Structured light illumination super-resolution microscopic imaging method based on combination of cross-correlation and principal component analysis

Abstract

The invention provides a structured light illumination super-resolution microscopic imaging method (Cr-PCA) based on the combination of cross-correlation and principal component analysis, illumination pattern vectors containing required parameters are obtained through different-frequency-spectrum self-correlation, and parameter estimation is carried out on the illumination pattern vectors through the principal component analysis technology. Besides, a double-window mask operator of a map base-blank window and M-estimation are introduced to carry out filtering and linear fitting respectively, so that the calculation amount is remarkably reduced, and the parameter estimation precision is greatly improved at the same time. Under the condition that high-frequency stripes and a high-power objective lens are used, high-fidelity, high-robustness and efficient real-time observation of living cells can still be achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of super-resolution fluorescence microscopy, and specifically relates to a structured illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis. Background Art

[0002] As one of the indispensable research tools in the modern biomedical field, the optical microscope is the "eye" for peeping into the mysteries of the microscopic world and the "key" to opening the door to the microscopic world. The traditional optical microscope mainly consists of optical lenses, which magnify or image objects using the lenses. However, the magnification of the optical microscope cannot be increased infinitely, as it is limited by the optical diffraction limit. Therefore, structures smaller than 200 nm cannot be observed. This means that scientists cannot distinguish viruses or individual protein molecules smaller than cells, seriously hindering the exploration of subcellular structures by humans. In the past 20-odd years, a variety of fluorescence super-resolution technologies have been successively proposed, breaking through the Abbe diffraction limit and visualizing biomolecules at the nanoscale to the single-molecule level, opening up a new path for observing life activities at the cellular and subcellular levels. Among many fluorescence super-resolution technologies, structured illumination microscopy (SIM) is more suitable for multi-color dynamic imaging of living cells due to its advantages such as fast imaging speed, large imaging field of view, low phototoxicity and photobleaching, and the need for no complex sample preparation [Li, D. et al. Extended-resolution structured illumination imaging of endocytic and cytoskeletal dynamics. Science 349, (2015).].

[0003] SIM realizes the expansion of the lateral resolution by using an interference pattern with a frequency close to the cut-off frequency of the optical transfer function (OTF) to shift the high-frequency information lost by the traditional fluorescence microscope to the detectable low-frequency region [Gustafsson, M. G. Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy. J. Microsc. 198, 82–87 (200

[0004] 0). However, to achieve high-quality SIM super-resolution reconstruction, rapid and extremely accurate estimation of illumination parameters (wave vector, initial phase, and modulation degree) is required. Currently, the more widely used sub-pixel parameter estimation algorithm is the iterative cross-correlation method (COR). This algorithm continuously calculates the cross-correlation value between the zero-order spectrum and the first-order spectrum through iteration and takes the parameter when the cross-correlation value is the largest as the estimation result [Gustafsson, M.G.L. et al. Three-Dimensional Resolution Doubling in Wide-Field Fluorescence Microscopy by Structured Illumination. Biophys. J. 94, 4957–4970 (2008).]. However, due to its inevitable iteration and the complexity of parameter adjustment in the code, the COR estimation process takes a lot of time, increasing phototoxicity and photobleaching, and is not suitable for observing living cells. To solve the above problems, the parameter estimation algorithm based on principal component analysis (PCA-SIM) efficiently performs sub-pixel parameter estimation through principal component analysis and achieves real-time, high-quality dynamic super-resolution imaging of the fine structure of living cells [Qian, J., Cao, Y., Bi, Y. et al. Structured illumination microscopy based on principal component analysis. eLight 3, 4 (2023).]. Although PCA-SIM can solve the parameter estimation problem in most experimental environments, its extreme dependence on the information around the peak point of the first-order spectrum greatly limits its resolution and flexibility, posing a severe challenge to achieving artifact-free dynamic super-resolution imaging of living cells in complex environments. Summary of the Invention

[0005] The object of the present invention is to provide a structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis, which is used to achieve high-fidelity, high-robustness, and efficient real-time observation of living cells in a complex experimental environment where the peak point of the first-order spectrum and the information around it are missing.

