Single-exposure multispectral computational imaging method, system and device based on metasurface

Abstract

The present invention belongs to the technical field of image processing and computational imaging, and discloses a single-exposure multi-spectral computational imaging method, system and device based on a metasurface. The method comprises: using a single-layer metasurface structure, placed in front of an array imaging sensor, collecting images of different spectral channels and the optical point spread function PSF of the corresponding channels through a single exposure; using the obtained exposure image and point spread function PSF information, reconstructing the imaging images of different channels based on the conjugate gradient method; the method of the present invention uses a single-layer metasurface structure as the front end of the optical imaging system, without the need to introduce traditional lenses and filters, and without the need to build a complex optical path system, and has the advantages of a simple optical system, small size and lightness; the method of the present invention reconstructs the multi-channel image collected by a single exposure through a conjugate gradient algorithm, and can accurately solve the "inverse scattering" problem, thereby improving the quality of the reconstructed image.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing and computational imaging, and in particular relates to a single-exposure multi-spectral computational imaging method, system and equipment based on a metasurface. Background Art

[0002] Traditional all-optical (lens) imaging systems map each point of an image sample onto a separate sensor pixel, directly producing a reliable spatial image. However, this method requires complex optical equipment (a series of lenses or lens groups), and without additional spectral filters or microlens arrays, it is usually unable to capture accurate spectral information, and it is even more difficult to achieve multispectral imaging.

[0003] In lensless computational imaging systems, a lensless optical system projects a blurred image of the target sample onto a sensor and reconstructs the target light field by solving a subsequent "inverse scattering" problem (via methods such as least squares fitting). However, the "inverse scattering" problem to be solved is usually ill-posed, sensitive to noise, and often requires high computational memory and computing power.

[0004] In recent years, an "end-to-end" optimization method has emerged in the field of computational imaging. This is a systematic method that combines the front-end analog optical system with the image processing digital back-end for optimization. It can be used for multi-channel multi-spectral single-exposure computational imaging. The performance of the imaging system designed based on this method is better than that of pure optical design and pure digital computational imaging design. The introduction of end-to-end optimization methods into the frontier fields of full-wave vector nanophotonics and optical metasurfaces has demonstrated strong physical data acquisition and manipulation capabilities.

[0005] Through the above analysis, the problems and defects of the prior art are as follows:

[0006] (1) Existing technologies require multiple exposures to reconstruct target samples on multiple spectra.

[0007] (2) The existing technology requires the construction of a complex optical path system, which is inconvenient to operate.

[0008] (3) The existing technology has poor reconstruction effect on images collected by single exposure, cannot accurately solve the "inverse scattering" problem, and the reconstructed image quality is low. Summary of the invention

[0009] In order to overcome the problems existing in the related art, the embodiments disclosed in the present invention provide a single-exposure multi-spectral computational imaging method, system and device based on a metasurface.

[0010] The technical solution is as follows: A single-exposure multi-spectral computational imaging method based on a metasurface comprises: an optimized metasurface optical system performs a one-time exposure on images of different channels, and collects point spreading functions (PSFs) of optical imaging systems of different channels;

[0011] In the image reconstruction stage, the conjugate gradient algorithm is used to restore the original sample image through the point spread function (PSF) of different channels and the imaging information on the sensor to obtain image restoration in multiple bands.

[0012] In one embodiment, the single-exposure multi-spectral computational imaging method based on a metasurface specifically comprises the following steps:

[0013] S1, build an image reconstruction model using a metasurface optical system and a conjugate gradient algorithm;

[0014] S2, randomly generating a series of two-dimensional code images as a training set, which is used to train the image reconstruction model built in step S1 to obtain the optimal optical metasurface structure parameters and conjugate gradient algorithm parameters;

[0015] S3, after completing the training, retain the optimal optical metasurface structure and conjugate gradient algorithm parameters in the image reconstruction model as the final image reconstruction experimental model;

[0016] S4, input the multi-channel image into the image reconstruction model in step S3 to obtain a restored image.

