Adaptive Quadtree-Based Hyperchaotic Image Encryption with RPMPFRHT

The traditional color image encryption techniques using fixed-size block partitioning and low-dimensional chaotic maps may have limited adaptability to local texture variations and lower security robustness. To address these issues, this paper presents an adaptive color image encryption scheme by combining the quadtree decomposition technique with the Reality-Preserving Multiple-Parameter Fractional Hartley Transform (RPMPFRHT) and the memristive hyperchaotic system. The quadtree decomposition-based adaptive image partitioning technique can dynamically divide image areas based on local statistical features, achieving texture-adaptive encryption. The fractional-order parameters of the RPMPFRHT are deterministically calculated from the SHA-512 hash value of the secret key, greatly increasing the effective key space and improving key sensitivity. The memristive hyperchaotic system produces high-entropy keystreams for global pixel scrambling, transform-domain permutation, and bidirectional diffusion, greatly improving the resistance to statistical attacks, differential attacks, and data loss attacks. Simulation experiments on the standard benchmark images show excellent security performance, with near-ideal NPCR (99.69%) and high UACI (37.07%) values, uniformly distributed histograms, low adjacent pixel correlation, and high resistance to noise and occlusion attacks. These simulation results demonstrate that the proposed scheme has better adaptability, higher security strength, and perfect reversibility for secure color image transmission and storage.