Research design, preparation and test method of wide-angle low-dimensional visible-near-infrared micro-nano photonic device for miniaturized satellite-borne spectral imager

Abstract

The visible-near infrared micro-nano photonic device designed by the invention has the advantages of polarization insensitivity, wide angle, high manufacturability, easiness in integration, wide spectrum transmission spectrum curve with strong spectrum curve controllability and the like. As a spectrum modulation element, band-pass imaging can be realized, the luminous flux can be enhanced, the view field range can be expanded, and the processing and preparation difficulty can be reduced. A researched and designed micro-nano photonic device and a spectrum detection system are fused and packaged at the front end of a large-area array detector to obtain a novel spectrum detector. When a target spectrum is modulated by a device, a gray value of each pixel can be obtained, and spectrum reconstruction comparative analysis is performed on the target spectrum in combination with a proposed low-rank logarithmic regularization matrix truncation spectrum reconstruction algorithm based on a micro-nano photonic device, so that high-precision spectrum reconstruction can be performed on the target spectrum in a large disturbance environment, and the accuracy of spectrum reconstruction is improved. And target spectral features can be stably captured. And an important research support is provided for subsequently realizing the development of a miniaturized satellite-borne spectrum reconstruction imager.

Description

Technical field:

[0001] The invention relates to a research, design, preparation and testing method of a wide-angle low-dimensional visible-near-infrared micro-nano photonic device for a miniaturized satellite-borne spectroscopic imager, and belongs to the technical field of optical devices and aerospace applications. Background technology:

[0002] With the continuous increase in the demand for remote sensing missions, satellite-borne spectroscopic imagers have attracted much attention in the past decade. It not only plays an important role in the military and national defense field, helping to detect and identify camouflaged targets, but is also widely used in agricultural surveys, forest and farmland monitoring, disaster prevention and mitigation, and environmental pollution monitoring, making a significant contribution to the rapid development of my country's national economy. The integration of spectral imaging technology with advanced technologies in the fields of optics, micro-nano photonics, optoelectronics, electronics, information processing, computer science, and precision machinery, and multidisciplinary cross-field research have driven the continuous development of satellite-borne spectral imaging technology in the field of aerospace applications, which is beneficial to the miniaturization of satellite-borne spectroscopic imagers. At present, photonic crystals and surface plasmons are considered to be the two most important means to realize micro-nano photonic devices. The application of micro-nano photonic devices in spectral imaging technology can effectively solve many problems faced by existing aerospace spectroscopic imagers, such as complex optical system design, small relative aperture, weak light collection ability, small field of view coverage, and sensitive application environment. This application can not only enhance the compactness of the optical-mechanical structure of the spectral imager, but also significantly reduce the structural complexity of the optical detection system, reduce its volume and weight, and thus make it easier to achieve integration and miniaturization. In addition, it can also effectively improve the light flux of the optical detection system and enhance the stability of the optical-mechanical structure.

[0003] Under the above research background, the present invention starts from the theoretical analysis of photonic crystals and plasma photonic crystals, establishes a photonic crystal structure model and a photonic crystal structure model coupled with a plasma grating waveguide, so as to realize the application research of micro-nano photonic devices in computational spectral imaging. Considering the application problem of the spectrum band of the satellite-borne spectral reconstruction imager, the micro-nano photonic devices in the visible-near infrared spectrum band are deeply studied and analyzed. Through these studies, a one-to-many complex mapping from the spectral dimension to the spatial dimension is realized, and the spectral response is effectively modulated, so as to obtain spectral information more efficiently and accurately, and at the same time, significant breakthroughs have been made in spatial resolution, spectral resolution and compactness of the instrument structure. In the demand for computational spectral reconstruction, a mathematical model for computational spectral reconstruction is established for analysis, and a low-dimensional broadband transmittance spectral curve constituting the micro-nano photonic device is selected. Combined with the optical properties of the proposed low-dimensional broadband transmittance spectral curve, based on the photonic crystal structure, the micro-nano photonic device is studied and designed in the visible-near infrared spectrum band (400-900nm). In addition, by combining with a typical spectral detection system and a computational spectral reconstruction algorithm, high-precision spectral reconstruction research and analysis of the detection target was carried out. Finally, a large-aperture, miniaturized spaceborne spectral reconstruction imager was developed, and its performance was verified through ground imaging experiments.

[0004] At the same time, in order to solve the problems of complex optical system design, small relative aperture, weak light collection ability, small field of view coverage, and sensitive application environment in existing aerospace satellite-borne spectral imagers. The micro-nano photonic devices designed by the research are applied to spectral imaging technology to reduce the structural complexity of the optical detection system, improve the light flux of the optical detection system, increase the angle range of the incident light, enhance the compactness and stability of the optical-mechanical structure, reduce its volume and weight, and facilitate integration and miniaturization. It is packaged in the detector and then applied to more complex environments. At the same time, the micro-nano photonic devices designed by the research must also ensure the feasibility of processing and preparation, improve the compatibility of the device and the detector, and reduce the difficulty of manufacturing.

[0005] The visible-near-infrared micro-nano photonic device designed by the present invention has the advantages of polarization insensitivity, wide angle, high manufacturability, easy integration, and a wide-band transmission spectrum curve with strong controllability of the spectrum curve. Using it as an optical spectroscopic element can realize bandpass imaging, enhance light flux, improve the field of view, and reduce the difficulty and complexity of processing and preparation. The micro-nano photonic device designed in the research is integrated with the spectral detection system (large array GSENSE5130 detector), and combined with the computational spectrum reconstruction algorithm proposed in this article, the spectrum reconstruction of typical ground objects is compared and analyzed. The low-rank logarithmic regularized matrix truncated spectrum reconstruction algorithm based on micro-nano photonic devices proposed in the present invention can realize high-precision spectrum reconstruction of the target spectrum in a large disturbance environment, and then stably capture the spectral characteristics of the ground object target. It provides important research support for the subsequent development of miniaturized satellite-borne spectrum reconstruction imagers. Summary of the invention:

[0006] The technical problem to be solved by the present invention is to provide a wide-angle low-dimensional micro-nano photonic device that can be used to miniaturize a satellite-borne spectral imager and is applied in the visible to near-infrared spectrum. The present invention starts from the structural research and design of the micro-nano photonic device, and then processes and prepares it. Then the processed and prepared visible-near-infrared micro-nano photonic device is tested and characterized to ensure that the device meets the spectral reconstruction requirements. Then it is encapsulated in the front end of the large-array GSENSE5130 detector to obtain a new spectral detector, and its spectral response performance and application environment are tested to ensure that the new spectral detector still meets the imaging requirements in a complex application environment. Then, the new spectral detector is applied to the satellite-borne spectral imager to realize the development of a miniaturized satellite-borne spectral imager, and imaging experiments are carried out. The image color restoration and spectral reconstruction of the scene detection target are experimentally verified in combination with the computational spectral reconstruction algorithm. The wide-angle low-dimensional visible-near-infrared micro-nano photonic device finally realized is from research and design to processing and preparation to testing, and finally applied to the satellite-borne spectral imager to realize the integrated method of miniaturization development.

[0007] The present invention adopts the following technical solutions to solve the above technical problems:

[0008] Step 1: Establishment and selection analysis of spectral reconstruction mathematical model

[0009] Since a large number of selected spectral curves will cause data redundancy between spectral curves, and then increase the morbidity of the linear equations, under a large disturbance environment, the accuracy of spectral reconstruction is affected. Therefore, in order to achieve high-precision spectral reconstruction, it is particularly important to reduce the number of spectral curves and study low-dimensional micro-nano photonic devices in spectral detection systems to address this problem. The present invention performs calculation and analysis through a spectral reconstruction mathematical model, and further proposes a truncated spectral reconstruction (IReg) algorithm based on a low-dimensional logarithmic regularized matrix to solve the problem of excessively high condition numbers. This method further uses the condition number as the objective function for optimization, and through the numerical analysis of the condition number under the above-mentioned different numbers of curves, it is possible to reduce the morbidity of the design matrix, improve the accuracy of spectral reconstruction, and fit the spectral discrete points to obtain a high-precision spectral reconstruction curve.

[0010] Step 2: Propose an algorithm based on low-dimensional logarithmic regularization matrix truncated spectrum reconstruction (IReg)

[0011] The standard regularization algorithm has the problem that the correction method for large singular values ​​is too strong and the correction for small singular values ​​is too weak. The proposed truncated spectrum reconstruction (IReg) algorithm based on low-dimensional logarithmic regularization matrix can achieve truncation and retention of large singular values ​​without correction, truncation logarithmic correction of singular values ​​between large and small singular values, and truncation and discarding of small singular values. At the same time, the proposed logarithmic regularization matrix gives different correction amplitudes to different singular values, giving smaller corrections to larger singular values ​​and larger corrections to smaller singular values.

[0012] Step 3: Research and design of wide-angle low-dimensional visible-near-infrared micro-nanophotonic devices

[0013] The research designs a composite grating array micro-nano photonic device for the visible-near infrared (400-900nm) spectrum band to meet the needs of spectral reconstruction. Based on the photonic crystal structure model, the micro-nano photonic crystal structure is used to adjust the structure and geometric parameters of the two-dimensional unit of the photonic crystal to achieve the regulation of the intensity and phase of the electromagnetic field. Many groups of different transmittance spectrum curves can be integrated on a small area to achieve spectral modulation effects, simplify the optical system, and obtain specific scene information with stronger adjustment capabilities, more efficiency, more accuracy, and more robustness. Each transmittance curve is designed and simulated through the micro-nano photonic crystal structure. The optical film of the structure is composed of many layers of dielectric films with different optical constants. The film system Essential Macleod design software of the transmission matrix method and the multi-physics simulation software COMSOL Multiphysics of the finite element method are used to design and simulate within the required spectral wavelength range. According to the required broadband transmittance spectrum curve, the boundary conditions of the micro-photonic crystal structure design are optimized, the dielectric material selection is completed, and the film system parameter optimization process is controlled. The two sets of simulation results are compared and analyzed based on the material stacking method and the material thickness of each layer of the film system for each broadband transmittance spectral curve. This enhances the reliability of the simulation results and determines the film thickness of the broadband transmittance spectral curve, laying the foundation for subsequent film system processing and preparation.

