A multi-channel spectral imaging system and method

Abstract

This invention discloses a multi-channel spectral imaging system and method, belonging to the field of spectral imaging technology. The system includes an array optical imaging lens, an array multi-channel spectral sampling filter, an array color camera, and a spectral processing circuit module. This system enables solid-state snapshot imaging without moving parts, featuring a compact structure, high stability, and suitability for dynamic scenes, vehicle-mounted / airborne, and other complex operating conditions. The method of this invention completes multi-channel spectral imaging by constructing a measurement matrix, acquiring and converting target image signals, establishing the correlation between spectral signals and acquired signals, sparse representation and optimization solutions, super-resolution reconstruction, and data cube construction. It exhibits excellent reconstruction accuracy and noise resistance, and can flexibly expand the array size to adapt to the spectral coverage requirements of different application scenarios.

Description

Technical Field

[0001] This invention relates to the field of spectral imaging technology, specifically to a multi-channel spectral imaging system and method. Background Technology

[0002] Hyperspectral imaging technology can simultaneously acquire spatial and continuous spectral information of a target scene, making it irreplaceable in applications such as material identification, environmental monitoring, and precision medicine. However, traditional spectral imaging equipment generally suffers from drawbacks such as large size, high cost, and slow acquisition speed, making it difficult to adapt to the needs of dynamic scenes or portable platforms.

[0003] Traditional multispectral imaging systems primarily acquire target spectral information through various spectral dispersive systems, such as narrowband filter type, dispersive spectrometers, array filters, tunable narrowband filter type, and interferometric imaging spectrometers. Core dispersive components include acousto-optic tunable filters, liquid crystal tunable filters, Fabry-Perot tunable filters, interferometric filters, linear graded filters, prisms, and gratings. While these systems achieve high spectral and spatial resolution, they sacrifice temporal resolution, making them unsuitable for applications with high real-time requirements.

[0004] To balance spectral, spatial, and temporal resolution, snapshot-based multichannel spectral imaging systems have become a research focus, and the mainstream approaches are currently divided into three categories:

[0005] 1. Aperture-splitting multichannel spectrophotometer: This type of camera uses a single monochrome camera and splits the optical path using special prisms, mirrors, and other components to distribute the light to the camera, achieving multichannel spectral imaging. However, optical path splitting easily leads to spatial information loss, the optical system design is complex, the stability of the mechanical structure is difficult to guarantee, and the spectral and spatial resolutions are relatively low due to the limited number of channels.

[0006] 2. Metasurface-based snapshot multispectral imaging: Integrating metasurface structures on the surface of a CMOS sensor simplifies traditional complex spectroscopic systems into monolithic thin-film devices, enabling the acquisition of complete cubic spectral data in a single exposure. This offers advantages such as compact structure and rapid imaging. However, this technology relies on high-precision nanofabrication processes, resulting in high manufacturing costs. Furthermore, the accompanying algorithms are complex, making real-time processing difficult. Essentially, it sacrifices spatial resolution for spectral resolution.

[0007] 3. Multispectral imaging with a monocular single-channel multi-camera architecture: This method uses multiple independent monochrome cameras, each equipped with filters for different wavelengths, placed in parallel to simultaneously image the target scene. It has the advantages of simultaneous imaging across multiple spectral channels and high spatial resolution. However, the spectral resolution is determined by the number of channels and is difficult to improve. Furthermore, the single-channel uses narrowband filtering technology, resulting in low spectral energy utilization.

[0008] In summary, existing snapshot multispectral imaging solutions have failed to achieve an ideal balance in terms of snapshot imaging capability, compact structure, spectral / spatial resolution, and light energy utilization. Aperture-splitting systems have complex optical paths and large volumes, metasurface solutions are costly and have complex algorithms, and multi-eye composite narrowband filter structures are limited by the number of channels and have low spectral resolution, which cannot meet the practical application requirements of dynamic, portable, and high-precision imaging. Summary of the Invention

[0009] This invention addresses the shortcomings of existing technologies by providing a multi-channel spectral imaging system and method. It solves the technical problems of low spectral resolution, low light energy utilization, complex structure, and difficulty in balancing snapshot imaging and compactness in existing spectral imaging systems, and achieves high light energy utilization, high spectral resolution, compact structure, and stable snapshot multi-channel spectral imaging.

