Pixel-level high-efficiency metasurface spectrum sorting device for multispectral imaging and design method

Abstract

The invention discloses a pixel-level high-efficiency metasurface spectrum sorting device for multispectral imaging and a design method. The device comprises a substrate, a metasurface filter and a metasurface optical router which are sequentially arranged in the direction of an optical path. And the metasurface filter structure and the metasurface optical router structure are subjected to optimization design respectively. According to the invention, the Fabry-Perot resonant cavity which is wide in free spectral range, small in full width at half maximum at a peak wavelength and high in transmittance and the free-shaped metasurface optical router structure which is high in routing efficiency are creatively combined, so that a pixel-level high-efficiency metasurface spectrum sorting structure is realized, and low crosstalk and high efficiency are realized. According to the invention, the contradiction that pixel-level integration and high imaging quality of a traditional multispectral imaging structure cannot be achieved at the same time is solved, and a novel structure and a novel design method are provided for pixel-level high-efficiency detection for multispectral imaging.

Description

Technical Field

[0001] The present invention relates to the fields of optical technology and integrated photonics technology, and in particular to a pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging and a design method thereof. Background Art

[0002] Traditional multispectral imaging architectures include a color filter array (CFA) for light separation and a microlens array (MLA) for light focusing. The CFA absorbs out-of-band light to achieve light separation, but exhibits high light transmission loss, making it difficult to operate in low-light conditions.

[0003] Pixel-level multispectral imaging has long been a key trend in image sensor development, but pixel shrinkage presents a host of thorny challenges. Smaller pixels reduce the photosensitive area, reducing the amount of collected light signals and increasing the impact of noise, leading to reduced image quality. The light-gathering ability of the microlenses is also weakened by pixel shrinkage, making it impossible to efficiently focus light onto the photosensitive pixels, and inter-pixel crosstalk becomes increasingly severe. Light interference between adjacent pixels significantly reduces image detail and color accuracy, making it impossible to simultaneously achieve both pixel-level imaging and high image quality.

[0004] As a new type of functional material, metasurfaces are composed of artificial micro-nanostructures distributed on their surfaces. This structural unit, which is close to the wavelength of light, enables unique interactions with the surrounding light field, exhibiting unusual light modulation effects. Traditional metasurface designs rely on periodically arranged regular shapes, which have relatively low degrees of freedom and make it difficult to achieve many complex functional requirements. With the gradual maturity of micro-nanostructure integration processes, some irregular-shaped structures can also be processed and manufactured, and topology optimization has begun to become a mainstream method for optical device design. It uses optimization targets to reverse-engineer free-form structures that meet functional requirements, broadening the degrees of freedom and allowing metasurfaces to play a greater role in optical design. Summary of the Invention

[0005] This invention proposes a high-efficiency, pixel-level metasurface spectral sorting device and design method for multispectral imaging. This device addresses the need for pixel-level imaging and high-quality image quality in multispectral sensors, and addresses the inability of existing sensor structures to achieve both multispectral imaging, a high signal-to-noise ratio, and pixel-level detection. The device utilizes a Fabry-Perot (FP) resonant cavity with a broadband free spectral range to achieve simultaneous multi-wavelength filtering, avoiding the light loss associated with single-wavelength filtering. The peak wavelength of the filtered light has a minimal full width at half maximum (FWHM) and extremely high transmittance, enhancing the signal-to-noise ratio. A metasurface optical router structure then separates the incident light of different wavelengths filtered by the FP resonant cavity and deflects it to different independent pixel regions, achieving efficient light separation and focusing, further enhancing the signal-to-noise ratio of a single pixel channel. The cascaded design avoids the light loss associated with traditional filters and enhances the signal-to-noise ratio. This approach meets the requirements for multispectral channels, high signal-to-noise ratio, and pixel-level detection. It solves the key technical difficulties in further realizing pixel-level integration of traditional multispectral sensor structures and provides a structural and design idea for pixel-level detectors for multispectral imaging.

[0006] The specific technical solution adopted in the present invention is:

[0007] A pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging, comprising:

[0008] A substrate layer that is transparent in the target wavelength band;

[0009] a bottom layer of a reflector for a Fabry-Perot cavity connected to the substrate layer;

[0010] a metasurface phase shifter for a Fabry-Perot cavity connected to the bottom layer of the reflector;

[0011] a top layer of a reflective mirror for a Fabry-Perot cavity connected to the metasurface phase shifter;

[0012] a transition layer connected to the top layer of the reflector;

[0013] A metasurface optical router connected to the transition layer.

