Metasurface dual-image encryption device and design method thereof

Abstract

The application provides a metasurface double-image encryption device and a design method thereof. Three groups of geometric parameters of nano-brick unit cells with high polarization conversion efficiency and corresponding different structural colors at a working wavelength are designed first. Based on the principle of single-pixel Fourier imaging, two target images are subjected to non-uniform Fourier sampling and four-step phase shift processing to generate respective bucket signals. The bucket signals corresponding to the two target images are processed and spliced into a composite encryption data matrix. The values of the composite encryption data matrix are split into digits, and first and second channel images are generated through structural color coding and gray scale coding, respectively. A third channel password matrix image recording the mapping relationship is generated. Three-channel metasurface arrays are designed and prepared according to the three channel images and the three groups of geometric parameters of the nano-brick unit cells with different structural colors. The application realizes efficient encryption and decryption of two images by using metasurfaces in combination with a single-pixel Fourier imaging algorithm.

Description

A metasurface dual-image encryption device and its design method Technical Field

[0001] This invention relates to the fields of micro-nano optics and optical information encryption technology, specifically a metasurface dual-image encryption device and its design method. Background Technology

[0002] In the current field of optical information encryption, traditional methods mainly rely on complex algorithmic encryption layers, which generally suffer from inherent limitations such as high computational resource consumption, large optical system size, and low integration, making it difficult to meet the urgent needs of modern information security for efficient and compact encryption schemes. Metasurfaces, as two-dimensional metamaterials, possess flexible light field manipulation characteristics, providing a new development path for optical encryption. Currently, there are three main types of information encryption methods based on metasurfaces:

[0003] The first type utilizes the multidimensional optical field manipulation capabilities of metasurfaces to construct a key space using optical parameters such as incident angle, wavelength, polarization, and orbital angular momentum mode, thereby increasing encryption complexity. This type of method has limited key dimensions and complexity, and when multiple optical parameters are manipulated in parallel, the modulation responses between different parameters are prone to coupling or crosstalk, leading to mixed information in each encryption channel and reducing the fidelity of the decrypted image.

[0004] The second category is high-dimensional optical field encryption methods based on cascaded metasurfaces. However, their practical application is limited by the current level of micro-nano fabrication technology, making it difficult to achieve large-scale, low-cost, and precise alignment of multiple metasurfaces. Furthermore, cascaded structures are sensitive to environmental vibrations and temperature changes, easily causing deviations in optical field modulation, leading to decryption failure. Simultaneously, the reflection / transmission losses present in each metasurface layer accumulate after cascading, resulting in insufficient final output light intensity and a reduced signal-to-noise ratio, making the decrypted image difficult to identify.

[0005] The third category combines computational imaging with metasurface encryption methods. Key images (such as QR codes or specific icons) are directly encoded into the metasurface structure, making it easily observable by optical systems or allowing feature extraction via image recognition technology, thus posing a potential risk of information leakage. The fourth category is based on dynamically controlled metasurface encryption methods. For example, by controlling the hydrogenation / dehydrogenation process of plasma nanostructures, the helicity of light, hydrogen, and oxygen, and the reaction duration are used as multiple keys to encrypt the metasurface. However, such dynamic metasurfaces that generate chemical reactions by introducing gases require precise control of environmental parameters such as temperature and concentration, which is often difficult to meet in practical applications.

[0006] In summary, existing encryption schemes based on metasurfaces have certain limitations. For example, some encryption methods that rely solely on the control of a single optical parameter hide high-resolution images within the polarization distribution of light. However, due to the limited number of working channels, the hidden information is vulnerable to theft by attackers, resulting in a high risk of information leakage. In contrast, encryption technologies based on multi-channel metasurfaces can store and encrypt information separately in different operating modes, increasing information storage capacity and significantly enhancing system security. However, some existing multi-channel metasurfaces still face several challenges when processing complex information: First, when processing complex information larger than 8 bits, electromagnetic crosstalk between channels degrades image quality, often requiring significantly increased processing precision to suppress crosstalk, thus significantly increasing manufacturing costs. Second, information stored in plaintext within specific channels is susceptible to direct leakage due to channel attacks. Third, encryption methods are limited by multi-image overlay mechanisms, introducing redundant information and reducing the storage density of effective information. Furthermore, some methods require significantly increased manufacturing costs to improve processing precision and reduce crosstalk between channels when processing complex information larger than 8 bits. Summary of the Invention

[0007] In existing metasurface-based encryption technologies, the lack of structural optimization for the Mie resonance characteristics of unit-cell nanostructures leads to low polarization conversion efficiency at specific operating wavelengths, affecting the brightness of the generated grayscale image and the fidelity of holographic image reconstruction. Furthermore, the traditional single-pixel imaging method fails to effectively utilize the frequency domain sampling and decoupling characteristics of the Fourier domain, making it susceptible to channel crosstalk during decryption, resulting in a low Structure Similarity Index (SSIM) and limiting the capacity and reconstruction quality of metasurfaces in information encryption. Currently, there is a lack of effective techniques to improve the polarization conversion efficiency of metasurfaces while combining single-pixel Fourier imaging algorithms to achieve efficient encryption and decryption of two images, simultaneously obtaining a high SSIM value. This invention provides a metasurface dual-image encryption device and its design method to address the problems of low polarization conversion efficiency, difficulty in independently encrypting two images, and low SSIM values ​​in existing encryption technologies. This invention fully leverages the advantages of metasurfaces in information storage and combines Fourier imaging methods with the anti-interference characteristics in the frequency domain to improve the overall performance and reliability of the encryption system.

[0008] To achieve the above-mentioned technical objectives, the technical solution adopted by the present invention is as follows:

[0009] On one hand, the present invention provides a design method for a metasurface dual-image encryption device, comprising the following steps:

[0010] S1. Electromagnetic simulation software was used to scan and optimize the geometric parameters of the nanobrick unit cell, and three sets of geometric parameters with high polarization conversion efficiency and corresponding to different structural colors were selected under the set working wavelength.

[0011] S2, For the two target images to be encrypted, perform non-uniform Fourier sampling and four-step phase shift processing respectively to generate bucket signals for each target image;

[0012] S3, rotate, stitch and fill the bucket signals corresponding to the two target images to form a composite encrypted data matrix;

[0013] S4, split the values ​​in the composite encrypted data matrix, generate a first-channel color image and a second-channel grayscale image through structured color encoding and grayscale encoding respectively, and generate a third-channel codebook image containing color-value mapping relationship;

[0014] S5 is a three-channel metasurface array designed based on the first channel color image, the second channel grayscale image, the third channel codebook image, and three sets of geometric parameters corresponding to different structural colors under the selected working wavelength.