[0010] To achieve the above-mentioned objectives, the present invention adopts the following technical solution: a multi-channel spectral imaging system, comprising, from the object side to the image side, an m×n array optical imaging lens, an array multi-channel spectral sampling filter arranged in a one-to-one correspondence with the optical axis of the array optical imaging lens, an array color camera, and a spectral processing circuit module electrically connected to the array color camera.

[0011] The array color camera is a CMOS or CCD type color camera, with a working wavelength covering 200nm-1100nm, replacing traditional black and white cameras. It combines its own RGB response characteristics to achieve multi-dimensional acquisition of spectral information. The array multi-channel spectral sampling filter has a working wavelength of 200nm-1000nm and is a customized multi-bandpass spectral filter. The number of passbands of a single filter is ≥3, and the full width at half maximum (FWHM) of a single bandpass is controlled between 1nm-50nm to ensure the accuracy of spectral sampling. At the same time, the passband positions of any two filters are uncorrelated, that is, the passband positions of one filter do not completely overlap with the passband positions of any other filter, avoiding spectral information redundancy and improving the coverage of spectral sampling.

[0012] The spectral processing circuit module is the core control and processing unit of the system. It is responsible for acquiring the CMOS electrical signals of the array color camera and, based on the bandpass characteristics of the array multi-channel spectral sampling filter, reconstructing the spectral image through compressed sensing algorithm, realizing the conversion from acquired signals to hyperspectral data.

[0013] Furthermore, the m×n array optical imaging lens can be a commercial surveillance lens or a custom optical lens. The focal length, aperture, and field of view can be adapted to the application scenario, and the image distance can be adjusted to ensure clear imaging and adapt to target detection at different distances.

[0014] Furthermore, the preferred operating wavelength of the array multi-channel spectral sampling filter is 400nm-1000nm, which covers the visible to near-infrared region and is suitable for most practical application requirements; the filter has a maximum transmittance of ≥98% and an average transmittance of 60%-65%, which greatly improves the utilization rate of spectral energy.

[0015] Furthermore, the CMOS sensor of the array color camera is preferably 1 / 2.8 inches or larger, with an effective pixel count ≥1980×1080, a pixel size ≥2.9μm×2.9μm, a signal-to-noise ratio ≥27dB, and a dynamic range ≥69dB, ensuring the clarity of image acquisition and the reliability of the signal. After multiplying and superimposing the transmittance curve of the filter with the RGB response curve of the color camera, at least 18 different spectral channels can be generated, achieving a significant improvement in spectral resolution.

[0016] Furthermore, the spectral processing circuit module incorporates a sparse matrix and greedy algorithm solution unit, which performs spectral imaging reconstruction based on compressed sensing algorithm, thereby improving reconstruction accuracy and noise resistance.

[0017] Based on the above-described multi-channel spectral imaging system, the present invention also provides a corresponding imaging method, comprising the following steps:

[0018] Step 1: Measurement Matrix Construction: Using a deterministic spectral dispersion filter structure with an array of multi-channel spectral sampling filters, transmittance curves for different spectral bands are obtained. The measurement matrix Φ is then constructed based on these transmittance curves. i (λ), where i is the channel number of the filter, which corresponds one-to-one with the channels of the array optical imaging lens and the array color camera.

[0019] Step 2: Target Signal Acquisition and Conversion: The optical signal of the target under test is acquired through an m×n array optical imaging lens. After being modulated by an array multi-channel spectral sampling filter, it is received by an array color camera and converted into an electrical signal. The modulated electrical signal of the target image is then converted into a grayscale signal I. i , i=1,2,3…m×n, to complete the acquisition and preprocessing of the original signal.

[0020] Step 3: Establishing the correlation: Define the spectral information of the target to be measured as S(λ,x,y), where λ is the spectral wavelength, and x and y are the spatial horizontal and vertical coordinates; the RGB response curves of the CMOS of the array color camera are respectively D R (λ), D G (λ) and D B (λ), let D be the denoted D RGB (λ) is the set of the three; establish the grayscale signal I. i With S(λ,x,y), measurement matrix Φ i (λ), D RGBIntegral correlation of (λ):

[0021]

[0022] The range of values ​​for λ is the spectral range [λ]. min ,λ max ], λ min and λ max The lower and upper limits of the spectral range are represented by I. i It is S(λ,x,y), Φ i (λ) and D RGB (λ) is the integral value within the spectral range, which realizes the mapping between the acquired signal and the target spectral information.

