A method for spatial-spectral aliasing correction of a DMD on-chip scanning hyperspectral imaging system

Abstract

The application discloses a method for correcting spatial-spectral aliasing of a DMD on-chip scanning type hyperspectral imaging system, which comprises the following steps: taking the geometric center of a DMD spatial light modulation micromirror array as a reference point to obtain a reference diffraction dispersion function matrix; after physically extrapolating other wave bands based on the reference diffraction dispersion function matrix, establishing a DMD spatial light modulation micromirror array dispersion function according to spatial shift invariance and dispersion law, and establishing a column diffraction dispersion function model corresponding to the case that all the micromirrors in a column are turned on at the same time and a system global observation column vector; based on the system global observation column vector, introducing spatial dimension total variation and spectral dimension total variation prior constraints, constructing a joint optimization objective function, solving the joint optimization objective function, and generating target three-dimensional hyperspectral data after spatial-spectral aliasing correction. The application improves the efficiency and precision of spatial-spectral aliasing correction, and can realize diffraction correction without changing the hardware structure of an existing DMD hyperspectral imaging system.

Description

Technical Field

[0001] This invention belongs to the field of hyperspectral imaging technology, specifically relating to a method for correcting spatial-spectral aliasing in a DMD on-chip scanning hyperspectral imaging system. Background Technology

[0002] Digital micromirror device (DMD) on-chip scanning hyperspectral imaging is an emerging hyperspectral imaging system. By controlling the micromirrors on the DMD chip to scan column by column, the target image can be scanned column by column on the DMD chip, thereby dividing the target's three-dimensional data cube into a series of collectable dispersive spectra. This technology overcomes the shortcomings of traditional spatial scanning hyperspectral imaging methods based on aperture or slit mechanical motion, such as the mutual constraints of temporal-spatial-spectral resolution and complex system structure, opening up a new solution for realizing miniaturized, high temporal-spatial-spectral resolution imaging technology. However, in the above-mentioned DMD on-chip scanning hyperspectral imaging system, a key physical problem has not yet been effectively solved: the diffraction effect of the DMD micromirrors causes the spatial and spectral dimensions to overlap, i.e., spatial-spectral aliasing.

[0003] Currently, correction schemes for DMD diffraction effects are mainly divided into two categories: one is hardware physical suppression methods, which avoid diffraction components by changing the optical system structure or modulating the DMD working mode; the other is software algorithm correction methods, which obtain the system point spread function through dense calibration or data-driven fitting of the entire field of view and the entire band, and then realize image restoration by solving the inverse problem.

[0004] However, existing solutions all have significant drawbacks: hardware physical suppression methods increase system complexity and cost, and cannot fundamentally solve the problem of spatial-spectral aliasing in pushbroom mode; software algorithm correction methods involve a huge amount of calibration work and cannot achieve joint correction of spatial and spectral dimensions, resulting in insufficient reconstruction accuracy and making it difficult to meet the requirements of engineering applications for high resolution and high spectral fidelity. Summary of the Invention

[0005] To address the aforementioned problems in the prior art, this invention provides a method for correcting spatial-spectral aliasing in a DMD on-chip scanning hyperspectral imaging system.

[0006] The technical problem to be solved by this invention is achieved through the following technical solution: A method for correcting spatial-spectral aliasing in a DMD on-chip scanning hyperspectral imaging system includes: Using the geometric center of the DMD spatial light modulation micromirror array as the reference point, the reference diffraction diffusion function matrix is ​​obtained; After physically extrapolating the reference diffraction diffusion function matrix to other bands, the diffusion function of the DMD spatial light modulation micromirror array is established based on the spatial shift invariance and dispersion law. Based on the diffusion function of the DMD spatial light modulation micromirror array, a column diffraction diffusion function model is established for the simultaneous activation of an entire column of micromirrors. Based on the column diffraction diffusion function model, a discrete equation for the light intensity contribution from the target voxel to the detector pixel in each pushbroom is established, and the equations are spliced ​​together to obtain the global observation column vector of the system. Based on the system's global observation column vector, and by introducing prior constraints of spatial and spectral total variation, a joint optimization objective function is constructed. After solving the joint optimization objective function, the target three-dimensional hyperspectral data after spatial-spectral aliasing correction is generated.

[0007] Optionally, using the geometric center of the DMD spatial light modulation micromirror array as a reference point, a reference diffraction diffusion function matrix is ​​obtained, including: After controlling the activation of a single micromirror located at the geometric center of the DMD spatial light modulation micromirror array, the original bright field image acquired by the area array detector under incoherent monochromatic light source illumination at a preset reference wavelength is obtained. After all the micromirrors of the DMD spatial light modulation micromirror array are turned off, a dark field image is acquired. After removing the dark field image from the original bright field image and performing energy normalization, a reference diffraction diffusion function matrix is ​​generated.

[0008] Optionally, physical extrapolation to other bands is performed based on the reference diffraction spread function matrix, including: Get the preset wavelength scaling factor Preset energy attenuation coefficient and the pre-calibrated system relative spectral response coefficient ; Based on a preset wavelength scaling factor, a preset energy attenuation coefficient, and the system's relative spectral response coefficient, the reference diffraction spread function matrix is ​​physically extrapolated to other bands to generate arbitrary wavelengths as shown in the following formula. Extrapolation model of diffusion function of single micromirror :

[0009] in, Represents the reference diffraction diffusion function matrix; Represents the pixel coordinates of the plane of the area array detector; Indicates the preset reference wavelength; This indicates that the system operates at any wavelength. The spectral response under the following conditions This indicates that the system is at a preset reference wavelength. The spectral response under [condition].

[0010] Optionally, the process of establishing the diffusion function of the DMD spatial light modulation micromirror array includes: Based on the spatial shift invariance and dispersion law of the front-mounted dual telecentric imaging optical path, the diffusion function of the DMD spatial light modulation micromirror array is established according to the following formula:

[0011] in, Indicates the number of columns and rows of the microscope; , Represents the physical coordinates of the micromirror; Indicates the horizontal magnification of the optical system; Indicates the vertical magnification of the optical system; This indicates the horizontal dispersion shift caused by the dispersive element; The diffusion function of the DMD spatial light modulation micromirror array is used to characterize the position of the first... i Column, No. j The absolute position response of the diffraction spot corresponding to the row of micromirrors on the array detector.

[0012] Optionally, based on the diffusion function of the DMD spatial light modulation micromirror array, a column diffraction diffusion function model is established for when an entire column of micromirrors is simultaneously activated, including: Obtain the preset spatial lighting intensity weighting factor; Based on the preset spatial illumination intensity weighting factor and the diffusion function of the DMD spatial light modulation micromirror array, the column diffraction diffusion function model corresponding to the simultaneous activation of an entire column of micromirrors is established according to the following formula:

[0013] in, Indicates wavelength as The light in the m Model of the diffraction diffusion function generated when multiple micromirrors are turned on simultaneously; This represents the preset spatial lighting intensity weighting factor; N Indicates the first m The number of rows contained in a column microscope.