[0015] On the other hand, a metasurface dual-image encryption device is provided, designed using the aforementioned metasurface dual-image encryption device design method, comprising:

[0016] Substrate;

[0017] A nanostructure layer is formed on the substrate, which constitutes a three-channel metasurface array. The three-channel metasurface array is a metasurface array that can realize three-channel functions in the same array in physical space.

[0018] In existing metasurface-based encryption technologies, the lack of structural optimization for the Mie resonance characteristics of unit cell nanostructures leads to low polarization conversion efficiency at specific operating wavelengths, affecting the brightness of the generated grayscale image and the fidelity of holographic image reconstruction. Furthermore, the traditional single-pixel imaging method fails to effectively utilize the frequency domain sampling and decoupling characteristics of the Fourier domain, making it susceptible to channel crosstalk during decryption, resulting in a low Structure Similarity Index (SSIM) and limiting the capacity and reconstruction quality of metasurfaces in information encryption. Currently, there is a lack of effective technical means to improve the polarization conversion efficiency of metasurfaces while combining single-pixel Fourier imaging algorithms to achieve efficient encryption and decryption of two images, simultaneously obtaining a high SSIM. This invention provides a metasurface dual-image encryption device and its design method to solve the problems of low polarization conversion efficiency, difficulty in independently encrypting two images, and low SSIM in decrypted images in existing encryption technologies. This invention has the following technical effects:

[0019] This invention optimizes the unit cell structure of a three-channel metasurface to achieve a polarization conversion efficiency of at least 50% at a set operating wavelength (e.g., in the specific embodiment provided later, the operating wavelength is set to 530nm, and the polarization conversion efficiency at 530nm is ≥50%). Combined with a single-pixel Fourier imaging algorithm, it enables efficient encryption and decryption of two target images. This invention fully utilizes the controllability of metasurfaces across multiple optical parameters and the imaging advantages of the FSPI algorithm to address numerous challenges in existing optical information encryption, achieving efficient and secure encryption and decryption of two images, and obtaining a high Structural Similarity Index (SSIM) in the decrypted image. This scheme can significantly improve the SSIM value of the decrypted image while avoiding crosstalk between channels, without increasing the complexity of the metasurface and algorithm structure. In summary, this invention fully leverages the advantages of metasurfaces in information storage and combines the anti-interference characteristics of Fourier imaging in the frequency domain, greatly improving the overall performance and reliability of the encryption system. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0021] Figure 1 is a flowchart of a metasurface dual-image encryption device design method provided in one embodiment;

[0022] Figure 2 is a flowchart of the encryption process for a target image in one embodiment;

[0023] Figure 3 shows the polarization conversion efficiency and co-polarization efficiency of three nanobrick unit cells with different geometric parameters selected in an embodiment at a working wavelength of 530 nm. Figure 3(a) is a schematic diagram of the polarization conversion efficiency of the three nanobrick unit cells at a working wavelength of 530 nm; Figure 3(b) is a schematic diagram of the co-polarization efficiency of the three nanobrick unit cells at a working wavelength of 530 nm.

[0024] Figure 4 is a schematic diagram of single-pixel Fourier imaging of a target image in an embodiment;

[0025] Figure 5 is a schematic diagram of the barrel signal rotation splicing process in one embodiment;

[0026] Figure 6 is a schematic diagram of encoding a composite encrypted data matrix into a channel image in one embodiment, wherein Figure 6(a) is the composite encrypted data matrix; Figure 6(b) is the first channel color image obtained by structural color encoding; Figure 6(c) is the first channel color image after adding a positioning identifier; Figure 6(d) is the structural color encoding codebook; Figure 6(e) is the four-level grayscale encoding codebook; Figure 6(f) is the second channel grayscale image obtained by grayscale encoding; Figure 6(g) is the first channel color image after adding a positioning identifier and obfuscation information.

[0027] Figure 7 is a schematic diagram of the third channel codebook in one embodiment;

[0028] Figure 8 is a schematic diagram of the arrangement of a three-channel metasurface array in one embodiment, wherein Figure 8(a) is a schematic diagram of the fixed array physical layout obtained after step S5.1; Figure 8(b) is a schematic diagram after determining the azimuth angle.

[0029] Figure 9 is a schematic diagram of the structure of a nanobrick unit cell in one embodiment;

[0030] Figure 10 shows the target and result diagrams of a three-channel metasurface array simulated by FDTD in one embodiment, where Figure 10(a) is the three-color image of the target in channel one; Figure 10(b) is the grayscale image of the target in channel two; Figure 10(c) is the holographic image of the target in channel three; Figure 10(d) is the simulation result of channel one; Figure 10(e) is the simulation result of channel two; and Figure 10(f) is the simulation result of channel three.

[0031] Figure 11 is a schematic diagram of the channel design and image reconstruction results of a three-channel metasurface array in one embodiment. Figure 11(a) shows the simulation result of the channel one image corresponding to the three-channel metasurface irradiated with unpolarized broadband white light; Figure 11(b) shows the simulation result of the channel two image obtained by irradiating the three-channel metasurface with 530nm horizontally polarized light; Figure 11(c) shows the simulation result of the channel three holographic image obtained by irradiating the three-channel metasurface with 530nm left-handed circularly polarized light; Figure 11(d) shows the image to be encrypted; Figure 11(e) shows the decryption result of the dual images; Figure 11(f) shows the azimuth distribution map; Figure 11(g) shows a magnified view of the azimuth distribution.

[0032] Figure 12 shows the reconstruction results of the decrypted image in the security verification experiment, where Figure 12(a) is the reconstruction result of the correctly decrypted image and Figure 12(b) is the reconstruction result of the incorrectly decrypted image.

[0033] The labels in Figure 9 are as follows: 1. Substrate; 2. Nanobrick unit cell. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0035] Referring to Figure 1, one embodiment provides a design method for a metasurface dual-image encryption device, including:

[0036] S1. Electromagnetic simulation software was used to scan and optimize the geometric parameters of the nanobrick unit cell, and three sets of geometric parameters with high polarization conversion efficiency and corresponding to different structural colors were selected under the set working wavelength.

