Palmprint Encryption and Authentication Method Based on Masked Autoencoders

With the rapid development of biometric recognition technology in the field of deep learning, such as face recognition, iris recognition and palmprint recognition, they are widely used in various fields, but with them comes the serious problem of information leakage and privacy protection. In this paper, we propose a palmprint encryption and authentication method based on the image recovery model MAE. Only 25% of the location information and features in the original image data are encrypted and saved separately, and then the encrypted data is used to generate an image and compare it with the original image for authentication purposes, and the encryption and authentication process is achieved with extremely low memory and resource consumption. The model has a very low image retention rate of 25% and uses de-imagination and fragmentation techniques to achieve privacy protection of palmprint data and ensure the security of biometrics through the irreversibility of data encryption.