[0014] Optionally, based on the column diffraction diffusion function model, a discrete equation for the light intensity contribution from the target voxel to the detector pixel in each pushbroom is established, and the equations are concatenated to obtain the global observation column vector of the system, including: The three-dimensional hyperspectral data to be reconstructed is flattened into a one-dimensional target vector, and the two-dimensional aliased image acquired by each push-broom exposure is flattened into a one-dimensional observation vector. Based on the column diffraction diffusion function model, a one-dimensional target vector, and a one-dimensional observation vector, a discrete light intensity contribution equation for the target voxel to the detector pixel is established. By concatenating the discrete light intensity contribution equations corresponding to the column diffraction diffusion function models at all push-broom times, a global observation column vector of the system is constructed.

[0015] Optionally, based on the system's global observation column vector, and by introducing prior constraints of spatial-dimensional total variation and spectral-dimensional total variation, a joint optimization objective function is constructed, including: Based on the system's global observation column vector, and introducing prior constraints of spatial-dimensional total variation and spectral-dimensional total variation, a joint optimization objective function is constructed according to the following formula:

[0016] in, This represents the target's three-dimensional hyperspectral data after spatial spectral aliasing correction; Represents the system's global observation column vector; The forward observation matrix is ​​obtained by synthesizing the diffraction diffusion function model across all columns. This represents a one-dimensional target vector after the three-dimensional hyperspectral data to be reconstructed has been flattened. Representation space total variation; Represents the total variation in the spectral dimension; Represents the total variation coefficients in the preset spatial dimension; This represents the total variation coefficients for the preset spectral dimension.

[0017] This invention also provides a DMD on-chip scanning hyperspectral imaging system, which is used to implement the spatial-spectral aliasing correction method of the above-mentioned DMD on-chip scanning hyperspectral imaging system. The system sequentially includes: an incoherent uniform illumination module, a front dual-telecentric imaging optical path, a DMD spatial light modulation micromirror array, a rear dual-telecentric spectral relay optical path, an area array detector, and a signal processor. A dispersive element is provided in the rear dual-telecentric spectral relay optical path, wherein: Incoherent uniform illumination module, used to provide a low-coherence, uniform incident light field at a preset reference wavelength; The front-mounted dual telecentric imaging optical path is used to image the target scene onto the surface of the DMD spatial light modulation micromirror array and provides spatial shift invariance. DMD spatial light modulation micromirror array is used to obtain spatially modulated beams by column sweeping and then transmit them to the rear dual telecentric spectral relay optical path. The rear dual telecentric spectral relay optical path is used to relay and spectrally disperse the spatially modulated beam reflected by the DMD spatial light modulation micromirror array and then transmit it to the area array detector. Area array detectors are used to acquire two-dimensional aliased images; A signal processor is used to process two-dimensional aliased images and execute the steps of a spatial-spectral aliasing correction method for a DMD on-chip scanning hyperspectral imaging system.

[0018] Alternatively, the incoherent uniform illumination module can be implemented using either a decoherent laser light source or by adding an ultra-narrow band filter in front of the white light source.

[0019] Optionally, the dispersive element is a transmissive bulk holographic blazed grating.

[0020] This invention provides a method for correcting spatial-spectral aliasing in a DMD on-chip scanning hyperspectral imaging system. The method includes: obtaining a reference diffraction spread function matrix using the geometric center of the DMD spatial light modulation micromirror array as a reference point; physically extrapolating the reference diffraction spread function matrix to other bands; establishing the DMD spatial light modulation micromirror array spread function based on spatial shift invariance and dispersion laws; establishing a column diffraction spread function model corresponding to the simultaneous activation of a whole column of micromirrors based on the DMD spatial light modulation micromirror array spread function model; establishing a discrete equation for the light intensity contribution from the target voxel to the detector pixel in each pushbroom based on the column diffraction spread function model, and concatenating them to obtain the system's global observation column vector; constructing a joint optimization objective function based on the system's global observation column vector and introducing prior constraints of spatial and spectral total variation; and generating the target's three-dimensional hyperspectral data after spatial-spectral aliasing correction after solving the joint optimization objective function. This invention improves the efficiency and accuracy of spatial-spectral aliasing correction and achieves diffraction correction without modifying the hardware structure of existing DMD hyperspectral imaging systems.

[0021] The present invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0022] Figure 1 This is an overall architecture diagram of a DMD on-chip scanning hyperspectral imaging system provided in an embodiment of the present invention; Figure 2 This is a schematic flowchart of a spatial-spectral aliasing correction method for a DMD on-chip scanning hyperspectral imaging system provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a process for obtaining a reference diffraction diffusion function matrix according to an embodiment of the present invention; Figure 4 This is a schematic diagram of a process for constructing a global observation column vector of a system, provided by an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the correspondence between a DMD column-wise push scan and the acquired two-dimensional aliased images provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the result after spatial spectral aliasing correction provided in an embodiment of the present invention. Detailed Implementation

[0023] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0024] A Discrete Motion Detector (DMD) is a spatial light modulator composed of millions of micrometer-scale tiltable micromirrors, widely used in projection displays, maskless lithography, confocal microscopy, and hyperspectral imaging. In a DMD on-chip scanning hyperspectral imaging system, the target is imaged on the surface of the DMD chip. The DMD sequentially opens the micromirrors column by column, reflecting the corresponding spatial pixels into the subsequent beam-splitting module. After dispersion, a two-dimensional image is acquired by an area array detector, and finally, a three-dimensional spectral data cube is obtained through push-broom scanning.

[0025] In existing DMD on-chip scanning hyperspectral imaging systems, a key physical problem remains unsolved: the diffraction effect of DMD micromirrors causes spatial and spectral information aliasing, i.e., spatial-spectral aliasing. DMD micromirror units, with dimensions on the micrometer scale and arranged periodically, inevitably experience diffraction. When a single micromirror is activated, its reflected light not only propagates along the geometric optical direction but also diffuses in all directions, forming complex diffraction patterns.

[0026] For on-chip scanning hyperspectral imaging systems using a DMD, during DMD scanning imaging, the diffracted light generated by the micromirrors in the currently activated column will diffuse to the spatial positions of adjacent columns. This means that during the system's push-broom imaging process, the light corresponding to the first micromirror on the detector target surface will diffuse to the spatial positions of adjacent columns. j The light signal received by the micromirror region does not originate solely from the target scene. j The pure geometric optical mapping of the column. Due to the depth modulation of the inherent two-dimensional blazed grating diffraction effect of the DMD micromirror, the light-gathering region inevitably mixes in the light-gathering region from the first column. j -1 column, number j The diffraction diffusion energy of the micromirrors in column +1 and even more distant adjacent columns. This high-frequency energy leakage causes severe cross-column spatial crosstalk, which is superimposed and coupled with the dispersion broadening effect of the spectrometer itself, ultimately leading to a dual degradation of the system's spatial resolution and spectral fidelity.

