Anti-interference image reconstruction methods, devices, media and electronic equipment
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- XIDIAN UNIV
- Filing Date
- 2026-02-04
- Publication Date
- 2026-06-02
Smart Images
Figure CN122134833A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of photoelectric imaging and target detection technology, and in particular to an anti-interference image reconstruction method, apparatus, medium and electronic equipment. Background Technology
[0002] In practical applications of photoelectric sensing technology, security and industrial inspections often face complex lighting interference such as fish-scale light and backlighting, leading to a sharp decline in the performance of traditional optical imaging systems. Traditional optical imaging systems suffer from bottlenecks such as low spectral resolution, difficulty in extracting polarization information, and weak anti-interference capabilities. Although infrared polarization multispectral imaging technology can acquire multidimensional information, traditional optical systems struggle to achieve dynamic control, and metasurfaces have fixed functions when used alone, making them unable to adapt to dynamic changes in complex lighting. Summary of the Invention
[0003] This application provides an anti-interference image reconstruction method, apparatus, medium, and electronic device, which combines phase change materials and intelligent algorithms to solve the problem of high-quality imaging in complex environments.
[0004] Firstly, this application provides an interference-resistant image reconstruction method, comprising: The infrared radiation light field of the target scene is obtained through an infrared optical lens; Interference light is filtered from the infrared radiation light field by a metasurface modulation module, and multidimensional light field information after passing through the metasurface modulation module is collected by an infrared imaging detector. The metasurface modulation module uses GSST phase change material and controls the infrared spectral band and polarization state by switching between crystalline and amorphous states. The multidimensional light field information collected is reconstructed by the multidimensional light field reconstruction module to obtain a high-resolution spectral image.
[0005] The anti-interference image reconstruction method provided in this embodiment reduces optical loss by using GSST material, enhances polarization response by utilizing chiral metasurfaces, and improves control efficiency by combining intelligent algorithms, enabling the output of high-quality images in complex environments such as fish scale light and backlight.
[0006] For example, the above method further includes: Determine the target circular dichroic response of the interfering light filter; Electromagnetic simulation was used to obtain the structural parameters of the metasurface and the corresponding response data, and a training dataset was obtained. The neural network was then trained using the training dataset. The circular dichroic response of candidate structure parameters is determined by the trained neural network; When the circular dichroic response does not meet the preset conditions, the candidate structural parameters are optimized by a genetic algorithm to obtain the target structural parameters when the circular dichroic response meets the preset conditions. The metasurface modulation module is constructed using the target structural parameters.
[0007] For example, optimizing the candidate structure parameters using a genetic algorithm includes: The optimization objective is minimized using a genetic algorithm. The optimization objective is:
[0008] in, This refers to the transmittance predicted by the neural network based on the candidate structure parameters, specifically the predicted transmittance value for right-handed circularly polarized light. This refers to the ideal transmittance of the candidate structural parameters. This refers to the transmittance prediction value of left-handed circularly polarized light based on the candidate structure parameters predicted by the neural network. This refers to the ideal transmittance of the candidate structural parameters, where M is the structural Boolean matrix of the candidate structural parameters.
[0009] For example, the above method further includes: The response calibration data of different wavelengths and polarization states of the original spectral image are obtained in advance by the metasurface modulation module; Based on the response calibration data, the point spread function for a specific polarization angle is calculated; The point spread function is convolved with the original spectral image to generate a blurred image; The blurred image is input into the decoding network to obtain a spectral image with a specific polarization angle.
[0010] For example, the decoding network is a multi-channel branch network based on the ResUNet architecture, including an expansion path, two contraction paths, and four output paths. The blurred image is segmented into image patches. The image patches and spatial location encoding information are input into the decoding network. The extended path upsamples the image patches to obtain upsampled features. The contracted path extracts multidimensional features through double residual convolutional blocks. After four downsampling operations, the features are fused with the upsampled features of the extended path to output spectral images with four specific polarization angles.