[0006] The technical solution for achieving the object of the present invention is as follows: A structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis, the steps are as follows:

[0007] Step 1: Obtain the original illumination images of three-step phase shift in three directions;

[0008] Step 2: Preliminarily frequency-shift the original illumination images;

[0009] Step 3: Adopt a heterospectrum autocorrelation strategy to extract the phase matrix;

[0010] Step 4: Obtain the illumination vector factor through the Tukey - Hamming window;

[0011] Step 5: Extract the principal components of the illumination vector factor;

[0012] Step 6: Introduce M - estimation to accurately estimate the illumination parameters from the extracted principal components;

[0013] Step 7: Accurately separate the spectrum and use Wiener deconvolution to reconstruct the super - resolution image.

[0014] Compared with the prior art, the remarkable advantages of the present invention are as follows: For the first time, the present invention realizes accurate estimation of illumination parameters and real - time, artifact - free super - resolution imaging of samples in complex and unstable experimental environments such as using high - frequency structured light and high - magnification objective lenses. It further improves the imaging resolution and flexibility, provides a real - time, efficient, highly robust and high - quality observation method for studying cell activities at the nanoscale, is expected to be used for rapid and real - time super - resolution reconstruction of living cells, helps to reveal the essence of biological processes, promotes the research and development of new drugs and the study of disease mechanisms, and plays an important role in promoting related fields such as life science, medical research and nanotechnology.

[0015] The present invention will be further described in detail below with reference to the accompanying drawings. Description of the Drawings

[0016] Figure 1 is a flow chart of the present invention.

[0017] Figure 2 is the result of super - resolution reconstruction of a fixed BPAE cell sample using the present invention and traditional methods.

[0018] Figure 3 is the result of real - time super - resolution reconstruction of COS - 7 cell mitochondria at different time points using the present invention. Detailed Embodiments

[0019] A structured - light illumination super - resolution microscopy imaging method based on the combination of cross - correlation and principal component analysis obtains the illumination phasor matrix through the heterospectral autocorrelation method, solving the problem of inaccurate parameter estimation that may be caused by the lack of first - order spectrum information. In addition, to solve the problem of time - consuming calculation of principal component analysis, the Tukey - Hamming window is introduced to perform principal component analysis on the filtered phasor matrix, greatly reducing the amount of calculation while filtering out a large amount of noise. To further accurately estimate the sub - pixel wave vector, M - estimation is used for linear fitting to reduce the influence of outliers on the fitting effect, significantly improving the super - resolution reconstruction performance. The flow chart of the inventive method is as Figure 1 shown, and the specific steps are as follows:

[0020] Step 1: Use a structured illumination microscopy (SIM) system to illuminate the sample from three different directions and acquire three-step phase-shifted sinusoidal illumination images for each direction. The three-step phase-shifted sinusoidal illumination images of the nth phase pattern (n = 1, 2, 3) in the same direction are represented as:

[0021]

[0022] where D is the original wide-field image acquired by the camera, r is the spatial coordinate of the image, S is the fluorescence intensity of the sample, is the convolution operation, P is the point spread function of the system, k ex , m, and are the wave vector, modulation depth, and initial phase of the modulated light, respectively.

