A method for Shack-Hartmann wavefront reconstruction with ultra large dynamic range

Abstract

The application discloses a kind of super large dynamic range Shack-Hartmann wavefront reconstruction methods, it is related to wavefront detection technical field, including: obtaining reference spot array image, and for each reference spot allocation ID;Obtain the spot array image sequence formed by wavefront to be measured;Perform global spot detection, obtain the centroid position set of all observation spots in frame;The observation spot of current frame is globally data associated with the historical spot trajectory established;For the observation spot matched successfully, with its centroid position as measurement value, update the state tracker of corresponding ID;For each spot successfully maintaining identity ID, calculate the displacement vector between its current centroid position and corresponding initial reference centroid position;Based on displacement vector, reconstruct the wavefront phase to be measured.The method can expand the determinant of dynamic range from sub-aperture geometric parameters to detector physical field of view while maintaining spatial resolution and sensitivity by algorithm upgrade only.

Description

Technical Field

[0001] This invention relates to the field of wavefront detection technology, and in particular to a Shaker-Hartmann wavefront reconstruction method with an ultra-large dynamic range. Background Technology

[0002] The Shaker-Hartmann wavefront sensor (SHWFS) is a core component of adaptive optics systems. It divides the incident wavefront into multiple sub-wavefronts using a microlens array, forming corresponding spot arrays on a detector. By measuring the displacement of each spot relative to its reference position, the wavefront slope can be calculated, and the wavefront phase can be reconstructed.

[0003] However, traditional SHWFS suffers from an inherent and irreconcilable contradiction: dynamic range, spatial resolution, and measurement sensitivity are mutually constrained. Dynamic range in SHWFS refers to the maximum wavefront distortion that a single sub-aperture can measure, primarily dependent on the ratio of microlens focal length to sub-aperture spacing. To increase dynamic range, the microlens focal length needs to be reduced or the sub-aperture spacing increased, but this directly leads to a decrease in spatial resolution or sensitivity. In scenarios involving strong turbulent atmospheres or high-power laser beams that produce wavefront distortions on the order of tens of wavelengths, the beam in traditional SHWFS often shifts out of its original sub-aperture region, even entering adjacent sub-apertures, resulting in "spot confusion" or "crossing." In such cases, traditional spot matching algorithms based on local neighborhood search, such as iterative extrapolation and sorting methods, completely fail, leading to wavefront reconstruction failure.

[0004] To extend the dynamic range, existing technologies mainly fall into two categories: hardware and software. Hardware methods, such as using spatial light modulators with switchable sub-apertures, increase system complexity, cost, and reduce reliability. While software processing methods offer some effectiveness, they essentially perform "rescue" matching of confused light spots within a single frame, failing to fundamentally address the issue of the continuity of light spot identity over time. Once the light spots undergo large displacements or trajectory intersections, the performance of existing software methods deteriorates sharply.

[0005] In summary, the limitation of existing methods is that the displacement and matching of the light spot must be carried out within a pre-divided sub-aperture grid framework. This paradigm physically leads to a contradiction between dynamic range and spatial resolution, which is the underlying bottleneck restricting performance improvement. Summary of the Invention

[0006] The purpose of this invention is to provide a wavefront reconstruction method based on multi-target tracking to break the above-mentioned constraints. By reconstructing spot matching into a global trajectory tracking problem, it achieves ultra-large dynamic range wavefront sensing that does not depend on sub-aperture boundaries.

[0007] The specific steps of the technical solution adopted in this invention are as follows:

[0008] A method for reconstructing the Shaker-Hartmann wavefront with an ultra-large dynamic range includes the following steps:

[0009] Step S110: Calibrate the Shak-Hartmann wavefront sensor, acquire the reference spot array image of the Shak-Hartmann wavefront sensor in the calibration state, assign a unique and permanent spot identification ID to each reference spot, and calculate the initial reference centroid position of each reference spot. The ID is initialized at a specific sub-aperture position in the microlens array and remains unchanged in subsequent tracking.

[0010] Step S120: Obtain the image sequence of the light spot array formed by the wavefront to be tested, which varies with the time series;

[0011] Step S130: For each frame of the light spot array image to be tested, perform global light spot detection to obtain the set of centroid positions of all observed light spots in that frame;

[0012] Step S140: Based on the multi-target tracking framework, globally associate the observed spot in the current frame with the established historical spot trajectories to maintain the continuity of the identity ID of each observed spot, where the historical spot trajectory is the motion state record of the observed spot corresponding to the ID in step S110; specifically including three sub-steps:

[0013] Step S140-1: Using the state tracker corresponding to each initialized ID, predict the most likely position of the spot in the current frame and the prediction uncertainty based on the historical motion state of the spot corresponding to that ID;

[0014] Step S140-2: Based on the prediction results of the historical trajectories corresponding to each initialized ID and the observed spot in the current frame, construct an association cost matrix. Each element of the matrix represents the matching cost between the predicted position of a specific ID and an observed spot in the current frame. The matching cost is a weighted combination of motion consistency cost and appearance similarity cost.