[0014] According to the current application demand analysis of large-field-of-view satellite-borne spectrometers, the field of view of miniaturized satellite-borne spectrometers is not only limited by the design of the optical system, but also by the size of the detector and the sensitivity of the micro-nano photonic devices packaged at the front end of the detector to the incident angle. Natural light first enters the optical system of the satellite-borne spectrometer. After the aberration correction and focusing of the optical system, the incident light hits the detector at a certain cone angle, so that the light incident on the front end of the detector containing the micro-nano photonic devices has a certain range of incident angles, thereby affecting the imaging effect of the spectrometer. Therefore, for the system in which the visible-near-infrared micro-nano photonic device is packaged at the front end of the detector as a spectral modulation element, when the light passing through the optical system is incident on the surface of the micro-nano photonic device, there is a certain incident angle, and the range of the incident angle is related to many factors, such as the optical system parameters and the optical-mechanical assembly accuracy. In response to this problem, the influence of the change of the incident angle on the optical characteristics of the spectral curve of the visible-near-infrared micro-nano photonic device is further simulated and analyzed, and it is judged whether the increase of the incident angle will affect the effect of spectral reconstruction. Visible-near-infrared micro-nanophotonic devices that allow large-angle light incidence are helpful for the development of miniaturized satellite-borne spectral imagers with large field of view.

[0015] When selecting the broadband transmittance spectrum curve that constitutes the micro-nano photonic device, it is combined with the computational spectrum reconstruction algorithm to enhance the controllability of the spectrum curve. It reduces the difficulty of processing, is easy to integrate and package into the front end of the detector, reduces the complexity of the optical system design, effectively reduces the impact of the external environment on the device performance, improves the stability and robustness of the system, and realizes the development of integration and miniaturization.

[0016] Step 4: Processing and preparation of visible-near infrared micro-nano photonic devices

[0017] The micro-nano photonic device designed in the research adopts a combination of lithography, coating and lift-off processes to process and prepare a wide-band filter film system arranged in a composite grating array, thereby realizing the preparation of visible-near-infrared micro-nano photonic devices.

[0018] First, through the photolithography process, the UV lithography machine (SUSS MA6) is used to transfer the pattern on the mask to the glass surface of the silicon dioxide (SiO2) substrate coated with photoresist through the exposure process. The photolithography process flow is ①HMDS→②Smoothing→③Exposure→④Development: ①HMDS is oven-coated, the substrate is placed in the HMDS professional oven, and HMDS is vaporized and deposited on the substrate surface; ②Smoothing uses a specific negative glue, and the photoresist is spin-coated on the substrate surface using the spinning equipment, and after baking, it flows into the next process; ③Exposure uses a UV lithography machine to expose the designed mask pattern, and the pattern on the mask is copied to the substrate surface coated with photoresist; ④The exposed substrate is placed in the developer, the photoresist and the developer will react chemically, and finally form a pattern opposite to the pattern on the mask, forming the strip gap that needs to be coated, and the developed substrate is baked to form a photolithography pattern sheet substrate.

[0019] Secondly, through the coating process, using the physical vapor deposition method (PVD), the prepared photolithography pattern substrate is placed in the vacuum coating equipment RSP800 magnetron sputtering coating machine, and the equipment is placed with SiO2 and niobium pentoxide (Nb2O5) coating targets. The target material and the gas introduced into the chamber are reacted and deposited, and then deposited layer by layer on the surface of the substrate. Then, by controlling the thickness of the SiO2 and Nb2O5 deposition, the designed thickness of each layer of film is achieved, thereby forming a substrate strip with a thin film. Then, using the lift-off process, the coated substrate is immersed in the degumming solution, and the thin films outside the film system are stripped off together with the photoresist, thereby completing the preparation of a set of thin films.

[0020] Finally, by repeating the photolithography, coating, and lift-off process, the required composite grating array wide-band filter film system is processed and prepared one by one.

[0021] Step 5: Testing and characterization of visible-near infrared micro-nanophotonic devices

[0022] The prepared micro-nano photonic devices are compared and analyzed by measuring the spectral curve of each group and the spectral curve calculated theoretically to determine the effect of the processed spectral curve and whether it meets the requirements of spectral reconstruction. It is verified that the spectral curve of the wide-band filter film system arranged in a composite grating array on the same substrate is processed by combining the lithography process, coating process and lift-off process. The similarity between the spectral curve and the spectral curve of the theoretical simulation design can reach more than 96%, which meets the processing error requirement range and realizes the preparation of visible-near-infrared micro-nano photonic devices.

[0023] At the same time, in order to verify whether the processed spectral curve meets the requirements of spectral reconstruction, the actual test transmittance curve is further combined with the proposed calculation spectrum reconstruction IReg algorithm to reduce the morbidity of the linear equations, and the spectral discrete points are fitted to obtain a high-precision spectral reconstruction curve to meet the requirements of spectral reconstruction.

[0024] The processed visible-near-infrared micro-nano photonic devices are first tested for the thickness of each film system and the parallelism between the film systems using a scanning electron microscope and a confocal microscope. Secondly, the thickness and width of the film system for each spectral curve are measured at multiple positions. Since one of the keys to spectral imaging lies in high-quality spectrally modulated optical films, the uniformity of the thickness distribution of optical films and the parallelism between the film systems of each group of optical films are important indicators of optical components. Poor uniformity will lead to differences in optical performance at different positions of the optical components, seriously affecting the transmission uniformity. If the film systems are not parallel to each other, the utilization rate of the imaging pixels in the middle area of ​​the detector pixels will be affected when they are integrated and packaged with the detector. When preparing optical thin film components, it is very important to control the uniform distribution of the deposition thickness on the component.

[0025] Step 6: Integrated packaging and performance testing of visible-near infrared micro-nano photonic devices

[0026] By cooperating with the 58th Institute of China Electronics Technology Group Corporation, the prepared composite grating array micro-nano photonic device was packaged at the front end of the large-array GSENS5130 detector in a dust-free nitrogen environment through the combination of Zeiss SteREO Discovery V8 optical microscope, achieving stable integrated packaging that meets aerospace-grade applications.

[0027] In order to ensure that the composite grating array micro-nano photonic device and the large array GSENSE5130 detector pixel are placed relatively parallel, and to ensure that the parallelism between the film system and the detector pixel reaches more than 95%. In order to effectively cover the imaging channel of the detector, to ensure the utilization rate of the imaging pixel in the middle area containing the film system, and then to ensure the imaging quality of the integrated miniaturized satellite-borne spectrometer. Therefore, two methods are used to monitor the integrated package and meet the parallelism and alignment requirements between the micro-nano photonic device and the detector pixel.

[0028] A new type of spectral detector is obtained by packaging a composite grating array micro-nano photonic device at the front end of the detector. The spectral response curve of the spectral sensitivity characteristics of the detector is tested in a dark box environment, that is, the quantum efficiency curve is tested.

[0029] To ensure that the new spectral detector can still work normally when used in complex environments, thermal vacuum experiments and thermal cycle experiments were carried out on the imaging module, and appearance inspections, electronic performance tests and imaging effect inspections were carried out on it, so as to early discover and eliminate defective components mixed in the product during the development process, ensure the reliability of the electronic stand-alone operation, as well as the reliability of the new spectral detector and other electronic components and related processes.

[0030] The imaging effect obtained after experimental testing is normal, without bad pixels, missing stripes, water ripples and other phenomena. This proves that the imaging quality can still be guaranteed in a relatively complex application environment and the expected imaging effect can be achieved. It also proves the reliability of the new spectral detector and other electronic components and related processes.

[0031] Step 7: Imaging experiment and reconstruction analysis of miniaturized spaceborne spectral imager

[0032] In order to ensure that the ground prototype of the miniaturized spaceborne spectral reconstruction imager can image actual scenes. We chose a room with a height of about 35m to conduct field imaging tests on ground scenes at a straight-line distance of about 5km. The miniaturized spaceborne spectral reconstruction imager adopts push-broom imaging, which is automatically controlled and rotated by a two-dimensional CNC turntable. TDI push-broom imaging is performed on ground scenes with a step length of 0.08°, and the push-broom imaging angle range is about 25°. At the same time, the Cameralink quick-view system of Wuhan Zhuomu Technology Co., Ltd. is used, which can display the received image data in real time and display part or all of it as required. It also detects the known content in the collected data in real time, finds error alarms, and supports preset storage channel data, which is stored while displaying. The imaging of distant scenes by the miniaturized spaceborne spectral reconstruction imager proves that the space camera design can provide high-quality images. During the field imaging process, the parameters such as focal length, TDI integration level and gain of the spectral reconstruction imager are adjusted and set, and multiple comparisons are made to find the best image point and obtain the best imaging effect. 25 groups of grayscale images under different spectral response curves taken by the miniaturized satellite-borne spectral reconstruction imager prepared through research.

[0033] The push-broom spectral reconstruction inversion of the miniaturized satellite-borne spectral reconstruction imager is performed under extremely fast relative motion, so the deviation of the reflected light or emitted light of the ground object to the detector is not large, that is, the light intensity of the same ground object is basically consistent when mapped to each film strip, which is the basis for correct spectral inversion. Due to the large amount of image data, we selected a part of the field image for spectral and image reconstruction and restoration. The test results show that a new detector obtained by integrating micro-nano photonic devices as spectral modulation elements and detectors, combined with the proposed computational spectral reconstruction algorithm, has been developed to develop a miniaturized satellite-borne spectral imager. Compared with the existing satellite-borne spectrometer, it has improved the light flux and field of view, enhanced stability, is insensitive to environmental influences, improved durability and reliability, and reduced the difficulty and cost of manufacturing. Finally, the main indicators of the miniaturized satellite-borne spectral imager developed by using micro-nano photonic devices were tested.

[0034] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0035] The present invention proposes a method for researching, designing, preparing and testing a wide-angle, low-dimensional visible-near-infrared micro-nano photonic device for a miniaturized satellite-borne spectroscopic imager. The method is an integrated, comprehensive and systematic method for solving and realizing the integration of micro-nano photonic devices from researching and designing to processing and preparing to testing, and finally applying them to satellite-borne spectroscopic imagers to realize miniaturized development.