[0014] This invention innovatively combines a Fabry-Perot resonant cavity with a wide free spectral range, a small full width at half maximum at its peak wavelength, and high transmittance with a free-form metasurface optical router structure with high routing efficiency, resulting in a high-efficiency, pixel-level metasurface spectral sorting structure that combines low crosstalk with high efficiency. This resolves the dilemma of achieving both pixel-level integration and high-quality imaging in traditional multispectral imaging architectures, providing a new structure and design approach for high-efficiency, pixel-level detection for multispectral imaging.

[0015] The substrate layer is a substrate transparent in the target wavelength band, such as a silicon dioxide substrate in the visible light near infrared band;

[0016] The bottom layer of the reflector can efficiently reflect light in the target wavelength band. Furthermore, the bottom layer of the reflector adopts a Bragg reflector structure, including: a pair of high and low refractive index media connected to the substrate layer;

[0017] The high- and low-refractive-index medium pair comprises: a high-refractive-index medium layer, which can be made of silicon, germanium, titanium oxide, etc. according to the target wavelength; and a low-refractive-index medium layer connected to the high-refractive-index medium layer, which can be made of silicon oxide, silicon nitride, etc. according to the target wavelength.

[0018] The metasurface phase shifter can change the relative refractive index of the layer, filter out specific wavelengths, and achieve simultaneous filtering of multiple wavelengths. Furthermore, the metasurface phase shifter is composed of square metasurface nanopillars;

[0019] The top layer of the reflector can efficiently reflect light in the target wavelength band. Furthermore, the top layer of the reflector adopts a Bragg reflector structure, including: a pair of high and low refractive index media connected to the metasurface phase shifter;

[0020] The high- and low-refractive-index medium pair comprises: a high-refractive-index medium layer, which can be made of silicon, germanium, titanium oxide, etc. according to the target wavelength; and a low-refractive-index medium layer connected to the high-refractive-index medium layer, which can be made of silicon oxide, silicon nitride, etc. according to the target wavelength.

[0021] The Fabry-Perot resonant cavity filter structure, comprised of the bottom reflector layer, metasurface phase shifter, and top reflector layer, is designed as a multi-passband structure corresponding to the detector pixel array, such as a four-passband structure corresponding to a 2×2 pixel array and a nine-passband structure corresponding to a 3×3 pixel array. Furthermore, the Fabry-Perot resonant cavity filter structure is a four-passband filter structure corresponding to the size of a 2×2 detector pixel array, allowing it to be combined with an optical routing device to route light of different wavelengths to different independent pixels for detection.

[0022] The metasurface optical router, which is the same size as a Fabry-Perot resonant cavity filter, routes incident light of different wavelengths within the filter's multi-peak transmission spectrum to corresponding pixel locations, significantly increasing the light flux per pixel. Materials such as silicon oxide and silicon nitride can be selected based on the target wavelength. Furthermore, the metasurface optical router corresponds to a 2×2 detector pixel array and comprises multiple square metasurface nanopillars with a specific side length of 20 nm. These nanopillars are arranged in a structure optimized by the metasurface optical router design to form the metasurface optical router.

[0023] A pixel-level, high-efficiency metasurface spectral sorting structure for multispectral imaging specifically comprises a substrate, a metasurface filter, and a metasurface optical router, sequentially arranged along an optical path. The metasurface filter comprises a Bragg reflector bottom layer, a metasurface phase shifter, and a Bragg reflector top layer, sequentially arranged along the optical path. Furthermore, the substrate layer, the bottom layer of the Bragg reflector, the metasurface phase shifter, the top layer of the Bragg reflector, the transition layer, and the metasurface optical router are sequentially arranged along the optical path, enabling pixel-level alignment and mounting or monolithic integration with an imaging device, with the optical path incident from the direction of the substrate transparent to the target wavelength band.

[0024] A pixel-level high-efficiency metasurface spectral sorting structure for multispectral imaging, the specific technical principles and implementation details are as follows. In order to achieve selective response to wavelength, the metasurface filter is first designed. In order to avoid the large amount of light absorption by traditional filters and ensure the surface flatness of the device structure, the present invention selects a filtering metasurface based on a Bragg reflector to complete the filtering function. Based on the principle of the Bragg reflector, the thickness of the selected dielectric material is calculated to meet the high reflection requirements within the wavelength channel. The resonance peak is adjusted by using the intermediate layer metasurface nanocolumns, and the required specific multi-peak resonant filtering is achieved by adjusting the side length of the nanocolumns and the period of the metasurface.