[0017] In one embodiment, in step S1, in building an image reconstruction model, the image v on the array sensor is defined as the convolution sum of multiple channel input images u and point spread function (PSF); the image is restored by computational imaging. The objective optimization function is

[0018]

[0019] Where N is the number of channels, spectral channels, or scene distance.

[0020] In one embodiment, in step S2, in obtaining the optimal optical metasurface structure, the metasurface is used as an optical device to perform forward solution optimization, including the following steps:

[0021] 1) Use the metasurface as the front end of the optical imaging system and calculate the point spread function (PSF) of the metasurface

[0022] PSF=|ntff(t(p)·E in )| 2 (2);

[0023] Where p is the structural parameter of the optical metasurface, t(p) is the transmission coefficient of the optical metasurface, and E in is the incident electric field, ntff(·) represents the conversion from near field to far field;

[0024] 2) The image v on the array sensor is obtained by convolution of the point spread function PSF of the metasurface and the real scene image u:

[0025]

[0026] 3) Reconstruction by solving the inverse scattering problem

[0027]

[0028] Using Tikhonov regularization to process equation (4), we get

[0029]

[0030] Among them, α is the constant to be optimized, and the convolution kernel function operator is defined as

[0031] Substituting equation (3) into (5) and making some transformations, we can obtain

[0032]

[0033] The conjugate gradient method is used to solve equation (6) to obtain the reconstructed image The objective function L is calculated using formula (1).

[0034] In one embodiment, after calculating the objective function L, it is also necessary to calculate the derivative of the objective function L with respect to the hypersurface structure p and the undetermined coefficient α of the reconstruction algorithm to implement the optimization process of back propagation.

[0035] In one embodiment, the back propagation optimization process includes the following steps:

[0036] (i) Calculate the derivative

[0037]

[0038] Combined with the derivative of formula (6), the above formula is replaced by:

[0039]

[0040] in,

[0041] The above equation is solved for Λ using the conjugate gradient method;

[0042] (ii) Calculate the derivative

[0043]

[0044] Combined with the derivative of formula (6), the above formula is replaced by:

[0045]

[0046] After obtaining the derivative of the objective function with respect to the optimization parameter, the system is optimized in combination with the gradient descent algorithm.

[0047] In one embodiment, the system is optimized in combination with a gradient descent algorithm including:

[0048] The updating method of the parameters to be optimized is as follows:

[0049]

[0050] Among them, P is the variable to be optimized, including p and α, and β is the dynamic learning rate. The specific value depends on the selected stochastic gradient descent method and Adam algorithm.

[0051] Another object of the present invention is to provide a single-exposure multi-spectral computational imaging system based on a metasurface for implementing the single-exposure multi-spectral computational imaging method based on a metasurface, wherein the single-exposure multi-spectral computational imaging system based on a metasurface comprises:

[0052] An image reconstruction model building module is used to build an image reconstruction model using a metasurface optical system and a conjugate gradient algorithm, and to train the image reconstruction model using a gradient descent method to obtain optimal metasurface structural parameters and conjugate gradient algorithm parameters;

[0053] A point spread function recording module, which is used to collect multi-channel images through the optimized optical metasurface using an array sensor and record the point spread function (PSF) of different channels;

[0054] The sample image reconstruction module is used to reconstruct the sample image using the optimized conjugate gradient algorithm combined with the point spread functions PSF of different channels.

[0055] Another object of the present invention is to provide a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the single-exposure multi-spectral computational imaging method based on the metasurface.

[0056] Another object of the present invention is to provide an all-optical imager, characterized in that the all-optical imager executes the metasurface-based single-exposure multi-spectral computational imaging method.

[0057] Combining all the above technical solutions, the advantages and positive effects of the present invention are as follows:

[0058] First, in view of the technical problems existing in the above-mentioned prior art and the difficulty of solving the problems, the technical solutions to be protected by the present invention and the results and data during the research and development process are closely combined to analyze in detail and deeply how the technical solutions of the present invention solve the technical problems, and some creative technical effects brought about after solving the problems. The specific description is as follows:

[0059] By combining optimized optics and appropriate computational imaging algorithms to reconstruct the target sample over multiple spectra, the detector can be spectrally responsive but can be achieved with just a single exposure.