[0023] Step 2: Preliminarily frequency-shift the original illumination image from the illumination images in Step 1. The specific steps are as follows:

[0024] Step 2.1: Perform a Fourier transform on the original structured light illumination image obtained in Step 1, which can be represented in the frequency domain as:

[0025]

[0026] where the superscript ~ represents the Fourier transform of the original object, n is the nth phase pattern in the same direction, k is the spatial frequency coordinate of the image, O is the optical transfer function of the system, the subscripts 0 and ±1 are the orders of the spectrum, k ex , m, and are the wave vector, modulation depth, and initial phase of the modulated light, respectively, is the ±1 order spectrum information modulated by the structured light;

[0027] Step 2.2: Obtain the integer-pixel wave vector k int through calibration;

[0028] Step 2.3: Using the integer-pixel wave vector k int obtained in Step 2.2, the spectrum information of the original illumination image can be represented as after preliminary frequency-shifting:

[0029]

[0030] where the subscript shift is used to distinguish the spectrum information after frequency-shifting, and k sub is the sub-pixel wave vector.

[0031] Step 3: Adopt a heterospectrum autocorrelation strategy to extract the phase matrix, specifically:

[0032] Step 3.1: Perform a linear calculation on the spectrum image obtained in Step 2 to preliminarily separate the 0th order spectrum of the sample Can be expressed as:

[0033]

[0034] where O is the optical transfer function of the system, k is the spatial frequency coordinate of the image, is the 0th order spectrum of the sample. The 1st order spectrum after integer pixel frequency shift Can be expressed as:

[0035]

[0036] where the superscript' is used to distinguish the spectrum after preliminary separation, m and are the modulation depth and initial phase of the modulated light respectively, k sub is the sub-pixel wave vector, k int is the integer pixel wave vector, is the 1st order spectrum of the sample after integer pixel frequency shift;

[0037] Step 3.2: Obtain the heterospectrum autocorrelation expression from the 0th order spectrum of the sample and the 1st order spectrum after integer pixel frequency shift obtained in Step 3.1, which can be expressed as:

[0038]

[0039] where, is the inverse Fourier transform operation, S 0 (r) and S 1 (r) are the spatial domain images of the 0th order spectrum and the 1st order spectrum of the sample respectively;

[0040] Step 3.3: Extract the phase matrix through the heterospectrum autocorrelation expression in Step 3.3, specifically:

[0041]

[0042] where exp is the exponential function with base e, and angle is the function that returns the phase. Step 3 adopts the heterospectrum autocorrelation strategy to construct an illumination phasor matrix applicable to any high-frequency fringe illumination image scene, thus avoiding the problem that the illumination parameters cannot be correctly estimated due to the lack of information around the peak point.

[0043] Step 4: Use the Tukey - Hamming window to extract the phase matrix obtained in Step 3 to obtain the illumination vector factor, specifically:

[0044]

[0045] where k is the spatial frequency coordinate of the image, NaN is the invalid point, k x,min 、k y,minare respectively the left or lower boundary of the Tukey window in the Tukey-blank window in the horizontal x or vertical y direction, k x,max and k y,max are respectively the right or upper boundary of the Tukey window in the Tukey-blank window in the horizontal x or vertical y direction, R is the size of the blank window in the Tukey-blank window in the horizontal and vertical directions, k x and k y are respectively the frequency coordinates in the horizontal x or vertical y direction, k sub and are respectively the sub-pixel wave vector and the initial phase of the modulated light, and w(k) is specifically expressed as:

[0046]

[0047] where r is the ratio of the cosine segmentation length to the entire window length (usually determined by experience, and 0 ≤ r ≤ 1). Among them, the cosine lobe is responsible for adjusting the transition region of the window function, so that the window function smoothly transitions at its edges. Step 4 uses the double-window masking operator of the Tukey-blank window to filter the inverse Fourier transform of the phasor matrix (which is essentially a two-dimensional Dirichlet kernel function with energy concentrated in the middle region), significantly reducing the computational amount while greatly improving the accuracy of parameter estimation.

