Multimodal Biometrics Authentication Based On Lip Movements And Voice

With the advent of the information age,how to ensure the security of information is a problem that every country and individual have to face.Authentication,as the key technology to connect the "identity" of the real world and the accessing entity of the digital world,is the basis for realizing the key links of information security such as permission confirmation and permission grant.Compared with the traditional knowledge-based and possession-based authentication,biometric-based authentication relies on the uniqueness of biometrics to authenticate users,since biometrics will not bring extra burdens for users to carry and remember,they are more convenient and reliable for authentication.Compared with other biometrics,it is low-cost to collect voice.In addition,in recent years,voice authentication has matured as a reliable and stable method for authentication in a wide range of applications.At the same time,lip movements closely related to the users voice have also been proven to be able to authenticate users.However,unimodal authentication based on either voice or lip movements has limitations:voice-based authentication is vulnerable to spoofing attacks and suffers from serious performance degradation in noisy environment;authentication based on lip movements is sensitive to the speakers’ physical or psychological conditions,which leads to the instability of the authentication performance.Although some researches exploit both voice and lip movements to build a multimodal authentication system,they fail to implement efficient fusion of the two biometrics since they use different signals to carry the two biometrics.Based on the above background,this paper proposes a safe,efficient,practical,and userfriendly multimodal biometrics authentication technology based on lip movements and voice,and implements the corresponding system on existing commercial smartphones.This technical solution uses acoustic signals to sense the users lip movement s,exploits audio as the carrier of the lip motion biometric and voice biometric,making it possible to fuse the two biometrics efficient,and also improves the security and stability of the corresponding system through the complementary advantages of the two biometrics.In order to sense lip movements with acoustic signals,this paper proposes a fine-grained lip movements sensing approach based on acoustic signals,which realizes the extraction of subtle lip motion features.In addition,this technical solution can collect these two biometrics at the same time,which greatly reduces the users burden of operation,and the two biometrics can be collected merely with the speakers and microphones commonly configured on existing commercial smartphones,which ensures lowcost data collection.Finally,a large number of experiments are conducted to prove that this technical solution can provide users with convenient and reliable authentication.