[0060] The method of the present invention uses a single-layer metasurface structure as the front end of the optical imaging system, without the need to introduce traditional lenses (or lens groups) and filters, and without the need to build a complex optical path system. It has the advantages of a simple optical system, small size and light weight;

[0061] The method of the present invention reconstructs the multi-channel image acquired by a single exposure through a conjugate gradient algorithm, and can accurately solve the "inverse scattering" problem, thereby improving the quality of the reconstructed image.

[0062] Second, considering the technical solution as a whole or from the perspective of the product, the technical effects and advantages of the technical solution to be protected by the present invention are described in detail as follows:

[0063] The present invention provides a method for realizing multi-spectral computational imaging based on a single exposure of an optical metasurface, comprising the following steps: (1) constructing an image reconstruction model by using the optimized design of an artificial microstructure of an optical metasurface and a conjugate gradient algorithm, such as Figure 2 As shown. (2) A series of two-dimensional code images are randomly generated as training sets for training the image reconstruction model constructed in (1) to obtain the optimal optical metasurface structure parameters and free parameters in the conjugate gradient optimization algorithm; (3) After the training is completed, the optimal optical metasurface structure and conjugate gradient algorithm parameters in the image reconstruction model are retained as the final image reconstruction experimental model; (4) The multi-channel image is input into the reconstruction experimental model in (3) to obtain a restored image. The method of the present invention uses a single-layer metasurface structure as the analog front end of the optical imaging system, without the need to introduce traditional lens groups and filters, and has the advantages of a simple optical system, small size and light weight; the multi-channel image acquired by a single exposure is reconstructed by a reconstruction algorithm, and the "inverse scattering" problem can be accurately solved, thereby improving the quality of the reconstructed image. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0065] Figure 1 is a flow chart of a single-exposure multi-spectral computational imaging method based on a metasurface provided in an embodiment of the present invention;

[0066] Figure 2 is a principle diagram of an image reconstruction model provided by an embodiment of the present invention;

[0067] Figure 3 is a diagram of the principle of image reconstruction training provided by an embodiment of the present invention;

[0068] Figure 4 Schematic diagram of a single-exposure multi-spectral computational imaging system based on a metasurface provided in an embodiment of the present invention;

[0069] Figure 5 is a diagram of an optimized supersurface structure provided by an embodiment of the present invention;

[0070] Figure 6 is a diagram of the three-wavelength channel reconstruction result provided by an embodiment of the present invention;

[0071] Figure 7 is a diagram of the three-depth channel reconstruction result provided by an embodiment of the present invention;

[0072] In the figure: 1. Image reconstruction model building module; 2. Point spread function recording module; 3. Sample image reconstruction module. DETAILED DESCRIPTION

[0073] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific implementation disclosed below.

[0074] 1. Explanation of the embodiment:

[0075] The single-exposure multi-spectral computational imaging method based on a metasurface provided in an embodiment of the present invention includes: the design of a metasurface optical system and the optimization of a conjugate gradient reconstruction algorithm. The optimized metasurface optical system performs a one-time exposure of images of different channels and collects the point spread functions (PSFs) of the optical imaging systems of different channels. In the image reconstruction (image restoration) stage, the conjugate gradient algorithm is used to restore the original sample image through the point spread functions (PSFs) of different channels and the imaging information on the sensor, thereby finally obtaining image restoration of multiple bands.

[0076] like Figure 1 As shown, the single-exposure multi-spectral computational imaging method based on a metasurface provided by an embodiment of the present invention comprises the following steps:

[0077] S101, using the metasurface optical system and the conjugate gradient algorithm to build an image reconstruction model, using the gradient descent method to train the image reconstruction model, and obtaining the optimal metasurface structure parameters and conjugate gradient algorithm parameters, such as Figure 2 shown.