[0036] The technical problem to be solved is to provide a wide-angle low-dimensional micro-nano photonic device that can facilitate the miniaturization of satellite-borne spectral imagers and apply it in the visible to near-infrared spectrum.

[0037] The present invention starts with structural research and design of micro-nano photonic devices, then processes and prepares them, and then tests and characterizes the processed visible-near-infrared micro-nano photonic devices to ensure that the devices meet the requirements of spectral reconstruction.

[0038] Furthermore, through stable integrated packaging for aerospace-grade applications, it is packaged at the front end of the large-array GSENSE5130 detector to obtain a new spectral detector, and its spectral response performance and application environment are tested to ensure that the new spectral detector can still meet the imaging requirements in complex application environments.

[0039] Then, the new spectral detector was applied to the satellite-borne spectral imager, realizing the development of a miniaturized satellite-borne spectral imager. Imaging experiments were carried out, and the image color restoration and spectral reconstruction of the scene detection target were experimentally verified in combination with the computational spectral reconstruction algorithm.

[0040] Ultimately, the wide-angle, low-dimensional visible-near-infrared micro-nano photonic devices were developed from research and design to processing and preparation to testing, and finally applied to satellite-borne spectroscopic imagers to achieve miniaturized integrated development. Description of the drawings:

[0041] Figure 1 Spectral reconstruction and analysis steps of visible-near infrared micro-nanophotonic devices

[0042] Figure 2 Theoretical flow chart of mathematical model for computational spectrum reconstruction

[0043] Figure 3 Schematic diagram of the composite grating array arrangement and combination of visible-near infrared micro-nano photonic devices

[0044] Figure 4 Comparison of spectral reconstruction results of different ground objects under (a)~(h)30dB, (i)~(p)20dB and (q)~(x)17dB large disturbance environments, with 100 random errors applied, using IReg, L1 and L2 algorithms

[0045] Figure 5 Evaluation index of spectral reconstruction accuracy of ground objects

[0046] Figure 6 Schematic diagram of the design of SiO2 / Nb2O5 photonic crystal multilayer structure

[0047] Figure 7 A1-A24 are the film thickness diagrams of 24 photonic crystal structures

[0048] Figure 8 M1-M24 and C1-C24 are two simulation results of Essential Macleod and COMSOL Multiphysics, respectively. Comparison of 24 sets of broadband spectral transmittance curves obtained

[0049] Fig. 9 Under TEM conditions, when the incident angle α = 0 ~ 40°, the changes of 24 broadband transmittance spectrum curves

[0050] Fig.10 Processing process flow chart for preparing 24 composite grating array wide-band filter film system

[0051] Fig.11 The actual picture of the composite grating array filter film system after processing and preparation

[0052] Fig.12 24 spectral curves of the composite grating array micro-nano photonic device Comparison of the theoretical simulation spectrum curve (S1~S24) and the actual test spectrum curve (E1~E24) of the film system

[0053] Fig.13 Confocal microscopy (CLSM) tests the width and parallelism of 24 sets of thin films of the filter optical device. The film width is 300μm, the interval width of the non-film system is 200μm, and the grating period is 500μm.

[0054] Fig.14 The top image shows the film thickness tested by CLSM, and the bottom image shows the film thickness tested by scanning electron microscopy (SEM), showing the thickness of the 21st film system.

[0055] Fig.15 Scanning electron microscope SEM test of the same film system (film system 21), the film thickness image at different positions on the same cut surface

[0056] Fig.16 Mean and error range of film width of 34 groups of visible-near infrared micro-nano photonic devices

[0057] Fig.17 Process flow of packaging composite grating array micro-nano photonic devices in the front end of large array detectors

[0058] Fig.18 Film imaging of composite grating array micro-nano photonic devices

[0059] Fig.19 A physical picture of a composite grating array micro-nano photonic device packaged at the front end of the detector

[0060] Fig. 20 Quantum efficiency test principle block diagram

[0061] Fig.21 Quantum efficiency spectral response curve of the spectral sensitivity characteristics of the new spectral detector without film system

[0062] Fig. 22 Testing environment and equipment for spectral response curve of new spectral detector

[0063] Fig.23 Spectral response curves of 24 spectral film systems in the 400-900nm spectral range

[0064] Fig.24 Color spectral sensitivity characteristics of a new spectral detector

[0065] Fig.25 (a)(b)(c) are images of the new spectrometer detector before the test, and (d)(e)(f) are images of the new spectrometer detector after the test.

[0066] Fig.26 Miniaturized spaceborne spectroscopic imager, (a) optical system diagram, (b) ground prototype

[0067] Fig. 27 The miniaturized satellite-borne spectral imager captured 25 sets of images and local magnified images of the spectral response curves of different film systems.

[0068] Fig.28 Color restoration effect of images collected by the miniaturized space-borne spectral reconstruction imager

[0069] Fig.29 Reconstruction effect of some spectral curves in images collected by miniaturized spaceborne spectral reconstruction imager Fig.30 Main index test of miniaturized satellite-borne spectral imager based on micro-nano photonic devices Specific implementation method:

[0070] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings:

[0071] The technical problem to be solved by the present invention is to provide a wide-angle low-dimensional micro-nano photonic device that can be used to miniaturize a satellite-borne spectral imager and is applied in the visible to near-infrared spectrum. The present invention starts from the structural research and design of the micro-nano photonic device, and then processes and prepares it. Then the processed and prepared visible-near-infrared micro-nano photonic device is tested and characterized to ensure that the device meets the spectral reconstruction requirements. Then it is encapsulated in the front end of the large-array GSENSE5130 detector to obtain a new spectral detector, and its spectral response performance and application environment are tested to ensure that the new spectral detector still meets the imaging requirements in a complex application environment. Then, the new spectral detector is applied to the satellite-borne spectral imager to realize the development of a miniaturized satellite-borne spectral imager, and imaging experiments are carried out. The image color restoration and spectral reconstruction of the scene detection target are experimentally verified in combination with the computational spectral reconstruction algorithm. Finally, the wide-angle low-dimensional visible-near-infrared micro-nano photonic device is realized from research and design to processing and preparation to testing, and finally applied to the satellite-borne spectral imager to realize the integrated method of miniaturization development.

[0072] The visible-near-infrared micro-nano photonic device designed in the study is integrated with the large-array GSENSE5130 spectral detection system. Each strip can cover one or more rows of pixels. The target spectrum is modulated by the visible-near-infrared micro-nano photonic device, and the grayscale value of each pixel is obtained on the detector. The spectrum reconstruction algorithm is further combined to perform high-precision spectrum reconstruction analysis on the detected target spectrum. The spectrum reconstruction analysis steps of the visible-near-infrared micro-nano photonic device are shown in the attached figure. Figure 1 shown.

[0073] Step 1: Establishment and selection analysis of spectral reconstruction mathematical model

[0074] At present, the new computational imaging spectrometer is different from the traditional imaging spectrometer in that a number of spectral discrete points are obtained by solving a set of linear equations to obtain a broadband target spectral curve. At present, the number of broadband spectral transmittance curves selected for computational spectral reconstruction in the visible spectrum band is more than 100. Among them, Bao Jie's team at Tsinghua University used glue quantum dots to prepare 195 spectral curves in the 350-750nm spectral band and achieved spectral reconstruction under 30dB noise. However, due to the large number of selected spectral curves, data redundancy between spectral curves will be caused, which will increase the pathological nature of the linear equations and affect the accuracy of spectral reconstruction in a large disturbance environment. Therefore, it is particularly important to study this problem, reduce the number of spectral curves, study low-dimensional micro-nano photonic devices, and apply them to spectral detection systems to achieve high-precision spectral reconstruction.

[0075] The present invention uses a spectrum reconstruction mathematical model to perform calculation and analysis, passes the target spectrum X(λ) through the filter array P(λ) at the front end of the detector, and modulates the detector quantum efficiency Q(λ) to obtain a gray value B. λ is the working wavelength range of the detector. The theoretical flow chart of the mathematical model for spectrum reconstruction is shown in the attached figure. Figure 2 shown.

[0076]

[0077] Let A(λ)=P(λ)·Q(λ), then equation (3.1) can be simplified to

[0078]

[0079] After discretizing equation (3.2), we can get:

[0080] A·X=B (3)

[0081] In order to solve the target spectral curve more accurately, equation (1) is discretized to obtain equation (4). At this time, the problem of spectral reconstruction can be summarized as the problem of accurately solving a set of linear equations. Since the quantum efficiency of different optical detection systems is different, the degree of linear equations will increase and the robustness will decrease. Therefore, solving the ill-conditioned equations becomes the core of spectral reconstruction.

[0082]

[0083] According to the linear equation system, A·X=B, which can be written as A m×n ·X=B. According to formula (4), it can be decomposed into:

[0084]

[0085] When m>n, it is called an overdetermined system of equations; when m<n, it is called an indeterminate system of equations; when m=n, and |A|=0, it is called a singular system of equations; when m=n, and cond(A)>>1, it is called an ill-conditioned system of equations. Therefore, the ill-posed problem that the present invention needs to solve is mainly the problem of solving ill-posed linear algebraic equations. Among them, the essential difficulty faced in solving ill-posed problems is the instability of approximate solutions. In order to discuss the stability of numerical methods, this paper proposes the concept of condition number. The larger the condition number, the more ill-conditioned the problem. Due to the diversity of spectral curves, there is no way to establish a mathematical relationship between the condition number and the optimal solution curve. Therefore, the current main goal is to reduce the degree of ill-conditioning of the matrix, and the optimization goal is to take the condition number of 1 as the optimization goal. Obtaining a small degree of ill body reduces the difficulty of spectral reconstruction and improves the accuracy of spectral reconstruction.

