Research On Biometric Authentication Methods Supporting Privacy Protection And Efficient Retrieval

Biometrics technology is widely used in finance,public security and other fields to provide authentication technology support for new informatization and intelligent Internet.The privacy of biometrics is highly tied to personal identity and is not easy to change.Its leakage seriously threatens the vital interests of users.It has been protected by legislation in China,the United States and other countries.Existing biometric privacy protection schemes need to store biometric keys,and their security and privacy rely too much on the trusted end.However,the P2 P scenario has the characteristics of peer and end roles that do not trust each other,which makes the existing privacy protection schemes no longer applicable.In the process of biometric retrieval,the encryption and hiding of biometric data lead to difficulties in indexing and low retrieval efficiency.This article focuses on privacy protection in biometrics,constructs a biometric privacy protection model in P2 P scenarios,and designs a general and efficient retrieval scheme for hidden or encrypted biometrics.Aiming at the privacy protection problem of biometric identification in P2 P scenarios,the fuzzy extractor is used to bind the biometric key and public information,and a biometric privacy protection scheme based on fuzzy extraction is proposed.The biometric key is accurately reproduced according to the public information and biometric characteristics when it is used.,To overcome the shortcomings of traditional biometric privacy protection schemes that biometric keys must be stored.Use the biological key to construct a one-way function to generate permanent keys and temporary keys to prevent attackers from cracking biological privacy information in reverse and improve the privacy of the scheme.Use the blockchain to build a trusted mechanism for public information and public keys to prevent attackers from forging and tampering with them and resist man-in-the-middle attacks.Experiments show that the secondary authentication success rate of the scheme reaches99.0%,the average authentication time is 4.4ms,and a single physical server can support 20 million users,which meets the application scenarios of biometrics.Aiming at the problems of hiding or encrypting biometric data that cannot build an effective index and low retrieval efficiency,the approximate retrieval method is used to find multiple possible results that are closest to the target,and a biometric retrieval scheme based on perceptual hash is proposed to avoid one-by-one traversal and comparison..For hidden or encrypted biological characteristics,the perceptual hash distance is calculated to determine the similarity between the retrieval target and each biological feature,and the set of approximate nearest neighbors closest to the target is found,reducing the most timeconsuming comparison operation in the retrieval link.Utilize the one-way nature of perceptual hash to prevent attackers from reversing the analysis of biological characteristics based on perceptual hash,and improve the privacy of the solution.Experiments show that the average retrieval time of perceptual hash is only 0.7% of the ordinary linear retrieval time.Compared with the classic E~2 LSH and random KD tree retrieval schemes,the average retrieval time is shortened by 9.3% and 18.0%,respectively.Based on the researched architecture and methods,this paper designs and implements a biometric identification system that supports privacy protection and efficient retrieval.The system completes biometric identification in P2 P scenarios and meets the needs of biometric privacy protection and efficient retrieval.The application of the system to the cloud computing platform verifies the feasibility and efficiency of the researched scheme.