Image Encryption Algorithm Based on A Multilayer Broad Echo State Network and Hyperchaotic Chen System

作者：Bufeng Zhang, Yan Cao, Tao Xu, Abid Hussain, Xiaoqiang Zhu, Yuan Liu · 发表于：2026 International Conference on Artificial Intelligence and Materials (ICAIM) · 年份：2026 · DOI：10.1109/ICAIM69488.2026.11601840

The rapid growth of multimedia communication has made secure image transmission increasingly important in applications such as medical imaging, remote sensing, and intelligent surveillance. However, due to the inherent characteristics of digital images, including high redundancy and strong pixel correlation, many conventional encryption schemes still suffer from limited adaptability, insufficient diffusion, or high deployment costs. To address these issues, we propose the Broad Echo State Network (BESN), a training-free image encryption scheme that integrates a hyperchaotic Chen system. The BESN is constructed from a key-derived seed without any training process. For efficiency improvement, a $4-4$ high/low bit plane decomposition is employed together with a differential permutation strategy. To provide strong nonlinearity and sensitivity to tampering, diffusion is implemented using bidirectional variablestep ciphertext feedback, combined with an accumulator and an AES S-box. The results demonstrate that the ciphertext exhibits near-ideal statistical characteristics, with an average information entropy of 7.9993 and NPCR/UACI values approaching their theoretical optima.