[0011] For example, the step of convolving the point spread function with the original spectral image to generate a blurred image includes: The point spread function is convolved with the original spectral image to obtain an optical response image at a specific polarization angle; A noisy, blurred image is generated based on the optical response image and combined with the camera's spectral sensitivity function.
[0012] The interfering light is fish-scale light in the 3.8-4.2 µm band.
[0013] Secondly, this application provides an interference-resistant image reconstruction apparatus, comprising: An infrared image acquisition module is used to acquire the infrared radiation light field of a target scene through an infrared optical lens; The control module is used to filter interference light from the infrared radiation light field through the metasurface modulation module and to collect multidimensional light field information after passing through the metasurface modulation module through the infrared imaging detector. The metasurface modulation module uses GSST phase change material and controls the infrared spectral band and polarization state by switching between crystalline and amorphous states. The image reconstruction module is used to reconstruct the acquired multidimensional light field information through the multidimensional light field reconstruction module to obtain a high-resolution spectral image.
[0014] Thirdly, this application provides an electronic device including a memory and one or more processors. The memory stores one or more computer programs, each including instructions that, when executed by the processor, cause the electronic device to perform the anti-interference image reconstruction method as described in the first aspect.
[0015] Fourthly, this application provides a computer-readable storage medium storing instructions that, when executed on an electronic device, cause the electronic device to perform the interference-resistant image reconstruction method as described in the first aspect.
[0016] Fifthly, this application provides a computer program product that, when run on an electronic device, causes the electronic device to perform the anti-interference image reconstruction method as described in the first aspect.
[0017] Understandably, the beneficial effects achieved by the anti-interference image reconstruction apparatus, electronic equipment, computer-readable storage medium, and computer program products provided above can be referred to the beneficial effects in the first aspect, and will not be repeated here. Attached Figure Description
[0018] Figure 1 A schematic flowchart illustrating the anti-interference image reconstruction method provided in this application embodiment; Figure 2 This is a structural diagram of the metasurface modulation module in the anti-interference image reconstruction method provided in the embodiments of this application; Figure 3 A schematic diagram of the system architecture for constructing the metasurface modulation module in the anti-interference image reconstruction method provided in this application embodiment; Figure 4This is a schematic diagram of the anti-interference image reconstruction apparatus provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation
[0019] To facilitate a clear description of the technical solutions in the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish identical or similar items with substantially the same function and effect. For example, "first chip" and "second chip" are only used to distinguish different chips and do not limit their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" do not necessarily imply that they are different. It should be noted that in the embodiments of this application, the words "exemplary" or "for example" are used to indicate that they are examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in this application should not be construed as being better or more advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner. In the embodiments of this application, "at least one" means one or more, and "more than one" means two or more.
[0020] It should be noted that "at the time of..." in the embodiments of this application can be either at the instant when a certain situation occurs, or for a period of time after the occurrence of a certain situation. The embodiments of this application do not make specific limitations on this.
[0021] The implementation of this embodiment will now be described in detail with reference to the accompanying drawings.
[0022] This embodiment provides an anti-interference image reconstruction method. For example, this anti-interference image reconstruction method can be applied to various electronic devices such as computers (PCs), tablets, virtual reality / augmented reality devices, wearable devices, industrial computers, and vehicle systems; it can also be applied to servers, cloud computing, server clusters, etc. This embodiment does not impose any special limitations on it.
[0023] Figure 1 A schematic flowchart of the anti-interference image reconstruction method provided in the embodiments of this application is shown.
[0024] like Figure 1 As shown, the interference-resistant image reconstruction method may include the following steps: Step 101: Obtain the infrared radiation light field of the target scene using an infrared optical lens; Step 102: The infrared radiation light field is filtered for interference by the metasurface modulation module, and the multidimensional light field information after passing through the metasurface modulation module is collected by the infrared imaging detector; the metasurface modulation module adopts GSST phase change material and controls the infrared spectral band and polarization state by switching between crystalline and amorphous states. Step 103: Reconstruct the acquired multidimensional light field information using the multidimensional light field reconstruction module to obtain a high-resolution spectral image.