[0023] Step 4: Construction of Optimization Equation: The spectral information S(λ,x,y) of the target to be measured is sparsely represented, and the optimization equation is constructed using a greedy algorithm based on compressed sensing:

[0024]

[0025] Where P(λ) is the spectral range weighting matrix, used to assign weights to spectral signals of different wavelengths; D(λ) is the D... RGB (λ) is the discretized matrix, adapted to the computational requirements of the algorithm; Ψ is the sparse matrix, α is the sparse coefficient, and k is the sparsity threshold. This equation is used to optimize the solution of spectral information.

[0026] Step 5: Single-unit spectral reconstruction: Solve the optimization equation using a greedy algorithm to obtain sparse coefficients α. Based on these sparse coefficients, perform super-resolution reconstruction of the target spectrum S(λ,x,y) of a single coding unit to obtain the spectral information of a single channel. The greedy algorithm can be one or more combinations of orthogonal matching pursuit algorithm, compressed sampling matching pursuit algorithm, and subspace pursuit algorithm to adapt to different accuracy and speed requirements.

[0027] Step 6: Hyperspectral Data Cube Construction: Repeat steps 4-5 to solve the spectrum of the target to be measured for all coding units of the array multi-channel spectral sampling filter in sequence. Integrate the spectral information of all units according to the spatial and spectral dimensions to construct a complete hyperspectral three-dimensional data cube (spectral dimension - spatial horizontal axis dimension - spatial vertical axis dimension) and complete the super-resolution reconstruction imaging of the multi-channel spectrum.

[0028] Beneficial effects: Compared with the prior art, the beneficial effects of the present invention are reflected in:

[0029] 1. Innovative multi-channel filtering technology enhances imaging efficiency and stability: By employing a single custom-designed multi-bandpass filter to achieve multi-channel spectral separation, information from three or more independent spectral bands can be acquired simultaneously in a single imaging session. This simplifies the optical structure and imaging process of traditional spectral imaging systems, avoids registration errors and timing delays caused by mechanical switching or complex optical paths, and enables solid-state snapshot imaging without moving parts. The system stability and imaging efficiency are significantly improved, making it particularly suitable for dynamic scenes or rapid spectral detection applications.

[0030] 2. High light energy utilization efficiency, improving imaging quality and data reliability: By utilizing the multi-bandpass characteristics of the filter, the light flux loss during the spectral sampling process is significantly reduced, making the spectral energy utilization rate entering the camera 3 times higher than that of traditional multi-channel spectral cameras; under the same lighting conditions, images with higher signal-to-noise ratio and more sufficient brightness can be acquired, effectively improving the imaging quality in low-light environments and providing a reliable data foundation for subsequent spectral reconstruction and quantitative analysis.

[0031] 3. High spectral and spatial resolution, while maintaining detection accuracy: The passband positions of the array multi-channel spectral sampling filters are independent and uncorrelated. The transmittance curve of the filter can be superimposed with the RGB response curve of the color camera to generate no less than 18 spectral channels. The number of spectral channels and spectral resolution are 3 times higher than those of traditional black and white camera architecture. At the same time, each unit of the system uses a full aperture for light collection, with no spatial information loss, ensuring high spatial resolution.

[0032] 4. High integration and flexibility, adaptable to multiple application scenarios: Adopting an integrated architecture of "array optical lens - multi-bandpass filter - color camera - spectral processing circuit", the structure is compact with no moving parts, has strong environmental adaptability, and is suitable for complex working conditions such as vehicle and airborne applications; the system scale is an m×n array, and the lens matrix arrangement can be adopted according to different application scenarios to ensure that the image overlap area occupies 20%-25% of the image width. The plane of the arranged color camera is fixed by a mechanical bracket, and the power supply output can be expanded according to the camera power and the number of cameras. After the number of channels increases, the response matrix corresponding to the added color camera is vertically stitched in the measurement matrix Φi(λ), and the processor specifications are adapted according to the computing power. It can be flexibly expanded to realize customized optimization of spectral coverage and material recognition capabilities, and adapt to the needs of multiple fields such as material recognition, environmental monitoring, and precision medicine.