[0015] Step S140-3: Input the above cost matrix into the global optimal allocation algorithm to solve for the "ID-observation" pairing scheme that minimizes the overall matching cost, and output the matching results, including: successfully matched ID-observation pairs, unmatched historical trajectory ID set, and unmatched observation spot set;

[0016] Step S150: Based on the matching results of steps S140-3, for successfully matched observed spots, the state tracker with the corresponding ID is updated using its centroid position as the measurement value, and its unmatched count is reset; for unmatched historical trajectories, its unmatched count is incremented by 1; when the number of consecutive unmatched frames reaches a preset threshold, it is determined that the spot corresponding to the unmatched trajectory has disappeared, its tracker is terminated, and the historical spot trajectory corresponding to the disappeared spot is deleted from the system; for unmatched observed spots, they are directly filtered out as noise, and no new trajectory is created.

[0017] Step S160: For each successfully matched ID-observation pair, calculate the displacement vector between its current centroid position and the initial reference centroid position corresponding to it in step S110 for the target spot.

[0018] Step S170: Based on the displacement vector, a slope vector is formed, and the complete wavefront phase to be measured is reconstructed using a wavefront restoration algorithm.

[0019] A computing device includes: at least one processor and a memory storing program instructions; when the program instructions are read and executed by the processor, the computing device performs the method.

[0020] A readable storage medium storing program instructions that, when read and executed by a computing device, cause the computing device to perform the method.

[0021] A computer program product includes a computer program that, when executed by a processor, implements the method.

[0022] Beneficial effects:

[0023] (1) Significant improvement in dynamic range: By changing the spot matching from "local search within sub-aperture" to "target tracking within global field of view", the traditional method is freed from the constraints of the physical boundary of the sub-aperture, and the effective measurement range is extended to the entire field of view of the detector, which significantly enhances the system's ability to measure large-amplitude wavefront distortion.

[0024] (2) High compatibility and practicality: This invention does not require any modification to the SHWFS hardware. It can achieve a significant performance improvement simply by upgrading the algorithm. It is low in cost and easy to integrate into existing adaptive optics systems, providing urgently needed technical means for cutting-edge fields such as astronomical observation, laser engineering, and free space communication.

[0025] (3) Solved the problem of mismatched spot: By introducing a multi-target tracking paradigm, a unique and continuous identity trajectory is established for each spot in the time series, which solves the fundamental problem that traditional methods fail when faced with large displacement, intersection and overlap of spot.

[0026] (4) Significantly enhanced robustness: The global data association strategy that integrates motion prediction and multi-feature matching enables the algorithm to have strong adaptability and stability to noise, short-term occlusion of light spots, fast motion and complex motion patterns.

[0027] (5) High precision preservation: Within the range of small to medium amplitude aberrations, the method of the present invention can maintain measurement accuracy comparable to that of the traditional local centroid method, ensuring the effectiveness and practicality of the method. Attached Figure Description

[0028] Figure 1 This is a detailed flowchart of the ultra-large dynamic range Shake-Hartmann wavefront reconstruction method of the present invention.

[0029] Figure 2 The image shows three example frames from the dynamic large-amplitude turbulent wavefront sequence generated in the embodiment.

[0030] Figure 3 This is a schematic diagram illustrating the tracking effect of the method in this embodiment.

[0031] Figure 4 This is a diagram showing the phase reconstruction effect of the method in this embodiment.

[0032] Figure 5 This is a comparison chart of the wavefront residuals of the traditional algorithm and the proposed method in the same test frame in the embodiment. Detailed Implementation

[0033] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0034] This invention provides an ultra-large dynamic range Shak-Hartmann wavefront reconstruction method. Addressing the limitations of traditional Shak-Hartmann wavefront sensors (SHWFS) due to large spot displacement and mismatching, this method abandons the traditional "spot-sub-aperture" constraint, treating the spot as an independent moving target within the global field of view. By establishing a motion prediction model for each spot and using a globally optimal matching algorithm to achieve data association, this invention transforms the foundation of wavefront sensing from sub-aperture slope measurement to global trajectory tracking.

