A Research on On-line Biometric Authentication System with Privacy Protection on the Internet

生物识别技术在这样的社会中迅速盛行。在这些例子中，生物识别系统应用于封闭的环境中，然而，社会对生物识别系统应用于互联网服务的认证有强烈的需求，如网上银行、电子政务、公司审批等。另一方面，生物识别系统也带来了一些隐私问题。也就是说，在某些情况下，诸如病史之类的重要隐私信息会从生物识别信息中泄露，这些生物识别信息包括从生物识别传感器设备获取的生物识别原始数据、从相应生物识别原始数据提取的特征信息以及注册模板。因此，我们从隐私保护的角度研究新的生物特征认证框架和模板数据库。一次性生物特征认证提出了一种新的生物特征认证信息保护技术，特别是特征信息和样本信息的保护。我们建议的要点是，提取的特征和注册的模板通过在每次身份验证中生成的一次性转换进行转换。转换后的特征和模板通过互联网等不安全的通信线路传输，并用于匹配过程。由于每次传输的特征信息和模板都不相同，因此可以防止互联网上的窃听和重放攻击。生物特征模板数据库的安全分布式存储方案我们提到每个生物特征模板由多个元素组成，如指纹细节。我们提出了一种模板元素的分布式存储方案，通过对每个注册人员的存储地址索引加散列表进行置乱。索引由注册模板元素的二进制数据组合的哈希值和基于所有者个人数据的种子值的伪随机数确定。在此解决方案中，存储设备中的相邻数据不是来自唯一的人员模板。因此，除非存储地址的映射被窃取，否则攻击者不会重建每个注册人员的模板。此外，该程序对伪造品的干扰也有一定的作用。利用已登记模板的相互关联建立生物特征模板数据库我们假设生物特征模板在数据库中的注册者之间是相互关联的。我们提出了一种生物特征数据库的存储方案，该方案将模板分离为所有注册模板的平均数据。平均数据和差值数据分开存储。平均数据用所有入组图像的平均图像表示，或者用矢量形式数据的矢量量化指标表示。由于模板的重建既需要平均数据，也需要差异数据，如果攻击者没有获得平均数据的信息，该方案能够防止对注册模板的恶意操作。少

The biometrics rapidly prevails in society like this. In these examples, the biometric systems are applied in closed environments, however, there are fervent social demands about the biometric systems applied to the authentication on internet services such as internet banking, electrical government, approval in company, etc.On the other hands, the biometric system causes some privacy issues. That is, in some cases, significant privacy information like a medical history is compromised from the biometric information which is including biometric raw data acquired from biometric sensor devices, the feature information extracted from the corresponding biometric raw data, and enrolled templates.Therefore, we investigate novel biometric authentication framework and template database in the viewpoint of privacy protection.1. One-Time Biometric AuthenticationWe proposed a novel protection technique for the information of biometric authentication, especially the feature information and the templ … More ates. The point of our proposal is that the extracted features and the enrolled templates are transformed by one-time transformation that is generated in each authentication. The transformed features and templates travel through insecure communication line like the internet, and they are used in matching process. This technique causes security against eavesdropping and replay attacks on the internet, because the transmitted feature information and the templates are different every time.2. Secure Distributed Storage Scheme in Biometric Template DatabaseWe mentioned the each biometric template consists of multiple elements like a fingerprint minutiae. We proposed a distributed storage scheme of the elements of the templates by scrambling indices of storage address with hash table for each enrolled person. The indices are determined from the hash value of combination of binary data of the elements of the enrolled template, and pseudo random number with a seed value based on the owner's individual data. In this solution, the adjacent data in the storage device are not from a unique person's template. Therefore, no adversary reconstructs each enrolled person's template, unless the map of the storage address is stolen. In addition, this procedure has an effect on disturbance of the forgery.3. Biometric Template Database by using Mutual Correlation of Enrolled TemplatesWe supposed the biometric templates have mutual correlation between the registrants on the database. We proposed the storage scheme for the biometric database which separates the templates to average data of the all enrolled templates. The average data and the difference data are stored separably. The average data are denoted by average images of all enrolled images, or indices of vector quantization for vector form data. Because reconstruction of the templates requires both the average data and the difference data, if attacker obtains no information of the average data this scheme is able to prevents malicious manipulation against the enrolled templates. Less