[0025] The metasurface is constructed using GSST phase change material. Specifically: the target circular dichroic response for interfering light filtering is determined; the metasurface structural parameters and corresponding response data are obtained through electromagnetic simulation calculations to obtain a training dataset, which is then used to train a neural network; the trained neural network determines the circular dichroic response of candidate structural parameters; when the circular dichroic response does not meet preset conditions, the candidate structural parameters are optimized using a genetic algorithm to obtain the target structural parameters when the circular dichroic response meets the preset conditions; and a metasurface modulation module is constructed using the target structural parameters.
[0026] In this embodiment, the interfering light is fish-scale light in the 3.8-4.2 µm wavelength band. The metasurface has the following modulation methods: Table 1: Metasurface Control Methods
[0027] The training dataset includes structural parameters and their corresponding response data. The response data includes the ideal transmittance of right-handed and left-handed circularly polarized light corresponding to the structural parameters. Candidate structural parameters can be determined empirically first, and then optimized using a genetic algorithm. When optimizing the candidate structural parameters using the genetic algorithm, the optimization direction is determined by minimizing the optimization objective. The optimization objective is as follows:
[0028] in, The transmittance predicted by the neural network for candidate structure parameters is the right-hand circularly polarized light transmittance. This refers to the ideal transmittance of candidate structure parameters for right-handed circularly polarized light. This refers to the transmittance prediction value of left-handed circularly polarized light based on the candidate structure parameters predicted by the neural network. This refers to the ideal transmittance of the candidate structure parameters for left-handed circularly polarized light, where M is the structure Boolean matrix of the candidate structure parameters.
[0029] The multidimensional light field reconstruction module is a pre-trained deep learning network. By inputting multidimensional square information into this network, high-resolution spectral images of different spectra can be reconstructed. The training process includes: pre-acquiring response calibration data of different wavelengths and polarization states of the original spectral image through a metasurface modulation module; calculating the point spread function at a specific polarization angle based on the response calibration data; convolving the point spread function with the original spectral image to generate a blurred image; and inputting the blurred image into the decoding network to obtain a spectral image at a specific polarization angle.
[0030] The process of convolving the point spread function with the original spectral image to generate a blurred image specifically involves: convolving the point spread function with the original spectral image to obtain an optical response image at a specific polarization angle; and generating a noisy blurred image based on the optical response image and the camera's spectral sensitivity function. The formula is as follows:
[0031] in, This is the optical response image obtained from convolution at a specific polarization angle. Let be a function of camera spectral sensitivity. The intensity of light at that wavelength. λ is the wavelength, and n is the noise signal.
[0032] The decoding network is a multi-channel branch network based on the ResUNet architecture, including an expansion path, two contraction paths, and four output paths. It segments the blurred image into image patches, inputs the image patches and spatial location encoding information into the decoding network, upsamples the image patches in the expansion path to obtain upsampled features, and extracts multidimensional features through double residual convolutional blocks in the contraction path. After four downsampling operations, these features are fused with the upsampled features from the expansion path to output spectral images with four specific polarization angles.
[0033] In this embodiment, an infrared optical lens receives and converges the infrared radiation of the target scene; a metasurface modulation module based on the phase change material GSST is used to perform spectral gating and polarization modulation on the incident light field to suppress fish-scale light and backlight interference; an infrared imaging detector is used to collect multidimensional light field information modulated by the metasurface; a multidimensional light field reconstruction processing module is used to process the collected light field signal through a control strategy that integrates deep learning and heuristic algorithms to reconstruct high-resolution spectral and polarization images. The metasurface modulation module suppresses complex lighting conditions such as fish-scale light and backlight while simultaneously acquiring the spectral and polarization information of the target; the multidimensional light field reconstruction processing module uses a deep learning-based spectral polarization image reconstruction network to achieve anti-fish-scale light and anti-backlight imaging under complex lighting conditions.