[0086] Since each image contains millions of pixels, each pixel corresponds to a spectral curve, but these spectral curves have high redundancy, in order to reduce redundancy. Therefore, it is necessary to select a sample library with representative low redundancy and high diversity. Assuming that the image resolution is F, it consists of N spectra forming an N×F matrix S. In order to test the diversity of N samples, the volume function is used

[0087]

[0088] Among them, i is the maximum singular value of the matrix S. The volume maximization method is widely used in the selection of column subsets in classic clustering and dimensionality reduction tasks due to its effectiveness. The n spectral samples of the maximum volume function are selected to ensure low sample correlation. In this way, 100 groups of sample curves are selected in the 400-900nm spectral range using a genetic algorithm.

[0089] In the field of remote sensing, spectral angles are widely used in object recognition and classification. In spectral analysis, the cosine value reflects the similarity between two spectral vectors. Then, based on the combination of K-means clustering algorithm and particle swarm algorithm, multiple groups of spectral curves that meet the spectral angle range and matrix full rank are optimized and screened. Since the spectral angle cosine value equation (5) measures the similarity of A(λ), it satisfies the full rank of matrix A.

[0090]

[0091] Among them, x i With y i They are two sets of energy values ​​of A(λ) within the working wavelength range of the detector.

[0092] According to the literature analysis on the visible spectrum, the minimum number of curves for spectrum reconstruction in the visible band with a spectral width of 140nm to 350nm is 20. Therefore, when the spectral width is increased to 500nm, the number of spectral curves is also set to 20 in order to achieve the research and design of low-dimensional micro-nano photonic devices in the 400-900nm spectral band. According to the above correlation analysis, 20 broadband transmittance spectral curves (A1-A20) with spectral angle cosine values ​​less than 0.9982 and full matrix rank are selected.

[0093] Through analysis, we can see that the accuracy and stability of solving the linear equation system (3) are not only related to the full rank, but also to the design matrix condition number.

[145] The equation is

[0094]

[0095] Among them, σ max With σmin are the maximum singular value and the minimum singular value of the design matrix A respectively.

[0096] The larger the condition number, the higher the ill-conditioning degree of the design matrix A. At this time, the condition number of the design matrix of 20 broadband transmittance spectral curves is 6.6237e+4, which makes the condition number too high, which will cause the linear equations to be very ill-conditioned and affect the reconstruction accuracy. By solving the condition number under different numbers of spectral curves, it is found that when the number of spectral curves is between 5 and 9, the condition number increases slowly from small to large, and the value is about 100; when the number of spectral curves is 10, the value of the condition number increases to about 150, and when the number of spectral curves is between 11 and 20, the condition number increases exponentially, which will cause the ill-conditioning degree of the linear equations to increase greatly and affect its solution accuracy. In order to further reduce the influence of the condition number on the ill-conditioning degree of the design matrix A and improve the solution accuracy of the linear equations. Therefore, a truncated spectral reconstruction (IReg) algorithm based on a low-dimensional logarithmic regularization matrix is ​​further proposed to solve the problem of too high condition number. This method further uses the condition number as the objective function for optimization. Through the numerical analysis of the condition number under different numbers of curves, when the matrix condition number is within 100, it is considered that the design matrix A is less ill-conditioned and can achieve high-precision spectral reconstruction. Therefore, in view of this situation, the spectral discrete points obtained by the 10×10 matrix are used to capture the curve shape of the target spectrum, and the condition number is reduced from the initial 150.2105 to 25.8340; then, the spectral discrete points of the 20×20 matrix are further solved with this as a constraint; finally, the morbidity of the design matrix A is reduced, the accuracy of spectral reconstruction is improved, and the spectral discrete points are fitted to obtain a high-precision spectral reconstruction curve.

[0097] In order to achieve stability and reliability in engineering applications, four groups of monochromatic broadband spectral curves (A21-A24) of red, green, blue and near-infrared are further proposed as comparison and verification schemes to qualitatively measure the color of the ground objects. The visible-near-infrared micro-nano photonic device designed in the study adopts a composite grating array arrangement and combination method, which is designed on the same substrate for subsequent processing and preparation. At the same time, it is also used for TDI push-scanning imaging in the developed miniaturized satellite-borne spectroscopic imager to enhance the integral series and gain and improve the imaging quality. Among them, the schematic diagram of the composite grating array arrangement and combination method of the visible-near-infrared micro-nano photonic device is shown in the attached figure. Figure 3 shown.

[0098] Step 2: Propose an algorithm based on low-dimensional logarithmic regularization matrix truncated spectrum reconstruction (IReg)

[0099] The L1 regularization algorithm can better solve the problem of finding the minimum value in equation (8), while ensuring that the residual is small, it also makes the vector X as small as possible. The residual is measured by ||AX-B||2, and the regularization term is represented by the sum of the absolute values ​​of each parameter ||X||1. Among them, λ is the regularization parameter, which has the ability to suppress noise.

[0100] min{||AX-B||2+λ||X||1} (8)

[0101] The L1 regularization algorithm can also be considered as a least squares method restricted by constraints. It has the same penalty for all parameters, which can make some weights zero, thus generating a small number of features and then a sparse model. Therefore, L1 regularization is mostly used for sparse solutions, but it can also prevent overfitting to a certain extent. However, when the noise is greater than a certain critical value, the noise suppression effect of L1 regularization is not significant.

[0102] The regularization term of the L2 regularization algorithm is the two-norm of each parameter ||RX||2. The L2 regularization algorithm will not make the weight become 0, that is, it will not produce a sparse model. It has the characteristics of continuous smoothness, so more features can be obtained. In essence, it is to solve the optimal function formula (9). After differentiation, formula (10) can be obtained.

[0103]

[0104] x=(A T A+λR T R) -1 A T B (10)

[0105] The difference from the least squares method is that λR is added T R, which plays an important role in noise suppression. R is the regularization matrix, also the unit matrix. R corrects the singular values ​​of the design matrix A. The regularization parameter λ determines the correction amplitude of the singular value, which can be obtained by the L curve method. R determines the correction method of the singular value. Perform singular value decomposition on the design matrix A, and we can get U = [u1,u2,u3,…,u n ], S=[σ1,σ2,σ3,…,σn], V=[v1,v2,v3,…,v n ]. Among them, u i With v i are the left and right singular vectors of the design matrix A, σ i are the singular values ​​of the design matrix A, is the filtering function, and equation (10) is simplified as:

[0106]

[0107] In the field of computational spectral reconstruction, the size of the singular value can represent the information of the detected target spectrum. The larger the singular value, the more information of the target spectrum it contains. Among them, R can be a unit matrix, so the filter function corrects the singular value to the same extent. This correction method has no significant effect on the improvement of the condition number of the design matrix A, and the suppression of large noise is not obvious. The L2 regularization algorithm does not fundamentally solve the problem of too large condition number, but the L2 regularization algorithm can make the parameter value in the model as small as possible, so that the model is not dependent on a few special features as much as possible, but each feature is given a weight that is as balanced as possible. Therefore, it has the advantage of solving ordinary fitting problems very well. In a small noise environment, relatively smooth and less jittery spectral discrete points can be obtained. However, in a large disturbance environment, when there are small singular values, the obtained spectral discrete points will have large jitter. The essence of the regularization matrix is ​​to correct the singular values. The R unit matrix in the L2 regularization algorithm actually corrects all singular values ​​with the same amplitude to suppress the influence of noise on the solution. This correction method will cause the correction of larger singular values ​​to be too strong and the correction of smaller singular values ​​to be too weak, thus affecting the accuracy of the solution.

[0108] The current research on spectral reconstruction shows that the effect of spectral reconstruction is better when the signal-to-noise ratio is 80dB, and the curve of spectral reconstruction is distorted and locally convex when the signal-to-noise ratio is 40dB. When the signal-to-noise ratio is lower than 30dB, the reconstructed spectral curve is quite different from the target spectral curve. However, due to the coupling of multiple errors such as the detector's own noise, stray light, and optical system aberrations, the error generated by the gray value B is greater than 30dB. It is necessary to propose an algorithm with strong robustness that can reconstruct the target's spectrum with high precision in a large disturbance environment. Therefore, this paper proposes a truncated spectral reconstruction (IReg) algorithm based on a low-dimensional logarithmic regularization matrix.

[0109] The standard regularization algorithm has the problem that the correction method for large singular values ​​is too strong and the correction for small singular values ​​is too weak. The proposed truncated spectrum reconstruction (IReg) algorithm based on low-dimensional logarithmic regularization matrix can achieve truncation and retention of large singular values ​​without correction, truncation logarithmic correction of singular values ​​between large and small singular values, and truncation and discarding of small singular values. At the same time, the proposed logarithmic regularization matrix gives different correction amplitudes to different singular values, giving smaller corrections to larger singular values ​​and larger corrections to smaller singular values.

[0110] The log-regularization matrix is:

[0111]

[0112] Substituting equation (12) into equation (9), we can obtain the improved regularized solution equation:

[0113]

[0114] The regularized filter function is:

[0115]

[0116] Since the larger eigenvalues ​​represent the main information of the target spectrum and are less affected by errors, equation (14) is truncated twice according to the proportion of singular values ​​and decomposed into three-segment filtering functions.

[0117] The proportion of singular values ​​is:

[0118]

[0119] When the singular value ratio δ≥0.7, the first truncation is performed, and the tth singular value is the position of the first truncation. The singular values ​​in the range [1, t] are not corrected. When the singular value ratio δ≥0.95, the second truncation is performed, and the mth singular value is the position of the second truncation. The singular values ​​between [t+1, m] are corrected by logarithmic regularization, and the smaller singular values ​​in the range [m+1, n] are discarded.

[0120] After truncating the matrix, the new filter function f is:

[0121]

[0122] Substituting equation (16) into equation (11), we can get the truncated regularization algorithm. Solving the linear equation system (3) yields:

[0123]

[0124] The evaluation indexes MSE and ∈ of spectral reconstruction accuracy are expressed as follows:

[0125]

[0126] Among them, x i and x is the reconstruction value, x i With x being the standard value and n being the number of spectral segments.

[0127] In a large disturbance environment, the spectrum of four different ground objects selected from the ENVI remote sensing image processing platform was reconstructed by calculating the spectrum reconstruction algorithm, and a comparison chart of the spectrum reconstruction results was obtained, as shown in the attached figure. Figure 4 The evaluation index data of the spectral reconstruction accuracy of ground objects is shown in the attached Figure 5 shown.