[0027] Simultaneously, the diffraction pattern also broadens in the spectral dispersion direction, causing monochromatic light that should fall on a single detector pixel to diffuse to adjacent pixels, resulting in spectral aliasing. The overlapping of diffraction patterns of different wavelengths reduces spectral resolution. This crosstalk between the spatial and spectral dimensions couples together, forming spatial-spectral aliasing. Consequently, the spectral image acquired by the detector is not a pure "space-spectral" correspondence, but rather an aliasing result modulated by diffraction effects, severely reducing the accuracy and spatial resolution of spectral measurements.

[0028] Currently, industry solutions for eliminating and correcting DMD diffraction effects mainly follow two basic paths: hardware physical suppression and software image reconstruction, as detailed below: In hardware-based diffraction suppression, the core approach typically involves altering the optical system structure or modulating the physical operating mode of the DMD to circumvent or homogenize high-frequency diffraction components at the optical path level. For example, in research on DMD-based reflective coded aperture snapshot spectral imagers, existing techniques have improved the signal-to-noise ratio of reconstructed spectral images by deeply analyzing off-axis imaging mechanisms and introducing optical axis translation structures, utilizing precise hardware physical compensation methods. However, this approach has significant limitations in practical applications: as it is essentially a physical optical compensation, it heavily relies on substantial modifications to the system's optical path structure. This not only greatly increases the size and assembly redundancy of the optical system, leading to a significant increase in the processing and maintenance costs of precision optical components, but also causes the system to lose its flexibility for different imaging environments. Furthermore, this research focuses on aberration correction at the geometric optics level and does not address the modeling of diffraction effects caused by DMD micromirror units, making it difficult to fundamentally solve the high-frequency spectral blurring problem. Moreover, this hardware suppression scheme is completely unable to handle the spatial-spectral aliasing phenomenon generated during the operation of scanning systems.

[0029] Furthermore, during the system design process, researchers discovered that the diffraction effect generated by the microstructure of the DMD micromirrors was a core challenge affecting imaging quality. To suppress and manage this diffraction effect, researchers conducted a detailed experimental analysis of the diffraction pattern of the DMD chip, comparing the energy distribution of zero-order and first-order diffraction. Ultimately, this study achieved optimal modulation contrast by adjusting the optical configuration at the physical hardware level, actively discarding zero-order light and specifically selecting first-order diffracted light to construct the optical path. While this research solved the contrast reduction problem caused by diffraction in mid-infrared single-pixel microscopy systems using DMD, its core solution relies on spatial filtering and avoidance at the hardware optical path level. This method of extracting specific diffraction orders through physical optical paths has significant limitations: it not only sacrifices the overall light energy utilization of the system and requires extremely high precision in optical path assembly, but also fails to fundamentally solve the spatial-spectral aliasing problem caused by the activation of multiple micromirrors in on-chip scanning hyperspectral systems.

[0030] In diffraction cancellation based on algorithm correction and image reconstruction, such schemes do not alter the physical optical path. Instead, they obtain the system's prior diffusion function and combine it with an inverse problem-solving framework to numerically restore the degraded image. In the prior diffusion function acquisition stage, existing technologies generally employ exhaustive or fitting strategies for specific states. This involves setting up a large number of discrete field-of-view grids for ray tracing to extract the spatially varying diffusion function matrix carrying aberration and diffraction information, and then solving the full-field-of-view modeling problem through field-of-view interpolation and stitching. Existing technologies also employ a data-driven approach, relying on extremely large sets of standard training images repeatedly input into the system. Through batch gradient descent algorithms, a massive amount of computational power is consumed during backpropagation to iteratively fit the error convolution kernel for non-ideal imaging.

[0031] Although the existing technologies based on hardware physical control and software algorithm compensation have achieved expected benefits in specific scenarios, if they are forcibly transferred to DMD on-chip scanning hyperspectral imaging systems, they expose deep-seated mechanistic mismatches and insurmountable engineering barriers.

[0032] First, hardware suppression schemes based on micromirror merging or optical path diffusion inevitably sacrifice the system's Nyquist sampling frequency and energy transfer efficiency. For on-chip scanning hyperspectral systems that demand extreme spatial resolution and detection of weak light signals, the macroscopic scaling of micromirrors leads to severe loss of high-frequency spatial details of the target, and the customized optical path compensation structure greatly increases the complexity of the instrument's optomechanical design, making it difficult to achieve system lightweighting and cross-platform portability.

[0033] Secondly, in hyperspectral imaging architectures, DMD chips typically contain millions of physical micromirrors, and the spatial morphology, breadth, and energy distribution of their diffraction spots vary drastically and nonlinearly with the operating wavelength. If existing technological frameworks are used to perform physical calibration or simulation to obtain the diffusion function matrix across the entire field of view and wavelength band, the phase space formed by multiplying the spatial coordinate grid dimension by the continuous spectral channel dimension would be extremely large. Experiments involving tens or even hundreds of millions of single-mirror and single-wavelength cross-acquisitions would be completely impractical, both during instrument factory calibration and periodic calibration during use. The existing technological system severely lacks the mechanistic support for physical dimensionality reduction using the laws of optical diffraction and geometric transformation.

[0034] More critically, existing image reconstruction algorithms generally neglect the unique column diffraction superposition effect and the deep spatial-spectral coupling caused by grating dispersion in pushbroom scanning mode. Pushbroom systems deflect sequentially along physical columns. The detector, in a single exposure, does not acquire the independent response of isolated micromirrors, but rather a macroscopic column diffraction distribution formed by the superposition of hundreds of micromirrors under incoherent illumination. Existing models completely lack a mathematical description of this macroscopic synthesis mechanism. Furthermore, the spatial-spectral aliasing phenomenon induced by DMD diffraction in pushbroom systems not only manifests as cross-column pixel energy leakage in the spatial dimension, but also results in a nonlinear spectral drift in the horizontal direction of the target light wave under the influence of subsequent dispersive elements. Existing algorithms, when solving the inverse problem, generally adopt a strategy of dividing the broadband spectrum into independent channels for segment-by-segment peeling calculations. This approach artificially severs the coupling relationship between spatial crosstalk and spectral dispersion in the joint domain, failing to fundamentally achieve accurate inversion of the ill-conditioned matrix, ultimately leading to irreversible spectral peak shifts and spurious broadening in the reconstructed target spectral reflectance curve.