[0033] 5. Compressed sensing-based spectral reconstruction improves reconstruction accuracy and noise resistance: The greedy algorithm based on compressed sensing completes spectral reconstruction. Through sparse representation and optimization, the accuracy and noise resistance of spectral reconstruction are greatly improved while ensuring imaging speed. High-quality spectral data can still be obtained even in low light or complex environments. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0035] Figure 1 The diagram illustrates the structural principle of the multi-channel spectral imaging system provided in this embodiment of the invention. It clearly shows the device arrangement from the object side to the image side. The array optical imaging lens, the array multi-channel spectral sampling filter, and the array color camera CMOS are arranged coaxially in sequence. The spectral processing circuit module is electrically connected to the array color camera CMOS to achieve the integration of signal acquisition and reconstruction.

[0036] Figure 2 The transmittance curves of the array multi-channel spectral sampling filter provided in the embodiments of the present invention are shown. The horizontal axis represents wavelength (nm), and the vertical axis represents transmittance (%). The six curves are independent of each other, with no overlap in the passband positions, covering the 400nm-1000nm wavelength band.

[0037] Figure 3 The RGB response curves of the array color camera CMOS provided in the embodiments of the present invention; the horizontal axis is wavelength (nm) and the vertical axis is responsivity (au); the R, G, and B curves have high response in the red, green, and blue light bands, respectively.

[0038] Figure 4 This is a superposition curve of the filter transmittance curve and the RGB response curve of the array color camera CMOS provided in the embodiment of the present invention; the horizontal axis is the wavelength (nm) and the vertical axis is the superposition response value (au), forming a total of 18 independent response peaks, corresponding to 18 spectral channels.

[0039] The diagram is labeled as follows: 1-6-channel array optical imaging lens, 2-6-channel array multi-channel spectral sampling filter, 3-array color camera CMOS, 4-spectral reconstruction process. Detailed Implementation

[0040] The technical solution of the present invention will be described in detail below with reference to specific embodiments and accompanying drawings. These embodiments are only used to explain the present invention and are not intended to limit the scope of protection of the present invention.

[0041] Example: This example provides a 6-channel (m×n=2×3) multi-channel spectral imaging system and corresponding imaging method, which is suitable for dynamic spectral detection scenarios in portable environmental monitoring.

[0042] like Figure 1As shown, the multi-channel spectral imaging system of this embodiment includes, from the object side to the image side, the following components in sequence: a 6-channel array optical imaging lens 1, a 6-channel array multi-channel spectral sampling filter 2 corresponding one-to-one with the optical axis of the lens, a 6-channel array color camera CMOS 3, and a spectral processing circuit module electrically connected to the 6-channel array color camera CMOS 3. Figure 1 Mid-spectral reconstruction process 4).

[0043] The 6-channel array optical imaging lens 1 is a commercial surveillance lens with a focal length of 2.03mm, an aperture of F2.2, and a field of view of D: 117°, H: 110°, V: 74.6°. Distortion is >-2.5%. The image distance can be adjusted via a screw thread to ensure clear imaging within a distance of 0.5m-10m, meaning the image sharpness value acquired by the CMOS is ≥30. The optical axis parallelism of the lens, filter, and CMOS in each channel is ≤0.01°, and the coaxiality deviation is ≤0.05mm.

[0044] Furthermore, the 6-channel array multi-channel spectral sampling filter 2 operates in the 400nm-1000nm band. It consists of six independent multi-bandpass spectral filters. The filter substrate is quartz glass, fabricated using ion beam-assisted electron beam thermal evaporation technology. The filter dimensions are 6.7×6.7mm, consistent with the IR-CUT size of the camera lens used. Each filter has three passbands, with a full width at half maximum (FWHM) of 20nm-50nm for each passband. The passband positions of any two filters are completely non-overlapping. The 6-channel array multi-channel spectral sampling filter 2 has a maximum transmittance of 98.77%, a minimum transmittance of 11.3%, and an average transmittance of 60%-65%. The array filter device is fixed with an aluminum alloy bracket, which must ensure that the optical axis deviation remains ≤0.05mm after vibration of 5-2000Hz under automotive conditions.