[0048] Step 5: Extract the principal components of the illumination vector factor obtained in Step 4, specifically:

[0049] Step 5.1: Perform singular value decomposition on the illumination vector factor obtained in Step 4. The decomposed illumination vector factor can be expressed as:

[0050]

[0051] where is the inverse Fourier transform operation, C is the phase matrix extracted by the Tukey-blank window, U and V are respectively the left and right singular matrices of the illumination vector factor, Λ is the eigenvalue matrix of the illumination vector factor, and the superscript T is the transpose of the matrix;

[0052] Step 5.2: Extract the principal components of the illumination vector factor. The extracted principal components can be expressed as:

[0053]

[0054] where represents a matrix with the element in the first row and first column being 1 and other elements being 0. Among them, k x,min and k y,min are respectively the left or lower boundary of the Tukey window in the Tukey-blank window in the horizontal x or vertical y direction, k x,max and k y,maxare respectively the right boundary or the upper boundary of the Tukey window in the Tukey-blank window in the horizontal x or vertical y direction, R is the size of the blank window in the Tukey-blank window in the horizontal and vertical directions, k sub and are respectively the sub-pixel wave vector and the initial phase of the modulated light. In step 5, using the data reduction characteristic of the principal component analysis technique, the dimension of the vector matrix is compressed to two dimensions to extract the single principal component of the matrix, thereby filtering out high-frequency noise.

[0055] Step 6: Introduce M-estimation to accurately estimate the illumination parameters from the principal component extracted in step 5, specifically:

[0056] Step 6.1: Unwrap the phase of the first row elements of the left singular matrix U, and introduce M-estimation for linear fitting. The slope of the fitted line is the horizontal component of the sub-pixel wave vector k sub in the horizontal direction; unwrap the phase of the first column elements of the right singular matrix V, and introduce M-estimation for linear fitting. The slope of the fitted line is the vertical component of the sub-pixel wave vector k sub in the vertical direction;

[0057] Step 6.2: Take the phase value of the illumination vector factor principal component U(Λ·M 1 )V T at the spatial coordinate r = 0 of the image as the initial phase

[0058] Step 6.3: Perform sub-pixel level frequency shift on the first-order spectrum of the sample using the sub-pixel wave vector k sub obtained in step 6.1 and the integer pixel wave vector k int obtained in claim 2, and then perform deconvolution operations on the zero-order spectrum and the accurately frequency-shifted first-order spectrum respectively to obtain the modulation degree m, specifically:

[0059]

[0060] where the superscript * represents the conjugate of the original object, | | is the amplitude operation, S 0 (r) and S 1 (r) are respectively the spatial domain images of the zero-order spectrum and the first-order spectrum of the sample. Step 6 adopts the M-estimation method and fits the line by iteratively using the weighted least squares method. In each iteration, the weights of the inliers and outliers are adjusted according to the magnitude of the distance function until the value of the cost function (i represents the i-th fitting point, n represents the total number of fitting points) reaches the minimum. Among them, the selected distance function ρ(r i ) is:

[0061] ρ(r i ) = r i (13)

[0062] where r i represents the distance from the i-th fitting point to the fitting line. By using the weighted least squares method based on M-estimation and selecting an appropriate distance function, the influence of outliers on the fitting accuracy is greatly reduced, and the anti-interference ability and estimation accuracy of Cr-PCA are enhanced.

[0063] Step 7: The specific steps for accurately separating the spectrum and reconstructing the super-resolution image using Wiener deconvolution are as follows:

[0064] Step 7.1: Perform accurate spectrum separation based on the illumination parameters obtained in Step 6, specifically:

[0065]

[0066] where k is the spatial frequency coordinate, k ex , m, and are the wave vector, modulation depth, and initial phase of the modulated light, respectively, and k ex = k int + k sub . is the 0th-order spectrum information of the sample, is the ±1st-order spectrum information modulated by the structured light, and are the original image spectra of the 1st, 2nd, and 3rd phases collected by the structured light in the same direction, respectively;

[0067] Step 7.2: Separate and recombine the accurately separated spectra in Step 7.1, perform the same operations for the other two illumination directions, and then merge the obtained spectra through Wiener deconvolution to reconstruct the real-time super-resolution image, specifically:

[0068]

[0069] where is the inverse Fourier transform operation, the subscript d represents the d-th illumination direction, the subscript i represents the spectrum order, taking values (0, +1, -1), O is the optical transfer function of the system, k d,ex is the wave vector of the d-th illumination direction, is the i-th spectrum component of the d-th illumination direction without the optical transfer function and illumination parameter components, the superscript * represents the conjugate of the original object, || and Σ are the magnitude-taking and summation operations, respectively, and w is the Wiener constant (usually determined by experience).