[0025] To achieve pixel-level imaging in multispectral sensors, the different wavelengths of the filter need to be routed to the individual pixel areas of the detector as efficiently as possible. Traditional metasurface structures achieve specific optical properties through periodic arrangements of nanopillars. However, due to the miniaturization of pixel size, it is difficult to achieve complex optical routing functions within a small area with a simple arrangement of a few periods. To explore more design possibilities while maximizing optical routing efficiency, this paper uses a free-form metasurface based on topology optimization to achieve the target design requirements.

[0026] Based on the research on metasurface filters and metasurface optical routers, we further designed a high-efficiency metasurface spectral sorting structure at the pixel level to achieve pixel-level detection and high-quality imaging. Considering that the pixel size currently used in the near-infrared detector industry is generally around 10μm, and the minimum can reach 5μm. The present invention designs a pixel size of 2μm, an imaging wavelength of 1.2-1.6μm, and 16 imaging channels, which is composed of four four-channel metasurface spectral sorting structures. The size of a single four-channel metasurface spectral sorting structure is 4×4μm. 2 The four structures are arranged in a 2×2 square, with an overall period of 8μm in both the X and Y directions. Using a monolithic integration approach, two alignment lithography steps are used to combine the two functional devices, resulting in a metasurface spectral sorting structure that can be aligned at the pixel level.

[0027] A design method for a pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging includes the following steps:

[0028] 1) Select the operating band, pixel size, and required resonant wavelength of a pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging;

[0029] 2) Selecting a Fabry-Perot resonant cavity filter as a four-passband filter structure with a 2×2 detector pixel array size;

[0030] Selecting the structure, material, and thickness of the bottom and top layers of the reflector, and determining the distance between the bottom and top layers of the reflector, and controlling the free spectral range of the Fabry-Perot cavity through the distance so that the transmission peak periodically appears and multi-peak transmission is achieved;

[0031] Square metasurface nanopillars are selected as metasurface phase shifters. The side length of the nanopillars and the period of the metasurface are adjusted to adjust the equivalent refractive index of the nanopillar metasurface, complete the adjustment of the filtering wavelength, and realize multi-wavelength simultaneous filtering.

[0032] 3) Select the size of the metasurface optical router as 2×2 pixels and the focal length of the metasurface optical router;

[0033] 4) Select the dielectric material used in the metasurface optical router and optimize the binary material distribution to obtain the optimized structure of the metasurface optical router and complete the pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging.

[0034] 5) In step 4), the dielectric material used in the metasurface optical router is selected and the binary material distribution is optimized, specifically including:

[0035] The metasurface optical router after the conditions selected in step 3) is divided into regions, each corresponding to an independent pixel. An optimization objective function is selected to separate the incident light of different wavelengths filtered by the Fabry-Perot resonant cavity filter structure and deflect it to different regions. The transmittance of the light wavelength corresponding to each region is optimized and maximized, making the transmittance of the light wavelength corresponding to each region uniform, thereby achieving efficient light separation and single-pixel regional focusing.

[0036] The dielectric material used in the metasurface optical router is selected, and the density projection method is used to randomly initialize the material distribution of the medium. Then, the projection intensity is gradually increased in the iterative process, and finally a free-form binary material distribution that meets the required functions is obtained.

[0037] Compared with the prior art, the present invention has the following advantages:

[0038] 1. Conventional multispectral imaging sensor structures that want to achieve pixel-level detection will face the problems of reduced light transmittance due to reduced pixel area and increased crosstalk between adjacent pixels, which weakens the imaging quality. The structure proposed in the present invention combines the advantages of a high-transmittance, low-crosstalk metasurface filter and an efficient free-form metasurface optical routing device by cascading them. The optical routing device realizes the function of single-pixel focusing and significantly improves the light transmittance, and the filtering device significantly reduces crosstalk, providing ideas for further integration of multispectral imaging sensors.

