Cross-channel color image encryption scheme using a novel 3D neuron model and fast Josephus traversal

As digital communication continues to advance rapidly, ensuring the security of image data during transmission has become increasingly vital. To overcome the inherent limitations of traditional chaotic image encryption schemes, this paper introduces an innovative color image encryption algorithm that combines a Memristive Hindmarsh-Rose (MHR) neuron model and a Cyclically Maintained Josephus Algorithm (CMJA). Initially, the MHR model incorporates a magnetic flux variable and an adaptive control framework, employing a state-dependent switching mechanism to adjust feedback strength. This closed-loop behavior enhances the system’s nonlinearity and offers a wider range of tunable chaotic oscillations as a highly sensitive source of entropy. Next, to address the efficiency challenges of traditional Josephus scrambling in high-resolution scenarios, a novel CMJA is introduced. By utilizing “head-tail coverage and boundary contraction” strategies, the algorithm mitigates performance overhead from frequent element movement, effectively reducing the time complexity of index maintenance. In addition, a SHA-512 hash-based plaintext-dependent, one-cipher-per-image key generation mechanism and a cross-layer composite scrambling architecture are implemented, resulting in a highly nonlinear redistribution of pixels across the three-dimensional color components. Finally, experimental results show that the chi-square test values for all ciphertext channels remain below the critical threshold of 293.2478. When compared with state-of-the-art schemes, the proposed algorithm improves encryption efficiency while maintaining competitive security, indicating its potential applicability in secure communication.