Image transformation based on optical reservoir computing for image security

Abstract

Visually secure image encryption method can provide an added dimension of protection for image compared to conventional image encryption scheme that encrypts a plain image to a noise-like cipher image. Here we introduce a novel perspective based on image transformation to perform visually secure image encryption by constructing an optical reservoir computing (ORC) system. In this paper, we first design a novel optical dynamics model to construct a dual-channel signal injection ORC system, and to provide chaotic keys when the optical dynamics model operating at chaotic states. Then, the established ORC system is used to reveal the underlying transformation relationship between a plain image and a visually meaningful image both with the same size. One important role of such relationship is to provide evidence for the determination of cipher image that allows visually meaningful image to be directly set as cipher image, prompting a high visual security of that, and another role is to facilitate the calculation of necessary decryption keys in order to obtain a high-quality decrypted image. Compared with other existing encryption methods, this work overcomes a challenge of the inability in simultaneously achieving low burden cost, high visual security of cipher image, and high quality of decrypted image, and the simulation results and performance analysis demonstrate that the proposed scheme is effective and secure. Moreover, this study also offers an inspirational implication to develop novel alternative framework with respect to the study of visually secure image encryption and will help other researchers to develop new image security techniques based on transformation methods.