[0039] Second, traditional filter arrays absorb out-of-band light to achieve optical separation. Sixteen-channel filter arrays have an absorption rate of up to 90%, resulting in high optical transmission loss. The filter structure based on the FP resonant cavity achieves multi-band filtering through resonance between dielectrics, resulting in a broadband free spectral range. By adjusting the length, width, and period of the metasurface structure, the effective refractive index of the resonant structure is altered, thereby obtaining periodic resonant peaks at specific wavelengths. This enables simultaneous filtering of multiple wavelengths, significantly improving light utilization efficiency. Furthermore, crosstalk between specific wavelengths is minimized, facilitating subsequent wavelength channel separation using optical routing devices.

[0040] 3. The optical routing metasurface structure based on topology optimization of the present invention has a free shape, and has a larger search space and higher routing efficiency than the traditional metasurface based on nanopillar arrangement, which enables efficient optical routing at small pixel sizes, further improving the luminous flux and integrability of the device.

[0041] Fourth, the present invention is compatible with semiconductor technology, and the overall structure has the characteristics of surface flatness, miniaturization and lightness, and can be manufactured on a large scale. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic diagram of the three-dimensional structure of the pixel-level high-efficiency metasurface spectral sorting structure provided by an embodiment of the present invention.

[0043] Figure 2 It is a schematic diagram of the three-dimensional structure of the metasurface filter structure provided by an embodiment of the present invention.

[0044] Figure 3 This is a display of the optical transmission spectrum of the metasurface filter in an embodiment of the present invention.

[0045] Figure 4 It is a structural and functional schematic diagram of the metasurface optical router structure provided by an embodiment of the present invention.

[0046] Figure 5 This is an optimization flow chart of the metasurface optical router in an embodiment of the present invention.

[0047] Figure 6This is a display of the optical transmission spectrum of the metasurface optical router in an embodiment of the present invention.

[0048] Figure 7 This is a display of the optical transmission spectrum of the pixel-level high-efficiency metasurface spectral sorting structure in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The following describes a pixel-level high-efficiency metasurface spectral sorting structure proposed according to an embodiment of the present invention with reference to the accompanying drawings, and further elaborates on the principle design and technical effects of the present invention.

[0050] Figure 1 Schematic diagram of the three-dimensional structure of the pixel-level high-efficiency metasurface spectral sorting structure of the present invention.

[0051] A pixel-level high-efficiency metasurface spectral sorting structure for multispectral imaging, comprising:

[0052] A substrate layer 1 transparent to the target band, a bottom layer 2 of a reflector for a Fabry-Perot cavity connected to the substrate layer, a metasurface phase shifter 3 for a Fabry-Perot cavity connected to the bottom layer of the reflector, a top layer 4 of a reflector for a Fabry-Perot cavity connected to the metasurface phase shifter, a transition layer 5 connected to the top layer of the reflector, and a metasurface optical router 6 connected to the transition layer.

[0053] Figure 1 The bottom layer 2 of the reflector adopts a Bragg reflector structure. The Bragg reflector bottom layer 2 includes: a pair of high and low refractive index media connected to the substrate layer. The high and low refractive index media pair includes: a high refractive index medium layer 21, which can be selected from materials such as silicon, germanium, and titanium oxide according to the target band; and a low refractive index medium layer 22 connected to the high refractive index medium layer, which can be selected from materials such as silicon oxide and silicon nitride according to the target band.

[0054] Figure 1 The top layer 4 of the reflector adopts a Bragg reflector structure. The top layer 4 of the Bragg reflector includes: a high-refractive-index medium pair connected to the metasurface phase shifter. The high-refractive-index medium pair includes: a high-refractive-index medium layer 41, which can be selected from materials such as silicon, germanium, and titanium oxide according to the target band; and a low-refractive-index medium layer 42 connected to the high-refractive-index medium layer, which can be selected from materials such as silicon oxide and silicon nitride according to the target band.

[0055] The pixel-level high-efficiency metasurface spectral sorting structure generally includes: a substrate, a metasurface filter, and a metasurface optical router arranged in sequence along the optical path. The schematic diagram of the three-dimensional structure of the metasurface filter is shown in the figure. Figure 2 , including: a Bragg reflector bottom layer-2, a metasurface phase shifter-3 and a Bragg reflector top layer-4 arranged in sequence along the optical path direction.

[0056] The specific parameter design principles and optical performance of the metasurface filter are as follows:

[0057] The present invention selects a detection band of 1.2-1.6 μm, which is divided into 16 channels in total and is implemented using four filters with the same height but different metasurface side lengths. The specific structural design steps are as follows.