[0045] Furthermore, the array color camera CMOS 3 has a size of 1 / 2.8 inches, a maximum effective pixel count of 1980×1080, a pixel size of 2.9μm×2.9μm, a signal-to-noise ratio of 27dB, a dynamic range of 69dB, and an operating wavelength of 400nm-1000nm, which matches the operating wavelength of the 6-channel array multi-channel spectral sampling filter 2; the transmittance curve of the 6-channel array multi-channel spectral sampling filter 2 ( Figure 2 The RGB response curve of the array color camera CMOS 3 is multiplied and superimposed with the RGB response curve of the array color camera CMOS 3 (Figure 3), and the superposition method is as follows:

[0046] i=1,2,3,4,5,6; l=1,2,3,4…18;

[0047] Where Ti is the transmittance curve of the i-th filter, and D BGR D respectively B DG and D R , respectively represent the B / G / R corresponding curves of CMOS; the first channel is superimposed with the B channel of CMOS to form channel 1, superimposed with the G channel to form channel 2, superimposed with the R channel to form channel 3, and so on, to generate 18 different spectral channels (Figure 4).

[0048] Furthermore, the spectral processing circuit module incorporates an ARM Cortex-M4 core processor and a compressed sensing algorithm library, including an orthogonal matching pursuit algorithm solution unit, enabling real-time processing and spectral reconstruction of the acquired signals.

[0049] Based on the aforementioned 6-channel spectral imaging system, the multi-channel spectral imaging method of this embodiment includes the following steps:

[0050] Step 1: Using the deterministic spectral dispersion filtering structure of the 6-channel array multi-channel spectral sampling filter 2, the transmittance curves of the 6 filters in the 400nm-1000nm band are obtained by testing. Based on the transmittance curves, 6 measurement matrices Φ1(λ)-Φ6(λ) are constructed, where λ∈[400,1000]nm.

[0051] Step 2: The optical signal of the target under test is acquired by the 6-channel array optical imaging lens 1, modulated by the 6-channel array multi-channel spectral sampling filter 2, received by the array color camera CMOS 3 and converted into an electrical signal, and the 6 modulated target image electrical signals are converted into grayscale signals I1-I6 respectively.

[0052] Step 3: Define the spectral information of the target to be measured as S(λ,x,y), and the RGB response curve of the array color camera CMOS 3 as D. R (λ), D G (λ), D B (λ), establish grayscale signal I i (i=1-6) and S(λ,x,y),Φ i (λ), D RGB Integral correlation of (λ):

[0053]

[0054] I i Let S(λ,x,y), Φ i (λ) and D RGB (λ) is the integral value in the 400nm-1000nm band;

[0055] Step 4: Sparsely represent S(λ,x,y) and construct the optimization equation:

[0056]

[0057] Where P(λ) is the equal weight matrix for the 400nm-1000nm band, and D(λ) is the weight matrix for D. RGB (λ) is the discretized matrix, Ψ is the discrete cosine transform sparse matrix, k=5 is the sparsity threshold. The orthogonal matching pursuit algorithm is used to solve the problem: the residual is calculated using least squares, the atom index with the largest product of the residual and the atom is selected, the number of iterations does not exceed 50, and convergence occurs when the residual is less than 10^-3.

[0058] Step 5: Solve the above optimization equation using the orthogonal matching pursuit algorithm to obtain the sparse coefficient α, and reconstruct the target spectrum S(λ,x,y) of a single filter coding unit based on α super-resolution.

[0059] Step 6: Repeat steps 4-5 to solve the encoding units of the 6 filters in sequence. Integrate the spectral information of the 6 units according to the spatial x, y and spectral λ dimensions to construct a hyperspectral three-dimensional data cube in the 400nm-1000nm band. This completes the multi-channel spectral imaging of the environmental monitoring target. The reconstructed spectral data resolution can reach 20nm, and the imaging speed is ≤100ms / frame. Spectral calibration is performed using a monochromator and a total reflection plate, with a wavelength accuracy of ±0.5nm, meeting the real-time detection requirements of dynamic scenes.

[0060] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A multi-channel spectral imaging system, characterized in that, From the object side to the image side, the system includes, in sequence: an m×n array optical imaging lens, an array multi-channel spectral sampling filter arranged in a one-to-one correspondence with the optical axis of the array optical imaging lens, an array color camera, and a spectral processing circuit module electrically connected to the array color camera. The array color camera operates in the 200nm-1100nm band, and its imaging element is CMOS or CCD. The array multi-channel spectral sampling filter operates in the wavelength range of 200nm-1000nm. The array multi-channel spectral sampling filter is a multi-bandpass spectral filter, with a single filter having ≥3 passbands and a full width at half maximum (FWHM) of 1nm-50nm for each passband. The passband positions of any two array multichannel spectral sampling filters are uncorrelated, meaning that the passband positions of all multiband passes of one filter do not completely coincide with the passband positions of any other filter; The spectral processing circuit module is used to acquire the CMOS signal of the array color camera and reconstruct the spectral image according to the bandpass characteristics of the array multi-channel spectral sampling filter.

