Image encryption using dynamic S-boxes generated using elliptic curve points and chaotic system

Abstract

Images play a vital role in almost all sectors, including medical, video conferencing, weather forecasting, military applications, and most common social media applications. Today’s proliferating technology provides attackers an opportunity to breach and access confidential data that are not enciphered or weakly secured. There is a massive requirement for a robust image cryptosystem for the secure transmission of confidential images over an insecure network. Here, we present an encryption technique to encipher images using dynamic substitution boxes (S-boxes) generated by a chaotic system and the points from an elliptic curve over a finite field. The dynamic S-boxes generation depends on a key obtained from the hash value using Secure Hash Algorithm-512 ($SHA512$). Each different image generates different S-boxes making it immune to known-plaintext attacks. The input image pixels are permuted and substituted using dynamic permutation tables and dynamics S-boxes. The mean of S-box security analyses such as non-linearity, strict avalanche criterion, bit independent criterion, differential approximation probability, and linear approximation is obtained as 104, 0.510294, 0.51145, 0.046875, and 0.083252 respectively. The generated S-boxes using the proposed method meet the standards of secure S-boxes. The cipher output images produced by the proposed methodology are also examined using statistical and security analyses. The proposed method can resist cipher-text-only attacks and has a large keyspace of 512 bits to withstand a probabilistic or deterministic attack on the elliptic curve discrete logarithmic problems such as baby-step, giant-step attack, and Pollard’s rho attack. The strength of the generated cipher image using the proposed method is substantiated by tests such as the National Institute of Standards and Technology randomness test, the local entropy, correlation analysis, and differential analysis. Comparisons are made with some of the related existing state-of-the-art methods and turn out to be on par or better with the other compared methods.