Four-channel Imaging Based on Reconfigurable Metasurfaces: Hyperchaotic Encryption under Physical Protection

TL;DR

The paper tackles security vulnerabilities in multi-channel metasurface imaging by integrating cryptographic protection with physical-layer security. It proposes a four-channel polarization-multiplexed phase-change metasurface based on Sb2S3, combined with Chen hyperchaotic encryption and DNA encoding to securely serialize two near-field QR codes and two far-field holograms, plus a reversible crystalline-to-amorphous switch for content concealment. Key contributions include the double-cell meta-atom design enabling independent control of two Jones components for four-channel imaging at 633 nm, a chaotic-encoding workflow generating QR ciphertext with embedded decryption keys, and demonstration of adaptive physical concealment. The results show high imaging fidelity (near-field SSIM up to 99%, far-field SSIMs 72.04%–85.43%), strong robustness against cropping and statistical attacks, and effective physical-layer concealment. This integrated algorithm-physical co-security framework supports high-density optical information processing and secure, unclonable displays with potential applications in secure communications and data storage.

Abstract

Metasurfaces facilitate high-capacity optical information integration by simultaneously supporting near-field nanoprinting and far-field holography on a single platform. However, conventional multi-channel designs face critical security vulnerabilities for sensitive information due to insufficient encryption mechanisms. In this work, we propose a four-channel phase-change metasurface featuring algorithm-physical co-security-a dual-protection framework combining intrinsic metasurface physical security with chaotic encryption. Our polarization-multiplexed metasurface generates four optical imaging channels through meta-atom design, including two far-field holograms and two near-field patterns. To enhance system security, we apply Chen hyperchaotic encryption combined with the Logistic map and DNA encoding to convert near-field information into secure QR codes; far-field holograms are retained to demonstrate the metasurface's information capacity and for attack detection. Phase-change metasurface further provides physical-layer security by dynamically switching imaging channels via crystalline-to-amorphous state transitions, enhancing anti-counterfeiting and reliability. The proposed metasurface achieves high-fidelity imaging, robust anti-attack performance, and independent channel control. This integrated approach pioneers a secure paradigm for high-density optical information processing.

Figures (5)

• Figure 1: The metasurface generates two chaotic-encrypted QR codes in the near-field and two holograms in the far-field. Under x-polarized light, the far-field produces a "boat" hologram while the near-field reveals its corresponding chaotic-encrypted QR code. Under y-polarized illumination, the system displays an "NCU" emblem hologram in the far-field accompanied by another chaotic-encrypted QR code in the near-field. After transition to the crystalline state, the imaging functionality will be locked.
• Figure 2: Metasurface nanostructure design and chaotic system overview. (a) Double-cell nanopillar structure used: $\mathrm{Sb_2}$$\mathrm{S_3}$ nanobrick on $\mathrm{SiO_2}$ substrate. (b) Diagram of the Logistic chaotic map, showing system trajectories for different control parameters $\mu$. (c), (d) Transmittance and propagation phase for long axes of double-cell nanopillars in the amorphous state. (e), (f) Propagation phase and transmittance for long axes of double-cell nanopillars in the crystalline state.
• Figure 3: Chaotic encryption flowchart. The process starts by preprocessing the image and initializing the Logistic and Chen systems to generate chaotic sequences $L$, $X$, $Y$, $Z$, and $M$. The plaintext is encoded into nucleotide sequences (A, C, G, T) using DNA rules guided by $X$, then undergoes DNA operations with $L$ determined by $Z$. The results are decoded via rules from $M$ to form the initial ciphertext, which is further converted into binary and reorganized into a $3 \times 3$ matrix to produce the final QR ciphertext with embedded keys.
• Figure 4: Simulation results for the dual near-field and dual far-field channels. (a), (b) Under x-polarized illumination, near-field and far-field images. (d), (e) Under y-polarized illumination, near-field and far-field images. (c), (f) decrypted reconstructions from the near-field QR images.
• Figure 5: Chaos encryption performance evaluation. (a), (b) Cropping attacks performed on the encrypted QR code images. (c), (d) Decryption results from the cropped ciphertext. (e), (g) The adjacent pixel correlation analysis before and after encryption respectively. (f), (h) Grayscale histograms before and after encryption respectively. (i), (j) Comparative performance against Arnold encryption, including signal-to-noise ratio under cropping attacks and three-dimensional adjacent pixel correlation measurements across horizontal, vertical and diagonal directions.