Image Encryption Based on a Coined Chaotic System and High-Intensity Encryption Primitives

Abstract

Security and efficiency are two major themes of image encryption algorithm. To enhance the security together with improving the operational efficiency, this paper proposes an image encryption algorithm based on a novel chaos and high-strength encryption primitives. Under the framework of confusion-diffusion, three core elements affecting security, namely, random number generator, confusion primitive, and diffusion primitive are thoroughly investigated. Regarding random number generator, a new chaos, 2D-SCM, is constructed with the result of the 0-1 test is 0.9981. The confusion primitive adopts naturally uniformly distributed orthogonal Latin squares to achieve maximum uniform distribution of scrambled image. The diffusion primitive employs linear congruence transformation based on 65537 to elevate pixel transformation from modular addition to modular multiplication, effectively avoiding differential attack and equivalent key attack. In the pursuit of improving efficiency, a two-stage orthogonal Latin squares generation algorithm is introduced, specifically targeting the low efficiency associated with generating orthogonal Latin squares from finite fields. Security testing on the cipher image shows that NPCR and UACI closely align with expected values of 99.6094% and 33.4635%, respectively, and local information entropy stays within the expected range of (7.901515698,7.903422936), confirming the algorithm's robust defense against unauthorized attacks. Efficiency testing also reveals that the proposed algorithm matches the encryption efficiency of current excellent methods.