[0035] like Figure 1 As shown, the method includes the following steps:

[0036] Step S110: Calibrate the Shak-Hartmann wavefront sensor, acquire the reference spot array image of the Shak-Hartmann wavefront sensor in the calibration state, assign a unique and permanent spot identity (ID) to each reference spot, and calculate the initial reference centroid position of each reference spot. This ID is initialized at a specific sub-aperture position in the microlens array and remains unchanged in subsequent tracking.

[0037] Step S120: Obtain the image sequence of the light spot array formed by the wavefront to be tested, which varies with the time series;

[0038] Step S130: For each frame of the light spot array image to be tested, perform global light spot detection to obtain the set of centroid positions of all observed light spots in that frame;

[0039] In step S130, the methods for performing global spot detection include, but are not limited to, global thresholding and connected component analysis.

[0040] In step S130, the appearance features of each observation spot can also be optionally extracted. The appearance features of each observation spot include the integrated light intensity, point spread function (PSF) shape parameter, and observation spot size.

[0041] Step S140: Based on the multi-target tracking framework, globally associate the observed spot in the current frame with the established historical spot trajectories to maintain the continuity of the identity ID of each observed spot. The historical spot trajectory is the motion state record of the observed spot corresponding to the ID in step S110. This step includes three sub-steps:

[0042] Step S140-1: Using the state tracker corresponding to each initialized ID, predict the most likely position and prediction uncertainty of the spot in the current frame based on the historical motion state (position, velocity, etc.) of the spot corresponding to that ID; wherein, the state tracker can adopt a tracking algorithm based on Bayesian filtering, including but not limited to Kalman filter, particle filter or interactive multiple model filter; the state vector of the state tracker contains at least the position and velocity information of the observed spot, which is used to predict the motion of each spot trajectory.

[0043] Step S140-2: Based on the prediction results of the historical trajectories corresponding to each initialized ID and the observed spot in the current frame, an association cost matrix is ​​constructed. Each element of the matrix represents the matching cost between the predicted position of a specific ID and an observed spot in the current frame. This matching cost is a weighted sum of motion consistency cost and appearance similarity cost. The construction of the association cost matrix is ​​based on the distance between the predicted position and the centroid position of the observed spot, which can be represented by Mahalanobis distance or Euclidean distance, etc. The construction of the association cost matrix also incorporates the appearance feature distance of the observed spot, which includes at least one of the integrated light intensity, point spread function (PSF) shape parameter, and spot size of the observed spot.

[0044] Step S140-3: Input the above cost matrix into the global optimal allocation algorithm to solve for the "ID-observation" pairing scheme that minimizes the overall matching cost, and output the matching results, including: successfully matched ID-observation pairs, unmatched historical trajectory ID sets, and unmatched observation spot sets; wherein, the global optimal allocation algorithm can adopt an assignment algorithm based on bipartite graph matching, including but not limited to the Hungarian Algorithm, Auction Algorithm, or Jonker-Volgenant Algorithm (LAPJV).

[0045] Step S150: Based on the matching results of steps S140-3, for successfully matched observation spots, the state tracker (including position, velocity, and other state information) of the corresponding ID is updated using its centroid position as the measurement value, and its unmatched count is reset; for unmatched historical trajectories, the unmatched count is incremented by 1; when the number of consecutive unmatched frames reaches a preset threshold, it is determined that the spot corresponding to the unmatched trajectory has disappeared, its state tracker is terminated, and the trajectory is deleted from the system; for unmatched observation spots, they are directly filtered out as noise, and no new trajectory is created.

[0046] Step S160: For each successfully matched ID-observation pair, regardless of where it is currently located in the detector's field of view, calculate the displacement vector between its current centroid position and the initial reference centroid position corresponding to it in step S110.

[0047] Step S170: Based on the displacement vector, a slope vector is formed, and the complete wavefront phase to be measured is reconstructed using a wavefront restoration algorithm (such as the mode method, the region method, etc.).

[0048] The following is a specific example to illustrate the ultra-large dynamic range wavefront reconstruction method based on multi-target tracking of this invention. Assume the hardware parameters of SHWFS are: telescope aperture... Number of sub-apertures in microlens array Number of pixels per sub-aperture Operating wavelength Microlens focal length According to the formula Calculated (where Let d be the sub-aperture size, and d be the diameter of the diffraction spot formed by the sub-aperture on the detector focal plane. (The operating wavelength).

