A novel image encryption scheme based on Arnold cat map, Newton-Leipnik system and Logistic Gaussian map

In the existing literature, numerous chaos-based multimedia encryption schemes have been presented. Benefiting from inherent properties such as ergodicity and key-sensitivity; this can potentially improve non-linearity in encrypted data. This paper introduces a hybrid scheme that encrypt color images using multiple chaotic maps such as the two-dimensional Arnold cat map (ACM), Newton-Leipnik dynamic system (NLDS), and a modified version of the Logistic Gaussian chaotic system (LOGAS). To make the scheme plaintext dependent, the input parameters (keys) of the chaotic maps are determined by passing the plaintext color image layers through SHA-512. Initially, the color image layers are shuffled by applying two-dimensional Arnold map scrambling. The permuted pixels were encrypted through a pseudo-random number sequences (PRNS) generated from the NLDS. To strengthen the suggested approach's security level, the modified LOGAS were utilized to generate dynamic substitution boxes (S-boxes) to substitute the encrypted pixels. Moreover, the proposed approach is tested through various statistical tests. The adjacent pixel correlation and entropy value are close to 0 and 8, respectively. The NPCR and UACI tests values are 99% and 33%, respectively. The encrypted images histograms are flat which is close to optimal value, and the computational time is less than 2.0 sec for the proposed scheme. Thus, these findings indicate the effectiveness and better reliability of the proposed system.