[0035] Therefore, to improve the efficiency and accuracy of spatial-spectral aliasing correction without modifying the hardware structure of existing DMD hyperspectral imaging systems, this invention provides a spatial-spectral aliasing correction method for DMD on-chip scanning hyperspectral imaging systems. This method can be applied to DMD on-chip scanning hyperspectral imaging systems, such as... Figure 1 As shown, Figure 1 This is an overall architecture diagram of a DMD on-chip scanning hyperspectral imaging system provided in an embodiment of the present invention. The system may sequentially include: an incoherent uniform illumination module, a front dual telecentric imaging optical path, a DMD spatial light modulation micromirror array, a rear dual telecentric spectral relay optical path, an area array detector, and a signal processor (not shown). A dispersive element is provided in the rear dual telecentric spectral relay optical path.

[0036] The system includes: an incoherent uniform illumination module for providing a low-coherence, uniformly illuminated incident light field at a preset reference wavelength; a front dual-telecentric imaging optical path for imaging the target scene onto the surface of the DMD spatial light modulation micromirror array and providing spatial shift invariance; a DMD spatial light modulation micromirror array for transmitting the spatially modulated beam obtained by column-wise sweeping to the rear dual-telecentric spectral relay optical path; a rear dual-telecentric spectral relay optical path for relaying and spectrally dispersing the spatially modulated beam reflected by the DMD spatial light modulation micromirror array before transmitting it to the area array detector; an area array detector for acquiring two-dimensional aliased images; and a signal processor for processing the two-dimensional aliased images and executing the steps of the spatial-spectral aliasing correction method of the DMD on-chip scanning hyperspectral imaging system.

[0037] Combining the above-mentioned on-chip scanning hyperspectral imaging system of DMD, such as Figure 2 As shown, Figure 2 This is a flowchart illustrating a spatial-spectral aliasing correction method for a DMD on-chip scanning hyperspectral imaging system provided in an embodiment of the present invention, including: Step 201: Using the geometric center of the DMD spatial light modulation micromirror array as the reference point, obtain the reference diffraction diffusion function matrix.

[0038] One approach is to first select and activate a single micromirror in the central region of the DMD spatial light modulation micromirror array, extract the reference diffraction diffusion function matrix with the geometric center of the DMD spatial light modulation micromirror array as the origin, and use it as the basic input parameter for subsequent physical modeling.

[0039] The reference diffraction spread function matrix physically characterizes the spatial energy diffusion distribution footprint formed on the two-dimensional discrete target surface of the array detector when a single micromirror unit is in the on state at a preset reference wavelength, after passing through the system's rear optical components. Each pixel value within this reference diffraction spread function matrix represents the weight proportion of energy leaked from the total reflected energy of that single micromirror and falling into a specific pixel position on the array detector. It not only objectively records the higher-order diffraction stellar features caused by dispersive elements but also includes the inherent optical aberrations of the system itself.

[0040] It should be noted that the aforementioned reference diffraction diffusion function matrix was not set subjectively based on experience, but was obtained through rigorous physical calibration procedures. The calibration data came from the actual system, ensuring the consistency between the diffusion function and the actual system.

[0041] In some alternative embodiments, such as Figure 3 As shown, Figure 3 This is a schematic flowchart of obtaining a reference diffraction diffusion function matrix provided by an embodiment of the present invention, including: Step 301: After controlling the activation of a single micromirror located at the geometric center of the DMD spatial light modulation micromirror array, the original bright field image acquired by the area array detector under incoherent monochromatic light source illumination at a preset reference wavelength is obtained.

[0042] Step 302: After all the micromirrors in the DMD spatial light modulation micromirror array are turned off, acquire the dark field image.

[0043] Step 303: Remove the dark field image from the original bright field image and perform energy normalization to generate the reference diffraction diffusion function matrix.

[0044] Specifically, after controlling the activation of a single micromirror located at the geometric center of the DMD spatial light modulation micromirror array, uniform illumination can be achieved using an incoherent monochromatic light source with a known preset reference wavelength, and the original bright field image can be acquired by an area array detector. For example, the incoherent monochromatic light source with the preset reference wavelength can be 532nm monochromatic incoherent narrowband light, that is, the preset reference wavelength is 532nm.

[0045] Next, after all the micromirrors in the DMD spatial light modulation micromirror array are turned off, the system acquires a dark field image. The dark field image is then subtracted from the original bright field image to remove the dark current noise and readout stripe noise from the area array detector.

[0046] After energy normalization, the reference diffraction spread function matrix is ​​finally generated. Optionally, based on strictly adhering to the law of conservation of optical energy, the gray values ​​of all pixels within a preset window can be summed, and each pixel value can be divided by the sum to obtain the reference diffraction spread function matrix.

[0047] In this embodiment, by obtaining the reference diffraction diffusion function matrix after energy normalization, the system can calculate the cross-column crosstalk leakage ratio of any wavelength and any micromirror column through pure mathematical and physical deduction, thereby avoiding the catastrophic workload of scanning and calibrating millions of micromirrors one by one in the traditional exhaustive method.

[0048] Step 202: After physically extrapolating the reference diffraction diffusion function matrix to other bands, establish the diffusion function of the DMD spatial light modulation micromirror array based on the spatial shift invariance and dispersion law.

[0049] After obtaining the reference diffraction spread function matrix, a physical model can be established using scalar diffraction theory—that is, the spatial scale of the diffraction pattern is strictly proportional to the wavelength—to accurately derive the continuous diffraction response across the entire wavelength range from a single reference wavelength. Since the physical extrapolation is based on the wavelength scaling relationship of diffraction, the accuracy of the spread function across the entire wavelength range is guaranteed. Combined with the calibration process of the reference diffraction spread function matrix, this avoids both the deviations of purely theoretical calculations and the enormous workload of full-field calibration.

[0050] In some optional embodiments, physical extrapolation to other bands based on the reference diffraction spread function matrix includes: obtaining a preset wavelength scaling factor. Preset energy attenuation coefficient and the pre-calibrated system relative spectral response coefficient Based on a preset wavelength scaling factor, a preset energy attenuation coefficient, and the system's relative spectral response coefficient, the reference diffraction spread function matrix is ​​physically extrapolated to other bands to generate arbitrary wavelengths as shown in the following formula. Extrapolation model of diffusion function of single micromirror

[0051] ; in, Represents the reference diffraction diffusion function matrix; Represents the pixel coordinates of the plane of the area array detector; Indicates the preset reference wavelength; This indicates that the system operates at any wavelength. The spectral response under the following conditions This indicates that the system is at a preset reference wavelength. The spectral response under [condition].

[0052] In the process of physical extrapolation to other bands using the reference diffraction diffusion function matrix, the energy attenuation coefficient is the core correction parameter to ensure that the physical model conforms to objective laws. If the energy attenuation coefficient is not introduced and only spatial scale is scaled, the total energy of the calculation in the long band will be more than that in the short band, which seriously violates the law of conservation of energy.

