Reversible cross-channel coupling and dynamic diffusion for secure chaos-based color image encryption

Abstract The rapid growth of multimedia communication and cloud-based image transmission has increased the demand for secure and computationally efficient image encryption techniques. However, many existing chaos-based image encryption schemes rely on partially adaptive key generation, limited inter-channel interaction, and globally uniform diffusion mechanisms, which may reduce their resistance to statistical, differential, and chosen-plaintext attacks. To address these limitations, this paper presents an integrated chaos-based color image encryption framework, namely Reversible Cross-Channel Coupling and Dynamic Diffusion (RC3D), which combines SHA-256-based plaintext-dependent key generation, Piecewise Linear Chaotic Map (PWLCM)-driven permutation, reversible cross-channel coupling, and adaptive block-wise diffusion within a unified encryption architecture. Unlike conventional approaches that process RGB channels independently or employ fixed diffusion strategies, the proposed framework strengthens inter-channel dependency while preserving exact invertibility and enhances local randomness through adaptive diffusion without increasing computational complexity. Experimental results demonstrate that the proposed framework achieves information entropy exceeding 7.999, absolute correlation coefficients below 0.01, NPCR values greater than 99.6%, and UACI values exceeding 33.4%, indicating excellent resistance against statistical and differential cryptanalysis. Furthermore, the encrypted images exhibit uniform histogram distributions, high key sensitivity, and efficient computational performance, demonstrating that the proposed RC3D framework provides a practical and effective solution for secure real-time multimedia communication.