[0049] Step S110: Calibrate the Shaker-Hartmann wavefront sensor and acquire a reference image: Illuminate the SHWFS using a well-collimated laser source (approximately a plane wavefront) and acquire a spot array image as a reference image. .

[0050] Calculate the reference centroid: for Perform image processing to identify all effective sub-aperture regions. For each sub-aperture... The reference centroid coordinates of the light spot are calculated using threshold segmentation and centroid algorithm. All reference centroid coordinates constitute the reference centroid matrix. and .

[0051] Assign IDs and initialize trackers: for each valid sub-aperture Assign a unique, permanent identity. A Kalman filter (KF) is initialized for each ID as its state tracker. The initial state vector of the KF is set to... The initial position is the reference centroid, and the initial velocity is zero. The process noise covariance matrix Q and the measurement noise covariance matrix R are preset based on the possible acceleration of the light spot motion and the detector noise level.

[0052] Step S120: Assume the system begins receiving a continuous sequence of wavefront spot images of the target wavefront. In this embodiment, to verify the adaptability of the method of the present invention to large-amplitude dynamic wavefront distortion, the image sequence is composed of, as follows: Figure 2 The dynamic large-amplitude turbulent wavefront phase screen sequence shown is generated.

[0053] Step S130: For each frame Perform global spot detection in a loop:

[0054] For the current frame A global adaptive threshold segmentation combined with connected component analysis is used to detect all light spot regions in the image. For each detected light spot k, the following calculation is performed:

[0055] (1) Location of the center of mass , as observed values.

[0056] (2) Appearance characteristics This embodiment extracts two features: ① Total light intensity (Sum of pixel grayscale values); ② Spot radius (Obtained through Gaussian fitting or second-order moment calculation).

[0057] All observations are grouped into a set .

[0058] Step S140: Based on the multi-target tracking framework, globally associate the observed spot in the current frame with the established historical spot trajectories to maintain the continuity of the identity ID of each spot. This specifically includes three sub-steps:

[0059] Step S140-1: For all activated systems that have... Tracker (Kalman filter) ), perform the prediction step, and obtain the prediction state of its current frame. and predicted covariance Extract the predicted location ( ).

[0060] Step S140-2: Construct a cost matrix C with dimensions M×N (M trajectories, N observations). For the m-th trajectory and the k-th observation, calculate their association cost. :

[0061] ,

[0062] in:

[0063] Mahalanobis distance: , It is the coordinate difference vector between the predicted position and the observed position, which measures the consistency of motion.

[0064] Distance for appearance features: In this embodiment, it is defined as... This is a normalized weighted sum of light intensity and shape differences. It measures apparent similarity. This is the maximum reference light intensity value used to normalize the difference in integrated light intensity. This is the maximum reference size value used to normalize the differences in light spot size. The total light intensity (sum of pixel gray values) of the k-th observation. Let be the total light intensity of the m-th trajectory.

[0065] and For example, the weighting coefficient can be set as follows: , To gain more trust in sports predictions.

[0066] If calculated Greater than the preset maximum allowable cost (For example, 50), then set it to , indicates that matching is prohibited.

[0067] Step S140-3: Input the cost matrix C into the Hungarian algorithm for solving. The algorithm outputs the matching results, including: successfully matched ID-observation pairs, a set of unmatched historical trajectory IDs, and a set of unmatched observation spots. Successfully matched ID-observation pairs indicate the correct association between the current frame's observation spot and the historical trajectory; trajectories in the unmatched historical trajectory ID set will be marked as "unconfirmed" and will enter subsequent trajectory management; for unmatched observation spots, since the number of SHWFS spots should theoretically be constant, they are directly filtered out as noise and no new trajectories are created.

[0068] Step S150: Based on the matching results of steps S140-3, for successfully matched observation spots, their centroid position values ​​are... As a measurement, update the corresponding Kalman filter. and update its appearance features. for The moving average.

[0069] For each unmatched trajectory, its "loss count" is incremented by 1. If the loss count of a trajectory continuously exceeds a preset number of lost frames... (For example, 3 to 5 frames) If the spot has been permanently lost (possibly due to occlusion or sub-aperture failure), its tracker is terminated; for unmatched observations, in this embodiment, since the number of SHWFS spots should theoretically be constant, they are directly filtered out as noise and no new trajectory is created.

[0070] Through the Figure 2 The large-amplitude turbulent wavefront sequence shown is obtained by applying steps S140 and S150 to obtain... Figure 3 The diagram shows the light spot tracking effect.

