Dynamically scattering medium-based security-enhanced optical image encryption using orbital angular momentum and deep learning

Abstract

Scattering medium-based optical image encryption is a practical strategy for information security, but it is challenging to convert speckles caused by scattering into encrypted information. In addition, the security level of the encryption is an issue of concern. Here we propose and demonstrate a potential optical image encryption with high security, in particular using optical orbital angular momentums (OAMs) to carry plaintext information so that encrypted information can be coded and converted into speckles, besides a dynamic diffuser for encryption and a pre-trained convolutional neural network (CNN) model for decryption. The high level of security stems from the use of both the order key (one-to-one correspondence between the labelled OAMs and the speckles) and the model key (unduplicatable pre-trained CNN model). Proof-of-concept experiments to demonstrate the proposed method have been successfully carried out. The image encryption method can open up new prospects for authentication, anti-counterfeiting and encrypted communication.