Research Of Fingerprint Template Protection Algorithm Based On BioHashing

With the rapid development of Internet and networking technology, information security issues have received growing concern in the current biometric community. As one of the core technologies to achieve security of information system, authentication has been widely used in various applications in the real world. Compared with traditional authentication using the key or token, biometric has no need to carry or remember, and it is difficult to change and counterfeit, etc. These advantages offer the explanation why biometric authentication technology has been widely used in various fields. Given the unchangeable nature of the biometric feature, user’s biometrics cannot be reproduced or placed forever if they compromised. Thus, how to protect sensitive information of user biometrics is critical during the application of biometric identification, especially in protecting template data.This dissertation mainly studied fingerprint template protection methods based on BioHashing; we also proposed an alignment-free cancelable fingerprint templates generation algorithm. The specific work includes the following aspects:1. Current research of biometric template protection is summarized; the requirement which should be fulfilled in ideal template protection scheme is also analyzed. At the same time, this paper introduces several common template protection algorithms, and analyzes their advantages and disadvantages.2. On the basis of the analysis of BioHashing fingerprint template protection method and as for the main drawback of the original BioHashing, it presents two kinds of improved BioHashing fingerprint template protection algorithms. Combining the characteristics of the fingerprint feature data, we introduces majority voting algorithm to train the fingerprint template which can effectively reduce the differences between templates within a class and increase the recognition performance of the system. These two kinds of improved BioHashing algorithms are as follows:(1) On the basis of dislocation comparison and BioCode generation algorithm. During discrete process of fingerprint data, it used non-linear approach of dislocation comparison to generate a fixed-length binary sequence, which effectively improves the security of the algorithm. Experimental results show that the improved method for generating binary sequences can effectively protect sensitive data of fingerprint template and improve the recognition performance of the system.(2) On the basis of sliding window threshold quantization and BioCode generation algorithm. This algorithm uses a sliding window producing a fixed-length binary sequence in the quantization process, and combined with the sliding window with more key space, which effectively improve the fingerprint feature of the discrimination and security. Experimental results show that binary sequences generated by this method can not only effectively protect the original features of the fingerprint template, but improve the recognition performance of the system.3. Through the research of the cancelable fingerprint template generation methods, we designed a cancelable fingerprint template based on minutiae local information. First, we extract fingerprint minutiae feature after the preprocessing, then we use the improved Tico sampling structure to extract texture feature of minutiae neighboring region. Finally, the fingerprint template is generated by combining the user PIN. meanwhile, we design of fingerprint matching algorithm. The experiments results show that the algorithm not only achieve good recognition performance, but also fulfills revocability, diversity and non-invertibility, and has align-free fingerprint image by additional helper data, which achieves practicability of the algorithm.