[0053] The energy attenuation coefficient physically characterizes the energy conservation law in scalar diffraction theory. Specifically, when the wavelength of the target optical signal changes from a preset reference wavelength... Variation to any wavelength When this happens, the spatial scale of its diffraction spot will be proportional to... The scaling and broadening. To ensure the conservation of total photon energy reflected by the same micromirror at different wavelengths, as the light spot diffuses in two-dimensional space, i.e., the area of ​​the light spot increases... The local peak energy weight projected onto each discrete pixel of the array detector must be objectively attenuated accordingly.

[0054] This energy attenuation coefficient requires no additional experimental measurement; rather, it is derived through rigorous physical deduction. According to the two-dimensional optical diffraction scaling model, the value of this energy attenuation coefficient is strictly constrained by the wavelength scaling factor. Its mathematical value is .

[0055] The pre-calibrated system relative spectral response coefficient middle, For the system at any wavelength The spectral response under the following conditions For the system at a preset reference wavelength The spectral response is obtained by comprehensively considering information such as the reflectivity of the DMD micromirror, the diffraction efficiency of the dispersive element, and the quantum efficiency of the detector as a function of wavelength. This response can be obtained by performing a one-time broadband calibration of the system using a standard integrating sphere light source or a standard blackbody radiation source. The specific implementation process can be found in existing technologies and will not be elaborated here.

[0056] The final system relative spectral response coefficient physically characterizes the system's lateral linear magnification of the target scene. It determines the actual physical size of a certain microscopic physical dimension (such as the micromirror spacing d) on the DMD spatial light modulation micromirror array, which, after transmission through a complex optical lens group, is projected onto the target surface of the area array detector.

[0057] Next, the diffusion function of the DMD spatial light modulation micromirror array can be established based on the spatial shift invariance and dispersion law. In some optional embodiments, the process of establishing the diffusion function of the DMD spatial light modulation micromirror array includes: based on the spatial shift invariance and dispersion law of the front dual telecentric imaging optical path, the diffusion function of the DMD spatial light modulation micromirror array is established according to the following formula: ; in, Indicates the number of columns and rows of the microscope; , Represents the physical coordinates of the micromirror; Indicates the horizontal magnification of the optical system; Indicates the vertical magnification of the optical system; This indicates the horizontal dispersion shift caused by the dispersive element; The diffusion function of the DMD spatial light modulation micromirror array is used to characterize the position of the first... i Column, No. j The absolute position response of the diffraction spot corresponding to the row of micromirrors on the array detector.

[0058] The horizontal and vertical magnification of the optical system can be obtained through standard geometric distortion calibration. The specific procedure is as follows: First, place a standard high-precision micro / nano-fabricated grid target or checkerboard target in front of the system entrance pupil. Then, use a preset reference wavelength for monochromatic illumination and control the micromirrors of the DMD spatial light modulation micromirror array to be fully open. Obtain a clear image of the target on the area array detector, extract the pixel spacing of the target feature points on the area array detector, and divide it by the actual known physical spacing of the standard high-precision micro / nano-fabricated grid target or checkerboard target. The resulting ratio is the accurate actual magnification of the system.

[0059] The horizontal dispersion shift caused by the dispersive elements is not a variable that needs to be calculated or calibrated in real time before each imaging in this invention, but rather an inherent reference parameter of the hyperspectral imaging instrument. It characterizes the fixed physical dispersion displacement law of the dispersive elements used in the system after the design and assembly are established.

[0060] Step 203: Based on the diffusion function of the DMD spatial light modulation micromirror array, establish a column diffraction diffusion function model corresponding to the simultaneous activation of a whole column of micromirrors.

[0061] In push-broom imaging mode, the DMD independently opens an entire column of micromirrors each time. Let's assume the current position is the [number missing]. m Columns, containing N According to the principle of incoherent superposition, the total diffraction response of this column is the linear sum of the intensities of the diffusion functions of each micromirror. The column diffraction diffusion function model precisely encapsulates the diffraction interference superposition effect and the spatial-spectral coupling crosstalk mechanism of multiple micromirrors being turned on simultaneously from a macroscopic physical perspective.

[0062] In some optional embodiments, a column diffraction diffusion function model corresponding to the simultaneous activation of an entire column of micromirrors is established based on the diffusion function of the DMD spatial light modulated micromirror array, including: obtaining a preset spatial illumination intensity weighting factor; and establishing a column diffraction diffusion function model corresponding to the simultaneous activation of an entire column of micromirrors based on the preset spatial illumination intensity weighting factor and the DMD spatial light modulated micromirror array diffusion function, according to the following formula: ; in, Indicates wavelength as The light in the m Model of the diffraction diffusion function generated when multiple micromirrors are turned on simultaneously; This represents the preset spatial lighting intensity weighting factor; N Indicates the first m The number of rows contained in a column microscope.

[0063] The aforementioned preset spatial illumination intensity weighting factor is used to characterize the non-uniformity of the actual light source illuminance. In determining this parameter, a monochromatic pure light source can be first set to uniformly illuminate the system. Then, after all the micromirrors in the DMD spatial light modulation micromirror array are turned on, the area array detector captures an image at this point. The brightness distribution shown in the image is the true energy spectrum distribution of the system. It should be noted that because of micromirror diffraction, the image will have high-frequency ripples. A two-dimensional low-pass filter needs to be applied to this image to smooth out the high-frequency ripples, thereby obtaining a smooth energy envelope.

[0064] Finally, the global maximum light intensity value I is found on the extracted smooth energy envelope surface. max Divide the intensity of all pixels on the envelope plane by I. max This results in a two-dimensional weighted matrix with values ​​ranging from [0, 1]. Each value in this matrix corresponds to a physical coordinate. Preset spatial lighting intensity weighting factor .

[0065] Step 204: Based on the column diffraction diffusion function model, establish the discrete equation of light intensity contribution from the target voxel to the detector pixel in each pushbroom, and splice them to obtain the global observation column vector of the system.

[0066] In essence, true optical imaging is a continuous three-dimensional light field in multiple spaces and spectral integrals on a two-dimensional focal plane. In order for computer algorithms to process it, it must be discretized. Thus, based on the column diffraction diffusion function model, a discrete equation for the light intensity contribution from the target voxel to the detector pixel in each push-broom can be established and spliced ​​to obtain the global observation column vector of the system.

[0067] like Figure 4As shown, Figure 4 This is a schematic diagram of a process for constructing a system global observation column vector according to an embodiment of the present invention, including: Step 401: Flatten the three-dimensional hyperspectral data to be reconstructed into a one-dimensional target vector, and flatten the two-dimensional aliased image obtained from each push-broom exposure into a one-dimensional observation vector.

[0068] Step 402: Based on the column diffraction diffusion function model, the one-dimensional target vector, and the one-dimensional observation vector, establish the discrete light intensity contribution equation of the target voxel to the detector pixel.

