AntEater: When Arnold's Cat Meets Langton's Ant to Encrypt Images

Abstract

With the proliferation of digital images over open networks, secure and efficient encryption schemes are imperative for safeguarding image privacy. This paper proposes a novel 5-stage image encryption algorithm adhering to Shannon's confusion and diffusion paradigm. In the first stage, Langton's Ant is employed to induce chaos and perturb the pixel distribution of the original image. The second and fourth stages apply Mersenne Twister generated keys to confuse the image via XOR operations. An S-box substitution is utilized in the third stage to disrupt pixel statistical properties. Finally, Arnold's Cat map further scrambles and diffuses the pixel positions over the image. Extensive security analyses reveal the algorithm's robustness against various attacks such as visual, entropy, brute-force, statistical and differential attacks. Additionally, it successfully passes the NIST SP 800-22 test suite. Performance results demonstrate the proposed algorithm's efficacy for real-time secure image transmission with low computational overheads. The algorithm's security level combined with high-speed performance makes it well-suited for practical image encryption applications.