Article ID Journal Published Year Pages File Type
4947373 Neurocomputing 2017 14 Pages PDF
Abstract
Differing from the traditional permutation-diffusion architecture, an efficient symmetric image encryption algorithm is designed in this paper. To resolve the issue of low sensitivity of a plain-image, which is measured by a unified averaged changing intensity and the number of changing pixel rate, pre-modular operation is proposed to be used for pre-processing a classical encryption algorithm. Because of the invariance of pixel summation in the plain-image before and after pixel positions exchange, the keystream used for the permutation operation is designed to be dependent on the plain-image. Since for most encryption algorithms it is hard to implement the diffusion operation in relation with the plain-image dependence, a self-adaptive encryption scheme is proposed as a remedy. Consequently, the keystream for diffusion becomes dependent on the plain-image, which still satisfies the requirement as a symmetric cipher. To that end, by taking full advantage of the chaotic intertwining logistic map and its unpredictability, a new cipher with pre-modular, permutation and diffusion is designed for image encryption, which is then simulated and tested. Numerical experiments and security analysis demonstrate that the new cipher is secure enough for image communication over the open network.
Related Topics
Physical Sciences and Engineering Computer Science Artificial Intelligence
Authors
, ,