[0078] S102, using an array sensor to collect multi-channel images through the optimized optical metasurface, and recording the point spreading function (PSF) of different channels.

[0079] S103, using the optimized conjugate gradient algorithm combined with the point spread functions (PSFs) of different channels to reconstruct the sample image.

[0080] Example 1

[0081] The single-exposure multi-spectral computational imaging method based on a metasurface provided in an embodiment of the present invention comprises the following steps:

[0082] (1) Using a single-layer metasurface structure placed in front of an array imaging sensor, images of different spectral channels and the optical point spreading function (PSF) of the corresponding channels are collected through a single exposure;

[0083] (2) Using the exposure image and point spread function (PSF) information obtained in (1), the imaging images of different channels are reconstructed based on the conjugate gradient method.

[0084] Example 2

[0085] The single-exposure multi-spectral computational imaging method based on a metasurface provided in an embodiment of the present invention comprises the following steps:

[0086] (1) Use the metasurface optical system and conjugate gradient algorithm to build an image reconstruction model. Figure 2 shown.

[0087] (2) randomly generating a series of two-dimensional code images as training sets to train the image reconstruction model constructed in (1) to obtain the optimal optical metasurface structure parameters and conjugate gradient algorithm parameters;

[0088] (3) After the training is completed, the optimal optical metasurface structure and conjugate gradient algorithm parameters in the image reconstruction model are retained as the final image reconstruction experimental model;

[0089] (4) Input the multi-channel image into the reconstruction model in (3) to obtain the restored image.

[0090] Example 3

[0091] Based on the single-exposure multi-spectral computational imaging method based on the metasurface provided in Example 2 of the present invention, further, in the actual optimization design, the image v on the array sensor can be regarded as the convolution sum of multiple channel input images u and point spread functions (PSFs). Image is restored through computational imaging The objective optimization function is

[0092]

[0093] Where N is the number of channels (spectral channel or scene distance).

[0094] In the actual design, the embodiment of the present invention uses the metasurface as an optical device. For specific optimization ideas, see Figure 3 . The following steps are included:

[0095] First, the metasurface is used as the front end of the optical imaging system to calculate the point spread function (PSF) of the metasurface:

[0096] PSF=|ntff(t(p)·E in )| 2 (2)

[0097] Where p is the structural parameter of the optical metasurface, t(p) is the transmission coefficient of the optical metasurface, and E in is the incident electric field, and ntff(·) represents the conversion from near field to far field (the conversion from the electric field of the transmission surface of the metasurface to the electric field of the array sensor surface).

[0098] Next, the image v on the array sensor can be obtained by convolution of the PSF and the real scene image u:

[0099]

[0100] Reconstruction by solving the inverse scattering problem

[0101]

[0102] Here, the present invention uses the Tikhonov regularization to process equation (4), thereby obtaining

[0103]

[0104] Among them, α is the constant to be optimized, and the convolution kernel function operator is defined as

[0105] In order to solve equation (5), we substitute equation (3) into (5) and make some transformations to obtain

[0106]

[0107] Here, the conjugate gradient method is used to solve equation (6) to obtain the reconstructed image The objective function L can then be calculated (see equation (1)).

[0108] Among them, equations (1)-(6) give the basic process of the forward solution of the end-to-end design. Next, it is necessary to calculate the derivative of the objective function L with respect to the metasurface structure p and the unknown coefficient α of the reconstruction algorithm, so as to realize the optimization process of back propagation.

[0109] The specific steps include:

[0110] (1) Calculate the derivative

[0111]

[0112] Combined with the derivative of formula (6), the above formula can be written as:

[0113]

[0114] in,

[0115] The above equation can be solved for Λ using the conjugate gradient method.

[0116] (2) Calculate the derivative

[0117]

[0118] Combined with the derivative of formula (6), the above formula can be written as:

[0119]

[0120] Equations (7)-(11) give the inverse solution process of the end-to-end design. After obtaining the derivative of the objective function with respect to the optimization parameter, the system can be optimized by combining gradient descent algorithms, such as stochastic gradient descent and Adam algorithm. The update method of the parameters to be optimized is shown in the following formula:

[0121]

[0122] Among them, P is the variable to be optimized (i.e., p and α), and β is the dynamic learning rate, and the specific value depends on the selected algorithm.