[0070] Embodiment

[0071] To test the feasibility and real-time performance of the present invention, first, the super-resolution reconstruction of a fixed BPAE cell sample is performed using the method (Cr-PCA) described in the present invention. AsFigure 2 as shown, where the cell nucleus, α-actin, and mitochondria are labeled by DAPI, FITC, and MitoTracker TM Green FM, respectively. a Wide-field image and the reconstruction result of the present invention. b-c Comparison of the super-resolution reconstruction results of ROI 1 and ROI 2 in a by different methods. d Spectrum of the result obtained by super-resolution reconstruction using the present invention. e Time spent on parameter estimation of the wide-field image by different methods. f OTF of the simulated objective lens with 100X 1.45NA and the stripe structured light period of 5, and the first-order spectra in three directions with unclear peak points obtained after spectral separation of the collected wide-field image. g Intensity profiles along the gray dotted lines in b and c. This experiment was independently repeated 30 times.

[0072] As Figure 2 shown in f, since the modulation frequency of the stripe structured light is greater than the cut-off frequency of the OTF, the peak points of the first-order spectrum of the collected wide-field image are blurred. As can be seen from Figure 2 b-c, the mitochondrial image reconstructed by the parameter estimation algorithm based on principal component analysis (PCA-SIM) is blurred and has many artifacts, while the method described in the present invention and COR can still effectively achieve super-resolution reconstruction.

[0073] In addition, the time-consuming of parameter estimation for each algorithm was tested. Although the reconstruction effect of COR is similar to that of the method described in the present invention, its time-consuming is about 12 times that of the method described in the present invention (as Figure 3 shown in e). Therefore, compared with the other two algorithms, the method described in the present invention is more promising for real-time super-resolution observation of living cells in complex experimental environments such as high-frequency structured light and high-magnification objective lenses. This embodiment proves that the present invention has high-precision, high-efficiency, and high-robustness super-resolution reconstruction performance, and demonstrates its potential for long-term live cell imaging applications under complex experimental conditions.

[0074] An important application of the present invention is to monitor the dynamic changes of subcellular structures in real time in complex experimental environments. To verify this, the present invention was used to perform real-time observation on the mitochondrial dynamic tubule events (MDT) of living COS-7 cells. As Figure 3 shown, where the mitochondria are labeled by MitoTracker TM Green FM. a, d Wide-field images and the super-resolution images obtained by the present invention. The original images were acquired through a 60X 1.42NA objective lens. b, c Super-resolution images of the regions within the white and green boxes in a at different time points. e, f Super-resolution images of the regions within the blue and yellow boxes in d at different time points. As Figure 3a - 3b, the mitochondria on the left side in the ellipse gradually stretch towards the right and finally fuse with the mitochondria on the right side, forming a membrane bridge, which promotes the energy metabolism in mitochondria and the transmission of intracellular signals. These experimental phenomena are of great significance for studying the behavior and interaction of mitochondria in cells. This embodiment fully demonstrates the advantages of the present invention in studying the structure of living cells and their dynamic processes.

