| In recent years,mobile Internet has developed rapidly and mobile applications have emerged in an endless stream.However,user information security is also a constant problem.Major companies and enterprises have gradually strengthened their emphasis on security certification in order to further improve their products' security.With the improvement of mobile devices,face recognition as a new type of mobile authentication has gradually begun to enter people's daily life.This paper proposes a mobile authentication scheme based on face recognition,researches and designs a mobile face recognition authentication system with high fraud cost,strong universality and well user experience.This paper studies the current progress of related research at home and abroad,deeply analyzes the application scenarios,summarizes the overall requirements of the project,and devides into modules to design and implement.Firstly,in order to reduce the influence of facial eyeglasses on the accuracy of face recognition,researching the face image classifier based on convolutional neural network to judge whether or not a given facial image has eyeglasses;Second,an Enhanced Cycle-Consistent Generative Adversarial Networks(ECGAN)is proposed to implement a facial image converter for removing eyeglasses;Then implementing a detector to find facial region from facial image and calculating the difference between two faces to judge whether it is the same person or not;Finally,capturing face image and upload to service in mobile application to do authentication,and an anti-fraud face living body detection mechanism suitable for mobile device is analyzed and designed.The system is based on the idea of Service Oriented Architecture,can be simply transplanted to other systems and has a strong universality.It has been used in mobile applications of scientific research management systems,provides secondary authentication for user's important information,and greatly improves the security of related applications. |
PDF Full Text Request