[0034] Figure 2 A structural diagram of the metasurface modulation module of this embodiment is shown, as follows: Figure 2 As shown, the metasurface is designed with GSST phase change material nanofibers and a gold layer of Au as the substrate. By designing the two ends as gate voltages, when a voltage pulse is applied to the metasurface by the gate voltage, the crystalline and amorphous states of GSST can be realized.
[0035] Figure 3 The system architecture for constructing the metasurface modulation module in this embodiment is shown. For example... Figure 3 As shown, the structural design method of the metasurface modulation module is as follows: the meta-atomic topology is optimized by combining deep learning neural network and microbial genetic algorithm (MGA) to obtain the target circular dichroism (CD) response.
[0036] Includes the following steps: Step 1: Collect metasurface structure-response data through electromagnetic simulation calculations to construct a training dataset.
[0037] Step 2: Design and train a positive prediction neural network to establish a fast prediction model from structural parameters to optical response.
[0038] Step 3: Combine the trained neural network with the MGA algorithm, using the circular dichroic response (CD) as the optimization objective, to automatically search for the optimal structural parameters. CD refers to the absolute difference in transmittance between left-handed circularly polarized (LCP) incident and right-handed circularly polarized (RCP) incident.
[0039] Step 4: Adjust the metasurface structure parameters using an optimization framework consisting of a trained neural network and optimization modules. The optimization objective is to minimize the cost function.
[0040] in, The transmittance predicted by the neural network for candidate structure parameters is the right-hand circularly polarized light transmittance. This refers to the ideal transmittance of candidate structure parameters for right-handed circularly polarized light. This refers to the transmittance prediction value of left-handed circularly polarized light based on the candidate structure parameters predicted by the neural network. This refers to the ideal transmittance of the candidate structure parameters for left-handed circularly polarized light, where M is the structure Boolean matrix of the candidate structure parameters.
[0041] Step 5: Evaluate the performance of each generated structure through a cost function, driving the structure to evolve towards the optimal solution.
[0042] For example, the multidimensional light field reconstruction module is constructed through the following steps: Step 1: Obtain different wavelengths using a phase change metasurface module ( ) and polarization state ( The calibration data under ( ) is used to construct a dataset. where (i, j) are the superpixel coordinates; Step 2: Based on the Fresnel diffraction principle, calculate the dual series of the point spread function (PSF) at a specific polarization angle, where the PSF satisfies the formula:
[0043] in, Distinguish between 0°, 45°, 90°, and 135° polarization channels; variables This represents the radial coordinate within the PSF, where λ is the incident light wavelength. This represents the complex amplitude response of the metasurface under different polarization states; Applying the Fresnel diffraction principle, the dual series of a set of polarization-sensitive point spread functions were calculated, and then these... Convolving with an image can generate polarization-specific optical response data.
[0044] Step 3: Convolve the original spectral image with the PSF, and combine it with the camera's spectral sensitivity function. Generate a noisy, blurred image that satisfies the formula:
[0045] Where n is the noise signal; Step 4: The output of the multi-channel branch based on the ResUNet architecture is used as the decoding network to segment the image into image blocks. The spatial location coding information is then input into the network. The network contains two shrinking paths and four output paths. The shrinking path extracts multi-dimensional features through double residual convolutional blocks and fuses them with the upsampled features of the expansion path after four downsampling iterations. Step 5: The decoding network outputs multispectral images at four specific polarization angles to complete the synchronous reconstruction of spectral and polarization information.
[0046] The phase change metasurface module is fabricated using micro-nano fabrication technology, with strict control over etching precision and phase change material deposition quality to ensure that the deviation between structural parameters and design values is within the allowable range of the process.