Claims

1. A structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis, characterized in that: The specific process is: Step 1: Acquire the original illumination image with three-step phase shifts in three directions; Step 2: Preliminary frequency shift of the original illumination image; Step 3: Extract the phase matrix using the hetero-spectral autocorrelation strategy; Step 4: Obtain the lighting vector factor through the Tuji-blank window; Step 5: Extract the principal components of the lighting vector factors; Step 6: Introduce M estimation to accurately estimate the lighting parameters from the extracted principal components; Step 7: Accurately separate the spectrum and reconstruct the super-resolution image using Wiener deconvolution.

2. The structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis according to claim 1, characterized in that: The specific steps of obtaining the three-step phase-shifted original illumination images in three directions in step 1 are as follows: Using a structured light illumination microscopy system, the sample was illuminated from three different directions, and three-step phase-shifted sinusoidal illumination images were collected in each direction.

3. The structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis according to claim 2, characterized in that: The three-step phase-shifted sinusoidal illumination image of the nth phase mode (n=1, 2, 3) in the same direction is specifically: Where D is the original wide-field image captured by the camera, r is the spatial coordinate of the image, S is the fluorescence intensity of the sample, is the convolution operation, P is the point spread function of the system, k ex , m and are the wave vector, modulation depth and initial phase of the modulated light respectively.

4. The structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis according to claim 1, characterized in that: The specific steps of the preliminary frequency shift of the original illumination image in step 2 are: Step 2.1: Perform Fourier transform on the original structured light illumination image obtained in step 1. After the transform, it can be expressed in the frequency domain as: Where n is the nth phase mode, the superscript ~ is the Fourier transform of the original object, k is the spatial frequency coordinate of the image, O is the optical transfer function of the system, and the subscripts 0 and ±1 are the order of the spectrum, respectively. ex , m and are the wave vector, modulation depth and initial phase of the modulated light respectively; Step 2.2: Obtain the integer pixel wave vector k by calibration int ; Step 2.3: Use the integer pixel wave vector k obtained in step 2.2 int , after preliminary frequency shifting of the original illumination image spectrum information, it can be expressed as: The subscript shift is used to distinguish the frequency spectrum information after frequency shift, k sub is the sub-pixel wave vector.

5. The structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis according to claim 1, characterized in that: Step 3 uses the hetero-spectral autocorrelation strategy to extract the phase matrix, specifically: Step 3.1: Perform linear calculation on the spectrum image obtained in claim 2 to preliminarily separate the 0th order spectrum of the sample. It can be expressed as: Where O is the optical transfer function of the system, k is the spatial frequency coordinate of the image, is the 0-level spectrum information of the sample. The 1-level spectrum after the whole pixel frequency shift It can be expressed as: The superscript ' is used to distinguish the spectrum after preliminary separation, the subscript shift is used to distinguish the spectrum information after frequency shift, and m and are the modulation depth and initial phase of the modulated light, k sub is the sub-pixel wave vector, k int is the integer pixel wave vector, is the first-level spectrum of the sample after the whole pixel frequency shift; Step 3.2: The hetero-spectral autocorrelation expression is obtained from the sample 0-level spectrum obtained in step 3.1 and the 1-level spectrum after the integer pixel frequency shift, which can be expressed as: in, is the inverse Fourier transform operation, S0(r) and S1(r) are the spatial domain images of the 0th-order spectrum and the 1st-order spectrum of the sample, respectively; Step 3.3: Extract the phase matrix through the heterospectral autocorrelation expression in step 3.3, specifically: Where exp is the exponential function with base e, and angle is the function that returns the phase.

6. The structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis according to claim 1, characterized in that: Step 4 uses the Tukey-blank window to extract the phase matrix to obtain the illumination vector factor, specifically: Among them, k is the spatial frequency coordinate of the image, NaN is an invalid point, k x,min , k y,min are the left or lower boundary of the base window in the base-blank window in the horizontal x or vertical y direction, respectively, x,max , k y,max are the right boundary or upper boundary of the Tukey-blank window in the horizontal x or vertical y direction, respectively, R is the size of the blank window in the Tukey-blank window in the horizontal and vertical directions, k x , k y are the frequency coordinates along the horizontal x or vertical y direction, k sub and are the sub-pixel wave vector and initial phase of the modulated light respectively, and w(k) is specifically expressed as: Where r is the ratio of the cosine segment length to the entire window length (usually determined empirically, with 0≤r≤1).

