A Key Generation By Multibiometric Fusion In Encryption-Domain Level

As the information technology promoting processes of the world continuously,Network will be related to people's daily lives more and more tightly, and Information Security will lead to more widespread concern.Through the tireless efforts of many scholars at home and abroad have been taken, a number of the new Information Security Technologies emerge constantly, in which the program is the most representative—Biometric Encryption Technology. After long-term development and in-depth research, Biometric Encryption Technology, not only in theory had a major breakthrough, but in practice has also been fully affirmed. However, with in-depth research of technical means and social progress, we need a more secure and reliable solutions to meet the needs of practical application. After repeated demonstrations and innovation constantly, and ultimately Biometric Encryption are invariably tend to multi-feature, multi-modal integration of biometric encryption technology. In response to this new problem, on the basis of my group's research area,fixed-length feature extraction, key generation, and encryption domain matching,combined with some of our predecessors research results, this paper has proposed a key generation by multibiometric fusion in encryption-domain level, and a novel improved multi-level matching solution, the experimental results in the real fingerprint database seem to be satisfactory.In this paper, the supposed key generation method, using the fusion of fingerprint vector features and a collection of feature, and through the the coding method in Fuzzy Extractor , constructed three kinds of auxiliary data to deal with feature differences, then get the last fixed-length feature to form a stable and reliable security key. The experiment proved that the program is an effectively secure and reliable access to create key, at the same time, through adding a specific chaff point, and then ensuring the security of biometric templates. Through the introduction of multi-feature fusion method, increasing the difficulty of attack, thereby enhancing the safety performance of the system, and the integration of independent and complementary characteristics, to improve the recognition performance of the system. In addition, in order to solve the difficult problems of matching caused by adding a chaff point. In this paper , we also posed an improvement for the existing methods of encryption domain matching, the improved scheme would be based on distance metric ,and filtering out more chaff point by adding a local structure matching filtering method based on the minutiae-based, thus improving the accuracy of the system ,at the same time improve the system recognition performance.