[0128] The above spectral reconstruction results show that the designed composite grating array low-dimensional micro-nano photonic device, under the large disturbance environment of signal-to-noise ratio of 30dB, 20dB and 17dB, the IReg algorithm obtains the spectral reconstruction accuracy evaluation index MSE mean less than 1.4115×10 -4 ,1.5204×10 -3 and 2.9941×10 -3 , and the mean values ​​of ∈ are less than 0.034978, 0.087985 and 0.13071 respectively. After applying a hundred random errors, the L2 algorithm obtains that the mean values ​​of the MSE of the spectral reconstruction accuracy evaluation index are less than 3.6135×10 -4 ,7.5816×10 -3 and 0.012922, while the mean values ​​of ∈ are less than 0.040111, 0.15982 and 0.22973 respectively. After applying a hundred random errors, the L1 algorithm obtains that the mean values ​​of the MSE of the spectral reconstruction accuracy evaluation index are less than 3.8464×10 -4 8.2129×10 -3 and 0.018162, while the ∈ means are less than 0.042151, 0.17956, and 0.25033, respectively.

[0129] In an environment with a signal-to-noise ratio of 30 dB, the L1 and L2 algorithms can reconstruct the spectral curve of the ground object well, and the IReg algorithm can achieve high-precision reconstruction of the spectral curve of the ground object, with the MSE average of 100 times less than 10 -5 ~10 -4 The average value of ∈100 is 10 -2 However, in a large disturbance environment with a signal-to-noise ratio of 20 dB, the spectral curves reconstructed by the L1 and L2 algorithms show a small range of deformation. The MSE and ∈100-times average of the reconstructed curves by the IReg algorithm are smaller than those of the L1 and L2 algorithms, and the MSE 100-times average is within 10 -4 ~10 -3 The average value of ∈100 is 10 -2 To further verify the robustness and accuracy of the IReg algorithm, in a large disturbance environment with a signal-to-noise ratio of 17 dB, the spectral curves reconstructed by the L1 and L2 algorithms are deformed. However, the IReg algorithm can stably capture the target spectral features and reconstruct the target spectral curve relatively stably, with an MSE average of 100 times. -3 ,∈The average of 100 times is 10 -1. It can be seen that under large disturbance conditions, the spectral reconstruction accuracy of the IReg algorithm is higher than that of the L1 and L2 algorithms, which improves the stability and accuracy of the spectral reconstruction. Therefore, the spectral reconstruction accuracy calculated by the IReg algorithm is higher than that of the L1 and L2 algorithms, which meets the purpose of using low-dimensional composite grating array micro-nano photonic devices for high-precision spectral reconstruction of visible-near infrared spectra, and provides important research support for the development of miniaturized spectral imagers.

[0130] Step 3: Research and design of wide-angle low-dimensional visible-near-infrared micro-nanophotonic devices

[0131] According to the existing problems of high processing and preparation precision, high manufacturing cost, complex control system, environmental sensitivity and limited field of view in the optical spectroscopic elements of satellite-borne spectral imagers. Considering the detection of ground imaging, environmental pollution monitoring, resource investigation, military reconnaissance and camouflage identification, the visible spectrum range is selected. The current shortcomings in the visible spectrum are that the spectral band width is narrow, and the number of spectral curves required to achieve the same resolution within the same spectral width range is large, and it is sensitive to environmental influences and difficult to process. Therefore, in order to achieve higher detection and identification effects, the spectrum of micro-nano photonic devices is extended to the near-infrared spectrum during the research and design, and the spectral range is set between 400 and 900nm. The increase in the spectral band width has been applied to a wider band range.

[0132] The research designs a composite grating array micro-nano photonic device for the visible-near infrared (400-900nm) spectrum to meet the needs of spectral reconstruction. Based on the photonic crystal structure model, the micro-nano photonic crystal structure is used to adjust the structure and geometric parameters of the two-dimensional unit of the photonic crystal to achieve the regulation of the intensity and phase of the electromagnetic field. Many groups of different transmittance spectrum curves can be integrated on a small area to achieve spectral modulation effects, simplify the optical system, and obtain specific scene information with stronger adjustment ability, more efficiency, more accuracy and more robustness. Each transmittance curve is designed and simulated through the micro-nano photonic crystal structure. The optical film of the structure is composed of many layers of dielectric films with different optical constants. Two dielectric materials, niobium pentoxide (Nb2O5) and silicon dioxide (SiO2), are used. For the 24 sets of broadband spectral transmittance curves designed, the transmission matrix method film system Essential Macleod design software and the finite element method COMSOL Multiphysics multi-physics field simulation software are used to design and simulate in the wavelength range of 400-900nm. Schematic diagram of SiO2 / Nb2O5 photonic crystal multilayer structure design, as shown in the attached figure. Figure 6 shown.

[0133] When electromagnetic waves pass through multiple film layers, the transmission matrix method is used to connect the electromagnetic fields in two interfaces of the same film layer using the propagation phase difference. The refractive indices are The thickness of the i-th layer is Assuming that the photonic crystal is composed of multiple layers of thin films, the relationship between the tangential field of the first layer and the boundary of the Nth layer is expressed as:

[0134]

[0135] Among them, E1 and H1 are the tangential components of the electric field and magnetic field in the first layer, respectively. N and H N For the Nth layer.

[0136] When M0 represents a unit matrix, it is the characteristic matrix of the basic unit composed of two dielectric layers, and M is the transfer matrix expression of the photonic crystal structure:

[0137]

[0138] M j The transfer matrix expression of the j-th film layer is:

[0139]

[0140] Among them, ρ j =2πn j d j cosα j / λ,n j is the refractive index of the medium, α j is the incident angle of light in the medium, and λ is the wavelength of the incident light wave. j is the incident angle α in the medium j The relevant physical quantity can be taken as β in TM mode. j =n j / cosα j , TE mode β j =n j cosα j Here, the TM mode refers to the light wave propagation mode with only electric field components but no magnetic field components in the propagation direction, and the TE mode refers to the light wave propagation mode with only magnetic field components but no electric field components in the propagation direction. Therefore, the reflection coefficient and transmission coefficient are inferred as:

[0141]

[0142] In TE mode, In TM mode, M 11 、M 12 、M 21 and M 22is the matrix element of M. r is the reflectivity coefficient, t is the transmittance coefficient. ε is the dielectric constant and μ is the magnetic permeability, and are the dielectric constant and magnetic permeability of the left and right media in contact with the photonic crystal, respectively. Among them, ε0 is the relative dielectric constant of the first layer of medium, ε N is the relative permittivity of the Nth layer, μ0 is the magnetic permeability of the first layer medium, μ N is the magnetic permeability of the Nth layer of medium. Then we can get the reflectivity R = |r| 2 , the transmittance is T = q1 / q N |t| 2 Through the above analysis, the [SiO2 / Nb2O5] N It has a wide band gap and can cover the visible-near infrared spectrum (400-900nm) to obtain a broadband transmittance spectral curve that meets the needs of spectral reconstruction.

[0143] Essential Macleod and COMSOL Multiphysics software were used to design 24 sets of film systems with broadband transmittance curves. According to the required broadband transmittance spectrum curve, the boundary conditions of the micro-photonic crystal structure design were optimized, the dielectric material selection was completed, and the film system parameter optimization process was controlled. The two sets of simulation results were compared and analyzed based on the material stacking method and the thickness of each layer of the film system for each broadband transmittance spectrum curve to enhance the reliability of the simulation results, determine the film thickness of the broadband transmittance spectrum curve, and lay the foundation for the subsequent processing and preparation of the film system. Schematic diagram of the optical film thickness of the micro-nano photonic crystal structure, as shown in the attached figure Figure 7 Among them, the 24 broadband transmittance spectrum curves marked by symbols A1 to A24 represent the thickness of each layer of the film system.

[0144] The 24 broadband transmittance spectral curves obtained by simulation calculation using Essential Macleod software are spectral curves marked by M1 to M24; the 24 broadband transmittance spectral curves obtained by simulation calculation using COMSOL Multiphysics software are spectral curves marked by C1 to C24. The comparison diagram of the two sets of broadband transmittance spectral curves is shown in the attached figure. Figure 8 The results show that the two groups of broadband transmittance spectral curves are basically consistent, proving that the thickness of each layer of the designed broadband transmittance curve obtained by stacking SiO2 and Nb2O5 materials in the 24 designed micro-nano photonic crystal structures is stable and reliable, laying the foundation for the subsequent processing and preparation of each film system.

[0145] According to the current application demand analysis of large-field-of-view satellite-borne spectrometers, the field of view of miniaturized satellite-borne spectrometers is not only limited by the design of the optical system, but also by the size of the detector and the sensitivity of the micro-nano photonic devices packaged at the front end of the detector to the incident angle. Natural light first enters the optical system of the satellite-borne spectrometer. After the aberration correction and focusing of the optical system, the incident light hits the detector at a certain cone angle, so that the light incident on the front end of the detector containing the micro-nano photonic devices has a certain range of incident angles, thereby affecting the imaging effect of the spectrometer. Therefore, for the system in which the visible-near-infrared micro-nano photonic device is packaged at the front end of the detector as a spectral modulation element, when the light passing through the optical system is incident on the surface of the micro-nano photonic device, there is a certain incident angle, and the range of the incident angle is related to many factors, such as the optical system parameters and the optical-mechanical assembly accuracy. In response to this problem, the influence of the change of the incident angle on the optical characteristics of the spectral curve of the visible-near-infrared micro-nano photonic device is further simulated and analyzed, and it is judged whether the increase of the incident angle will affect the effect of spectral reconstruction. Visible-near-infrared micro-nanophotonic devices that allow large-angle light incidence are helpful for the development of miniaturized satellite-borne spectral imagers with large field of view.