[0058] Step 1: Select the materials for the two layers of the distributed Bragg reflector. Since the wavelength is near-infrared, silicon and silicon dioxide, commonly used in the near-infrared band, are selected. The Bragg reflector layer primarily serves as a reflector. Based on the principle of destructive interference, the layer thickness d of the medium is calculated using formula (1), where n represents the refractive index of the medium and λ is the center wavelength of the band.

[0059]

[0060] Calculations determined the silicon thickness to be 90nm and the silicon dioxide thickness to be 210nm. Two pairs of dielectric layers were chosen as the Bragg reflectors for the structure. The FP cavity height was chosen to be 5022nm to facilitate the generation of interference peaks corresponding to higher-order wavelengths, enabling multi-wavelength filtering.

[0061] Step 2: Select the height of the middle metasurface layer as 530 nm, adjust the side length and period of the metasurface to change the effective refractive index n of formula (2) to achieve filtering of a specific peak corresponding to the order m, where m represents the order (m=1,...,N), n represents the refractive index of the medium constituting the FP cavity, and d represents the length of the FP cavity. It represents the overall reflection phase (the overall understanding is that the light propagates back and forth once in the FP cavity, and the total phase changes by an integer multiple of 2π), which meets the constructive interference condition.

[0062]

[0063] Based on design requirements and practical structural constraints, this paper selected a 400nm period as the period of the metasurface unit structure. The nanopillar side lengths ranged from 50nm to 350nm. Using lumen FDTD to establish a simulation structure and scan the nanopillar side length parameters, four suitable sets of parameters were selected to form four filter groups, each of which achieved four-channel filtering, and the combined filter group achieved sixteen-channel filtering. The four sets of parameters and their corresponding resonant wavelengths are shown in Table 1.

[0064] Table 1 Four sets of parameters of the bandpass filter and their corresponding resonant wavelengths

[0065]

[0066] The transmission spectra obtained by simulation corresponding to the four sets of parameters are shown in Figure 3Compared with the maximum transmittance of one-sixteenth of traditional filters, it avoids single-wavelength filtering loss and greatly improves the light flux.

[0067] The three-dimensional structure and functional diagram of the metasurface optical router is shown in the figure. Figure 4 , the specific parameter design principles and optical performance are as follows:

[0068] In order to explore a larger design space, improve routing efficiency and achieve faster optimization convergence, the present invention selects gradient-based topology optimization for optical routing device design. The specific structural design steps are as follows.

[0069] Step 1: First, the optimized objective function (FOM) is selected so that the transmittance peaks (T1, T2, T3, T4) corresponding to the four wavelengths are as large as possible and relatively average, which is expressed by formula (3).

[0070]

[0071] Step 2: Divide the design area into tiny pixels. This design uses 20nm. Divide the design domain into a 201*201 matrix. The size of each matrix element is 0 or 1, representing that the material in this area is air, and its dielectric constant is recorded as ε min , or silicon nanopillars, whose dielectric constant is denoted as ε max First, randomize the initial matrix to a continuously varying value between 0 and 1, representing a continuous refractive index density distribution in the initial structure. This value is called the material density ρ. The dielectric constant ε distribution of the material is calculated using the formula (4).

[0072] ε=ε min +ρ(ε max -ε min ) (4)

[0073] Based on the principles of the adjoint method, forward and adjoint simulations of the structure were performed using Lumerical FDTD, a commercial micro-nano optical simulation software. The material gradients were obtained by multiplying the electric fields of the design region obtained from the two simulations. These material gradients were then used to update the structure using the gradient descent method. The Adam optimizer was selected as the gradient descent optimizer in this paper. This optimizer simultaneously calculates the velocity and momentum of the gradient, preserving the historical information of the gradient. This makes the gradient descent direction more precise and, to a certain extent, avoids falling into local optima.

[0074] During the optimization process, the density is first blurred and then projected using the density projection method. A cone filter ω(x) is used to blur the initial density, and the blur radius R is selected to be 100nm to prevent the generation of isolated structures that are incompatible with the process. In formula (5), N is the part within the circle with a radius of R and a center of x0. β and η in formula (7) are threshold parameters used to adjust the intensity of the projection, where η is generally set to 0.5. The value of β in formula (7) is gradually multiplied to gradually binarize the continuous refractive index distribution structure. ρ is the initial continuous density distribution, ρ' is the density distribution after filtering with the cone filter, and ρ is the final binarized density distribution, which follows a 0-1 distribution.