[0047] The method provided in this embodiment has the following effects: (1) In the spectral dimension, by controlling the switching of GSST between amorphous and crystalline states, the optical constant of the unit in the infrared band is changed, and selective absorption is achieved in the fish scale light interference band (such as 3.8-4.2 µm) to suppress high reflection interference; (2) In the polarization dimension, significant circular dichroism is generated through chiral structure design, which enables differentiated transmission of left- and right-hand circularly polarized light and enhances the polarization contrast between the target and the background. (3) Spectral and polarization modulation are synergistically achieved through the same physical structure: the GSST phase transition changes the response intensity of the chiral structure without changing its chiral sign, ensuring that the polarization selection function is maintained while suppressing interference spectra; (4) The overall system first identifies the interference features, then determines the optimal phase distribution and structural parameters through optimization algorithms, so that spectral gating and polarization modulation can be realized simultaneously on a single metasurface module. Finally, the image information of anti-fish scale light interference is recovered through the deep learning-based spectral polarization image reconstruction network, and multi-dimensional light field information is output while suppressing fish scale light and backlight interference in the scene.
[0048] Furthermore, this embodiment also provides an anti-interference image reconstruction apparatus, which can be used to perform the above-described anti-interference image reconstruction method. For example... Figure 4 As shown, the anti-interference image reconstruction device 300 specifically includes: an infrared image acquisition module 301, used to acquire the infrared radiation light field of the target scene through an infrared optical lens; a modulation module 302, used to filter interference light from the infrared radiation light field through a metasurface modulation module, and to collect multi-dimensional light field information after passing through the metasurface modulation module through an infrared imaging detector; the metasurface modulation module uses GSST phase change material, and controls the infrared spectral band and polarization state by switching between crystalline and amorphous states; and an image reconstruction module 303, used to reconstruct the collected multi-dimensional light field information through a multi-dimensional light field reconstruction module to obtain a high-resolution spectral image.
[0049] The specific details of each module or unit in the above-mentioned anti-interference image reconstruction device have been described in detail in the corresponding anti-interference image reconstruction method, so they will not be repeated here.
[0050] This application also provides an electronic device. Figure 5 A schematic diagram of the structure of an electronic device suitable for implementing embodiments of the present disclosure is shown. Figure 5 The electronic device 600 shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments disclosed herein.
[0051] like Figure 5 As shown, the electronic device 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes based on a program stored in a read-only memory (ROM) 602 or a program loaded from a storage section 608 into a random access memory (RAM) 603. The RAM 603 also stores various programs and data required for system operation. The CPU 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.
[0052] The following components are connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 610 as needed so that computer programs read from it can be installed into storage section 608 as needed.
[0053] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 609, and / or installed from removable medium 611. When the computer program is executed by central processing unit (CPU) 601, it performs the functions defined in the embodiments of this application.
[0054] For example, when the computer program is executed by the central processing unit (CPU) 601, it can perform the following: acquire the infrared radiation light field of the target scene through an infrared optical lens; filter the infrared radiation light field for interference light through a metasurface modulation module; and acquire the multidimensional light field information after passing through the metasurface modulation module through an infrared imaging detector; the metasurface modulation module uses GSST phase change material and controls the infrared spectral band and polarization state by switching between crystalline and amorphous states; and reconstruct the acquired multidimensional light field information through a multidimensional light field reconstruction module to obtain a high-resolution spectral image.
[0055] It should be noted that the computer-readable medium disclosed herein may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.
[0056] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0057] The units described in the embodiments of this disclosure can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the unit itself.
[0058] In another aspect, this application also provides a computer-readable medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device. The computer-readable medium carries one or more programs, which include instructions that, when executed by the electronic device, cause the electronic device to perform the methods described in the above embodiments.
[0059] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.