[0037] S2, For the two target images to be encrypted, perform non-uniform Fourier sampling and four-step phase shift processing respectively to generate bucket signals for each target image;

[0038] S3, rotate, stitch and fill the bucket signals corresponding to the two target images to form a composite encrypted data matrix;

[0039] S4, split the values ​​in the composite encrypted data matrix, generate a first-channel color image and a second-channel grayscale image through structured color encoding and grayscale encoding respectively, and generate a third-channel codebook image containing color-value mapping relationship;

[0040] S5 is a three-channel metasurface array designed based on the first channel color image, the second channel grayscale image, the third channel codebook image, and three sets of geometric parameters corresponding to different structural colors under the selected working wavelength.

[0041] Furthermore, in S1 of the above embodiment, the working wavelength and the height H and unit period C of the nanobrick unit cell are set, and the scanning range is set, including the scanning range of the long axis length L and the scanning range of the short axis length W of the nanobrick unit cell. Electromagnetic simulation software is used to screen out the three sets of geometric parameters with the highest polarization conversion efficiency under the set working wavelength within the scanning range.

[0042] As shown in Figure 2, in S2 of the above embodiment, non-uniform Fourier sampling and four-step phase shift processing are performed on the target image to generate the bucket signal of the target image, including the following steps:

[0043] S2.1 Perform a two-dimensional discrete Fourier forward transform on the target image to obtain the Fourier spectrum, center the Fourier frequency coordinates, and select key frequency points with concentrated energy in the frequency domain according to the set sampling path type to form a non-uniform Fourier sampling point set corresponding to the target image.

[0044] S2.2, For the Fourier frequency corresponding to each Fourier sampling point in the Fourier sampling point set, generate four sets of sinusoidal illumination light fields with different initial phases. The initial phases of the four sets of sinusoidal illumination patterns correspond to the set four-step phase shift phase sequence. By calculating the inner product of each sinusoidal illumination light field and the target image, the bucket signal of the four-step phase shift is simulated.

[0045] During image decryption, the three-channel metasurface is illuminated with broadband white light, horizontally polarized light of a set working wavelength, and left-handed circularly polarized light of a set working wavelength, respectively. The output images of the three channels are then read. Based on the color-value mapping relationship carried by the third-channel codebook image, the structural color and grayscale information in the first-channel color image and the second-channel grayscale image are decoded into bucket signal data matrices. The bucket signal data matrices are then inversely spliced ​​and subjected to inverse Fourier transform to reconstruct the two target images. Furthermore, the structural similarity index (SSIM) between the reconstructed target image and the original target image can be used as a fidelity evaluation function to assess the quality of the reconstructed target image.

[0046] The above embodiments first optimize the geometric parameters of the nanobrick unit cells. Nanobrick unit cells are fabricated using SOS (silicon-on-sapphire) material. Through parameter scanning, three sets of nanobrick unit cell geometric parameters with high polarization conversion efficiency and significant structural color differentiation at a set working wavelength (e.g., 530 nm) are selected. Then, a single-pixel Fourier imaging algorithm is used to achieve efficient encryption and decryption of two target images. This scheme can significantly improve the SSIM value of the decrypted image while avoiding inter-channel crosstalk, without increasing the complexity of the metasurface and algorithm structure.

[0047] Furthermore, in S1 of one embodiment, SOS (silicon-on-sapphire) material is selected, wherein crystalline silicon serves as the nanostructure layer and sapphire serves as the substrate. The geometric parameters of the nanobrick unit cell are scanned and optimized using CST STUDIO SUITE electromagnetic simulation software. The height H of the nanobrick unit cell is fixed at 200 nm, the unit period C is set to 300 nm, the working wavelength is set to 530 nm, and the scanning range is set as follows: the major axis length L scan range is 100-220 nm, and the minor axis length W scan range is 80-90 nm; the three sets of geometric parameters finally selected are:

[0048] The first set of geometric parameters (nano-Ⅰ): major axis length L1=112nm, minor axis length W1=82nm, polarization conversion efficiency at 530nm wavelength is 54%, corresponding to dark green;

[0049] The second set of geometric parameters (nano-Ⅱ): major axis length L2=170nm, minor axis length W2=82nm, polarization conversion efficiency at 530nm wavelength is 61%, corresponding to light green;

[0050] The third set of geometric parameters (nano-Ⅲ): major axis length L3=219.5nm, minor axis length W3=84.5nm, polarization conversion efficiency of 50% at a wavelength of 530nm, corresponding to yellow.

[0051] By optimizing the geometric parameters of the nanobrick unit cells of the three-channel metasurface array, a polarization conversion efficiency of no less than 50% (≥50% at 530nm working wavelength) is achieved, corresponding to three different structural colors, providing a physical basis for multi-channel information storage. The polarization conversion efficiency and co-polarization efficiency of the nanobrick unit cells with the above three different geometric parameters (nano-Ⅰ, nano-Ⅱ, nano-Ⅲ) at 530nm working wavelength are shown in Figure 3. Figure 3(a) is a schematic diagram of the polarization conversion efficiency of the three nanobrick unit cells at 530nm working wavelength; Figure 3(b) is a schematic diagram of the co-polarization efficiency of the three nanobrick unit cells at 530nm working wavelength, verifying the theoretical feasibility of using it as a physical carrier for multi-channel information encoding.

[0052] The encryption process of this invention is based on the principle of single-pixel Fourier imaging. By selective sampling and phase modulation in the Fourier domain, spatial image information is encoded into bucket signals, and then the image is restored through phase calculation and frequency domain reconstruction.

[0053] In the encryption process of the target image, the spatial resolution of the target image (such as M×N pixels) is first determined, and the core imaging parameters are set, including the phase sequence of four-step phase shift (usually 0°, 90°, 180°, 270°), the spectrum coverage ratio of Fourier sampling (i.e. the proportion of frequency domain points participating in sampling to the total frequency domain points), and the sampling path type (such as non-uniform paths such as spirals and circles that expand outward from the center of the frequency domain).

[0054] A two-dimensional discrete Fourier forward transform is performed on the target image to convert the spatial image into a Fourier spectrum in the frequency domain (containing coefficients at all frequency points), thus obtaining the Fourier spectrum of the target image. The Fourier frequency coordinates are centered, concentrating low-frequency components at the center and distributing high-frequency components around the periphery, conforming to the frequency domain energy distribution characteristics of natural images. The order and range of Fourier sampling are planned according to the set sampling path type (a spiral or circular non-uniform path extending outward from the center of the frequency domain): key frequency points with concentrated energy are preferentially selected in the frequency domain (primarily low-frequency points, while also considering some high-frequency points), forming a non-uniform Fourier sampling point set corresponding to the target image. Each Fourier sampling point in the Fourier sampling point set corresponds to a unique Fourier frequency. Includes horizontal frequency components With vertical frequency components The number of Fourier sampling points is determined by the spectral coverage ratio. Data compression is achieved by reducing redundant sampling, while improving imaging security. During reconstruction, inverse Fourier transform is used to convert the completed frequency domain spectrum back to the spatial domain image, which is used to reconstruct the target image from the acquired frequency domain information.

