The chosen meta-atom structure utilized TiO₂ as the medium due to its widespread availability in the visible-light band, specifically in the range of 400?700 nm, where it exhibits high transmittance. In addition, the optical loss was minimal, and the relevant metasurface fabrication technology has been fully developed. The meta-atom structure was arranged periodically, and the transmission phase was tailored by adjusting the diameter of the meta-atoms according to the phase theory. Subsequently, the metasurface phase was carefully controlled, and phase compensation corresponding to different positions was implemented to achieve the dispersion function of the metasurface. Our design ultimately chose a square shape as the substrate for the unit structure and a cylindrical shape with spatial symmetry for the meta-atoms. During the design process, increasing the height of the meta-atoms augmented the corresponding phase change, albeit at the expense of increased fabrication complexity. To realize a sufficient phase change while simplifying processing, striking a balance between the height of the meta-atoms and the desired phase became crucial. Due to the constraints of current fabrication technology, we opted for a uniform height for all the meta-atoms within the metasurface. Utilizing the finite-difference time-domain (FDTD) simulation method, we designed the geometric parameters of the meta-atoms, allowing us to obtain the transmittances and phases of the elemental structure with varying geometric parameters. Subsequently, we compiled a database incorporating data such as phase and transmittance.A numerical simulation method is utilized to simulate the single channel, as depicted in Fig. 3. Through a ray-tracing software simulation, we project the input image onto the position of the metasurface. To replace the designed metasurface, we opt for a combination of a grating and a mask layer. The spectral system dispersion results in a 12 μm length span across the bandwidth in the range of 400?700 nm. Additionally, we obtain the full spectral images by simulation (Fig. 4). In conclusion, a genetic algorithm is used to determine the light intensity coefficient of each channel to reconstruct the normalized spectrum (Fig. 5). The fitting results demonstrate the practical applicability of the proposed spectral imaging system.During the spectral reconstruction stage, we used a genetic algorithm, optimized over 30 generations, to obtain the normalized spectra. Through inversion on multiple multispectral datasets, all results exhibit perfect alignment with real spectra, achieving a spectral resolution of 30 nm. In summary, the integration of a spectroscopic metasurface into a charge-coupled device camera directly forms a spectral imaging system immune to the polarization state of incident light. This enhancement significantly broadens the applicability of the system. Furthermore, a high spectral resolution can be achieved by introducing different metasurface parameters. While our simulation verification is limited to the visible range, the design principles and methods of this system can be extrapolated to other bands, such as the near-infrared region.

Traditional spectral imaging systems rely on bulky optical components to achieve high spectral resolution, which pose challenges for miniaturization and portability. However, as a novel subwavelength artificial optical platform, a metasurface offers multi-degree-of-freedom control over light, including amplitude, phase, polarization, wavelength, and orbital angular momentum. Consequently, metasurfaces have emerged as an impressive advancement in the optical field leading to the development of numerous optical systems based on metasurfaces. These systems have found applications in various optical fields, such as imaging, holography, optical encryption, and quantum information. The planar structure of the metasurface enables systems to be small and light, providing a new solution to issues commonly encountered in traditional spectral imaging systems, such as large size, complex structure, limited functionality, and high cost.

We utilized TiO₂ to design the metasurface structure and successfully implemented a spectroscopic metasurface using the transmission phase principle. Our spectral imaging system comprised 10 channels, each offering a spectral resolution of 30 nm. Alongside the spectroscopic metasurface as the primary component, the system relied on a genetic algorithm to rebuild the spectral intensity.

The fixed meta-atom structure has a height (H) of 1000 nm and a period (p) of 220 nm. By maintaining H and p constants while varying the radius of the unit cell, we obtain corresponding phase distributions and transmittances for different radii. The nanopillar radius ranges from 80 to 180 nm, ensuring comprehensive 2π phase coverage and high transmission (Fig. 2).

We designed a spectroscopic metasurface utilizing the transmission phase principle, employing TiO₂ as the material. The working wavelength is selected in the range of 400?700 nm, and a dispersion-type spectral imaging system is established. The FDTD method is utilized to optimize the meta-atom diameter and establish the parameter values along with the corresponding arrangement of the metasurface. Subsequently, 11 spectral channels are selected as outputs within the visible range of 400?700 nm. To validate the performance of the spectral imaging system, we conducted full-image simulations using ray-tracing software.