Laminated microscopic imaging reconstruction method based on dynamic laminated iterative optimization

Abstract

The invention provides a laminated microscopic imaging reconstruction method based on dynamic laminated iterative optimization, which only introduces one tuning hyper-parameter of iterative reconstruction times, adaptively regulates two important factors of duration and direction in the reconstruction process, and reduces the dimensionality of the tuning hyper-parameter to 1; meanwhile, related detailed adjustment and optimization rules are given, so that the adjustment and optimization rules can be conveniently matched with factors such as actual samples and actual application environments, and high-quality reconstruction of the biological samples is realized; and finally, an optimization path adapted to the characteristics of each stage of laminated microscopic imaging reconstruction is planned, the path fuses respective advantages of the previous PIE design algorithm, the problem that the optimization speed and the stability are mutually balanced in the previous algorithm is solved, and rapid and stable reconstruction of the biological sample is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of ptychographic microscopy, and particularly relates to a ptychographic microscopy reconstruction method based on dynamic ptychographic iterative optimization. Background Art

[0002] Ptychographic microscopy is an important means to achieve quantitative phase imaging of biological samples, and is widely used in many fields such as biomedical imaging and cell mechanism research. According to the dimensional difference in implementing the ptychography technique, ptychographic microscopy can be divided into traditional ptychographic microscopy in the spatial domain and Fourier ptychographic microscopy in the frequency domain. Although the hardware composition and dimensional space for implementing ptychographic microscopy are different, ptychographic microscopy techniques all need to rely on the designed Ptychographical Iterative Engine (PIE) algorithm to quantitatively recover the complex amplitude information or spectral information of the sample from a series of collected intensity images. The update and proposal of each generation of PIE have significantly improved the overall performance of the ptychographic microscopy system.

[0003] PIE is a carefully designed algorithm framework. When the sample acquisition meets the overlap rate condition, PIE can quantitatively recover the phase information of the sample from the collected intensity information. The initial generation of PIE realizes the iterative convergence of sample information by minimizing the loss function with the sample intensity information and deriving the update formula of the sample function by the simple gradient descent method. However, the initial generation of PIE is easily interfered by factors such as noise and system errors, resulting in problems such as a decline in imaging quality. Moreover, the initial generation of PIE requires many ideal assumptions as prerequisites to be realized. For example: ignoring the thickness of the sample, accurately describing the position of the scanning probe, accurately describing the probe or pupil function, etc.

[0004] To improve the practical performance of PIE and expand its application scope, research scholars have successively proposed the ePIE (the extended PIE) and rPIE (the regularized PIE) algorithms. ePIE synchronously updates the probe function or pupil function and the sample function, liberating the precise description conditions of the probe or pupil function. Compared with the original PIE, ePIE has stable reconstruction performance and certain robustness to factors such as noise and system errors. However, due to the selection of the minimum step size, ePIE has a slow convergence speed and is easily trapped in local optima. rPIE is based on the concepts of ePIE and the original PIE, introducing two hyperparameters to be tuned to generalize the optimization direction of PIE, improving the optimization speed and quality of PIE. However, the optimization speed and quality of rPIE come at the cost of algorithm stability. When in environments with different noise and error levels, different parameter combinations of rPIE not only cause changes in the optimal quality and optimal timing of reconstruction, but also result in different timings of image degradation. Therefore, in most cases, researchers or users tend to use ePIE rather than rPIE to stably find a suboptimal solution to complete image reconstruction, thus avoiding the hyperparameter tuning and image degradation problems brought by rPIE.

[0005] To further improve the comprehensive performance of PIE, mPIE (momentum-accelerated PIE), which is designed based on rPIE and combines the advanced momentum acceleration concept, is proposed. Compared with previous PIE algorithms, mPIE has achieved multi-dimensional improvements in optimization ability, optimization speed, and optimization quality. However, to balance the optimization speed and stability, mPIE needs to tune seven additional hyperparameters according to the actual reconstruction effect. Moreover, similar to rPIE, the optimal hyperparameter combination of mPIE is not a fixed value and changes with the reconstruction sample and environment. Researchers can only adjust the hyperparameters according to the actual reconstruction effect within the tuning range recommended by the algorithm reference for different samples and usage environments. It can be said that the two problems of multi-parameter tuning and balancing the optimization speed and stability limit the extensive research and use of the current advanced PIE in actual bio-optical microscopy imaging tasks. Summary of the Invention

[0006] To solve the above problems, the present invention provides a ptychographic microscopy reconstruction method based on dynamic ptychographic iterative optimization, which fully combines the advantageous features of previous reconstruction algorithms, solves the problem of mutual restraint between optimization speed and stability existing in the current algorithms with a dynamic design idea, and only requires a single hyperparameter to achieve fast and high-quality reconstruction of microscopic samples.

