证性能来换取安全性的结论。Some problems of unimodal biometrics can be solved by multibiometrics,and it is valuable to implement the multibiometric system based on the lower quality dual iris,thermal face images.However,the threat is higher if leakage of the template of multibiometrcs.This paper converts the real thermal face image feature to binary initially,and then generates the secure sketch and biometric via binary thermal face image feature based on the fuzzy commitment scheme.The biometric key is used to encrypt the lower quality iris binary feature for securing the multibiometric template.The experimental results show that this approach can solve the problem of larger distance of intra-class of biometric feature template.And the system not only keeps higher verification performance,but also has the higher security.Furthermore,the paper also analyzes the security based on cross matching after the leakage of the part biometric thermal face feature.Meanwhile the paper obtains the conclusion of decreasing the error tolerance capability and sacrificing verification performance to enhance the system security.

关 键 词：多模态 低质量虹膜 热成像人脸 模糊承诺 模板保护 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TN911.73[自动化与计算机技术—计算机科学与技术]