[0055] The mathematical expression for the two-dimensional discrete Fourier forward transform is shown below:

[0056] ; ;

[0057] Where M and N are the target image sizes. The target image in spatial coordinates grayscale value at that location The value ranges are 0~M-1 and 0~N-1, respectively. , Fourier frequencies The horizontal and vertical frequency components, , The range of values ​​for are respectively , , For frequency domain coefficients, It is the imaginary unit.

[0058] Analysis using the two-dimensional discrete Fourier transform shows that each Fourier sampling point corresponds to a frequency domain coefficient. Frequency domain coefficients Including real and imaginary parts, the generated corresponding Fourier frequencies When a sinusoidal illumination light field interacts with a target image, the intensity of the light field modulation is weighted by frequency domain coefficients. The amplitude is determined by it.

[0059] For each Fourier frequency corresponding to a Fourier sampling point in the Fourier sampling point set, four sets of sinusoidal illumination light fields with different initial phases are generated. The initial phases of the four sets of sinusoidal illumination patterns are related to a set four-step phase shift sequence. Correspondingly, the frequency of each sinusoidal lighting pattern perfectly matches the Fourier frequency of the current Fourier sampling point.

[0060] The spatial intensity distribution of the sinusoidal illumination field is defined by the following formula:

[0061] ;

[0062] in Fourier frequency The corresponding spatial intensity distribution of the string illumination field, This is a constant term, representing the DC component of the light intensity; Intensity modulation coefficient; and Spatial coordinates Fourier frequency , Directional components; The initial phase is given by a value of This corresponds to the set four-step phase shift sequence.

[0063] The bucket signal is obtained through simulation calculation using the following formula:

[0064] ;

[0065] in, The target image in spatial coordinates The grayscale value at that location.

[0066] Specifically, at Fourier frequency Below, after sinusoidal illumination fields with different initial phases are projected onto the target image, a bucket signal with four phase shifts is simulated, as follows:

[0067] ; ; ; .

[0068] Based on the bucket signals of the four-step phase shift, the Fourier spectral coefficients of each Fourier sampling point are calculated to form the encrypted data matrix corresponding to the target image. For a target image to be encrypted (M×N image), a full sampling, i.e., M×N Fourier illumination light field, is required to reconstruct the image. Since natural images are real values, their Fourier spectra have symmetry. Therefore, the total light intensity value of the target image in the four-step phase shift (i.e., the bucket signals of the target image in the four-step phase shift) is a four-dimensional matrix of size M×N×4, where the size of the non-zero value matrix is ​​(M / 4+1)×(N / 2+1)×4. Figure 4 shows a schematic diagram of single-pixel Fourier imaging of the target image, where black represents zero values ​​and white represents non-zero values.

[0069] Based on the four-step phase-shifted bucket signal, the real and imaginary parts of the Fourier coefficients, i.e., the Fourier spectrum, can be recovered, which is needed in the decryption and reconstruction process. Specifically, the Fourier spectrum coefficients of each Fourier sampling point are calculated using the following formula:

[0070] ;

[0071] in , , , The initial phases are respectively for The corresponding bucket signal at that time It is the imaginary unit.

[0072] In one embodiment, in the proposed S3, the bucket signals corresponding to the two target images are rotated, stitched together, and filled to form a composite encrypted data matrix, including:

[0073] Take the non-zero value matrix regions in the bucket signals of the four-step phase shift of the two target images, and convert the bucket signals of the four-step phase shift of the two target images into a gray-level matrix through rotation and stitching operations. The upper and lower parts of the gray-level matrix can be stitched together to form a near-circular shape. The left and right parts of the gray-level matrix are stitched together by filling a column of zero matrix to obtain the first matrix.

[0074] The first matrix is ​​spliced ​​vertically, and a row of zeros is filled between the upper and lower parts to obtain a composite encrypted data matrix. The upper part of the composite encrypted data matrix corresponds to the first target image to be encrypted, and the lower part corresponds to the second target image to be encrypted.

[0075] Specifically, the two target images to be encrypted are the first target image T1 and the second target image T2, both with a size of M×N, and both are normalized to the [0,1] grayscale space. Using the method described in the above embodiment, the four-step phase-shifted bucket signals (each with four dimensions) of the two target images are obtained. Since the effective value of the bucket signal changes with the phase of the illumination field, its projection on the complex plane forms a semi-circular trajectory. For the two target images, the non-zero value matrix region (the size of the non-zero value matrix region is (M / 4+1)×(N / 2+1)×4) in their four-step phase-shifted bucket signals (both with a size of M×N×4) is taken. Through rotation and stitching operations, the four-step phase-shifted bucket signals of the two target images are converted into a grayscale matrix. The upper and lower parts of this grayscale matrix can be stitched together to form a near-circular shape, and the left and right parts of this grayscale matrix are stitched together by filling a column of zero matrices to obtain the first matrix. Figure 5 shows a schematic diagram of the bucket signal rotation and stitching process. Furthermore, the first matrix obtained by splicing the non-zero value matrix regions in the four-step phase-shifted bucket signals of the two target images is spliced ​​along the vertical direction, and a row of zero matrix is ​​filled in the middle of the two first matrices in the vertical direction to obtain a composite encrypted data matrix, wherein the upper half corresponds to the first target image T1 to be encrypted and the lower half corresponds to the second target image T2 to be encrypted.

[0076] In S4, the non-zero values ​​in the composite encrypted data matrix are split into four digits. The first and second digits are structurally encoded using a structured color-coded cipherbook to obtain the first channel color image, which is a three-color image. This facilitates the subsequent design process where the first color pixel corresponds to the first set of geometric parameters, the second color pixel corresponds to the second set of geometric parameters, and the third color pixel corresponds to the third set of geometric parameters. The third and fourth digits are grayscale encoded using a four-dimensional grayscale cipherbook to obtain the second channel grayscale image. The third channel cipherbook image, containing the color-value mapping relationship, is designed based on the index information of the structured color-coded cipherbook and the four-dimensional grayscale cipherbook.