[0069] Step 404: Concatenate the discrete light intensity contribution equations corresponding to the column diffraction diffusion function models at all push-broom times to construct the global observation column vector of the system.

[0070] In pushbroom imaging, a physically continuous three-dimensional target needs to be transformed into a computable discrete one-dimensional vector. Assume the three-dimensional hyperspectral data to be reconstructed is flattened into a one-dimensional target vector. At the same time, when the DMD is enabled... m During microscopy, a single push-broom exposure of the array detector captures a frame of a two-dimensional aliased image containing diffraction and dispersion aliasing. This image is then sequentially flattened into a one-dimensional column vector. .

[0071] The aforementioned one-dimensional target vector is the ultimate target variable obtained through inverse solving. It represents the pure spatial-spectral information of the target scene and, physically, represents the intrinsic three-dimensional hyperspectral data cube that the measured target should present under an ideal aberration-free and diffraction-free system. Among these, and This represents the discrete sampling resolution of the reconstruction result in a two-dimensional space, and its value is strictly consistent with the physical micromirror array size of the DMD spatial light modulator. For example, when using a 1080p resolution DMD chip, the value is set as follows: , . This represents the number of discrete bands in the spectral dimension of the reconstructed result, also known as the number of spectral channels. It is calculated from the system's operating wavelength range and the desired spectral resolution. For example, if the system operates between 400 nm and 700 nm and the inversion step size is set to 10 nm, the calculated number of spectral channels is... .

[0072] Since the actual physical imaging process involves multiple integral couplings of a continuous three-dimensional light field onto a two-dimensional array detector, this flattening vector is introduced... This is the mathematical prerequisite for successfully constructing the global observation column vector of the system. It makes it possible to encapsulate the complex physical diffraction and dispersion aliasing into a high-dimensional sparse positive observation matrix, thus completely opening up the path for solving algebraic inverse problems using proximal optimization algorithms.

[0073] Based on the aforementioned diffraction diffusion function model, one-dimensional target vector, and one-dimensional observation vector, the discrete intensity contribution equation of the target voxel to the detector pixel is... As shown below: ; In this context, the target voxel physically refers to an extremely small physical unit of the measured target in the three-dimensional space-spectral domain, and its coordinates are defined as follows: In physical terms, a detector pixel refers to a two-dimensional physical photosensitive unit on the target surface of an array detector, and its coordinates are defined as follows: ; The wavelength representing the target voxel; Indicates the total number of reconstructed bands; For index labels, the value is... ; Indicates the preset wavelength sampling step size; Represents detector pixels The system-integrated additive noise received in a single exposure is introduced to indicate that this model is a non-ideal observation model that conforms to the characteristics of real physical instruments.

[0074] As the DMD spatial light modulation micromirror array completes its full-array column-by-column sweep, the discrete light intensity contribution equations corresponding to the column diffraction diffusion function models at all sweep times are spliced ​​together end-to-end, ultimately constructing the system's global observation column vector.

[0075] During pushbroom imaging, the DMD sequentially activates different micromirror columns (a total of M scans) in a time sequence, and the area array detector subsequently acquires M frames of two-dimensional aliased images. These M independent column vectors are vertically concatenated in the time dimension to construct an ultra-high-dimensional global observation column vector for the system. The system's global observation column vector It can be defined as: ; in, This represents the forward observation matrix, which encapsulates the horizontal and vertical magnification of the optical system, the horizontal dispersion offset caused by the dispersive elements, and the preset row-column crosstalk spatial weights. This represents the system's overall additive noise vector, characterizing the sum of unavoidable Poisson shot noise, dark current of the array detector, and readout noise during actual detection.

[0076] This embodiment achieves a fundamental paradigm shift from "local isolated solution" to "global coupled solution" by concatenating the discrete light intensity contribution equations corresponding to the column diffraction diffusion function models at all push-broom times in the time series into a global observation column vector and constructing a forward observation matrix. This equation successfully transforms the extremely complex hardware degradation process of deep spatial-spectral aliasing, which cannot be directly eliminated at the physical optics level, into a standard large-scale linear algebraic inverse problem, thus completely opening up the path for high-fidelity digital dealiasing using near-end optimization algorithms.

[0077] Step 205: Based on the system's global observation column vector, and by introducing prior constraints of spatial and spectral total variation, a joint optimization objective function is constructed. After solving the joint optimization objective function, the target three-dimensional hyperspectral data after spatial-spectral aliasing correction is generated.

[0078] Based on the system's global observation column vector, and incorporating prior constraints from spatial and spectral total variation, the joint optimization objective function can be constructed according to the following formula: ; in, This represents the target's three-dimensional hyperspectral data after spatial spectral aliasing correction; Represents the system's global observation column vector; The forward observation matrix is ​​obtained by synthesizing the diffraction diffusion function model across all columns. This represents a one-dimensional target vector after the three-dimensional hyperspectral data to be reconstructed has been flattened. Representation space total variation; Represents the total variation in the spectral dimension; Represents the total variation coefficients in the preset spatial dimension; This represents the total variation coefficients for the preset spectral dimension.

[0079] In this embodiment, by introducing spatial total variation to suppress cross-row and column artifacts and spectral total variation to overcome spectral broadening caused by dispersion in the joint optimization objective function, and by using a near-end optimization algorithm to solve the objective function, the target three-dimensional hyperspectral data with thoroughly corrected spatial-spectral aliasing can be output.

[0080] This invention proposes a method for correcting spatial-spectral aliasing in a DMD on-chip scanning hyperspectral imaging system. By calibrating only the reference diffraction spread function matrix of a single micromirror at a single preset reference wavelength, and utilizing the periodic structure characteristics of the DMD micromirror array and the wavelength scaling relationship of diffraction, the spread function at any micromirror and any wavelength is extrapolated through a physical model. This reduces the calibration workload from millions of steps in traditional methods to a single step, solving the problem of the enormous calibration workload in existing technologies and making the diffraction correction method practical for engineering applications. It also improves the efficiency of spatial-spectral aliasing correction. Furthermore, by establishing a model from the reference diffraction spread function matrix of a single micromirror to the column diffraction spread function model, the method accurately describes the diffraction superposition effect when multiple micromirrors are simultaneously activated, specifically for the column-scanning operation mode of the DMD. This makes the correction model more closely reflect the actual imaging process, avoiding model inaccuracies caused by neglecting column diffraction effects and improving the accuracy of spatial-spectral aliasing correction. Moreover, it does not require modification to the existing hardware structure of the DMD hyperspectral imaging system, exhibiting good compatibility and promotional value. Users only need to add the algorithm module of this invention to their existing system to achieve diffraction correction without hardware modification.