[0075]

[0076] ρ'=ω(x)*ρ (6)

[0077]

[0078] After reaching the set number of iterations, the update is stopped and the final structure is obtained. The optimization process is as follows: Figure 5 As shown. Lumerical FDTD is used to simulate the optimized device to verify its optical performance. Its optical transmittance curve is as shown in Figure 6 As shown (taking w=277nm as an example).

[0079] After designing the metasurfaces with filtering and optical routing functions respectively, the two are connected through a transition layer to obtain a pixel-level high-efficiency metasurface spectral sorting structure. Simulation is performed and the optical performance is analyzed using lumenical FDTD software. The optical transmission spectrum of the simulated pixel-level high-efficiency metasurface spectral sorting structure is as follows: Figure 7 shown.

[0080] Calculation shows that the average transmittance of the sixteen wavelengths is 0.587, which far exceeds the luminous flux of a sixteen-channel ordinary filter. In addition, the crosstalk between the channels is very small, and the maximum transmittance of the non-main channel is no more than 0.1, which is expected to achieve further integration of multi-spectral imaging detectors.

Claims

1. A pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging, characterized in that: include: substrate layer; a bottom layer of a reflector for a Fabry-Perot cavity connected to the substrate layer; a metasurface phase shifter for a Fabry-Perot cavity connected to the bottom layer of the reflector; a top layer of a reflective mirror for a Fabry-Perot cavity connected to the metasurface phase shifter; a transition layer connected to the top layer of the reflector; A metasurface optical router connected to the transition layer.

2. The pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging according to claim 1, characterized in that: The substrate layer is a substrate that is transparent in the target wavelength band.

3. The pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging according to claim 1, characterized in that: The bottom layer of the reflector reflects light of the target wavelength band; The top layer of the reflector reflects light in the target wavelength band.

4. The pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging according to claim 1, characterized in that: The metasurface phase shifter is composed of a subwavelength periodic structure.

5. The pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging according to claim 3 or 4, characterized in that: The bottom layer of the reflector, the metasurface phase shifter and the top layer of the reflector together constitute a Fabry-Perot resonant cavity filter structure, and the Fabry-Perot resonant cavity filter structure is designed as a multi-band structure corresponding to the detector pixel array.

6. The pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging according to claim 5, characterized in that: The size of the metasurface optical router is the same as the structure of the Fabry-Perot resonant cavity filter.

7. The pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging according to claim 1, characterized in that: The transition layer is arranged between the top layer of the Bragg reflector and the metasurface optical router for transition connection, and the transition layer is a dielectric material transparent in the target wavelength band.

8. The pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging according to claim 1, characterized in that: The substrate layer, the bottom layer of the reflector, the metasurface phase shifter, the top layer of the reflector, the transition layer and the metasurface optical router are arranged in sequence along the light path direction.

9. The design method of a pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging according to any one of claims 1 to 8, characterized in that: The following steps are involved: 1) Select the operating band, pixel size, and required resonant wavelength of a pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging; 2) The size and number of passbands of the detector pixel array corresponding to the Fabry-Perot resonant cavity filter structure; Selecting the structure, material and thickness of the bottom and top layers of the reflector, and determining the distance between the bottom and top layers of the reflector; Select the subwavelength periodic structure that makes up the metasurface phase shifter, adjust the structural parameters to adjust the equivalent refractive index of the metasurface phase shifter layer, and complete the design of the metasurface phase shifter; 3) Select the size of the metasurface optical router and the focal length of the metasurface optical router; 4) Select the dielectric material used in the metasurface optical router and optimize the binary material distribution to obtain the optimized structure of the metasurface optical router and complete the pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging.

10. The design method of a pixel-level high-efficiency metasurface spectral sorting device for multispectral imaging according to claim 9, characterized in that: In step 4), optimizing the binary material distribution specifically includes: After the conditions are selected in step 3), the metasurface optical router is divided into regions, each region corresponding to an independent pixel. An optimization objective function is selected to separate the incident light of different wavelengths filtered by the Fabry-Perot resonant cavity filter structure and deflect it to different regions. The transmittance of the light wavelength corresponding to each region is optimized and made maximized and uniform.

Images