Multibiometric Cryptosystem Based On The Fingerprint AndIris Feature-Level Fusion

Recently, the biometric identification technology has developed fast, and some problems concerning security and privacy have been exposed in the applications which have caused people’s attention greatly. A new technology called biometric encryption technology which combined biometric and encryption has originated and has been gradually developed since then.Biometric include physical characteristics, such as fingerprint, iris, face etc. and behavior characteristics, such as signature, voice, gait, etc. Different biometric have its own advantages as well as special application environments. Fusing various biometric not only can obtain higher accuracy and better security compared to single biometric, but also can overcome some inherent drawbacks of some biometric, such as low quality of fingerprint. Therefore, it has obtained extensive study. But how to ensure the security of each biometric on the premise of high accuracy, thus, multi-biometric fusion encryption technology has triggered a heated research and has become a research focus in the field of biometric.In this paper, we combine with fingerprint and iris, as two single biometric that available easily, to discuss and study the multi-biometric fusion encryption technology. The main research work in this paper focuses on the aspects as follows:1. This paper discusses the relevant methods of biometric encryption. It introduces the most typical and the most representative method in the field of biometric encryption from two aspects as biometric encryption system and template transformation respectively.2. This paper explores the relevant methods of multi-biometric fusion. It concludes the existing research work in the related field of multi-biometric fusion from two aspects as multi-biometric fusion identification technology and multi-biometric fusion encryption technology respectively.3. This paper proposes a cancellable key binding scheme based on BioHashing and shuffling algorithm. Firstly, it modifies the way of doing random mapping and the way of calculating global threshold array to improve the performance of BioHashing method to obtain the required fingerprint feature BioCode. And then, it puts forward a general way to implement the irreversibility of the biometric. After that, for the high complexity problem of key recovery, it probes into a new method of binding key. Finally, the validity of this scheme is substantiated by simulation results.4. This paper puts forward a multiple biometric encryption model based on fingerprint and iris at feature level fusion. Firstly, it optimizes the fingerprint BioCode by the statistical method. And then, it investigates the relevant knowledge about iris feature extraction, and adopts the phase encoding method proposed by Daugman to obtain the iris feature IrisCode in the fusion model. After that, it puts forward the multi-biometric fusion encryption model. Finally, it verifies its performance by simulating and testing.In a word, this paper adopts the optimized global fingerprint feature to reduce the time complexity of feature extraction as well as key recovery so that to achieve a simple and practical fusion encryption architecture. Applying the improved shuffling algorithm to fingerprint and iris enhances the flexibility and security of the biometric greatly. The fingerprint and iris fusion encryption scheme not only overcomes the shortcoming that single biometric can be easily stolen, but also obtains a satisfying authentication effect.