An Efficient Medical Image Encryption Scheme Utilizing Nonuniform Cellular Automaton

Abstract

Encryption is one of the simplest and most efficient technologies to protect images from information leakage. Because medical images usually have great data volume and personal privacy, encryption methods with higher efficiency and security are expected. The nonuniform cellular automaton (CA), which has common advantages of CA such as high efficiency and discreteness, has a unique advantage in handling inputs with all zero pixels, and thus emerges as a promising candidate for medical image encryption. In this work, our proposal adopts the permutation–diffusion–diffusion structure rather than the classic permutation–diffusion structure to improve the efficiency and security. In the diffusion phase, the reversible nonuniform CA is adopted to encrypt the processed pixels, which can improve security against plaintext attacks. An efficient “divide-and-rule” diffusion method is designed for better diffusion performance and efficiency. The security and robustness analysis indicates that our method has satisfactory security against common attacks and robustness in resisting data loss and random noise.