基于神经网络的人脸识别模型研究OA

Biometric technology is often used in the process of identity identification for the purpose of authentication and authorized access in the field of network security,where the biometric data provided by users are processed and converted by the protocols adopted by the data security system,and then compared with the biometric data of submitted and authenticated authorized users.The result of the comparison determines how access is granted and granted to the user.In the field of information security,biometric identification is a mature and effective security verification mechanism with low error rate.In recent years,network security is facing a severe situation,and the biometric images pre-stored in various information security authentication systems are facing the risk of being stolen and abused by intruders.Therefore,this paper proposes a secure storage system which does not involve modifying and storing user biometric images,which generates irregular passwords for authentication and access authorization from the submitted biometric images.In order to ensure the accuracy of biometric extraction,this study also introduces a deep learning model to convert biometric images into binary strings for storage.In this study,the first-order and second-order error probabilities are calculated by experiments.The experimental results show that the encryption system proposed in this paper not only realizes the function of reliably extracting biometric features from the image,but also ensures the high security and recognition accuracy of the generated binary string.