[0123] Example 4

[0124] like Figure 4 As shown, an embodiment of the present invention provides a single-exposure multi-spectral computing imaging system based on a metasurface for implementing the single-exposure multi-spectral computing imaging method based on a metasurface, and the single-exposure multi-spectral computing imaging system based on a metasurface includes:

[0125] Image reconstruction model building module 1, used to build an image reconstruction model using a metasurface optical system and a conjugate gradient algorithm, and train the image reconstruction model using a gradient descent method to obtain optimal metasurface structure parameters and conjugate gradient algorithm parameters;

[0126] A point spread function recording module 2 is used to collect multi-channel images through the optimized optical metasurface using an array sensor and record the point spread function (PSF) of different channels;

[0127] The sample image reconstruction module 3 is used to reconstruct the sample image by using the optimized conjugate gradient algorithm combined with the point spread functions PSF of different channels.

[0128] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0129] Since the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present invention, their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0130] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0131] 2. Application examples:

[0132] Application Example 1

[0133] An application embodiment of the present invention provides a computer device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor implements the steps of any of the above-mentioned method embodiments when executing the computer program.

[0134] Application Example 2

[0135] The application embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0136] Application Example 3

[0137] The application embodiment of the present invention also provides an information data processing terminal, which is used to provide a user input interface to implement the steps in the above method embodiments when executed on an electronic device. The information data processing terminal is not limited to mobile phones, computers, and switches.

[0138] Application Example 4

[0139] The application embodiment of the present invention further provides a server, which is used to provide a user input interface to implement the steps in the above method embodiments when executed on an electronic device.

[0140] Application Example 5

[0141] An embodiment of the present invention provides a computer program product. When the computer program product runs on an electronic device, the electronic device can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0142] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device that can carry the computer program code to the camera / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk.

[0143] III. Evidence of the relevant effects of the embodiments:

[0144] By using the optimization method mentioned in the above embodiment, the present invention realizes multi-wavelength and multi-depth image reconstruction. The experimental results show that:

[0145] Figure 5 The optimized metasurface structure is given, with a radius distribution of a 3000×3000 cylinder array, and the radius of each cylinder is between [0.02, 0.15] (unit: micrometer).

[0146] Figure 6 The reconstruction results of three wavelength channel images are given. The first and third columns in the figure are the real image and the reconstructed image, respectively, and the corresponding pixel resolution is 128×128. The second column is the image on the sensor. Each row represents a channel. From the first to the third row, the corresponding wavelengths are 533nm, 558nm and 583nm, respectively, and the corresponding peak signal-to-noise ratios are 31dB, 33dB and 29dB. The corresponding object distance is 2cm and the image distance is 2mm.

[0147] Figure 7The reconstruction results of three depth channel images are given. The first and third columns in the figure are the real image and the reconstructed image, respectively, and the corresponding pixel resolution is 128×128. The second column is the image formed on the sensor. Each row represents a channel. From the first to the third row, the corresponding object distances are 100um, 120um and 140um, respectively, and the corresponding peak signal-to-noise ratios are 30dB, 28dB and 27dB. The corresponding wavelength is 533nm and the image distance is 2mm.

[0148] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principles of the present invention should be covered within the protection scope of the present invention.

