Deterministic Mamdani fuzzy-controlled KAA and hybrid logistic–sine chaotic image encryption with AES S-box substitution

In this digital era, the rapid growth of imaging applications in cloud, medical, and IoT environments requires secure and efficient encryption techniques capable of addressing pixel redundancy and statistical correlations. This study proposes a deterministic entropy-controlled image encryption framework that integrates Mamdani fuzzy inference, hybrid chaotic systems, and selective nonlinear substitution. The proposed framework utilizes image statistical characteristics, including entropy, mean, and contrast, to adaptively regulate chaotic initialization, substitution probability, and diffusion strength. A multi-stage permutation mechanism based on the KAA chaotic map and hybrid logistic-sine system enhances confusion, while a deterministic partial AES S-box substitution introduces reversible nonlinearity. Extensive experiments conducted on seven SIPI benchmark images demonstrate strong cryptographic performance, achieving NPCR values of 99.67–99.89%, UACI values of 33.11–33.66%, and entropy values of 7.984–7.995 bits. The encrypted images exhibit low adjacent pixel correlation values within approximately $$\pm 0.01$$ , a large key space of $$2^{356}$$ – $$2^{360}$$ , and NIST SP 800-22 randomness pass rates of 98.5–99.6%. Furthermore, the framework achieves average encryption and decryption times of 0.84 s and 0.79 s on PC platforms, demonstrating its suitability for secure multimedia transmission and resource-constrained IoT imaging applications.