[0081] This embodiment also provides a specific implementation process for a spatial-spectral aliasing correction method in a DMD on-chip scanning hyperspectral imaging system. The system operates in the visible light band (400 nm ~ 700 nm) and is pre-constructed. The specific hardware parameter configuration and full array implementation steps are as follows: DMD Spatial Light Modulation Micromirror Array: Employs a DMD chip with a micromirror array resolution of 1920×1080, i.e. , The physical side length and row / column spacing of the micromirrors are both The micromirror deflection angle is Optical System: Both the front imaging optical path and the rear spectral relay optical path adopt a dual telecentric optical path design. The system's lateral magnification configuration in both the horizontal and vertical directions is... , Dispersive element: A transmissive volume phase holographic (VPH) grating is installed in the rear optical path. Area array detector: Employs a resolution of... The sCMOS area array camera has a pixel size of The specific correction process is as follows: 1. Sparse calibration of the reference diffraction spread function matrix: In actual acquisition, the system first records the dark field image. The control unit drives the DMD to deflect only the center coordinates. i =960, j A single micromirror with a wavelength of 540 nm, using a preset reference wavelength. Illumination is provided by an incoherent monochromatic light source. After exposure and acquisition by the area array detector, the signal processor subtracts the dark background and performs energy integration normalization, thereby extracting the reference diffraction spread function matrix with the geometric center as the origin. . 2. Full-band physical extrapolation of the diffusion function of a single micromirror: using a standard mercury-xenon lamp uniform illumination system, a reference center wavelength is set. The linear dispersion coefficient was measured. That is, horizontal dispersion offset For wavelengths ranging from 400 nm to 700 nm, with a preset wavelength sampling step size... The signal processor sequentially calculates the wavelength scaling factor for each of the 31 target spectral bands. The relative spectral response weighting coefficients are pre-calibrated using the standard integrating sphere at the system entrance pupil. The reference diffraction distribution is rigorously extrapolated to the entire wavelength range using the following formula. Extrapolation model of diffusion function of single micromirror : ; 3. Macroscopic synthesis modeling of column diffraction in push-broom mode: The DMD spatial light modulation micromirror array is illuminated with a 532 nm narrowband decoherent light source. After two-dimensional low-pass filtering and normalization, the true two-dimensional spatial flat-field illumination weight matrix of 1920×1080 is extracted. Its elements are denoted as When the system performs column push scan, the current column... Columns contain The micromirrors are activated simultaneously. Based on the principle of superposition of incoherent light source intensities, the array of micromirrors at the wavelength is synthesized. The diffraction diffusion function model below is: ; 4. Construction of the joint spatial-spectral forward observation matrix: The signal processor flattens the ideal, non-aliased 1920×1080×31 three-dimensional hyperspectral data cube to be reconstructed into a one-dimensional target vector. When performing the... During the subsequent pushbroom, the two-dimensional aliased image containing spatial-spectral aliasing captured by the detector in a single exposure is flattened into a one-dimensional observation vector, such as... Figure 5 As shown, Figure 5This is a schematic diagram illustrating the correspondence between a DMD column-wise scan and the acquired two-dimensional aliased images provided in an embodiment of the present invention. The diagram shows the target broadband images formed on the working surface of the DMD corresponding to the scene in columns 1 to N, and also shows the two-dimensional aliased images acquired by the detector in a single exposure. The horizontal axis represents wavelength, and the vertical axis represents the spatial height of the detector, corresponding to the row number of the DMD spatial light modulation micromirror array. Subsequently, the image sequences from all 1920 scans are stitched together end-to-end to form the system's global observation column vector.

[0082] 5. Inverse Inversion with Physical Prior Constraints: Using a standard color calibration chart as the target, substituting it into the forward equation and introducing 30dB Gaussian white noise, 1920 frames of pushbroom observation images with aliasing are generated, corresponding to the aforementioned global observation vector. A joint optimization objective function with spatial and spectral total variation is then constructed: ; Among them, the preset spatial total variation coefficients Preset spectral dimension total variation coefficients The Alternating Direction Method of Multipliers (ADMM) proximal optimization algorithm framework is invoked to iteratively solve the problem until convergence. The final output is the target 3D hyperspectral data that has been thoroughly corrected for spatial crosstalk and spectral aliasing, such as... Figure 6 As shown, Figure 6 This is a schematic diagram of the result after spatial-spectral aliasing correction according to an embodiment of the present invention. The horizontal axis represents wavelength, and the vertical axis represents the spatial height of the detector, corresponding to the row number of the DMD spatial light modulation micromirror array. The diagram shows the ideal pure spectra corresponding to columns 1 through N, i.e., the results after spatial-spectral aliasing correction.

[0083] Based on the same inventive concept, this invention also provides a DMD on-chip scanning hyperspectral imaging system. This system is used to implement the spatial-spectral aliasing correction method of the aforementioned DMD on-chip scanning hyperspectral imaging system. The system sequentially includes: an incoherent uniform illumination module, a front dual-telecentric imaging optical path, a DMD spatial light modulation micromirror array, a rear dual-telecentric spectral relay optical path, an area array detector, and a signal processor. A dispersive element is provided in the rear dual-telecentric spectral relay optical path, wherein: Incoherent uniform illumination module, used to provide a low-coherence, uniform incident light field at a preset reference wavelength; The front-mounted dual telecentric imaging optical path is used to image the target scene onto the surface of the DMD spatial light modulation micromirror array and provides spatial shift invariance. DMD spatial light modulation micromirror array is used to obtain spatially modulated beams by column sweeping and then transmit them to the rear dual telecentric spectral relay optical path. The rear dual telecentric spectral relay optical path is used to relay and spectrally disperse the spatially modulated beam reflected by the DMD spatial light modulation micromirror array and then transmit it to the area array detector. Area array detectors are used to acquire two-dimensional aliased images; A signal processor is used to process two-dimensional aliased images and execute the steps of a spatial-spectral aliasing correction method for a DMD on-chip scanning hyperspectral imaging system.

[0084] Alternatively, the incoherent uniform illumination module can be implemented using either a decoherent laser light source or by adding an ultra-narrow band filter in front of the white light source.

[0085] Optionally, the dispersive element is a transmissive bulk holographic blazed grating.

[0086] It should be noted that, for the embodiments of the DMD on-chip scanning hyperspectral imaging system, since they are basically similar to the method embodiments, the description is relatively simple. For relevant parts, please refer to the description of the method embodiments. All embodiments of the above-mentioned method for correcting spatial-spectral aliasing of a DMD on-chip scanning hyperspectral imaging system are applicable to the DMD on-chip scanning hyperspectral imaging system and can achieve the same or similar beneficial effects.

