A Review of Image Encryption Driven by Fractional-Order Neural Network Models

Traditional integer-order chaos-based encryption systems often exhibit relatively limited dynamical behavior and may suffer from complexity degradation or short periodic trajectories with finite-precision implementation, thereby weakening their resistance to cryptanalysis. Fractional-Order Neural Networks (FONNs), owing to their inherent memory and hereditary properties as well as their effectively infinite-dimensional dynamics, provide a promising approach to alleviating these limitations. This paper presents a systematic review of Fractional-Order Neural Network (FONN)-based image encryption methods and their recent advances. The novelty of this review lies in its unified classification framework that categorizes FONN-based encryption schemes into three representative architectures: fractional-order Hopfield neural networks, fractional-order memristive neural networks and fractional-order cellular neural networks. It provides a comparison of their dynamical characteristics, encryption mechanisms and security performance. This review provides a coherent theoretical framework and practical reference for further research on FONN-based image encryption.