[0007] A method for reconstructing laminar microscopy imaging based on dynamic laminar iterative optimization, which is applied to a laminar microscopy imaging system, and the laminar microscopy imaging system includes overlapping sub-apertures arranged in M rows and N columns. The method is as follows:

[0008] Perform sub-spectrum reconstruction operations on each sub-aperture in sequence. And each time a sub-spectrum reconstruction operation is performed, a sample spectrum is generated correspondingly until all sub-apertures are traversed to complete one iteration of reconstruction. Among them, except that the input variable for the first sub-spectrum reconstruction operation is a preset value, when any sub-aperture performs any number of sub-spectrum reconstruction operations, the input variable is the spectrum estimation value corresponding to this sub-aperture after the last sub-spectrum reconstruction operation in the previous iteration of reconstruction. At the same time, after any sub-aperture performs a sub-spectrum reconstruction operation, the spectrum estimation values corresponding to this sub-aperture and other sub-apertures overlapping with this sub-aperture will all change;

[0009] After each iteration of reconstruction, judge whether the error value between the sample spectrum obtained after the last sub-spectrum reconstruction operation in this iteration of reconstruction and the sample spectrum finally obtained in the previous iteration of reconstruction meets the set convergence condition or whether the number of iterations of reconstruction reaches the preset maximum number. If so, the sample spectrum formed by the spectrum estimation values corresponding to each sub-aperture after the last sub-spectrum reconstruction operation in this iteration of reconstruction is used as the final reconstruction result. If both are not, the spectrum estimation values corresponding to each sub-aperture after the last sub-spectrum reconstruction operation in this iteration of reconstruction are used as new input variables to enter the next iteration of reconstruction.

[0010] Furthermore, after any sub-aperture performs a sub-spectrum reconstruction operation in any iteration of reconstruction, the calculation method of the sample spectrum formed by the spectrum estimation values corresponding to each sub-aperture is as follows:

[0011]

[0012] where k represents the wave vector in the frequency domain, and * represents the conjugate operator, is the sample spectrum after the sub-spectrum reconstruction operation of the sub-aperture in the m-th row and n-th column in the i-th iteration of reconstruction, is the sample spectrum before the sub-spectrum reconstruction operation of the sub-aperture in the m-th row and n-th column in the i-th iteration of reconstruction, is the pupil function of the laminar microscopy imaging system before the sub-spectrum reconstruction operation of the sub-aperture in the m-th row and n-th column in the i-th iteration of reconstruction shifted to the left by k m,n units of the pupil function, k m,n is the illumination wave vector used when the laminar microscopy imaging system acquires the intensity image of the sub-aperture in the m-th row and n-th column, f(P (m,n)-1 (k + k m,n )) is the core term related to the pupil function before the sub-spectrum reconstruction operation of the sub-aperture in the m-th row and n-th column, The spectral estimation value before performing the sub-spectrum reconstruction operation for the sub-aperture at the m-th row and n-th column in the i-th iterative reconstruction, The spectral estimation value after performing the sub-spectrum reconstruction operation for the sub-aperture at the m-th row and n-th column in the i-th iterative reconstruction.

[0013] Furthermore, the core term related to the pupil function before the sub-aperture at the m-th row and n-th column performs the sub-spectrum reconstruction operation is calculated as follows:

[0014]

[0015] where and are both structure functions that vary with the number of iterative reconstructions, |·| represents the modulus operation, and max represents obtaining the maximum value of the squared modulus of the pupil function in the form of a two-dimensional complex matrix after the modulus operation, and and are calculated as follows:

[0016]

[0017] where e represents the base of the natural logarithm, and α i represents the direction adjustment parameter related to the iterative reconstruction number i, and there is:

[0018]

[0019] where i Emax is the preset tuning parameter.

