Research And Application Development Of Face Authentication For Mobile

Face verification technology can be widely used in applications requiring authentication such as access control,attendance,and government convenience services.With the widespread use of mobile devices and the continuous development of mobile Internet,various applications based on mobile intelligent terminals are more in line with the usage habits of the public.Therefore,how to realize convenient and efficient face-to-face authentication for mobile terminals and related statistical analysis is an urgent problem that needs to be solved in current automation office.This thesis focuses on the needs analysis of face authentication on the mobile side.Based on face recognition technology,this paper designs and implements an efficient face authentication system.The main work includes:1)Aiming at the problem that the similar feature spacing is not compact in the face image classification problem,the loss function is improved,and the speed and accuracy of mobile face recognition are improved.A loss function combined with cosine space and angular space is used.The method is based on the MobileFaceNet network foundation and uses a deep separable convolution method to reduce a large amount of calculation and parameter quantities.At the same time,by constructing an inverted residual structure,the input and output dimensions are further reduced to reduce resource consumption on the mobile side.Aiming at the importance of reducing the network performance by focusing the various points of the face image on the average pooling layer,the deep convolution layer is used to extract the image features.Finally,the optimal network model is trained by the combined loss function.In the experimental part,the cleaned,aligned and cropped MS-Celeb-1M data set is normalized,and the processed sample is used as the input of the deep neural network.The features are extracted by the MobileFaceNets network,and finally passed the combined loss function.To increase the difference characteristics between classes,and reduce the intra-class difference features,the model is encouraged to learn more deep features to improve the accuracy of face recognition.2)Based on the research of face recognition,the requirements analysis of mobile face authentication is given,and the system framework is given.The functional structure of the mobile face authentication system,the system user use case diagram and the corresponding use case description are designed to provide a basis for system development and implementation.3)Based on the above requirements analysis,the mobile face authentication system is designed and implemented.With Android Studio as a development tool,the face authentication system was designed and implemented.The system mainly includes mobile platform image acquisition module,image filtering and identification module,network communication module,data persistence module and data acquisition and analysis module.The system completes the transmission problem of standard photos through the OKHttp-based Retrofit network communication.The deployment of the face recognition module is realized by the dynamic link library technology,and the event burying point is added by the key functions of the system.Use AspectJ for AOP monitoring to obtain user-used,interesting functional operations and related data for data analysis.