7. The structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis according to claim 1, characterized in that: The specific method of extracting the principal components of the lighting vector factor in step 5 is: Step 5.1: Perform singular value decomposition on the lighting vector factor, which can be expressed as: in, is the inverse Fourier transform operation, C is the phase matrix extracted by the Tukey-blank window, U and V are the left and right singular matrices of the illumination vector factor, Λ is the characteristic matrix of the illumination vector factor, and the superscript T is the transpose of the matrix; Step 5.2: Extract the principal components of the lighting vector factor, which is specifically expressed as: in, Represents a matrix in which the first row and first column are all 1 and the other elements are all 0, where k x,min , k y,min are the left or lower boundary of the base window in the base-blank window in the horizontal x or vertical y direction, respectively, x,max , k y,max are the right boundary or upper boundary of the Tukey-blank window in the horizontal x or vertical y direction, respectively, R is the size of the blank window in the Tukey-blank window in the horizontal and vertical directions, k sub and are the sub-pixel wave vector and initial phase of the modulated light, respectively.

8. The structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis according to claim 7, characterized in that: Step 6 introduces M estimation. The specific steps for accurately estimating the lighting parameters from the extracted principal components are as follows: Step 6.1: Perform phase unwrapping on the first row of the left singular matrix U and introduce M estimation for linear fitting. The slope of the fitted line is the sub-pixel wave vector k sub Component in the horizontal direction; perform phase unwrapping on the first column elements of the right singular matrix V, and introduce M estimation for linear fitting. The slope of the fitted line is the sub-pixel wave vector k sub Component in the vertical direction; Step 6.2: Take the phase value of the principal component of the illumination vector factor at r = 0 as the initial phase Step 6.3: Use sub-pixel wave vector k sub and the integer pixel wave vector k int The sample level 1 spectrum is subjected to sub-pixel frequency shift, and then the level 0 spectrum and the level 1 spectrum after precise frequency shift are deconvolved to obtain the modulation degree m, specifically: Among them, the superscript * represents the conjugate of the original object, || is the amplitude operation, is the initial phase, S0(r) and S1(r) are the spatial domain images of the 0th and 1st order spectra of the sample respectively, and r is the spatial coordinate of the image.

9. The structured light illumination super-resolution microscopy imaging method based on the combination of cross-correlation and principal component analysis according to claim 8, characterized in that: Step 7: The specific steps of accurately separating the spectrum and reconstructing the super-resolution image using Wiener deconvolution are as follows: Step 7.1: Perform accurate spectrum separation using the lighting parameters obtained in step 6, specifically: Among them, k is the spatial frequency coordinate, k ex , m and are the wave vector, modulation depth and initial phase of the modulated light, and k ex =k int +k sub , is the 0-level spectrum information of the sample, is the ±1-level spectrum information after structured light modulation, and They are the original image spectra of the 1st, 2nd and 3rd phases collected by structured light in the same direction; Step 7.2: Separate and reassemble the spectrum accurately separated in step 7.1, and perform the same operation on the other two lighting directions, and then merge the acquired spectrum through Wiener deconvolution to reconstruct a real-time super-resolution image, specifically: in, is the inverse Fourier transform operation, subscript d is the d-th illumination direction, subscript i is the spectral order, which takes values ​​of (0, +1, -1), O is the optical transfer function of the system, and k d,ex is the wave vector of the d-th illumination direction, is the i-th spectral component of the d-th illumination direction without optical transfer function and illumination parameter components, the superscript * represents the conjugate of the original object, || and ∑ represent the amplitude and summation operations respectively, and w is the Wiener constant.

Images