[0077] In one embodiment, the target images T1 and T2 to be encrypted are both M×N in size, with an initial size of 80×80. The size of the non-zero value matrix in the four-step phase-shifted bucket signal is 21×41×4. Through rotation and splicing operations, the four-step phase-shifted bucket signals of the two target images are converted into a 41×83 grayscale matrix, i.e., the first matrix. The first matrix is ​​spliced ​​vertically, with a row of zeros filled in the middle to obtain an 83×83 composite encrypted data matrix. The upper half corresponds to the first target image T1 to be encrypted, and the lower half corresponds to the second target image T2 to be encrypted, as shown in Figure 5, which is a schematic diagram of the bucket signal rotation and splicing process. Specifically, Figure 6 is a schematic diagram of encoding the composite encrypted data matrix into the channel image, where Figure 6(a) is the composite encrypted data matrix. The non-zero values ​​in the composite encrypted data matrix shown in Figure 6(a) are split into four digits. The first two digits are encoded and stored in the three-color image using the structured color-coded cipher book shown in Figure 6(d), resulting in the first channel color image shown in Figure 6(b). Figure 6(b) shows the first channel color image obtained by structural color encoding, representing the first two digits of the non-zero value of the encryption result. For easier experimental observation, a positioning marker is added to Figure 6(b), as shown in Figure 6(c), which is the first channel color image after adding the positioning marker. The non-zero value in the composite encrypted data matrix is ​​split into four digits, with the last two digits stored in the grayscale image using a combination encoding of the four-step grayscale encoding codebook shown in Figure 6(e), resulting in the second channel grayscale encoded image shown in Figure 6(f), representing the third and fourth digits of the non-zero value of the encryption result. This generates a first channel color image of 83×166 pixels and a second channel grayscale image of 166×166 pixels. To fit the three-channel metasurface array design, the even-numbered rows of the first channel color image are expanded, resulting in a 166×166 three-color image as the first channel color image. To enhance the security of the encrypted information, obfuscation information is added to the even-numbered rows of the encrypted information, resulting in the final image shown in Figure 6(g), which is the first channel color image after adding the positioning marker and obfuscation information. A 166×166 third-channel cipher image is designed based on the index information (17-23) of the structured color-coded cipher and the four-dimensional grayscale-coded cipher, as shown in Figure 7. It includes color-value mapping relationships, which include color-value mapping relationships and grayscale-value mapping relationships.

[0078] The above embodiments utilize the symmetry of the Fourier spectrum of natural images to process only the bucket signals in the non-zero value matrix region (avoiding redundant calculations). By "rotating the non-zero value matrix → filling the zero matrix → splicing the top and bottom", the bucket signals of the two images are integrated into a composite encrypted data matrix, and then split and encoded into different channels, which not only improves information storage efficiency, but also enhances the system's anti-interference capability.

[0079] This invention utilizes a four-step phase-shifting method to acquire bucket signals of the target image in the Fourier domain. By extracting non-zero values ​​of the bucket signals, matrix rotation and splicing, and multi-channel encoding, it achieves independent encryption and storage of two images on the same metasurface.

[0080] S5, based on the first channel color image, the second channel grayscale image, the third channel codebook image, and the different geometric parameters corresponding to different structural colors at the selected working wavelength, designs a three-channel metasurface array, including:

[0081] S5.1, each pixel in the first channel color image is mapped to a set of corresponding geometric parameters selected in S1 according to its color value, where the first color corresponds to the first set of geometric parameters, the second color corresponds to the second set of geometric parameters, and the third color corresponds to the third set of geometric parameters. This determines the geometric size of the nanobrick unit cell corresponding to each pixel on the three-channel metasurface array, forming a fixed array physical layout.

[0082] Specifically, for each pixel in the first channel color image, the geometric parameters of the corresponding nanobrick unit cell in the three-channel metasurface array are determined based on the pixel's color, thus completing the determination of the geometric dimensions of each nanobrick unit cell in the three-channel metasurface array. For example, in S1, the first set of geometric parameters, the second set of geometric parameters, and the third set of geometric parameters corresponding to dark green, light green, and yellow are selected respectively. If the current pixel is dark green, the geometric parameters of the nanobrick unit cell corresponding to the current pixel are the first set of geometric parameters; if the current pixel is light green, the geometric parameters of the nanobrick unit cell corresponding to the current pixel are the second set of geometric parameters; and if the current pixel is yellow, the geometric parameters of the nanobrick unit cell corresponding to the current pixel are the third set of geometric parameters. In this way, the geometric dimensions of the nanobrick unit cell corresponding to each pixel position on the three-channel metasurface array are uniquely determined, forming the physical layout basis of the entire three-channel metasurface array. This layout directly presents the target color pattern under unpolarized white light.

[0083] S5.2, based on the array physical layout determined in S5.1, according to Malus's law, the normalized gray value of each pixel in the second channel grayscale image is inverted and calculated into the azimuth angle of the corresponding nanobrick unit cell. .

[0084] S5.3, based on the array physical layout determined in S5.1, uses the fixed transmission phase generated by the nanobrick unit cell corresponding to each pixel. and azimuth Obtain the total phase modulation amount Using the third-channel codebook image as the target image for optimization, a simulated annealing algorithm is employed to perform global iterative optimization of the azimuth angle of the nanobrick unit cell corresponding to each pixel on the metasurface array. This minimizes the error evaluation function between the reconstructed holographic image and the optimized target image, thereby obtaining the final azimuth angle distribution that simultaneously satisfies the requirements of grayscale modulation and holographic phase modulation. This completes the determination of the geometric dimensions and azimuth angle distribution of all nanobrick unit cells on the three-channel metasurface array. Figure 8 shows a schematic diagram of the arrangement of the three-channel metasurface array, where Figure 8(a) is a schematic diagram of the fixed array physical layout obtained after step S5.1; Figure 8(b) is a schematic diagram after determining the azimuth angle.

[0085] In one embodiment, a metasurface dual-image encryption device based on the metasurface dual-image encryption device design method provided in the above embodiments is provided, comprising:

[0086] Substrate;

[0087] A nanostructure layer is formed on the substrate, and the nanostructure layer constitutes a three-channel metasurface array; the three-channel metasurface array is a metasurface array that can realize three-channel functions in the same array in physical space.

[0088] The nanostructure layer can be composed of any dielectric or semiconductor material with high refractive index (n>2.0) and low optical loss in the visible light operating band, specifically including but not limited to: metallic materials, other group IV or group III-V semiconductors, such as silicon, germanium, gallium nitride, gallium arsenide, etc.; high refractive index media, such as titanium oxide, titanium nitride, silicon nitride, aluminum oxide, etc.; two-dimensional materials, such as graphene, transition metal sulfides, etc.