Claims

1. A single-exposure multispectral computational imaging method based on a metasurface, characterized in that: The single-exposure multi-spectral computational imaging method based on the metasurface includes: the optimized metasurface optical system performs a one-time exposure on images of different channels, and collects point spread functions (PSFs) of optical imaging systems of different channels; In the image reconstruction stage, the conjugate gradient algorithm is used to restore the original sample image through the point spread function PSF of different channels and the imaging information on the sensor to obtain image restoration in multiple bands; The single-exposure multi-spectral computational imaging method based on the metasurface specifically comprises the following steps: S1, build an image reconstruction model using a metasurface optical system and a conjugate gradient algorithm; S2, randomly generating a series of two-dimensional code images as a training set, which is used to train the image reconstruction model built in step S1 to obtain the optimal optical metasurface structure parameters and conjugate gradient algorithm parameters; S3, after completing the training, retain the optimal optical metasurface structure and conjugate gradient algorithm parameters in the image reconstruction model as the final image reconstruction experimental model; S4, inputting the multi-channel image into the image reconstruction model in step S3 to obtain a restored image; In step S1, the image reconstruction model is constructed. The image v on the array sensor is defined as the convolution sum of multiple channel input images u and the point spread function (PSF). The image is restored by computational imaging. The objective optimization function is Where N is the number of channels, spectral channels or scene distance; In step S2, in order to obtain the optimal optical metasurface structure, the metasurface is used as an optical device and forward optimization is performed, including the following steps: 1) Use the metasurface as the front end of the optical imaging system and calculate the point spread function (PSF) of the metasurface PSF=|ntff(t(p)·E in )| 2 (2); Where p is the structural parameter of the optical metasurface, t(p) is the transmission coefficient of the optical metasurface, and E in is the incident electric field, ntff(·) represents the conversion from near field to far field; 2) The image v on the array sensor is obtained by convolution of the point spread function PSF of the metasurface and the real scene image u: 3) Reconstruction by solving the inverse scattering problem Using Tikhonov regularization to process equation (4), we get Among them, α is the constant to be optimized, and the convolution kernel function operator is defined as Substituting equation (3) into (5) and making some transformations, we can obtain The conjugate gradient method is used to solve equation (6) to obtain the reconstructed image The objective optimization function L is calculated using formula (1).

2. The single-exposure multi-spectral computational imaging method based on a metasurface according to claim 1, characterized in that: After calculating the target optimization function L, it is also necessary to calculate the derivative of the target optimization function L with respect to the structural parameter p of the optical metasurface and the constant α to be optimized to realize the optimization process of back propagation.

3. The single-exposure multi-spectral computational imaging method based on a metasurface according to claim 2, characterized in that: The back propagation optimization process includes the following steps: (i) Calculate the derivative Combined with the derivative of formula (6), the above formula is replaced by: in, The above equation is solved for Λ using the conjugate gradient method; (ii) Calculate the derivative Combined with the derivative of formula (6), the above formula is replaced by: After obtaining the derivative of the target optimization function relative to the optimization parameter, the system is optimized in combination with the gradient descent algorithm.

4. The single-exposure multi-spectral computational imaging method based on a metasurface according to claim 3, characterized in that: The system is optimized by combining the gradient descent algorithm including: The updating method of the parameters to be optimized is as follows: Among them, β is the dynamic learning rate, and its specific value depends on the selected stochastic gradient descent method and Adam algorithm.

5. A single-exposure multi-spectral computational imaging system based on a metasurface for implementing the single-exposure multi-spectral computational imaging method based on a metasurface according to any one of claims 1 to 4, characterized in that: The single-exposure multi-spectral computational imaging system based on the metasurface includes: An image reconstruction model building module (1) is used to build an image reconstruction model using a metasurface optical system and a conjugate gradient algorithm, and train the image reconstruction model using a gradient descent method to obtain optimal metasurface structure parameters and conjugate gradient algorithm parameters; A point spread function recording module (2), used for collecting multi-channel images through the optimized optical metasurface using an array sensor, and recording point spread functions (PSF) of different channels; The sample image reconstruction module (3) is used to reconstruct the sample image by using the optimized conjugate gradient algorithm combined with the point spread functions (PSFs) of different channels.

6. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the single-exposure multi-spectral computational imaging method based on the metasurface as described in any one of claims 1 to 4.

7. An all-optical imager, characterized in that: The all-optical imager executes the metasurface-based single-exposure multi-spectral computational imaging method as described in any one of claims 1 to 4.

Images