[0087] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A method for correcting spatial-spectral aliasing in a DMD on-chip scanning hyperspectral imaging system, characterized in that, The method includes: Using the geometric center of the DMD spatial light modulation micromirror array as the reference point, the reference diffraction diffusion function matrix is ​​obtained; After physically extrapolating the reference diffraction diffusion function matrix to other bands, the diffusion function of the DMD spatial light modulation micromirror array is established according to the spatial shift invariance and dispersion law. Based on the diffusion function of the DMD spatial light modulation micromirror array, a column diffraction diffusion function model is established for the simultaneous activation of an entire column of micromirrors. Based on the aforementioned column diffraction diffusion function model, a discrete equation for the light intensity contribution from the target voxel to the detector pixel in each pushbroom is established, and the equations are spliced ​​together to obtain the global observation column vector of the system. Based on the system's global observation column vector, and by introducing prior constraints of spatial and spectral total variation, a joint optimization objective function is constructed. After solving the joint optimization objective function, the target three-dimensional hyperspectral data after spatial-spectral aliasing correction is generated.

2. The method according to claim 1, characterized in that, The process of obtaining the reference diffraction diffusion function matrix, using the geometric center of the DMD spatial light modulation micromirror array as the reference point, includes: After controlling the activation of a single micromirror located at the geometric center of the DMD spatial light modulation micromirror array, the original bright field image acquired by the area array detector under incoherent monochromatic light source illumination at a preset reference wavelength is obtained. After all the micromirrors of the DMD spatial light modulation micromirror array are turned off, a dark field image is acquired; The dark field image is removed from the original bright field image, and after energy normalization, the reference diffraction diffusion function matrix is ​​generated.

3. The method according to claim 1 or 2, characterized in that, The physical extrapolation to other bands based on the reference diffraction spread function matrix includes: Get the preset wavelength scaling factor Preset energy attenuation coefficient and the pre-calibrated system relative spectral response coefficient ; Based on the preset wavelength scaling factor, the preset energy attenuation coefficient, and the system relative spectral response coefficient, the reference diffraction spread function matrix is ​​physically extrapolated to other bands to generate arbitrary wavelengths as shown in the following formula. Extrapolation model of diffusion function of single micromirror : in, Represents the reference diffraction diffusion function matrix; Represents the pixel coordinates of the plane of the area array detector; Indicates the preset reference wavelength; This indicates that the system operates at any wavelength. The spectral response under the following conditions This indicates that the system is at a preset reference wavelength. The spectral response under [condition].

4. The method according to claim 3, characterized in that, The process of establishing the diffusion function of the DMD spatial light modulation micromirror array includes: Based on the spatial shift invariance and dispersion law of the front dual telecentric imaging optical path, the diffusion function of the DMD spatial light modulation micromirror array is established according to the following formula: in, Indicates the number of columns and rows of the microscope; , Represents the physical coordinates of the micromirror; Indicates the horizontal magnification of the optical system; Indicates the vertical magnification of the optical system; This indicates the horizontal dispersion shift caused by the dispersive element; The diffusion function of the DMD spatial light modulation micromirror array is used to characterize the position of the first... i Column, No. j The absolute position response of the diffraction spot corresponding to the row of micromirrors on the array detector.

5. The method according to claim 4, characterized in that, The step of establishing a column diffraction diffusion function model corresponding to the simultaneous activation of an entire column of micromirrors based on the diffusion function of the DMD spatial light modulation micromirror array includes: Obtain the preset spatial lighting intensity weighting factor; Based on the preset spatial illumination intensity weighting factor and the diffusion function of the DMD spatial light modulation micromirror array, a column diffraction diffusion function model corresponding to the simultaneous activation of an entire column of micromirrors is established according to the following formula: in, Indicates wavelength as The light in the first m Model of the diffraction diffusion function generated when multiple micromirrors are turned on simultaneously; This represents the preset spatial lighting intensity weighting factor; N Indicates the first m The number of rows contained in a column microscope.

6. The method according to claim 1 or 2, characterized in that, Based on the column diffraction diffusion function model, a discrete equation for the light intensity contribution from the target voxel to the detector pixel in each pushbroom is established and concatenated to obtain the system's global observation column vector, including: The three-dimensional hyperspectral data to be reconstructed is flattened into a one-dimensional target vector, and the two-dimensional aliased image acquired by each push-broom exposure is flattened into a one-dimensional observation vector. Based on the aforementioned column diffraction diffusion function model, the aforementioned one-dimensional target vector, and the aforementioned one-dimensional observation vector, a discrete light intensity contribution equation for the target voxel to the detector pixel is established. By concatenating the discrete light intensity contribution equations corresponding to the column diffraction diffusion function models at all push-broom times, the global observation column vector of the system is constructed.

7. The method according to claim 1 or 2, characterized in that, The joint optimization objective function is constructed based on the system's global observation column vector and by introducing prior constraints of spatial-dimensional total variation and spectral-dimensional total variation, including: Based on the system's global observation column vector, and introducing prior constraints of spatial-dimensional total variation and spectral-dimensional total variation, a joint optimization objective function is constructed according to the following formula: in, This represents the target's three-dimensional hyperspectral data after spatial spectral aliasing correction; Represents the system's global observation column vector; The forward observation matrix is ​​obtained by synthesizing the diffraction diffusion function model across all columns. This represents a one-dimensional target vector after the three-dimensional hyperspectral data to be reconstructed has been flattened. Representation space total variation; Represents the total variation in the spectral dimension; Represents the total variation coefficients in the preset spatial dimension; This represents the total variation coefficients for the preset spectral dimension.

8. A DMD on-chip scanning hyperspectral imaging system, characterized in that, The system is used to implement the spatial-spectral aliasing correction method of the DMD on-chip scanning hyperspectral imaging system according to any one of claims 1-7. The system sequentially comprises: an incoherent uniform illumination module, a front dual-telecentric imaging optical path, a DMD spatial light modulation micromirror array, a rear dual-telecentric spectral relay optical path, an area array detector, and a signal processor. The rear dual-telecentric spectral relay optical path is provided with a dispersive element, wherein: The incoherent uniform illumination module is used to provide a low-coherence, uniformly illuminated incident light field at a preset reference wavelength. The aforementioned front-mounted dual telecentric imaging optical path is used to image the target scene onto the surface of the DMD spatial light modulation micromirror array and provides spatial shift invariance. The DMD spatial light modulation micromirror array is used to obtain a spatially modulated beam by column sweeping and then transmit it to the rear dual telecentric spectral relay optical path. The rear dual telecentric spectral relay optical path is used to relay and spectrally disperse the spatially modulated beam reflected by the DMD spatial light modulation micromirror array and then transmit it to the area array detector. The area array detector is used to acquire two-dimensional aliased images; The signal processor is used to process the two-dimensional aliased image and execute the steps of the spatial-spectral aliasing correction method of the DMD on-chip scanning hyperspectral imaging system.

9. The system according to claim 8, characterized in that, The incoherent uniform illumination module is implemented using either a decoherent laser light source or by adding an ultra-narrow band filter in front of a white light source.

10. The system according to claim 8, characterized in that, The dispersive element is a transmissive bulk holographic blazed grating.

Images