[0060] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. An anti-interference image reconstruction method, characterized in that, include: The infrared radiation light field of the target scene is obtained through an infrared optical lens; Interference light is filtered from the infrared radiation light field by a metasurface modulation module, and multidimensional light field information after passing through the metasurface modulation module is collected by an infrared imaging detector. The metasurface modulation module uses GSST phase change material and controls the infrared spectral band and polarization state by switching between crystalline and amorphous states. The multidimensional light field information collected is reconstructed by the multidimensional light field reconstruction module to obtain a high-resolution spectral image.
2. The anti-interference image reconstruction method according to claim 1, characterized in that, Also includes: Determine the target circular dichroic response of the interfering light filter; Electromagnetic simulation was used to obtain the structural parameters of the metasurface and the corresponding response data, and a training dataset was obtained. The neural network was then trained using the training dataset. The circular dichroic response of candidate structure parameters is determined by the trained neural network; When the circular dichroic response does not meet the preset conditions, the candidate structural parameters are optimized by a genetic algorithm to obtain the target structural parameters when the circular dichroic response meets the preset conditions. The metasurface modulation module is constructed using the target structural parameters.
3. The anti-interference image reconstruction method according to claim 2, characterized in that, The candidate structure parameters are optimized using a genetic algorithm, including: The optimization objective is minimized using a genetic algorithm. The optimization objective is: in, This refers to the transmittance predicted by the neural network based on the candidate structure parameters, specifically the predicted transmittance value for right-handed circularly polarized light. This refers to the ideal transmittance of candidate structure parameters for right-handed circularly polarized light. This refers to the transmittance prediction value of left-handed circularly polarized light based on the candidate structure parameters predicted by the neural network. This refers to the ideal transmittance of the candidate structure parameters for left-handed circularly polarized light, where M is the structure Boolean matrix of the candidate structure parameters.
4. The anti-interference image reconstruction method according to claim 1, characterized in that, Also includes: The response calibration data of different wavelengths and polarization states of the original spectral image are obtained in advance by the metasurface modulation module; Based on the response calibration data, the point spread function for a specific polarization angle is calculated; The point spread function is convolved with the original spectral image to generate a blurred image; The blurred image is input into the decoding network to obtain a spectral image with a specific polarization angle.
5. The anti-interference image reconstruction method according to claim 4, characterized in that, The decoding network is a multi-channel branch network based on the ResUNet architecture, including an expansion path, two contraction paths, and four output paths. The blurred image is segmented into image patches. The image patches and spatial location encoding information are input into the decoding network. The extended path upsamples the image patches to obtain upsampled features. The contracted path extracts multidimensional features through double residual convolutional blocks. After four downsampling operations, the features are fused with the upsampled features of the extended path to output spectral images with four specific polarization angles.
6. The anti-interference image reconstruction method according to claim 1, characterized in that, The step of convolving the point spread function with the original spectral image to generate a blurred image includes: The point spread function is convolved with the original spectral image to obtain an optical response image at a specific polarization angle; A noisy, blurred image is generated based on the optical response image and combined with the camera's spectral sensitivity function.
7. The anti-interference image reconstruction method according to claim 1, characterized in that, The interfering light is fish-scale light in the 3.8-4.2 µm wavelength band.
8. An anti-interference image reconstruction device, characterized in that, include: An infrared image acquisition module is used to acquire the infrared radiation light field of a target scene through an infrared optical lens; The control module is used to filter interference light from the infrared radiation light field through the metasurface modulation module and to collect multidimensional light field information after passing through the metasurface modulation module through the infrared imaging detector. The metasurface modulation module uses GSST phase change material and controls the infrared spectral band and polarization state by switching between crystalline and amorphous states. The image reconstruction module is used to reconstruct the acquired multidimensional light field information through the multidimensional light field reconstruction module to obtain a high-resolution spectral image.
9. A computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the interference-resistant image reconstruction method as described in any one of claims 1 to 7.
10. An electronic device, characterized in that, The device includes a processor and a memory, the memory storing one or more computer programs, the one or more computer programs including instructions that, when executed by the electronic device, cause the electronic device to perform the interference-resistant image reconstruction method according to any one of claims 1-7.