[0020] Furthermore, the greater the imaging error of the ptychographic microscopy system, the smaller the value of i Emax will be.

[0021] Furthermore, the spatial complex amplitude is calculated as follows:

[0022]

[0023] where r represents the spatial coordinate in the spatial domain, F -1 represents the inverse Fourier transform, is the sample spectrum shifted k m,n units to the right to obtain the sample spectrum, is the pupil function of the ptychographic microscopy system before the sub-aperture at the m-th row and n-th column in the i-th iterative reconstruction performs the sub-spectrum reconstruction operation.

[0024] Furthermore, the spectral estimation value is calculated as follows:

[0025]

[0026] Among them, r represents the spatial domain spatial coordinate, F represents the Fourier transform, and I m,n (r) represents the intensity image acquired by the sub-aperture in the m-th row and n-th column, is the spatial domain complex amplitude of the spectral estimate before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction, represents replacing the spatial domain complex amplitude with the amplitude in the intensity image to obtain the spatial domain complex amplitude.

[0027] Furthermore, the pupil function of the ptychographic microscopy system after the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction is calculated as follows:

[0028]

[0029] Among them, is the pupil function of the ptychographic microscopy system before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction, is the sample spectrum obtained by shifting the sample spectrum, which is the sample spectrum before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction, k m,n units to the right, is the core term related to the sample spectrum before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column.

[0030] Furthermore, the core term related to the sample spectrum before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column

[0031]

[0032] Among them, and are both structure functions that change with the number of iterative reconstructions. |·| represents the modulus operation, and max represents obtaining the maximum value of the squared modulus of the sample spectrum in the form of a two-dimensional complex matrix after the modulus operation, and and are calculated as follows:

[0033]

[0034] Among them, e represents the base of the natural logarithm, and α i represents the direction adjustment parameter related to the iterative reconstruction number i, and there is:

[0035]

[0036] wherein, i Emax is a preset tuning parameter.

[0037] Furthermore, the greater the imaging error of the stacked microscopic imaging system, the smaller the value of i Emax will be.

[0038] Furthermore, the stacked microscopic imaging system is a spatial-domain stacked microscopic imaging system or a Fourier ptychographic microscopy system.

[0039] Beneficial effects:

[0040] The present invention provides a reconstruction method for stacked microscopic imaging based on dynamic stacked iterative optimization, which only introduces one tuning hyperparameter, i.e., the number of iterative reconstructions, adaptively regulates two important factors, namely duration and direction, during the reconstruction process, and reduces the dimension of the tuning hyperparameter to 1. At the same time, the present invention gives relevant detailed tuning rules to make it convenient to adapt to factors such as actual samples and actual application environments, so as to achieve high-quality reconstruction of biological samples. Finally, the present invention plans an optimization path adapted to the characteristics of each stage of stacked microscopic imaging reconstruction. This path integrates the respective advantages of previous PIE design algorithms, solves the problem of mutual trade-off between optimization speed and stability in previous algorithms, and realizes fast and stable reconstruction of biological samples. Therefore, compared with the existing stacked imaging reconstruction algorithms, the design method of the present invention not only has the characteristics of fewer tuning hyperparameters and strong algorithm robustness, but also has many advantages in terms of the quality of reconstructed images, the convergence speed of the algorithm, the optimization speed of the algorithm, and the stability of the algorithm. Description of the Drawings

[0041] Figure 1 is a flowchart of the reconstruction method for stacked microscopic imaging based on dynamic stacked iterative optimization provided by the present invention. Detailed Embodiments

[0042] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application.

[0043] In Fourier ptychography and traditional ptychographic microscopy, PIE is used to reconstruct the sample. There is only a spatial conversion between the frequency domain and the spatial domain, and they have the same execution concept. Based on this, a ptychographic microscopy reconstruction method based on dynamic ptychographic iterative optimization is applied to a ptychographic microscopy system. The ptychographic microscopy system is an in-space ptychographic microscopy system or a Fourier ptychographic microscopy system, and the ptychographic microscopy system includes overlapping sub-apertures arranged in M rows and N columns. It should be noted that during the ptychographic microscopy acquisition stage, the ptychographic imaging system has acquired M×N intensity images of the biological sample. Then, during the reconstruction stage, the dynamic ptychographic iterative engine will use these intensity images to restore the high-resolution and wide-field-of-view phase information of the sample.

[0044] As Figure 1 shown, taking the reconstruction of Fourier ptychographic microscopy in the frequency domain as an example, the ptychographic microscopy reconstruction method provided by the present invention will be described in detail as follows:

[0045] Perform sub-spectrum reconstruction operations on each sub-aperture in sequence. And each time a sub-spectrum reconstruction operation is performed, a sample spectrum is generated correspondingly until all sub-apertures are traversed to complete one iterative reconstruction. Among them, except that the input variable for the first sub-spectrum reconstruction operation is a preset value, when any sub-aperture performs any sub-spectrum reconstruction operation, the input variable is the spectrum estimate corresponding to this sub-aperture after the last sub-spectrum reconstruction operation in the previous iterative reconstruction. At the same time, after any sub-aperture performs a sub-spectrum reconstruction operation, the spectrum estimates corresponding to this sub-aperture and other sub-apertures overlapping with this sub-aperture will all change;

[0046] After each iterative reconstruction, judge whether the error value between the sample spectrum obtained after the last sub-spectrum reconstruction operation in this iterative reconstruction and the sample spectrum finally obtained in the previous iterative reconstruction meets the set convergence condition or whether the number of iterative reconstructions reaches the preset maximum number. If so, the sample spectrum formed by the spectrum estimates corresponding to each sub-aperture after the last sub-spectrum reconstruction operation in this iterative reconstruction is used as the final reconstruction result. If both are not, the spectrum estimates corresponding to each sub-aperture after the last sub-spectrum reconstruction operation in this iterative reconstruction are used as new input variables to enter the next iterative reconstruction.

[0047] Among them, after any sub-aperture performs a sub-spectrum reconstruction operation in any iterative reconstruction, the calculation method of the sample spectrum formed by the spectrum estimates corresponding to each sub-aperture is:

[0048]

[0049] where k represents the wave vector in the frequency domain, and * represents the conjugate operator, is the sample spectrum after the sub-spectrum reconstruction operation of the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction, The sample spectrum before performing the sub-spectrum reconstruction operation for the sub-aperture at the m-th row and n-th column in the i-th iterative reconstruction, The pupil function of the ptychographic microscopy system before performing the sub-spectrum reconstruction operation for the sub-aperture at the m-th row and n-th column in the i-th iterative reconstruction Shifted to the left by k m,n units of the pupil function, where k m,n is the illumination wave vector used when the ptychographic microscopy system acquires the intensity image of the sub-aperture at the m-th row and n-th column, f(P (m,n)-1 (k + k m,n )) is the core term related to the pupil function before performing the sub-spectrum reconstruction operation for the sub-aperture at the m-th row and n-th column, The spectrum estimate value before performing the sub-spectrum reconstruction operation for the sub-aperture at the m-th row and n-th column in the i-th iterative reconstruction, The spectrum estimate value after performing the sub-spectrum reconstruction operation for the sub-aperture at the m-th row and n-th column in the i-th iterative reconstruction.

[0050] Furthermore, the core term related to the pupil function before performing the sub-spectrum reconstruction operation for the sub-aperture at the m-th row and n-th column is calculated as follows:

[0051]

[0052] Where, and are both structure functions that vary with the number of iterative reconstructions, |·| is the modulus operation, max represents obtaining the maximum value after squaring the modulus of the pupil function in the form of a two-dimensional complex matrix and and are calculated as follows:

[0053]

[0054] Where, e represents the base of the natural logarithm, and α i represents the direction adjustment parameter related to the iterative reconstruction number i, and there is:

[0055]

[0056] Where, i Emax is the preset tuning parameter. The larger the imaging error of the ptychographic microscopy system, the smaller the value of.

[0057] Furthermore, the spatial domain complex amplitude is calculated as follows:

[0058]

[0059] Among them, r represents the spatial coordinate in the spatial domain, and F -1 represents the inverse Fourier transform, is the sample spectrum shifted k m,n units to the right to obtain the sample spectrum, is the pupil function of the ptychographic microscopy system before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction.

[0060] Furthermore, the calculation method of the spectrum estimation value is as follows:

[0061]

[0062] Among them, r represents the spatial coordinate in the spatial domain, F represents the Fourier transform, and I m,n (r) represents the intensity image obtained by the sub-aperture in the m-th row and n-th column, is the spatial complex amplitude of the spectrum estimation value before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction, represents the spatial complex amplitude obtained after replacing the amplitude in the with the intensity image.

[0063] Furthermore, the calculation method of the pupil function of the ptychographic microscopy system after the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction is as follows:

[0064]

[0065] Among them, is the pupil function of the ptychographic microscopy system before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction, and its initial value is generally set to a low-pass circular filtering function with a 0-1 distribution related to the numerical aperture of the system imaging lens, is the sample spectrum obtained by shifting the sample spectrum before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction to the right by k m,n units, is the core term related to the sample spectrum before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column.

[0066] Furthermore, the calculation method of the core term related to the sample spectrum before the sub-spectrum reconstruction operation is performed on the sub-aperture in the m-th row and n-th column is as follows:

[0067]

[0068] Among them, and They are all constructor functions that vary with the number of iterative reconstructions. |·| represents the modulo operation, and max represents obtaining the sample spectrum in the form of a two-dimensional complex matrix. The maximum value after squaring the modulo operation, and and The calculation methods are as follows:

[0069]

[0070] where e represents the base of the natural logarithm, and α i represents the direction adjustment parameter related to the iterative reconstruction times i, and there is:

[0071]

[0072] where i Emax is a preset tuning parameter. The greater the imaging error of the tomographic microscopy system, the smaller the value of i Emax will be.

[0073] It should be noted that the tomographic microscopy system is an in-space tomographic microscopy system or a Fourier ptychographic microscopy system; however, since the in-space tomographic imaging reconstruction process can be regarded as the in-space dual form of the Fourier ptychographic microscopy reconstruction, the in-space tomographic imaging reconstruction process of the present invention will not be elaborated herein.

[0074] The following takes the iterative update of a sub-spectrum with an arbitrary number (m,n) in Fourier ptychographic microscopy as an example to further illustrate the tomographic microscopy reconstruction method of the present invention, specifically as follows:

[0075] Step 1: Assume that the current iteration number is i (i≥1), and take the sample spectrum estimate O i -1 (k) generated in the (i - 1)-th iteration as the initial value of the current iteration number. When updating the sample sub-spectrum with the number (m,n), take out the spectral information of the sub-aperture at this position and convert it to the in-space complex amplitude

[0076]

[0077] Step 2: Apply the actual intensity measurement value limit to the current in-space complex amplitude to update the in-space complex amplitude information at this sub-spectrum position. Specifically, replace the amplitude part of the in-space complex amplitude with the intensity image I m,n (r) collected at this numbered position and retain its phase:

[0078]

[0079] Step 3: Convert the in-space complex amplitude to the frequency domain to obtain the updated spectral information of the sub-aperture at this position

[0080]

[0081] Step 4: Update the sample spectrum using the designed dynamic stacked iteration concept. First, introduce the iteration number i to plan the optimization path of PIE. Referring to the construction forms of ePIE and rPIE, the core term f(P (m,n)-1 (k + k m,n )) in the denominator position of the dynamic stacked iteration engine has the initial form of:

[0082]

[0083] Combined with the sigmoid smoothing function, and have the following construction forms:

[0084]

[0085] where α i represents the direction adjustment parameter related to the reconstruction iteration number i. In the present invention, introduce a hyperparameter i Emax to be tuned, and the constructed form of α i is as follows:

[0086]

[0087] Among all the above-mentioned hyperparameters, only i Emax needs to be tuned. The above form realizes the design of the dynamic stacked iteration engine under a single tunable hyperparameter. The recommended value of i Emax is 20 under the condition of no actual experimental system hardware error. When there is a large amount of noise or optical error in the imaging environment, i Emax can be appropriately reduced to make it have the stable characteristics of ePIE faster. When the error in the imaging environment is small, the value of i Emax can be appropriately increased to make it more biased towards the fast optimization characteristics of rPIE and have the stable performance of ePIE in the later stage. Finally, use the designed denominator core term to update the spectrum function of the sample and the pupil function of the system simultaneously:

[0088]

[0089] where * represents the conjugate operator, and respectively represent the sub-spectrum of the sample after and before updating the complex amplitude with the actual intensity measurement value.

[0090] Repeat steps 1-4 according to the predetermined sub-spectrum reconstruction order until all sub-spectrum position numbers (m,n) have been traversed once, which is regarded as completing one iteration reconstruction process, and the iteration count i is incremented by 1. When the reconstruction result reaches the convergence condition or the iteration count i reaches the preset maximum iteration count, the dynamic ptychographic iterative engine terminates the iteration and outputs the final reconstructed result image.

[0091] In summary, based on the ptychographic microscopy technology, the present invention addresses two problems existing in the current ptychographic reconstruction algorithm framework, namely, multi-parameter tuning and the mutual restraint between the optimization speed and stability of the algorithm. A ptychographic microscopy reconstruction method based on dynamic ptychographic iterative optimization is proposed.

[0092] The present invention only introduces one tuning hyperparameter, the number of iterative reconstructions, to achieve the adaptive regulation of two important factors, the optimization duration and the optimization direction, during the reconstruction process, reducing the dimension of the overall tuning hyperparameters to 1. The present invention also plans an optimization path adapted to the characteristics of each stage of ptychographic microscopy reconstruction, integrating the respective advantages of previous PIE design algorithms, and solving the problem of the mutual trade-off between the optimization speed and stability in previous algorithms. Compared with the existing ptychographic imaging reconstruction algorithms, the design method of the present invention has the characteristics of fewer tuning hyperparameters and strong algorithm robustness, and has many advantages in terms of the quality of the reconstructed image, the algorithm convergence speed, the algorithm optimization speed, and the algorithm stability.

[0093] Of course, the present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can certainly make various corresponding changes and deformations according to the present invention. However, these corresponding changes and deformations should all fall within the protection scope of the appended claims of the present invention.

Claims

1. A stacked microscopic imaging reconstruction method based on dynamic stacking iterative optimization, applied to a stacked microscopic imaging system, wherein the stacked microscopic imaging system comprises M rows and N columns of overlapping sub-apertures, characterized in that: The method is: The sub-spectrum reconstruction operation is performed on each sub-aperture in turn, and each time a sub-spectrum reconstruction operation is performed, a sample spectrum is generated correspondingly, until all sub-apertures are traversed and an iterative reconstruction is completed; wherein, except that the input variable of the first sub-spectrum reconstruction operation is a preset value, when any sub-aperture performs any number of sub-spectrum reconstruction operations, the input variable is the spectrum estimation value corresponding to the sub-aperture after the last sub-spectrum reconstruction operation in the previous iterative reconstruction; at the same time, after any sub-aperture performs a sub-spectrum reconstruction operation, the spectrum estimation values ​​corresponding to the sub-aperture and other sub-apertures overlapping with the sub-aperture will change; After each iterative reconstruction is completed, it is determined whether the error value between the sample spectrum obtained after the last sub-spectrum reconstruction operation in this iterative reconstruction and the sample spectrum finally obtained in the previous iterative reconstruction meets the set convergence condition or whether the number of iterative reconstructions reaches the preset maximum number. If so, the sample spectrum formed by the spectrum estimation values ​​corresponding to each sub-aperture after the last sub-spectrum reconstruction operation in this iterative reconstruction is used as the final reconstruction result. If both are no, the spectrum estimation values ​​corresponding to each sub-aperture after the last sub-spectrum reconstruction operation in this iterative reconstruction are used as new input variables to enter the next iterative reconstruction.

2. A stacked microscopic imaging reconstruction method based on dynamic stacking iterative optimization as claimed in claim 1, characterized in that: After any sub-aperture performs a sub-spectrum reconstruction operation in any iterative reconstruction, the calculation method of the sample spectrum formed by the spectrum estimation values ​​corresponding to each sub-aperture is: Where k represents the frequency domain wave vector, * represents the conjugate operator, The sample spectrum after performing the sub-spectrum reconstruction operation on the sub-aperture of the mth row and the nth column in the i-th iterative reconstruction, The sample spectrum before the sub-spectrum reconstruction operation is performed for the sub-aperture in the mth row and nth column in the i-th iterative reconstruction, Pupil function of the stacked microscopy system before performing the sub-spectral reconstruction operation for the sub-aperture in the mth row and nth column in the i-th iterative reconstruction Shift left by k m,n The pupil function of units, k m,n is the illumination wave vector used by the stacked microscopy imaging system to collect the intensity image obtained by the subaperture in the mth row and nth column, f(P (m,n)-1 (k+k m,n )) is the core term related to the pupil function before the sub-aperture of the mth row and nth column performs the sub-spectrum reconstruction operation, The spectrum estimation value before the sub-spectrum reconstruction operation is performed for the sub-aperture in the mth row and nth column in the i-th iterative reconstruction, The spectrum estimation value after performing the sub-spectrum reconstruction operation on the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction.

3. A stacked microscopic imaging reconstruction method based on dynamic stacking iterative optimization as claimed in claim 2, characterized in that: The core term related to the pupil function before the sub-aperture in the mth row and nth column performs the sub-spectrum reconstruction operation The calculation method is as follows: in, and are constructors that vary with the number of iterative reconstructions, |·| is a modulo operation, and max represents the pupil function in the form of a two-dimensional complex matrix. The maximum value after the modulo operation is squared, and and The calculation method is as follows: Among them, e represents the base of natural logarithm, α i represents the direction adjustment parameter related to the number of iterative reconstructions i, and has: Among them, i Emax are preset tuning parameters.

4. A stacked microscopic imaging reconstruction method based on dynamic stacking iterative optimization as claimed in claim 3, characterized in that: The larger the imaging error of the stacked microscopy system, the Emax The smaller the value of .

5. The stacking microscopy imaging reconstruction method based on dynamic stacking iterative optimization according to claim 2, characterized in that: Spatial complex amplitude The calculation method is as follows: Among them, r represents the spatial coordinate, F -1 represents the inverse Fourier transform, The sample spectrum Shift right k m,n The sample spectrum obtained after units is The pupil function of the stacked microscopy system before performing the sub-spectral reconstruction operation for the sub-aperture in the m-th row and n-th column in the i-th iterative reconstruction.

6. The stacking microscopy imaging reconstruction method based on dynamic stacking iterative optimization according to claim 2, characterized in that: Spectral estimate The calculation method is as follows: Among them, r represents the spatial coordinate, F represents Fourier transform, I m,n (r) represents the intensity image acquired by the sub-aperture in the mth row and nth column, is the spatial domain complex amplitude of the spectrum estimation value before the sub-spectrum reconstruction operation is performed on the sub-aperture of the mth row and the nth column in the i-th iterative reconstruction, Indicates that the spatial domain complex amplitude is replaced by the intensity image The spatial domain complex amplitude is obtained by compressing the amplitude in .

7. The stacking microscopy imaging reconstruction method based on dynamic stacking iterative optimization according to claim 2, characterized in that: Pupil function of the stacked microscopy imaging system after performing sub-spectral reconstruction operation on the sub-aperture of the mth row and nth column in the i-th iterative reconstruction The calculation method is as follows: in, The pupil function of the stacked microscopy system before performing the sub-spectral reconstruction operation for the sub-aperture at the m-th row and n-th column in the i-th iterative reconstruction, The sample spectrum before the sub-spectrum reconstruction operation is performed for the sub-aperture in the mth row and nth column in the i-th iterative reconstruction and then shifted right by k m,n The sample spectrum of units, The kernel term associated with the sample spectrum before performing the sub-spectrum reconstruction operation for the sub-aperture in the mth row and nth column.

8. The stacking microscopy imaging reconstruction method based on dynamic stacking iterative optimization according to claim 2, characterized in that: The core term related to the sample spectrum before the sub-aperture of the mth row and nth column performs the sub-spectrum reconstruction operation The calculation method is as follows: in, and All of them are constructors that change with the number of iterative reconstructions, |·| is a modulo operation, and max means obtaining the sample spectrum in the form of a two-dimensional complex matrix. The maximum value after the modulo operation is squared, and and The calculation method is as follows: Among them, e represents the base of natural logarithm, α i represents the direction adjustment parameter related to the number of iterative reconstructions i, and has: Among them, i Emax are preset tuning parameters.

9. The stacking microscopy imaging reconstruction method based on dynamic stacking iterative optimization according to claim 8, characterized in that: The larger the imaging error of the stacked microscopy system, the Emax The smaller the value of .

10. The stacking microscopy imaging reconstruction method based on dynamic stacking iterative optimization according to claim 1, characterized in that: The stacked microscopy imaging system is a spatial stacked microscopy imaging system or a Fourier stacked microscopy imaging system.

Images