[0089] The substrate can be made of any insulating material with high transparency, high insulation, and high thermal conductivity in the visible light band, including but not limited to: silicon, silicon dioxide / quartz, sapphire, etc. The substrate is a composite substrate or a thin film transfer structure, such as an SOI (Silicon-on-Insulator) structure.

[0090] The above extensions of crystalline silicon nanostructure layers and substrate materials can be freely combined. For example, titanium dioxide can be used as the nanostructure layer and silicon dioxide as the substrate (TOS material), or crystalline silicon can be used as the nanostructure layer and sapphire as the substrate (SOS material), etc.

[0091] In one embodiment, SOS (silicon-on-sapphire) material was selected, with crystalline silicon as the nanostructure layer and sapphire as the substrate. The geometric parameters of the nanobrick unit cell were optimized using CST STUDIO SUITE electromagnetic simulation software. The height H of the nanobrick unit cell was fixed at 200 nm, the unit period C at 300 nm, and the working wavelength at 530 nm. The scanning range was set as follows: major axis length L scan range 100-220 nm, minor axis length W scan range 80-90 nm. The three sets of geometric parameters selected were:

[0092] The first set of geometric parameters: major axis length L1=112nm, minor axis length W1=82nm, polarization conversion efficiency at 530nm wavelength is 54%, corresponding to dark green;

[0093] The second set of geometric parameters: major axis length L2=170nm, minor axis length W2=82nm, polarization conversion efficiency at 530nm wavelength is 61%, corresponding to light green;

[0094] The third set of geometric parameters: major axis length L3 = 219.5 nm, minor axis length W3 = 84.5 nm, polarization conversion efficiency of 50% at a wavelength of 530 nm, corresponding to yellow.

[0095] The feasibility and effectiveness of this invention have been verified through joint simulations using CST, FDTD, and MATLAB. This embodiment uses a cuboid nanobrick unit cell as an example. The length, width, and height of the nanobrick unit cell are all subwavelengths. As shown in Figure 9, the nanobrick unit cell consists of substrate 1 and nanobrick unit cell 2 etched onto substrate 1. An xyz rectangular coordinate system is established. The major axis of nanobrick unit cell 2 is L, the minor axis is W, and the height in the z-direction is H. The side lengths of substrate 1 in both the x and y directions are C, and the angle between L and the x-direction is the azimuth angle. .

[0096] The first-channel color image, the second-channel grayscale image, and the third-channel codebook image generated during the encryption stage are jointly encoded into the same monolayer metasurface array to obtain a three-channel metasurface array. The geometric parameters (major axis length L and minor axis length W) of each nanobrick unit cell in the metasurface array are selected from the three sets of geometric parameters obtained by optimization in step S1 (the corresponding geometric parameters are selected according to the color of the pixel), and its azimuth angle must simultaneously meet the intensity modulation requirements of the second channel and the phase modulation requirements of the third channel.

[0097] The correctness of the three-channel metasurface array scheme was verified by FDTD simulation, and the results are shown in Figure 10. Figure 10(a) is the three-color image of the target in channel one; Figure 10(b) is the grayscale image of the target in channel two; Figure 10(c) is the holographic image of the target in channel three; Figure 10(d) is the simulation result of channel one; Figure 10(e) is the simulation result of channel two; and Figure 10(f) is the simulation result of channel three. The simulation results show that the output images of the three channels are highly consistent with the design target, the grayscale steps of channel two are clear, and the holographic image of channel three is completely reconstructed.

[0098] The simulation parameters used are shown in Table 1:

[0099] Table 1. Simulation parameters of radar and target

[0100]

[0101] Each pixel in the first channel color image has a specific color, and the color of each pixel is mapped to a set of geometric parameters selected in step S1. In this embodiment, dark green pixels correspond to the first set of geometric parameters (L=112nm, W1=82nm), light green pixels correspond to the second set of geometric parameters (L2=170nm, W2=82nm), and yellow pixels correspond to the third set of geometric parameters (L3=219.5nm, W3=84.5nm). Thus, the geometric dimensions of each pixel position (corresponding to a nanobrick unit cell) on the metasurface array are uniquely determined, forming the physical layout basis of the device. This layout can directly display the structural color pattern of the first channel color image under unpolarized broadband white light illumination.

[0102] Given that the geometric parameters of each nanobrick unit cell in the metasurface array are fixed, to achieve a four-step grayscale image in the second channel, the emitted light intensity needs to be discretely modulated using Malus's law. When linearly polarized light of the working wavelength (e.g., 530 nm) is incident, and the incident polarization direction is orthogonal to the transmission axis of the analyzer, the normalized emitted light intensity I and the azimuth angle of the nanobrick unit cell are... satisfy:

[0103] ;

[0104] in , The values ​​represent the transmittance or reflectance of a nanobrick unit cell along its long and short axes, respectively, and satisfy the energy conservation constraint. .

[0105] For the target gray value of each pixel in the second channel grayscale image, the azimuth angle of the corresponding nanobrick unit cell can be derived. Because the cosine function has periodicity and symmetry, a specific outgoing light intensity often has multiple... The resolved value, this "one-to-many" mapping characteristic, can be called the orientation degeneracy of the nanobrick unit cell. This embodiment uses the azimuth angle... The principal value range is limited to Inside, four gray levels are assigned. The values ​​are four candidate azimuth angles, thus obtaining a set of candidate azimuth angles (including four candidate azimuth angles) that meet the grayscale modulation requirements, which serve as an important constraint for subsequent optimization.

[0106] The goal of the third channel is to reconstruct the codebook image through far-field diffraction under circularly polarized light illumination. Each nanobrick, under incident circularly polarized light, emits a reverse-polarized beam carrying geometric phase. Furthermore, when using several nanobrick unit cells with different size parameters for phase modulation, in addition to the geometric phase modulation caused by the azimuth angle, it is also necessary to consider the fixed transmission phase generated by the nanostructures themselves of different sizes. :

[0107] ;

[0108] in, The incident wavelength, The equivalent refractive index of a nanobrick unit cell is given. The thickness of the nanobrick unit cell. When fixed, by changing its equivalent refractive index This allows for phase modulation of the light wave, and the emitted light field can then be expressed as:

[0109] ;

[0110] Variations in the size parameters of nanobrick unit cells directly alter their electromagnetic response characteristics, manifesting as changes in the transmission / reflection coefficients. , The change in phase. Therefore, the reverse polarizer beams emitted from nanobrick unit cells of different geometric sizes have different transmission phase modulations. And its geometric phase term It is still only related to the azimuth angle of the nanobrick unit cell.