[0146] When TM waves and TE waves are incident from the air at a certain angle α, different electromagnetic field distribution characteristics are formed, but the transmittance effect is consistent, which shows that the proposed visible-near infrared micro-nano photonic device has polarization-insensitive characteristics. Therefore, under the condition of transverse electromagnetic mode (TEM), this chapter analyzes the changes in the optical characteristics of the spectral curve when the incident angle changes from 0° to 90°. Through simulation calculation, it is obtained that when the incident angle α increases from 0° to 40°, the changes in the 20 broadband transmittance spectral curves are obtained, as shown in the attached figure. Fig. 9 The results show that when the incident angle reaches 30°, the change trend of the spectral curve is basically the same, which can meet the application of miniaturized space-borne spectral imagers in large fields of view.

[0147] Based on the photonic crystal structure, a micro-nano photonic device suitable for the visible-near infrared (400-900nm) spectral band is designed. The device has the advantages of polarization insensitivity, wide angle, high manufacturability, easy integration, and a broadband transmission spectral curve with strong controllability of the spectral curve. As a new type of spectral modulation element, it can realize bandpass imaging, enhance the width and light flux of the applicable spectral band, and reduce the difficulty and complexity of processing and preparation. Under TEM conditions, when the incident angle is within 30°, the broadband transmittance spectral curve of the micro-nano photonic device is basically consistent. This feature enhances the field of view of the system. When selecting the broadband transmittance spectral curve that constitutes the micro-nano photonic device, it is combined with the computational spectrum reconstruction algorithm to enhance the controllability of the spectral curve. The processing difficulty is reduced, it is easy to integrate and package into the front end of the detector, reduce the complexity of the optical system design, effectively reduce the impact of the external environment on the device performance, improve the stability and robustness of the system, and realize the development of integration and miniaturization.

[0148] Step 4: Processing and preparation of visible-near infrared micro-nano photonic devices

[0149] The micro-nano photonic devices designed in the research are manufactured by combining photolithography, coating and lift-off processes to process and prepare 24 sets of wide-band filter films arranged in a composite grating array to achieve the preparation of visible-near-infrared micro-nano photonic devices.

[0150] First, through the photolithography process, the UV lithography machine (SUSS MA6) is used to transfer the pattern on the mask to the glass surface of the silicon dioxide (SiO2) substrate coated with photoresist through the exposure process. The photolithography process flow is ①HMDS→②Smoothing→③Exposure→④Development: ①HMDS is oven-coated, the substrate is placed in the HMDS professional oven, and HMDS is vaporized and deposited on the substrate surface; ②Smoothing uses a specific negative glue, and the photoresist is spin-coated on the substrate surface using the spinning equipment, and after baking, it flows into the next process; ③Exposure uses a UV lithography machine to expose the designed mask pattern, and the pattern on the mask is copied to the substrate surface coated with photoresist; ④The exposed substrate is placed in the developer, the photoresist and the developer will react chemically, and finally form a pattern opposite to the pattern on the mask, forming the strip gap that needs to be coated, and the developed substrate is baked to form a photolithography pattern sheet substrate.

[0151] Secondly, through the coating process, using the physical vapor deposition method (PVD), the prepared photolithography pattern substrate is placed in the vacuum coating equipment RSP800 magnetron sputtering coating machine, and the equipment is placed with SiO2 and niobium pentoxide (Nb2O5) coating targets. The target material and the gas introduced into the chamber are reacted and deposited, and then deposited layer by layer on the surface of the substrate. Then, by controlling the thickness of the SiO2 and Nb2O5 deposition, the designed thickness of each layer of film is achieved, thereby forming a substrate strip with a thin film. Then, using the lift-off process, the coated substrate is immersed in the degumming solution, and the thin films outside the film system are stripped off together with the photoresist, thereby completing the preparation of a set of thin films.

[0152] Finally, by repeating the photolithography, coating, and lift-off process, the required 24 composite grating array wide-band filter film systems are processed and prepared one by one. The process flow chart for preparing 24 composite grating array wide-band filter film systems is shown in the attached figure. Fig.10 shown.

[0153] During the processing and preparation process, the process problems that may be encountered are as follows:

[0154] (1) In the photolithography process, firstly, the photoresist may be unevenly coated during spin coating; secondly, debonding may occur during the chemical reaction between the photoresist and the developer after exposure; finally, cracks or curling may occur during the baking of the developed substrate if the temperature is too high; the above problems can be improved through multiple process trials to ensure the integrity and accuracy of the structure. The detailed process steps are as follows:

[0155] ①HMDS: Due to the poor adhesion between the photoresist and the substrate itself, HMDS is used as a thickener for the photoresist. Take out the clean substrate, put it into the HMDS oven, and evacuate it to a vacuum state. HMDS is a volatile liquid. The oven pipeline is connected to the HMDS bottle. After heating to the set temperature, HMDS will evaporate to form steam. The steam is drawn into the HMDS oven and coated on the surface of the substrate in the vapor phase to form an HMDS adhesion promoter layer with a surface thickness of nanometer level.

[0156] ② Glue spreading: Take out the substrate treated with HMDS, put it on the manual glue spreading table after cooling, turn on the vacuum adsorption to adsorb the substrate on the vacuum suction cup, suck a certain amount of photoresist with a dropper, drop it on the center of the substrate surface, start the setting glue spreading program, and the photoresist dropped on the center of the substrate surface will be evenly coated on the substrate surface through the spin coating process, thus forming a substrate with evenly coated photoresist; since the coated photoresist is in a molten state, the subsequent exposure process cannot be carried out, and the coated substrate needs to be placed on a hot plate for baking to dry the photoresist; since the subsequent process requires coating and retention in the non-photoresist area, the coated photoresist is selected to use negative photoresist (i.e., the photoresist retained in the exposed area after development).

[0157] ③ Development: Place the exposed substrate into a developer. The photoresist that has been exposed to light will not react chemically with the developer and will be retained, while the photoresist that has not been exposed to light will react chemically with the developer and be removed, thereby forming a first stripe pattern, wherein the first stripe is the area that needs to be coated later, with no obstruction on the surface, and the area other than the first stripe is a photoresist protection area. In the subsequent coating process, a film layer of the coated first stripe and a film layer of the coated first stripe with photoresist underneath will be formed, and the development process requires precise control of the development time.

[0158] ④Put the developed substrate on the hot plate for drying again to dry the surface moisture and some residual organic solvents in the photoresist. After the above HMDS→smearing→exposure→development process, the photolithography graphic substrate with the first stripe pattern is completed. Among them, the mask plate is arranged and marked for engineering in the mask lithography so as to determine the effect of each spectral curve later. According to the pixel size (4.25μm) and pixel array (5056×2968) of the large array GSENSE5130 detector, it can be concluded that the size range to be processed is 21.488×12.614mm 2 In view of the size requirements of the detector and the actual application, the length and width are increased during processing and preparation. The length is increased by 5mm, and the width is increased by 5 film coating widths (300μm) and interval widths (200μm) above and below. A total of 34 spectral curves of the film are processed and prepared to ensure that there will be no missing due to offset during the later packaging.

[0159] (2) In the coating process, the prepared photolithography pattern substrate is placed in the vacuum coating equipment, and the required coating target material is placed in the equipment. The target material and the gas introduced into the chamber react to form the required dielectric film and metal film, which are deposited layer by layer on the substrate surface. By controlling the deposition thickness of the dielectric film and metal film, the required designed filter film thickness is achieved, thereby completing the preparation of the structure. If the ventilation flow rate and time are not well controlled, the thickness of the coating will deviate from the simulation results. Therefore, in this process, the ventilation flow rate and time should be controlled through multiple process explorations.

[0160] In the coating process, the prepared photolithography pattern substrate is applied to the tray of the vacuum coating equipment using high-temperature tape, the tray is placed in the vacuum coating equipment, the equipment is turned on to evacuate, and the coating equipment chamber reaches the vacuum degree of the equipment. At the same time, the SiO2 and Nb2O5 targets required for coating have been placed in the equipment, and the ion source of the equipment is turned on. Oxygen is introduced. For example, when a SiO2 film layer is required to be coated, the silicon target is turned on, and the sputtered silicon atoms will quickly form a layer of SiO2 film with the ionized oxygen atoms and deposit on the surface of the substrate. Through this process, layers of SiO2 films can be formed, and the thickness of the SiO2 film can be controlled by setting the coating time to achieve the pre-designed film thickness. Conversely, when a Nb2O5 film layer is required to be coated, the target is turned on, and the sputtered niobium atoms will quickly form a layer of Nb2O5 film with the ionized oxygen atoms and deposit on the surface of the substrate. Through this process, layers of Nb2O5 films can be formed, and the thickness of the Nb2O5 film can be controlled by setting the coating time to achieve the pre-designed film thickness. Finally, by controlling the film thickness of each layer of SiO2 and Nb2O5, the film thickness of the pre-designed filter strip is achieved, thereby forming a coated substrate with a thin film. After the above coating process, a coated substrate is formed. The photolithography part uses a UV SUSS MA6 photolithography machine, and the coating part uses a vacuum coating RSP800 magnetron sputtering coating machine.

[0161] (3) In the lift-off process, the coated substrate structure is immersed in a degumming solution, and the thin film outside the structure is stripped off together with the photoresist, thereby forming a thin film structure. Incomplete stripping may occur in this process, resulting in an incomplete structure. It is necessary to ensure that the chemical agent can completely corrode and strip off the photoresist, while having no effect on the dielectric film and the metal film. The coated substrate is placed in the degumming solution, and the degumming solution will react chemically with the photoresist. The photoresist will dissolve in the degumming solution, and the thin film area without photoresist support will separate from the substrate surface due to the lack of support. The part not covered by the photoresist is retained because the degumming solution will not react with the film layer formed by the coating. After this stripping process, a substrate with the 24th thin film is formed.

[0162] (4) Substrate material test: The uniformity and roughness of the film were tested, and the results showed that the deformation caused to the wavefront was not large; the stress of the film layer was small, so the wavefront distortion caused by stress deformation was also small. The uniformity of the thickness distribution of the optical film and the parallelism between the various film systems of the optical film are important indicators of optical components. Poor uniformity will lead to differences in optical performance at different positions of the optical component, seriously affecting the transmission uniformity. If the various film systems are not parallel, the utilization rate of the imaging pixels in the middle area of ​​the detector pixel will be affected when the detector is integrated and packaged. The actual picture of the processed and prepared composite grating array wide-band filter film system is attached. Fig.11 shown.

