Researches On Fuzzy Vault-Based Fingerprint Keys

In the field of biometric encryption, the fuzzy vault algorithm is a classical algorithm, it makes good combination of the fuzziness of biological features and the accuracy of crytographic algorithm, and achieve the purpose of binding personal identity and key information. The algorithm compensates the insufficiency of the general information security system. And the general mode of "user name & password" is replaced by biological authentication in the algorithm, which avoiding the short board of identity authentication, such as the unauthorized user taking identity fraud and information shared etc. At the same time, comparing to the traditional cryptography, the fuzzy vault algorithm made some improvement for the situation such as limitation of the password length, inconvenience of storage and management, memory difficulties etc. As a result, the security of the information transfer and storage process is enhanced.In the biological recognition domain and many applications, fingerprint has been chosen as a core of biometric encryption combined technology due to its maturity in terms of availability, uniqueness, permanence, feasibility, ease use and acceptance, forming a certain advantage in both the technology and the market. This thesis mainly discusses fuzzy vault algorithm which based on the fingerprint features, and based on a lot of previous work, a kind of pseudo-random sequence is introduced to construct our randomness fuzzy vault algorithm is proposed. The main work in this thesis can be described as follows:Firstly, local feature-based fingerprint recognition algorithm is discussed and the original Fuzzy vault algorithm as a kind of fingerprint key algorithm is introduced. It is the foundation of our proposed algorithm in this thesis, which using the collected information of fingerprint features to encrypting the secret information. The overall performance of the fuzzy vault algorithm can be improved by extracting reliable fingerprint features information through the fingerprint recognition algorithm;Secondly, the improved fuzzy vault algorithms which were proposed by U.Uludag and Shenglin Yang were discussed. The performances of those algorithms were analyzed and some problems existed in the algorithm design, analyses and experiments were indicated; Lastly, an improved randomness fuzzy vault algorithm is given, which is our main algorithm proposed in this thesis. In order to transform the fingerprint features, a pseudo-random sequence is introduced in this algorithm, which increases the randomness of the experimental data. The improved algorithm changes the original fuzzy vault's polynomial to a linear function, in which the length of function is not related with the key information any more. By introducing the ideal of key separation, the key information privacy is improved. At the same time, the key information and personal identity are well bound. Our improved algorithm, can not only bind the key information well but also save the user's fingerprint features.