[0111] Therefore, the total phase modulation amount at each pixel position is determined based on the corner degeneracy. , For the azimuth angle to be optimized, the simulated annealing algorithm is used to determine the final azimuth angle distribution. To achieve high-fidelity holographic reconstruction of the codebook image, it is necessary to find the optimal azimuth angle for each pixel in the entire array. This ensures that the reconstructed holographic image obtained after Fourier transform of the total phase distribution is as consistent as possible with the optimized target image (i.e., the third channel codebook image).

[0112] This embodiment employs simulated annealing for global optimization. First, the third-channel codebook image is designed off-axis. For example, after off-axis design, the maximum diffraction angle of the target holographic image in the x and y axes is 62°. Then, distortion correction and energy compensation are performed on the image to obtain the optimized target.

[0113] First, set the evaluation function as follows:

[0114] ;

[0115] in, This refers to the optimized target image, specifically the third-channel codebook image. This indicates that the holographic image is currently being reproduced. This represents the total number of pixels in the target image. , These represent the first and second elements in the target image that are being optimized. The, the 1 pixel, , These represent the first digit in the currently reconstructed holographic image. The, the 1 pixel.

[0116] In the azimuth candidate set (containing four candidate azimuth angles) of the nanobrick unit cell corresponding to each pixel, an initial azimuth angle is randomly assigned to the nanobrick unit cell corresponding to each pixel. Combined with the initial azimuth angle of the nanobrick unit cell corresponding to each pixel With fixed transmission phase Generate initial phase distribution Calculate the corresponding current reconstructed holographic image According to the current reconstructed holographic image With the optimized target image (i.e., the third-channel codebook image) Calculate the initial evaluation function ,like Less than the evaluation threshold If the output is not found, the iteration process continues. For each pixel, a new phase distribution is generated by randomly selecting one of the four candidate azimuth angles from the corresponding azimuth angle candidate set. And recalculate the evaluation function. , and evaluation function If a comparison is made, Less than Then accept the phase distribution under the new azimuth angle distribution. Conversely, based on the Monte Carlo acceptance criterion, if the conditions are met... The phase distribution under the new azimuth angle distribution is also acceptable. Otherwise, continue iterating until the pre-specified maximum number of iterations is reached or the evaluation function falls below a preset evaluation threshold. It outputs the azimuth angle distribution of each pixel corresponding to the final phase distribution.

[0117] This optimization process takes into account both the grayscale modulation constraints of the second channel (azimuth angle is taken from the candidate set) and the holographic imaging quality of the third channel, so that the final determined azimuth angle distribution can support the correct decoding of both channels at the same time.

[0118] By merging the spatial distribution of the determined geometric parameters of the nanobricks with the optimized azimuth angle distribution, the complete design parameters of the nanobricks at each pixel position (498×498 in total) are obtained. Based on this, it can be directly used for nanofabrication processes such as electron beam exposure to fabricate a three-channel metasurface encryption device integrating structural color display, grayscale light intensity modulation, and holographic phase encoding on an SOS substrate.

[0119] During the image decryption process, the three-channel metasurface is illuminated with broadband white light, 530nm horizontally polarized light, and 530nm left-handed circularly polarized light, respectively, and the output images of the three channels are read. Based on the color-value mapping relationship carried by the third channel codebook image, the structural color and grayscale information in the first channel color image and the second channel grayscale image are decoded into a bucket signal data matrix. The data matrix is ​​then inversely spliced ​​and subjected to inverse Fourier transform to reconstruct the original two target images. The simulation results based on MATLAB are shown in Figure 11. Figure 11 is a schematic diagram of the channel design and image reconstruction results of the three-channel metasurface array. Figure 11(a) is the simulation result of the channel one image corresponding to the three-channel metasurface irradiated with unpolarized broadband white light. Figure 11(b) is the simulation result of the channel two image obtained by irradiating the three-channel metasurface with 530nm horizontally polarized light. Figure 11(c) is the simulation result of the channel three holographic image obtained by irradiating the three-channel metasurface with 530nm left-handed circularly polarized light. Figure 11(d) is the image to be encrypted. Figure 11(e) is the decryption result of the dual images. Figure 11(f) is the azimuth distribution diagram. Figure 11(g) is a magnified view of the azimuth distribution.

[0120] To verify the security of the encryption scheme, a security verification experiment was conducted. Figure 12 shows the image reconstruction results after decryption in the security verification experiment, where Figure 12(a) shows the image reconstruction result after correct decryption, and Figure 12(b) shows the image reconstruction result after incorrect decryption. The image reconstruction results are expressed as a structural similarity index. This indicates that, under the correct key condition, the SSIM values ​​of T1 and T2 are 0.7929 and 0.8831, respectively; while under the incorrect key decryption condition, the SSIM value drops to around 0.149832, indicating that the system has good encryption security and reconstruction robustness.

[0121] This invention, through the screening of unit cell structure parameters of SOS material, achieves a polarization conversion efficiency of up to 61% at a working wavelength of 530nm. Combining Malus's law and the principle of geometric phase modulation, it significantly improves the brightness uniformity of the grayscale image in channel two and the reproduction fidelity of the holographic image in channel three. FDTD simulation verifies that the grayscale steps in channel two are clear and the "T" shaped pattern in channel three is free from defocus.

[0122] This invention achieves independent encryption of two images and high-capacity storage, overcoming the limitations of existing single-image encryption: existing metasurface encryption schemes often can only encrypt a single image and require the transmission of a large number of random key matrices. Based on a single-pixel Fourier imaging algorithm, this invention splits and stores the bucket signals of two target images in channels one and two. Independent encryption of the two images is achieved by encoding the first two bits of the structure color and the last two bits of the grayscale, eliminating the need to transmit a large number of keys (the keys are integrated into the metasurface physical structure and the holographic image), thus improving information capacity.

[0123] The image reconstruction effect of this invention is good and the SSIM value is high: This invention uses four-step phase shift to suppress noise and inverse Fourier transform to accurately reconstruct the image. MATLAB simulation shows that the SSIM value reaches 0.883153 when decrypting correctly, and only about 0.15 when decrypting incorrectly. This ensures both decryption quality and improves encryption security.

