Research And Implementation Of Encryption Authentication Technology For Face Recognitio

Face recognition,as an artificial intelligence technology,is widely used in many social fields,but its research mainstream lies in improving recognition accuracy,ignoring the potential security risks in the development of technology.The first is that face images are vulnerable to attacks and tampering after camera capture and before feature extraction;The second is to complete facial feature matching in the clear text domain,which poses a security threat to the feature information,while protection measures are difficult to balance data security and recognition performance.To address the above issues,this paper proposes a face image authentication scheme based on the Least Significant Bit(LSB)blind watermark and a face ciphertext feature recognition scheme based on Face Net and State Security,and designs and implements an encryption and authentication system for face recognition.The specific research work is as follows:(1)A face image authentication scheme based on LSB blind watermark is proposed.To address the issue of criminals being able to attack or tamper to obtain illegal permissions during the transmission of face images from the camera to the feature extraction module,the LSB watermark algorithm is embedded in the camera chip.Once the face image is captured,a preset watermark is embedded,and then the watermark face image is transmitted to the feature extraction module.Before feature extraction,the watermark is extracted for authentication using Peak Signal to Noise Ratio(PSNR),and can be extracted if the authentication is successful,Instead,discard the problem image.Experiments show that the proposed watermarking algorithm has good vulnerability and invisibility,and can effectively protect the security of face image data.(2)A face ciphertext feature recognition scheme based on Face Net and State secrets is proposed.In order to solve the problem that the security and recognition performance cannot be achieved simultaneously during the transmission,storage,and comparison of facial feature data,a multi task convolutional neural network and Face Net algorithm are used to detect and extract facial features from legitimate facial images.The CKKS homomorphic encryption and the national secrets SM4 algorithm are combined to encrypt facial features,complete the transmission,storage,and Hamming distance comparison of facial features in the encrypted text domain,and use the national secrets SM9 algorithm to manage the SM4 key.Experiments show that this scheme improves security while ensuring face recognition performance.(3)Design and implement an encryption and authentication system for face recognition.Aiming at the problem that the security and recognition performance of face images and facial feature data cannot be achieved simultaneously in face recognition,based on the above scheme and combined with the software design and development process,digital watermarking algorithms,deep learning face recognition algorithms,and biometric encryption and authentication algorithms are integrated to design and develop corresponding functional modules,thereby designing and implementing an encryption and authentication system for face recognition.Tests show that the system improves data security while ensuring face recognition performance.