[0163] Step 5: Testing and characterization of visible-near infrared micro-nanophotonic devices

[0164] The prepared micro-nano photonic devices are compared and analyzed by measuring the spectral curve of each group and the theoretically calculated spectral curve to determine the effect of the processed spectral curve and whether it meets the needs of spectral reconstruction. Among them, two types of spectrophotometers, Lambda1050+ and UV-2550, are used to test the transmittance curves of 24 spectral curve film systems respectively, so as to better determine the accuracy and stability of the test. The comparison chart of the actual test transmission spectral curve of the 24 spectral curve film systems and the theoretical simulation calculation transmission spectral curve is as shown in the attached figure. Fig.12 As shown. Among them, symbols S1~S24 represent the transmission spectrum curves calculated by theoretical simulation, and symbols E1~E24 represent the transmission spectrum curves of actual test. The two sets of spectral curves were compared and analyzed, and it was verified that the spectral curves of 24 wide-band filter films arranged in a composite grating array on the same substrate prepared by combining photolithography, coating and lift-off processes are more than 96% similar to the spectral curves designed by theoretical simulation, which meets the required range of processing error and realizes the preparation of visible-near-infrared micro-nano photonic devices.

[0165] At the same time, in order to verify whether the processed spectral curve meets the requirements of spectral reconstruction. The actual test transmittance curve is further combined with the proposed calculation spectrum reconstruction IReg algorithm to reduce the influence of the condition number on the morbidity of the design matrix A. The spectral discrete points obtained by the 10×10 matrix are also used to capture the curve shape of the target spectrum. The obtained design matrix condition number can be reduced to 26.7801. Then, the spectral discrete points of the 20×20 matrix are further solved based on this constraint, which can reduce the morbidity of the linear equations. The spectral discrete points are fitted to obtain a high-precision spectral reconstruction curve to meet the requirements of spectral reconstruction.

[0166] The processed visible-near-infrared micro-nano photonic devices are first tested for the thickness of each film system and the parallelism between the film systems using a scanning electron microscope and a confocal microscope. Secondly, the thickness and width of the film system for each spectral curve are measured at multiple positions. Since one of the keys to spectral imaging lies in high-quality spectrally modulated optical films, the uniformity of the thickness distribution of optical films and the parallelism between the film systems of each group of optical films are important indicators of optical components. Poor uniformity will lead to differences in optical performance at different positions of the optical components, seriously affecting the transmission uniformity. If the film systems are not parallel to each other, the utilization rate of the imaging pixels in the middle area of ​​the detector pixels will be affected when they are integrated and packaged with the detector. When preparing optical thin film components, it is very important to control the uniform distribution of the deposition thickness on the component.

[0167] The spectral curves of the micro-nano photonic devices arranged in a composite grating array were tested by confocal microscopy (CLSM). The width of the film system was 300μm, the interval width of the non-film system was 200μm, and the grating array period was 500μm. The test effect diagram is shown in the attached figure. Fig.13 The film thickness was tested by scanning electron microscopy (SEM), and the thickness of the 21st spectrum film was used as an example to show the test results. The film thickness diagram of the 24 spectrum films tested by CLSM is also shown, as shown in the attached figure. Fig.14 The images of film thickness at different locations under the same film system tested by SEM are shown in the attached figure. Fig.15 To avoid the offset phenomenon during packaging, 5 repeated film systems are added above and below the film system, so the filter has 34 film systems as a whole.

[0168] The mean and error range of the width of 34 films of the micro-nano photonic device arranged in a composite grating array are shown in the attached figure. Fig.16 As shown. At the same time, to ensure the parallelism between the film systems, the width of 34 sets of film systems was tested at ten positions, and the average width error of the ten sets of data was less than ±5μm, and the parallelism error was less than 1.89μm. The parallelism and error range meet the application requirements and are suitable for packaging to the front end of the detector.

[0169] Step 6: Integrated packaging and performance testing of visible-near infrared micro-nano photonic devices

[0170] In cooperation with the 58th Institute of China Electronics Technology Group Corporation, the composite grating array micro-nano photonic device was packaged in the front end of the large array GSENS5130 detector in a dust-free nitrogen environment through the Zeiss SteREO Discovery V8 optical microscope, achieving stable integrated packaging that meets aerospace-grade applications. The process flow of packaging the composite grating array micro-nano photonic device in the front end of the large array detector is shown in the attached figure. Fig.17 shown.

[0171] In order to ensure that the composite grating array micro-nano photonic device and the large-area GSENSE detector pixel are placed relatively parallel, and to ensure that the parallelism between the film system and the detector pixel reaches more than 95%. In order to effectively cover the imaging channel of the detector, to ensure the utilization rate of the imaging pixel in the middle area containing the film system, and then to ensure the imaging quality of the integrated miniaturized satellite-borne spectrometer. Therefore, two methods are used to monitor the integrated package and meet the parallelism and alignment requirements between the micro-nano photonic device and the detector pixel.

[0172] (1) Process monitoring: The parallelism between the pixel and the line width of the film system in the micro-nano photonic device is determined by using a Zeiss SteREO Discovery V8 optical microscope. This method ensures that the first pixel and the pixels at both ends are aligned as much as possible in terms of process, thereby ensuring parallelism during electrical installation. At the same time, the designed identification symbols are used to ensure that the coverage between the 24 film systems and the pixels is within the effective identification range, and to ensure their alignment and parallelism.

[0173] (2) Imaging monitoring: through actual imaging, the parallel state between the pixel and the line width of the film system in the micro-nano photonic device is determined. In terms of imaging effect, this method uses Matlab binarization processing to find the difference between the left and right pixels under the same film system, that is, the number of imaging difference pixels. The parallelism is adjusted according to the obtained deviation ratio to ensure the pixel utilization rate and imaging quality, and the film system imaging diagram is obtained, as shown in the attached figure. Fig.18 shown.

[0174] The composite grating array micro-nano photonic device is packaged at the front end of the detector to obtain a new type of spectral detector, as shown in the attached Fig.19 As shown. Then combined with the designed area array CMOS imaging circuit to realize imaging electronic control, the imaging circuit mainly includes the area array CMOS-GSENSE5130 and power supply circuit, SPI configuration timing, DDR mode LVDS image data transmission, digital circuit power supply circuit and camera controller communication, etc. The main functions are realized by the information and timing processor FPGA.

[0175] The composite grating array micro-nano photonic device is packaged at the front end of the detector to obtain a new type of spectral detector, which can be used in miniaturized satellite-borne spectral imagers to achieve spectral reconstruction of the target spectrum. The large-array GSENSE5130 model detector selected can meet the needs of scientific research, optoelectronics and other applications. The high-speed CMOS image sensor is designed, and the frame rate can reach 100FPS at full resolution. The chip supports vertical axis windowing, and the frame rate is increased proportionally. In addition, the dark current of the work is only 3e- / s@25℃, and a customized ceramic PGA package is used, with space reserved at the bottom of the ceramic for chip cooling.

[0176] In the 400-900nm band, the spectral response curve of the detector's spectral sensitivity characteristics is tested in a dark box environment, which is the test of the quantum efficiency curve. The specific spectral response test principle and method are as follows:

[0177] (1) Spectral response test principle

[0178] The total energy method is used to test the quantum efficiency of the detection device. Based on the principle of the device photoelectric effect at a given wavelength, the device quantum efficiency is calculated according to formulas (23) and (24) in %:

[0179]

[0180] Among them, P is the light energy, the unit is W; T is the exposure time, the unit is s; h is Planck's constant, h = 6.6260755×10-34Js; c is the light beam, c = 2.99792458×108m / s; λ is the wavelength, the unit is m.

[0181] (2) Spectral response test steps:

[0182] ① Press the attachment Fig. 20 Link test system and perform instrument presets;

[0183] ② At wavelength λ, test the output energy value P of the standard detector;

[0184] ③ At wavelength λ, the sum of the output signal values ​​of the test device is y L . And in dark field, with the same exposure time, the total dark field output signal value y of the test device is d The formula for obtaining the total number of electrons is Where K is the conversion gain in DN / e.

[0185] ④ Calculate the quantum efficiency of the device at the wavelength λ according to the quantum efficiency calculation formula (in the test principle);

[0186] ⑤ Switch to different wavelengths λ to test the quantum efficiency of the device. With wavelength λ as the horizontal axis, the quantum efficiency spectral response curve of the device without film system is shown in the attached figure. Fig.21 shown.

[0187] ⑥ According to formula (25), the sensitivity of the device at the wavelength λ is calculated as:

[0188]

[0189] The test environment and test equipment for the spectral response curve of the new detector are as follows: Fig. 22It is worth noting that, first, the signal value of the bright field output should be kept within the linear range of the new detector output as much as possible; second, the new detector photosensitive surface should receive all the monochromatic light spots; finally, the influence of light from other spectral bands or stray light should be avoided during the test.

[0190] Further testing revealed that the new spectral detector had 24 spectral response curves of the spectral sensitivity characteristics of the spectral film system in the 400-900nm band, as shown in the attached figure. Fig.23 The spectral response curves of the blue, green, red and near-infrared color spectral sensitivity characteristics of the new spectral detector composed of the composite grating array micro-nano photonic device and the GSENSE5130 detector are shown in the attached figure. Fig.24 shown.

[0191] To ensure that the new spectral detector can still work normally when used in complex environments, thermal vacuum experiments and thermal cycle experiments were carried out on the imaging module, and high and low temperature tests were carried out on the focusing mechanism. It was also subjected to appearance inspection, electronic performance test and imaging effect inspection, so as to early discover and eliminate defective components mixed in the product during the development process, to ensure the reliability of the electronic stand-alone operation and the reliability of the new spectral detector and other electronic components and related processes.