[0124] This invention features a highly concealed key and improved resistance to cracking: Existing solutions' key images are easily observed directly, allowing attackers to extract features through image recognition. In contrast, the key of this invention is integrated into a three-channel holographic image, requiring specific circularly polarized light for observation. Furthermore, the ciphertext is split and stored across two channels, requiring simultaneous access to information from channels one, two, and three for decryption, thus increasing the difficulty of brute-force attacks.

[0125] This invention does not increase design and manufacturing complexity, and balances practicality and cost: Existing cascaded metasurfaces require multiple sheets to be precisely aligned, which is costly and susceptible to environmental interference. This invention uses a single three-channel metasurface, which does not require additional processing steps, has a lower manufacturing cost than cascaded metasurfaces, and is easy to scale up for production and application.

[0126] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A design method for a metasurface dual-image encryption device, characterized in that, include: S1. Using electromagnetic simulation software, the geometric parameters of the nanobrick unit cell are scanned and optimized to select three sets of geometric parameters with high polarization conversion efficiency and corresponding to different structural colors at a set working wavelength. The working wavelength, the height H and unit period C of the nanobrick unit cell are set, and the scanning range is set, including the scanning range of the major axis length L and the minor axis length W of the nanobrick unit cell. The electromagnetic simulation software is used to select the three sets of geometric parameters with the highest polarization conversion efficiency at the set working wavelength within the scanning range. S2. For the two target images to be encrypted, non-uniform Fourier sampling and four-step phase shift processing are performed respectively to generate bucket signals for each target image, including the following steps: S2.1, performing two-dimensional discrete Fourier transform on the target images. Forward transform yields the Fourier spectrum. The Fourier frequency coordinates are centered, and based on the set sampling path type, key frequency points with concentrated energy are preferentially selected in the frequency domain to form a non-uniform Fourier sampling point set corresponding to the target image. S2.2: For each Fourier sampling point in the Fourier sampling point set, four sets of sinusoidal illumination light fields with different initial phases are generated. The initial phases of the four sets of sinusoidal illumination patterns correspond to a set four-step phase shift phase sequence. By calculating the inner product of each sinusoidal illumination light field and the target image, the four-step phase shift bucket signal is simulated. S3: The bucket signals corresponding to the two target images are rotated, spliced, and filled to form a composite encrypted data matrix, including: taking four... The non-zero value matrix region in the four-step phase-shifted bucket signal of the two target images is converted into a grayscale matrix through rotation and stitching operations. The upper and lower parts of the grayscale matrix can be stitched together to form a near-circular shape. The left and right parts of the grayscale matrix are stitched together by filling a column of zeros to obtain the first matrix. The first matrix is ​​then stitched together vertically, with a row of zeros filling the gap between the upper and lower parts to obtain the composite encrypted data matrix. The upper half of the composite encrypted data matrix corresponds to the first target image to be encrypted, and the lower half corresponds to the second target image to be encrypted. S4, the values ​​in the composite encrypted data matrix are split, and a first-channel color image and a second-channel grayscale image are generated respectively through structured color encoding and grayscale encoding. S5. A third-channel codebook image containing color-value mapping relationships is generated; S6. Based on the first-channel color image, the second-channel grayscale image, the third-channel codebook image, and three sets of geometric parameters corresponding to different structural colors at the selected working wavelength, a three-channel metasurface array is designed, including the following steps: S5.

1. Each pixel in the first-channel color image is mapped to a set of corresponding geometric parameters selected in S1 according to its color value, wherein the first color corresponds to the first set of geometric parameters, the second color corresponds to the second set of geometric parameters, and the third color corresponds to the third set of geometric parameters, thereby determining the geometric size of the nanobrick unit cell corresponding to each pixel on the three-channel metasurface array, forming a fixed array physical layout; S5.

2. In S5.Based on the established array physical layout, and according to Malus's law, the normalized gray value of each pixel in the second channel grayscale image is inverted and calculated into the corresponding nanobrick unit cell azimuth angle. S5.3, based on the array physical layout determined in S5.1, and based on the fixed transmission phase generated by the nanobrick unit cell corresponding to each pixel itself. and azimuth Obtain the total phase modulation amount Using the third-channel codebook image as the target image for optimization, the simulated annealing algorithm is used to perform global iterative optimization of the azimuth angle of the nanobrick unit cell corresponding to each pixel on the metasurface array. This minimizes the error evaluation function between the reconstructed holographic image and the optimized target image, thereby obtaining the final azimuth angle distribution that simultaneously satisfies the grayscale modulation requirements and the holographic phase modulation requirements. Thus, the determination of the geometric dimensions and azimuth angle distribution of all nanobrick unit cells on the three-channel metasurface array is completed.

2. The design method for a metasurface dual-image encryption device according to claim 1, characterized in that, During the image decryption process, the three-channel metasurface is illuminated with broadband white light, horizontally polarized light with a set working wavelength, and left-hand circularly polarized light with a set working wavelength, respectively. The output images of the three channels are read. Based on the color-value mapping relationship carried by the third channel codebook image, the structural color and grayscale information in the first channel color image and the second channel grayscale image are decoded into a bucket signal data matrix. The bucket signal data matrix is ​​then inversely spliced ​​and subjected to inverse Fourier transform to reconstruct the two target images.

3. The design method for a metasurface dual-image encryption device according to claim 1, characterized in that, In S2.1, the sampling path type is a spiral or circular non-uniform path that extends outward from the center of the frequency domain.

4. The design method for a metasurface dual-image encryption device according to claim 1, characterized in that, In S2.2, the spatial intensity distribution of the sinusoidal illumination field is defined by the following formula: in Fourier frequency The corresponding spatial intensity distribution of the string illumination field, This is a constant term, representing the DC component of the light intensity; Intensity modulation coefficient; and Spatial coordinates Fourier frequency 、 Directional components; The initial phase is given by a value of This corresponds to the set four-step phase shift sequence.

5. The design method for a metasurface dual-image encryption device according to claim 4, characterized in that, In S2.2, the bucket signal of the four-step phase shift is obtained by simulation calculation using the following formula: in, The target image in spatial coordinates grayscale value at that location 、 、 、 The initial phases are respectively for The corresponding bucket signal at that time.

6. A metasurface dual-image encryption device, designed using the metasurface dual-image encryption device design method as described in any one of claims 1 to 5, characterized in that, include: Substrate; A nanostructure layer is formed on the substrate, which constitutes a three-channel metasurface array. The three-channel metasurface array is a metasurface array that can realize three-channel functions in the same array in physical space.

Images