2. The multi-channel spectral imaging system according to claim 1, characterized in that, The m×n array optical imaging lens is a commercial surveillance lens or a custom optical lens, and the image clarity can be ensured by adjusting the image distance.

3. The multi-channel spectral imaging system according to claim 1, characterized in that, The array of multi-channel spectral sampling filters operates in the 400nm-1000nm range, with a maximum transmittance of ≥98% and an average transmittance of 60%-65%.

4. The multi-channel spectral imaging system according to claim 1, characterized in that, The array color camera has a CMOS size of 1 / 2.8 inches or larger, an effective pixel count of ≥1980×1080, a pixel size of ≥2.9μm×2.9μm, a signal-to-noise ratio of ≥27dB, and a dynamic range of ≥69dB.

5. The multi-channel spectral imaging system according to claim 1, characterized in that, The transmittance curve of the array multi-channel spectral sampling filter is multiplied and superimposed with the RGB response curve of the array color camera to generate no less than 18 different spectral channels.

6. The multi-channel spectral imaging system according to claim 1, characterized in that, The spectral processing circuit module completes spectral imaging reconstruction based on the compressed sensing algorithm, and has a built-in sparse matrix and greedy algorithm solution unit.

7. A multi-channel spectral imaging method based on the system described in any one of claims 1-6, characterized in that, Includes the following steps: Step 1: Obtain transmittance curves for different spectral bands using a deterministic spectral dispersion filtering structure with an array of multi-channel spectral sampling filters, and construct the measurement matrix Φ. i (λ), where i is the filter channel number; Step 2: Acquire the image signal of the target under test through an m×n array optical imaging lens and an array color camera, and convert the image signal of the target under test, which has been modulated by a filter, into a grayscale signal I. i , i = 1, 2, 3…m×n; Step 3: Define the spectral information of the target to be measured as S(λ,x,y), and the RGB response curves of the CMOS of the array color camera are respectively D R (λ), D G (λ) and D B (λ), establish I i With S(λ,x,y), Φ i (λ), D RGB The integral correlation of (λ), where D RGB (λ) is D R (λ), D G (λ), D B The set of (λ), λ∈[λ] min ,λ max ], λ min and λ max These represent the lower and upper limits of the spectral range; Step 4: Sparsely represent the spectral information S(λ,x,y) of the target to be measured, and construct the optimization equation using a greedy algorithm based on compressed sensing: Where P(λ) is the spectral range weight matrix, and D(λ) is the D... RGB (λ) is the discretized matrix, Ψ is the sparse matrix, α is the sparsity coefficient, and k is the sparsity threshold. Step 5: Solve the optimization equation using a greedy algorithm to obtain the sparse coefficients α, and reconstruct the target spectrum S(λ,x,y) of a single coding unit based on the sparse coefficients. Step 6: Repeat steps 4-5 to solve the spectrum of the target under test for all coding units of the array multi-channel spectral sampling filter in sequence, construct a complete hyperspectral three-dimensional data cube, and complete the super-resolution reconstruction imaging of the multi-channel spectrum.

8. The multi-channel spectral imaging method according to claim 7, characterized in that, In step 3, the integral correlation relationship is the grayscale signal I. i The target spectral information S(λ,x,y) and the measurement matrix Φ are the measurement matrix. i (λ) and RGB response curve D RGB (λ) in the spectral range [λ] min ,λ max The integral value within ], that is: 。 9. The multi-channel spectral imaging method according to claim 7, characterized in that, In step 4, the greedy algorithm is one or more combinations of orthogonal matching pursuit algorithm, compressed sampling matching pursuit algorithm, and subspace pursuit algorithm.

10. The multi-channel spectral imaging method according to claim 7, characterized in that, In step 6, the dimensions of the hyperspectral three-dimensional data cube are spectral dimension - spatial horizontal axis dimension - spatial vertical axis dimension.

Images