[0192] The thermal vacuum experiment was cycled 3.5 times in a vacuum tank system with a vacuum pressure of 6.65×10-3Pa and a temperature of -20℃~+50℃, of which the first 2 cycles were the main work and the last 1.5 cycles were the backup work. The main test process is: first, at room temperature, test the main performance parameters and check whether the temperature sensor and data acquisition system are working properly; second, heat up to 50℃ in the vacuum tank for 6 hours, then cool down to -20℃ for 6 hours; finally, heat up to room temperature and then carry out the next cycle, a total of 3.5 cycles. When the thermal vacuum cycle is over, the vacuum tank returns to normal pressure, and after the component is taken out of the test container, a detailed performance test should be carried out. The new spectral detector is in normal state and the thermal vacuum test is over. Among them, it should be powered on during the entire test process, and three hot starts and cold starts should be performed in the first and last cycles, and performance monitoring should be carried out. Only the main performance monitoring should be carried out in the middle cycle. The redundant circuits or channels in the components should undergo various working conditions, and sensitive parameters should be detected as much as possible to check for faults and intermittent phenomena.

[0193] The thermal cycle test is carried out in a high and low temperature box with a temperature range of -25℃ to +45℃, where the temperature change rate is greater than or equal to 7℃ / min, and 10.5 cycles are carried out at the same time, of which the first 6 cycles are the main work and the last 4.5 cycles are the backup work. The main test process is: first, at room temperature, test the main performance parameters and check whether the temperature sensor and data acquisition system are working properly; second, heat up to 45℃ in the high and low temperature box and stay for 1.5h, then cool down to -25℃ and stay for 2h; finally, heat up to room temperature and then carry out the next cycle, a total of 10.5 cycles. After the temperature cycle is over and the room temperature is restored, take out the test component and carry out detailed performance testing. The new spectral detector is in normal state and the thermal cycle test is over. Among them, in order to prevent condensation and moisture from forming on the surface and inside of the component at low temperature, the test box should be filled with dry air or nitrogen, and the last half cycle is a high temperature half cycle, and the component should be powered on. The whole test process should be powered on, and three hot starts and cold starts should be performed in the first and last cycles, and performance monitoring should be performed. Only major performance monitoring should be performed in the middle cycles. The redundant circuits or channels in the components should be subjected to various working conditions, and sensitive parameters should be tested as much as possible to check for faults and intermittent phenomena. The temperature control point of the thermal cycle test should be selected on the leeward side of the test component, and the heat radiation from the external heat source cannot directly reach it.

[0194] After the new spectral detector has passed the above application environment test, the imaging effect of the new detector is further tested and compared with the imaging effect of the detector before the test. Fig.25 As shown in the figure. The imaging effect obtained after the test is normal, without bad pixels, missing stripes, water ripples and other phenomena, which proves that the imaging quality can still be guaranteed in a relatively complex application environment and the expected imaging effect can be achieved. It also proves the reliability of the new spectral detector and other electronic components and related processes.

[0195] Step 7: Imaging experiment and reconstruction analysis of miniaturized spaceborne spectral imager

[0196] Traditional satellite-borne spectroscopic imagers include a front-end telescope system, a spectroscopic system, and a detector. In this study, a coaxial two-mirror optical system is used as the front-end telescope system of the spectroscopic imager. The micro-nano photonic device packaged at the front end of the detector replaces the complex spectroscopic system of the traditional satellite-borne spectroscopic imager, thus realizing the miniaturization and integration of the satellite-borne spectroscopic imager. After the detection target passes through the coaxial two-mirror optical system and then the micro-nano photonic device, it is imaged on the array detector. Then, the spectral information is effectively extracted in combination with the computational spectrum reconstruction algorithm, which provides important technical support for the development of miniaturized satellite-borne spectroscopic imagers. The optical system of the miniaturized satellite-borne spectroscopic imager is shown in the attached figure. Fig.26 (a). After processing, assembly and testing, the ground prototype is shown in the attached figure. Fig.26 (b) as shown.

[0197] In order to ensure that the ground prototype of the miniaturized spaceborne spectral reconstruction imager can image actual scenes. We chose a room with a height of about 35m to conduct field imaging tests on ground scenes at a straight-line distance of about 5km. The miniaturized spaceborne spectral reconstruction imager adopts push-broom imaging, which is automatically controlled and rotated by a two-dimensional CNC turntable. TDI push-broom imaging is performed on ground scenes with a step length of 0.08°, and the push-broom imaging angle range is about 25°. At the same time, the Cameralink quick-view system of Wuhan Zhuomu Technology Co., Ltd. is used, which can display the received image data in real time and display part or all of it as required. It also detects the known content in the collected data in real time, finds error alarms, and supports preset storage channel data, which is stored while displaying. The imaging of distant scenes by the miniaturized spaceborne spectral reconstruction imager proves that the space camera design can provide high-quality images. During the field imaging process, the parameters such as focal length, TDI integration level and gain of the spectral reconstruction imager are adjusted and set, and multiple comparisons are made to find the best image point and obtain the best imaging effect. Through the research and preparation of the miniaturized satellite-borne spectral reconstruction imager, 25 groups of grayscale images under different spectral response curves were taken, as shown in the attached figure. Fig. 27 shown.

[0198] The push-broom spectral reconstruction and inversion of the miniaturized satellite-borne spectral reconstruction imager performs imaging operations under extremely fast relative motion, so the deviation of the reflected light or emitted light of the ground object to the detector is not large, that is, the light intensity of the same ground object mapped to each film strip is basically consistent, which is the basis for correct spectral inversion. Due to the large amount of image data, we selected a part of the field image for spectral and image reconstruction and restoration. The color restoration of the image collected by the miniaturized satellite-borne spectral reconstruction imager, in which the scene with typical characteristics was selected for spectral curve reconstruction, as shown in the attached figure. Fig.28 The reconstruction effect of some spectral curves in the image collected by the miniaturized space-borne spectral reconstruction imager is shown in the attached figure. Fig.29 shown.

[0199] The test results show that a new detector obtained by integrating micro-nano photonic devices as spectral modulation elements and detectors, combined with the proposed computational spectrum reconstruction algorithm, has been developed to develop a miniaturized satellite-borne spectroscopic imager. Compared with existing satellite-borne spectrometers, it has improved the light flux and field of view, enhanced stability, is insensitive to environmental influences, improved durability and reliability, and reduced the difficulty and cost of manufacturing. Finally, the main indicators of the miniaturized satellite-borne spectroscopic imager developed using micro-nano photonic devices were tested, and the test results are shown in the attached figure. Fig.30 shown.

Claims

1. A method for researching, designing, preparing and testing a wide-angle low-dimensional micro-nano photonic device for a miniaturized satellite-borne spectroscopic imager, characterized in that: The micro-nano photonic device mainly includes: In the visible-near-infrared spectral range (400-900nm), a visible-near-infrared micro-nano photonic device with polarization insensitivity, wide angle, high manufacturability, easy integration and packaging, and strong controllability of spectral curves was studied and designed. Combined with the proposed computational spectral reconstruction IReg algorithm, 20 broadband transmittance spectral curves that meet the spectral reconstruction requirements were selected, which enhanced the controllability of the spectral curves. Under TEM, the broadband transmittance spectral curves that make up the micro-nano photonic device are basically consistent within the incident angle of 30°, which is beneficial to improving the field of view. At the same time, the research and development on the same substrate reduces the processing difficulty of the micro-nano photonic device, makes it easier to integrate and package it into the front end of the detector, reduces the complexity of the optical system design, effectively reduces the impact of the external environment on the device performance, enhances the compactness and stability of the optomechanical structure, reduces the volume and weight, and makes it easy to integrate and miniaturize the spectral imager. The visible-near-infrared micro-nano photonic device is integrated with the large-array GSENSE5130 detector as a spectral modulation element, and combined with the spectral reconstruction IReg algorithm, the spectral reconstruction of typical ground objects is compared and analyzed. The results show that the MSE / ∈ index of low-dimensional visible-near infrared micro-nano photonic devices is within 10 under a large disturbance environment with a signal-to-noise ratio of 20 dB. -3 / 10 -2 It can achieve high-precision spectral reconstruction in the visible-near infrared spectrum, providing important support for the development of miniaturized spectral imagers. Among them, the integration method used to realize miniaturization in the satellite-borne spectral imager has the following main steps: Step 1: Establishment and selection analysis of spectral reconstruction mathematical model; Step 2: Propose a truncated spectrum reconstruction (IReg) algorithm based on a low-dimensional logarithmic regularization matrix; Step 3: Research and design of wide-angle low-dimensional visible-near-infrared micro-nano photonic devices; Step 4: Processing and preparation of visible-near infrared micro-nano photonic devices; Step 5: Testing and characterization of visible-near infrared micro-nano photonic devices; Step 6: Integrated packaging and performance testing of visible-near infrared micro-nano photonic devices; Step 7: Imaging experiment and reconstruction analysis of miniaturized spaceborne spectroscopic imager.

2. The novel detector prepared by processing and packaging the micro-nano photonic device according to claim 1 is characterized in that: Based on the lithography process, coating process and lift-off process, a composite grating array low-dimensional visible-near-infrared micro-nano photonic device was processed and prepared on the same substrate. It was used as a new spectral modulation element, which simplified the design of the optical system, reduced the difficulty of optomechanical assembly and improved the stability. This micro-nano photonic device was integrated and packaged on the front end of the large-array GSENSE5130 detector to obtain a new spectral detector. The process and imaging tests ensured that the parallelism between the film system and the pixel was above 95%.

3. The new spectrum detector application environment test according to claim 2 is characterized in that: This new spectral detector is placed in a vacuum pressure of ≤6.65×10 -3 Under the environment of Pa and temperature range of -20℃~+50℃, no bad pixels, missing stripes and water ripples occurred, which proved that the imaging quality can still be guaranteed and the expected imaging effect can be achieved in relatively complex application environments. It also proved the reliability of new spectral detectors and other electronic components and related processes.

4. The novel spectral detector according to claim 3 is applied to a satellite-borne spectral imager to achieve miniaturization, characterized in that: The new spectral detector is applied to the satellite-borne spectral imager, and the aperture is Φ126mm, the spatial resolution is 2.96m, the spectral resolution is 10nm, the weight is 6.65kg and the volume is 248.5×246.4×415mm. 3 The miniaturized satellite-borne spectral imager was used to conduct imaging experiments and achieved good color image restoration and spectral reconstruction.

Images