Answers are grounded in this paper and cite their sources.
This paper presents a hybrid image encryption framework based on a differential photonic quantum memristor-modulated Hopfield neural network (D-PQM–HNN). The four-dimensional D-PQM–HNN exhibits rich and complex chaotic dynamics with a borderline hyperchaotic tendency. The network is used as the entropy source of the framework, and the bitstream generated from the network passes the National Institute of Standards and Technology (NIST) SP 800-22 statistical test suite at the adopted significance level. The bitstream drives a multilayer encryption pipeline comprising bidirectional cumulative XOR diffusion, permutation-based confusion, an extended 16-symbol DNA-domain layer with rule-selected mixing, multi-point crossover and mutation, and a terminal Advanced Encryption Standard-256-in-counter-mode (AES-256-CTR) stage. The complete algorithm is implemented on an NVIDIA Jetson AGX Orin using channel-wise Compute Unified Device Architecture (CUDA) parallelization. For 512 × 512 color images, the implementation reduces the total encryption time from approximately 1.5 s on the reference desktop CPU to 0.22–0.24 s on the embedded platform. The experiments on four benchmark images include histogram and chi-square analyses, adjacent-pixel correlation, number of pixels change rate/unified average changing intensity (NPCR/UACI), information entropy, key sensitivity, and robustness against noise and data loss. The results show near-uniform cipher image statistics, high plaintext and key sensitivity, and recoverable image information under the tested corruption conditions, while demonstrating the feasibility of the proposed framework for edge-oriented image encryption.
No takes yet. Share an insight, caveat, or question.
Sağbaş et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: