Wavefront reconstruction method for improving dynamic range of Shack-Hartmann wavefront sensor

Abstract

The invention discloses a wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor. Belongs to the technical field of wavefront detection. According to the method, a new Shack-Hartmann wavefront detection method is needed, the effective dynamic range of a Shack-Hartmann sensor can be expanded on the premise that the hardware complexity is not remarkably increased, local mismatching accumulation can be avoided through a reasonable matching strategy, and the detection accuracy is improved. Therefore, high-precision and high-reliability wavefront reconstruction can be realized in a large-amplitude wavefront distortion scene. Comprising the following steps: S1, detecting by a Shack-Hartmann wavefront sensor to obtain a light spot distribution diagram H (x, y); s2, extracting a light spot centroid pixel-level position through a matched filtering algorithm; s3, calculating the sub-pixel-level position of the mass center of the light spot through Gaussian interpolation; s4, a mass center-subaperture matching relation is established through a Kuhn-Munkres algorithm; and S5, calculating the wavefront slope according to the corresponding relationship between the centroid and the sub-aperture, and then recovering the wavefront according to the wavefront slope.

Description

Technical Field

[0001] The invention belongs to the technical field of wavefront detection, and in particular relates to a wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor. Background Art

[0002] Wavefront detection technology is an important method in the field of optics for measuring and analyzing light wavefront distortion. It is widely used in astronomy, adaptive optics, laser systems, microscopic imaging, and ophthalmology. Among them, the Shack–Hartmann Wavefront Sensor (SHWS) is widely used due to its simple structure and fast response speed.

[0003] The basic principle of a Shack-Hartmann wavefront sensor (SHWS) is to decompose the incident wavefront into multiple sub-wavefronts using a microlens array. The shape of the entire wavefront is reconstructed by measuring the offset of each sub-wavefront. When the wavefront distortion is large, the focal spot may shift outside the corresponding sub-aperture, resulting in incorrect matching and affecting the accuracy of wavefront reconstruction. Although the dynamic range can be improved by increasing the sub-aperture or shortening the focal length, this comes at the expense of spatial sampling rate and measurement accuracy, resulting in a trade-off with the dynamic range. Currently, methods for extending the dynamic range of SHWSs are mainly divided into hardware-based and algorithm-based approaches. Hardware-based approaches enhance information acquisition through structural adjustments, such as shifting the position of the microlens array or detector or introducing auxiliary sensors or masks, but this sacrifices measurement speed and increases system complexity. Algorithm-based approaches rely on optimizing the spot matching strategy to improve performance. These methods require no hardware modifications to the existing system and are low-cost, but they suffer from issues such as reliance on initial matching and error accumulation, which affect overall robustness.

[0004] Therefore, there is an urgent need for a new Shack–Hartmann wavefront detection method that can not only expand the effective dynamic range of the Shack–Hartmann sensor without significantly increasing the hardware complexity, but also avoid the accumulation of local mismatches through a reasonable matching strategy, thereby achieving high-precision and high-reliability wavefront reconstruction in scenarios with large wavefront distortion, and meeting the practical application needs in fields such as astronomy, adaptive optics, and industrial detection. Summary of the Invention

[0005] In order to solve the above technical problems, a new Shack-Hartmann wavefront detection method is needed, which can not only expand the effective dynamic range of the Shack-Hartmann sensor without significantly increasing the hardware complexity, but also avoid the accumulation of local mismatches through a reasonable matching strategy, thereby achieving high-precision and high-reliability wavefront reconstruction in scenarios with large wavefront distortion, and meet the practical application requirements of fields such as astronomy, adaptive optics and industrial detection. The present invention provides a wavefront reconstruction method for improving the dynamic range of the Shack-Hartmann wavefront sensor. The method uses a centroid search algorithm to obtain the absolute position of the centroid on the entire Hartmann spot image, eliminating the dependence on sub-apertures. Then, a global matching algorithm is used to match the extracted centroids to the sub-apertures one-to-one. Finally, wavefront reconstruction is performed to achieve high-reliability and large dynamic range wavefront detection.

[0006] The method comprises the following steps:

[0007] S1, Shack-Hartmann wavefront sensor detects and obtains the light spot distribution map H(x,y);

[0008] S2, extracting the pixel-level position of the centroid of the light spot through the matched filtering algorithm;

[0009] S3, calculating the sub-pixel position of the centroid of the light spot by Gaussian interpolation;

[0010] S4, establishing the centroid-subaperture matching relationship through the Kuhn-Munkres algorithm;

[0011] S5. Calculate the wavefront slope based on the correspondence between the centroid and the sub-aperture, and then restore the wavefront based on the wavefront slope.

[0012] Further, step S2 is specifically as follows:

[0013] Execute the matched filtering algorithm, calculate the correlation function between the Hartmann spot image and the preset mask, and take the maximum value of the correlation function as the pixel-level absolute position of the spot centroid on the entire spot distribution map; then set all the pixel values ​​of the area with the same centroid position and spot size on the Hartmann spot image to zero, and then perform the correlation function calculation to find the next spot centroid, and use the segmentation threshold of the grayscale image as the judgment of whether it is a valid centroid. Iterate this process until all the spot centroids are searched.

[0014] Furthermore, the related function is:

[0015] Among them, N mask is the size of the preset mask, and g(i,j) is the matched filter mask.

[0016] Furthermore, the grayscale image segmentation threshold is used as the judgment of whether it is a valid centroid:

[0017] The correlation function C(x1,y1) at the first extracted spot mass center position (x1,y1) is used as the judgment threshold. Since the correlation function of the subsequent complete spot mass center is not greater than C(x1,y1), when a spot mass center is missing, it is considered that more than half of the entire spot distribution is out of the boundary. At this time, its correlation function must be less than half of C(x1,y1). C(x k ,y k )>C(x1,y1) / 2 and k≤N are used as the termination conditions for judging the centroid extraction; among them, C(x k ,y k ) is the correlation function at the kth centroid, and N is the number of microlenses in the microlens array.

[0018] Further, step S3 is specifically: performing Gaussian interpolation, obtaining the sub-pixel offset of the spot centroid by fitting interpolation according to the correlation function of the neighborhood pixels, and performing this process on all spot centroids to obtain the sub-pixel position estimation of all spot centroids.

[0019] Furthermore, Gaussian interpolation is:

[0020] The sub-pixel offset of the spot mass center is obtained by fitting interpolation, and finally the sub-pixel position estimation centroid sequence {(x k +Δx k ,y k +Δy k )|k=1,2,..,N c}, where N c It represents the number of actual spot centroids that can be matched by the matched filtering algorithm in step S2.

[0021] Further, step S4 is specifically as follows:

[0022] The overall offset is estimated by the deviation between the overall center of mass of the spot distribution and the overall center of mass of the sub-apertures. The sub-apertures whose corresponding spots exceed the detector target range are estimated. After removing these sub-apertures, the center of mass of the remaining sub-apertures is calculated. The deviation between the center of mass of the remaining sub-apertures and the global center of mass of the spot distribution is then used to further determine the sub-apertures whose corresponding spots will exceed the detection range. This process is repeated until the remaining sub-aperture distribution remains unchanged. The global offset (dx, dy) is estimated by the above method.

[0023] Eliminate the global offset of the spot mass center, that is, {(x k +Δx k ,y k +Δy k )-(dx,dy)|k=1,2,..,N c}, and then use the Kuhn-Munkres algorithm to match the translated center of mass with the sub-aperture, and finally obtain the sub-aperture {(x SA(k) ,y SA(k) )|k=1,2,..,N c}.

[0024] Furthermore, when executing the Kuhn-Munkres algorithm, a virtual center of mass is set, and the weight corresponding to the virtual center of mass and the geometric center of any sub-aperture is set to a value greater than 10. 9 Set N s -N c The total number of centroids is equal to the number of subapertures. The Kuhn-Munkres algorithm is executed and the virtual centroids are excluded. Finally, the subapertures corresponding to the centroids are obtained. s Indicates the number of subapertures.

[0025] Furthermore, the wavefront slope is given by:

[0026] Get, where k represents the subaperture index, S x_k 、S y_k They represent the average wavefront slopes in the x and y axes within the kth sub-aperture, c xk 、c yk They respectively represent the horizontal and vertical coordinates of the geometric center position of the kth sub-aperture.

[0027] Furthermore, the wavefront is restored according to the wavefront slope as follows:

[0028] The relationship between the wavefront slope and the wavefront can be expressed by the first-order partial derivative of the Zernike orthogonal polynomial

[0029]

[0030] Where M is the Zernike order, a i is the coefficient of the i-th Zernike term, Δ k is the distribution of the subaperture transmission function corresponding to the kth mass center in the coordinate system, A SA is the area of ​​the subaperture, S x (x,y), S y (x, y) are the x-axis and y-axis wavefront slopes at the position (x, y), Z i (x,y) is the i-th Zernike polynomial; where, Therefore, the relationship between the wavefront slope and the wavefront can be written in matrix form as

[0031]

[0032] The quantities are represented by Z, A, and S respectively, and we get:

[0033] S = Z·A;

[0034] The approximate solution of the Zernike coefficients by the least squares method is:

[0035] A=(Z T Z) -1 Z T ·S;

[0036] According to the characteristics of the Zernike polynomials themselves, Z is a singular matrix, and the SVD decomposition method is used to find the least squares solution, and finally the wavefront is restored according to the Zernike coefficients.

[0037] The beneficial effects of the method of the present invention are:

[0038] (1) Based on the traditional Hartmann wavefront detection system, no additional hardware changes are required, and the advantages of simple structure and fast response speed are maintained. The centroid extraction and spot matching strategy are used to achieve accurate measurement of large displacements, while improving the dynamic range without increasing additional costs.

[0039] (2) A matched filter algorithm is used to globally search for the centroid, which is independent of the sub-aperture and can work normally even in the case of large wavefront distortion where the spot exceeds the sub-aperture, thus achieving wavefront detection with a large dynamic range.

[0040] (3) The Kuhn-Munkres algorithm is used to achieve the matching between the center of mass and the sub-aperture, which does not rely on prior conditions such as the known spot correspondence relationship. Moreover, as an overall matching strategy, there will be no error accumulation problem and it is more robust.

[0041] (4) The Hartmann wavefront reconstruction method can achieve high-precision wavefront reconstruction even when strong turbulence or large tilt causes part of the light spot to be missing or cannot be correctly matched. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a typical schematic diagram of the detection principle of the Hartmann wavefront sensor;

[0043] Figure 2 This is a flow chart of a large dynamic range wavefront reconstruction method proposed in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0044] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0045] Example 1

[0046] like Figure 1 As shown in FIG. 1 , a typical schematic diagram of the detection principle of a Hartmann wavefront sensor is provided. This embodiment provides a wavefront reconstruction method for improving the dynamic range of the Shack-Hartmann wavefront sensor. Figure 2 As shown, the method includes the following steps:

[0047] S1. The Shack-Hartmann wavefront sensor detects and obtains a light spot distribution diagram H(x, y), where (x, y) represents the horizontal and vertical coordinates in a rectangular coordinate system.

[0048] S2. Extract the pixel-level position of the centroid of the light spot through the matched filtering algorithm: execute the matched filtering algorithm, calculate the correlation function between the Hartmann spot image and the preset mask, and take the maximum value of the correlation function as the pixel-level absolute position estimate of the centroid of the light spot on the entire Hartmann spot image; then set all the pixel values ​​of the area where the centroid position and the spot size are the same on the Hartmann spot image to zero, and then perform the correlation function calculation to find the next centroid of the light spot, and use the segmentation threshold of the grayscale image as the judgment of whether it is a valid centroid. Iterate this process until all the centroids of the light spots are searched.

[0049] S3. Calculate the sub-pixel position of the light spot centroid by Gaussian interpolation: Perform Gaussian interpolation and obtain the sub-pixel offset of the light spot centroid by fitting interpolation based on the correlation function of the neighboring pixels. Perform this process on all light spot centroids to obtain the sub-pixel position estimation of all light spot centroids.

[0050] S4. Establish a centroid-subaperture matching relationship through the Kuhn-Munkres algorithm: Execute the centroid-subaperture matching algorithm, first estimate the overall offset through the deviation between the overall centroid of the spot pattern and the overall centroid of the subaperture, and estimate the subaperture whose corresponding spot exceeds the detector target surface range; set the virtual centroid to match the number of spot centroids and subapertures, execute the Kuhn-Munkres algorithm, and use the distance between the spot centroid and the geometric center of the subaperture as the cost function (the virtual centroid uses a sufficiently large constant as the cost function) to solve the optimal weighted bipartite matching problem and establish a matching relationship between the spot and the subaperture.

[0051] S5. Calculate the wavefront slope based on the correspondence between the center of mass and the sub-aperture, and then restore the wavefront based on the wavefront slope: Execute the wavefront reconstruction algorithm, calculate the local average wavefront slope and Zernike partial derivative matrix corresponding to each light spot, apply the least squares method to solve the Zernike coefficients, and thus restore the wavefront.

[0052] Example 2

[0053] This embodiment further limits the embodiment 1 and further explains step S2.

[0054] Estimate the spot size based on the diameter and focal length of the microlens, select the appropriate mask width, and determine the matched filter mask in the form of a Gaussian function:

[0055] Here, σ represents the standard deviation of the Gaussian mask.

[0056] Calculate the correlation function C(x,y) between the spot distribution map H(x,y) and the matched filter mask g(x,y):

[0057]

[0058] where N mask is the size of the mask. The point with the largest correlation function is considered to be the pixel closest to the center of mass of the light spot (x k ,y k ),Right now:

[0059] (x k ,y k )=argmaxC(x,y);

[0060] (x k ,y k ) can be regarded as the pixel-level position estimation of the center of mass of the light spot. Once the center of mass of a light spot is determined, the local area of ​​the point in the correlation function C(x,y) is set to zero to shield the light spot from being repeatedly detected in subsequent processing. The specific method is: k ,y k ) as the center and the full width at half maximum (FWHM) of the spot as the radius are set to zero. Then, the centroid position of the next spot is calculated. This process is iterated until the centroids of all N spots are extracted, where N is the number of microlenses in the microlens array. In particular, considering the possibility of missing spots in the case of large wavefront distortion (for example, the overall wavefront has a large tilt, causing the spot of the edge sub-aperture to exceed the detector target area), it is necessary to estimate the number of spots.

[0061] A simple and effective method is to perform a threshold judgment on the size of C(x,y), such as using the Otsu method to perform threshold segmentation. The Otsu method is to select a threshold to minimize the intra-class variance of the thresholded black and white pixels, which can separate the information from the background noise as much as possible. Let T = Otsu(H(x,y)), when C(x k ,y k )>T is considered to be the effective spot centroid, otherwise it is background noise. Therefore, C(xk , y k ) and k≤N as the termination condition of judging the extraction of the centroid.

[0062] Optionally, the correlation function C(x1,y1) at the first extracted spot centroid position (x1,y1)=argmax C(x,y) is used as the judgment threshold. Since the correlation function of the centroid of the subsequent complete spot is not greater than it, when the spot centroid is missing, it can be considered that more than half of the entire spot distribution is outside the boundary, at which time its correlation function must be less than half of C(x1,y1). Therefore, C(x k , y k ) and k≤N as the termination condition of judging the extraction of the centroid.

[0063] Finally, the centroid sequence {(x k , y k )|k=1,2,..,N c} of the pixel-level position estimation can be obtained, L is the number of actual spot centroids that can be matched, and N c ≤N.

[0064] Embodiment 3,

[0065] This embodiment is a further limitation of embodiment 1, and further describes step S3. The pixel-level position estimation of the spot centroid cannot meet the accuracy of the centroid required for reconstructing the wavefront information, and further sub-pixel level offset estimation is required. Through Gaussian interpolation of the 3x3 adjacent region of the maximum value of the correlation function, the offset estimation accuracy of the spot image can reach sub-pixel. The Gaussian interpolation can be expressed as

[0066]

[0067] Optionally, the values of the four pixels on the diagonal of the 3x3 neighborhood or the values in a larger neighborhood range can also be used for interpolation. The interpolation finally obtains the centroid sequence {(x k +Δx k , y k +Δy k )|k=1,2,..,N c}.

[0068] Embodiment 4,

[0069] This embodiment is a further limitation of embodiment 1, and further explains step S4. Considering that there is a situation where the spot is missing in the case of large wavefront distortion, if the Kuhn-Munkres algorithm is used directly, a large number of spots will be preferentially matched to the nearby sub-apertures, and the overall offset will be misplaced. In view of this, the present invention adopts a method of preprocessing using the overall offset estimation. First, it is necessary to estimate the overall offset to roughly align the spot with the sub-aperture. The following method is used: First, the global centroid (G Hx ,G Hy ) and the center of the detector (G Sx ,G Sy ), use the deviation between the two (G Hx -G Sx ,G Hy -G Sy ) to estimate which rows and columns of sub-apertures correspond to spots that will exceed the detection range, and after removing these sub-apertures, calculate the center of mass of the remaining sub-apertures to update (G sx ,G sy ), and then use its deviation from the global centroid of the spot distribution to further determine the sub-aperture where the corresponding spot will exceed the detection range, and iterate until the remaining sub-aperture distribution remains unchanged. This method can estimate the global offset (dx, dy) and the sub-aperture where the corresponding spot exceeds the detection range. After estimating the global offset, the spot centroid is eliminated from the global offset, that is, {(x k +Δx k ,y k +Δy k )-(dx,dy)|k=1,2,..,N c}, and then use the Kuhn-Munkres algorithm to match the translated center of mass with the sub-aperture.

[0070] In Hartmann wavefront detection, the matching between the centroid and the subaperture can be regarded as a weighted bipartite graph matching problem. The matching effect between the centroid and the subaperture is measured by the cost function, that is, the total weight of the matching. k +Δx k ,y k +Δy k )-(dx,dy)|k=1,2,..,N c} and the subaperture geometric center {(x SA_l ,y SA_l )|l=1,2,..N s The square of the Euclidean distance (x k +Δx k -dx-x SA_l ) 2 +(y k +Δy k-dy-y SA_l ) 2 As the weight between the centroid and the sub-aperture, the sum of the weights under the current matching is the cost function. The Kuhn-Munkres algorithm efficiently finds a perfect match in the weighted bipartite graph by finding an augmenting path, that is, minimizing the matching cost, thereby achieving the correct matching of the spot and the sub-aperture. In practice, due to local wavefront distortion, some spots in the same row and column will be within the detector range, while others will be outside the detection range. Therefore, the number of remaining sub-apertures N after the overall centroid offset estimation is s It will still be greater than the number of spots N c , in this case, it is necessary to set virtual nodes (virtual centroids) and set their weights corresponding to the geometric centers of any sub-apertures to a sufficiently large number (for example, 10 9 ). Set N s -N c The total number of centers of mass is equal to the number of subapertures. The Kuhn-Munkres algorithm is executed and the virtual centers of mass are excluded. Finally, the center of mass {(x k +Δx k ,y k +Δy k )|k=1,2,..,N c}Establish the corresponding sub-aperture {(x SA(k) ,y SA(k) )|k=1,2,..,N c}.

[0071] Example 5

[0072] This embodiment further limits the embodiment 1 and further explains step S5.

[0073] Step 5: Calculate the wavefront slope based on the correspondence between the centroid and the sub-aperture, and then restore the wavefront based on the wavefront slope. The wavefront slope calculation method is:

[0074]

[0075] Where k represents the subaperture index, S x_k 、S y_k They represent the average wavefront slopes in the x and y axes within the kth sub-aperture, c xk 、c yk The relationship between the wavefront slope and the wavefront can be expressed by the first-order partial derivative of the Zernike orthogonal polynomial.

[0076] Here M is the Zernike order, a i is the coefficient of the i-th Zernike term, Δk is the distribution of the subaperture transmission function corresponding to the kth mass center in the coordinate system, A SA is the area of ​​the subaperture, S x (x,y), S y (x, y) are the x-axis and y-axis wavefront slopes at the position (x, y), Z i (x,y) is the i-th Zernike polynomial. Generally, all subapertures of Hartmann are of the same size, that is, The relationship between the wavefront slope and the wavefront is expressed in matrix form as:

[0077]

[0078] Zernike partial derivative matrix (including constant coefficient 1 / A SA ), the Zernike coefficient vector and the wavefront slope vector are represented by Z, A, and S respectively, and we get:

[0079] S = Z·A;

[0080] The approximate solution of the Zernike coefficients by the least squares method is:

[0081] A=(Z T Z) -1 Z T ·S;

[0082] Due to the characteristics of the Zernike polynomials, Z is a singular matrix. To avoid large errors caused by pathological conditions, SVD decomposition can be used to find the least squares solution. Finally, the wavefront is restored based on the Zernike coefficients.

Claims

1. A wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor, characterized in that: The method comprises the following steps: S1, Shack-Hartmann wavefront sensor detects and obtains the light spot distribution map H(x,y); S2, extracting the pixel-level position of the centroid of the light spot through the matched filtering algorithm; S3, calculating the sub-pixel position of the centroid of the light spot by Gaussian interpolation; S4, establishing the centroid-subaperture matching relationship through the Kuhn-Munkres algorithm; S5. Calculate the wavefront slope based on the correspondence between the centroid and the sub-aperture, and then restore the wavefront based on the wavefront slope.

2. The wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor according to claim 1, characterized in that: Step S2 is specifically as follows: Execute the matched filtering algorithm, calculate the correlation function between the Hartmann spot image and the preset mask, and take the maximum value of the correlation function as the pixel-level absolute position of the spot centroid on the entire spot distribution map; then set all the pixel values ​​of the area with the same centroid position and spot size on the Hartmann spot image to zero, and then perform the correlation function calculation to find the next spot centroid, and use the segmentation threshold of the grayscale image as the judgment of whether it is a valid centroid. Iterate this process until all the spot centroids are searched.

3. The wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor according to claim 2, characterized in that: The related function is: Among them, N mask is the size of the preset mask, and g(i,j) is the matched filter mask.

4. The wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor according to claim 3, characterized in that: The specific determination of whether the grayscale image is a valid centroid is as follows: The correlation function C(x1,y1) at the first extracted spot mass center position (x1,y1) is used as the judgment threshold. Since the correlation function of the subsequent complete spot mass center is not greater than C(x1,y1), when a spot mass center is missing, it is considered that more than half of the entire spot distribution is out of the boundary. At this time, its correlation function must be less than half of C(x1,y1). C(x k ,y k )>C(x1,y1) / 2 and k≤N are used as the termination conditions for judging the centroid extraction; among them, C(x k ,y k ) is the correlation function at the kth centroid, and N is the number of microlenses in the microlens array.

5. The wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor according to claim 4, characterized in that: Step S3 specifically includes: performing Gaussian interpolation, obtaining the sub-pixel offset of the spot centroid by fitting interpolation according to the correlation function of the neighborhood pixels, and performing this process on all spot centroids to obtain the sub-pixel position estimation of all spot centroids.

6. The wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor according to claim 5, characterized in that: Gaussian interpolation is: The sub-pixel offset of the spot mass center is obtained by fitting interpolation, and finally the sub-pixel position estimation centroid sequence {(x k +Δx k ,y k +Δy k )|k=1,2,..,N c }, where N c It represents the number of actual spot centroids that can be matched by the matched filtering algorithm in step S2.

7. The wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor according to claim 6, characterized in that: Step S4 is specifically as follows: The overall offset is estimated by the deviation between the overall center of mass of the spot distribution and the overall center of mass of the sub-apertures. The sub-apertures whose corresponding spots exceed the detector target range are estimated. After removing these sub-apertures, the center of mass of the remaining sub-apertures is calculated. The deviation between the center of mass of the remaining sub-apertures and the global center of mass of the spot distribution is then used to further determine the sub-apertures whose corresponding spots will exceed the detection range. This process is repeated until the remaining sub-aperture distribution remains unchanged. The global offset (dx, dy) is estimated by the above method. Eliminate the global offset of the spot mass center, that is, {(x k +Δx k ,y k +Δy k )-(dx,dy)|k=1,2,..,N c }, and then use the Kuhn-Munkres algorithm to match the translated center of mass with the sub-aperture, and finally obtain the sub-aperture {(x SA(k) ,y SA(k) )|k=1,2,..,N c }.

8. The wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor according to claim 7, characterized in that: When executing the Kuhn-Munkres algorithm, set the virtual center of mass and the weight corresponding to the geometric center of any sub-aperture to a value greater than 10. 9 Set N s -N c The total number of centroids is equal to the number of subapertures. The Kuhn-Munkres algorithm is executed and the virtual centroids are excluded. Finally, the subapertures corresponding to the centroids are obtained. s Indicates the number of subapertures.

9. The wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor according to claim 8, characterized in that: The wavefront slope is determined by: k=1,2,…,N c ; Get, where k represents the subaperture index, S x_k 、S y_k They represent the average wavefront slopes in the x and y axes within the kth sub-aperture, c xk 、c yk They respectively represent the horizontal and vertical coordinates of the geometric center position of the kth sub-aperture.

10. The wavefront reconstruction method for improving the dynamic range of a Shack-Hartmann wavefront sensor according to claim 9, characterized in that: The wavefront is restored according to the wavefront slope as follows: The relationship between the wavefront slope and the wavefront can be expressed by the first-order partial derivative of the Zernike orthogonal polynomial Where M is the Zernike order, a i is the coefficient of the i-th Zernike term, Δ k is the distribution of the subaperture transmission function corresponding to the kth mass center in the coordinate system, A SA is the area of ​​the subaperture, S x (x,y), S y (x, y) are the x-axis and y-axis wavefront slopes at the position (x, y), Z i (x,y) is the i-th Zernike polynomial; where, Therefore, the relationship between the wavefront slope and the wavefront can be written in matrix form as and The vectors are represented by Z, A, and S respectively, and we get: S = Z·A; The approximate solution of the Zernike coefficients by the least squares method is: A=(Z T Z) -1 Z T ·S; According to the characteristics of the Zernike polynomials themselves, Z is a singular matrix, and the SVD decomposition method is used to find the least squares solution, and finally the wavefront is restored according to the Zernike coefficients.

Images