[0071] Step S160: For all successfully matched and updated trajectories Its corresponding sub-aperture index It is known that, among which, and Let represent the row and column indices of the sub-aperture to which the target spot belongs in the reference state, respectively. Calculate the displacement vector of this spot in the current frame:

[0072] ,

[0073] ,

[0074] in, This is the KF-updated state estimate, which is smoother than the original observations. This displacement vector... That is, the sub-aperture The measured value of the local slope of the wavefront.

[0075] Step S170: Collect slope data for all effective sub-apertures to form a slope vector S. Utilize the wavefront reconstruction matrix pre-constructed during the system calibration phase. (For example, obtained through the Zernike pattern method) , For the influence matrix, (This represents a pseudo-inverse operation on matrix D), calculating the wavefront expansion coefficients: ,in It is a slope vector formed by arranging the wavefront local slope measurements corresponding to all effective sub-apertures in a predetermined order.

[0076] Finally, the complete wavefront phase distribution can be reconstructed using these coefficients (such as the Zernike coefficients). .like Figure 4 As shown in the figure, this diagram illustrates... Figure 2 A comparison is made between the actual wavefront phase distribution of the first frame of the sequence and the wavefront phase distribution reconstructed using the method of this invention. The two exhibit a high degree of consistency in phase distribution morphology and aberration characteristics, intuitively demonstrating that this method can still achieve high-fidelity wavefront reconstruction under large dynamic range conditions.

[0077] The key parameters and effects of this example are as follows:

[0078] Dynamic range improvement: The dynamic range of traditional methods is limited by ,in It is the maximum measurable wavefront tilt angle of the light spot within a single sub-aperture. It is the sub-aperture size. This refers to the microlens focal length. In this method, as long as the light spot is within the detector's field of view and can be continuously tracked, its displacement can be measured. Under the parameters set in this example, the conventional dynamic range is approximately... , The operating wavelength is [value]. After applying this method, the wavefront distortion was successfully tracked and accurately reconstructed in simulation tests, exceeding [value]. In such scenarios, the dynamic range is significantly improved. For example... Figure 5 As shown in the figure, this figure illustrates the distribution of wavefront reconstruction residuals (i.e., the difference between the reconstructed phase and the true phase) of the conventional method and the method of this invention under the same test frame.

[0079] Anti-crossing capability: In the test simulating the intersection of light spot paths, only the positional distance is used. Simple tracker Frequent exchanges. Introducing appearance features ( After that, near the intersection The exchange rate decreased significantly.

[0080] Real-time performance: On a computer equipped with a processor, the average time to complete all the above steps S120-S170 for a single frame of 288×288 pixel image (12x12 sub-aperture) is about 15 milliseconds, which meets the requirements of real-time closed-loop control of most adaptive optics systems (further optimization is needed for frame rate > 1kHz).

[0081] Those skilled in the art will understand that the specific parameters (such as weights) in the above embodiments are... , Maximum number of lost frames The selection of appearance features, etc., can be adjusted according to different SHWFS models, application scenarios, and performance priorities. These adjustments are all conventional optimizations based on the core idea of ​​this invention—that is, maintaining spot identity to extend dynamic range through global data association of multi-target tracking—and all fall within the protection scope of this invention.

[0082] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for reconstructing a Shaker-Hartmann wavefront with an ultra-large dynamic range, characterized in that, Includes the following steps: Step S110: Calibrate the Shak-Hartmann wavefront sensor, acquire the reference spot array image of the Shak-Hartmann wavefront sensor in the calibration state, assign a unique and permanent spot identification ID to each reference spot, and calculate the initial reference centroid position of each reference spot. The ID is initialized at a specific sub-aperture position in the microlens array and remains unchanged in subsequent tracking. Step S120: Obtain the image sequence of the light spot array formed by the wavefront to be tested, which varies with the time series; Step S130: For each frame of the light spot array image to be tested, perform global light spot detection to obtain the set of centroid positions of all observed light spots in that frame; Step S140: Based on the multi-target tracking framework, globally associate the observed spot in the current frame with the established historical spot trajectories to maintain the continuity of the identity ID of each observed spot, where the historical spot trajectory is the motion state record of the observed spot corresponding to the ID in step S110; specifically including three sub-steps: Step S140-1: Using the state tracker corresponding to each initialized ID, predict the most likely position of the spot in the current frame and the prediction uncertainty based on the historical motion state of the spot corresponding to that ID; Step S140-2: Based on the prediction results of the historical trajectories corresponding to each initialized ID and the observed spot in the current frame, construct an association cost matrix. Each element of the matrix represents the matching cost between the predicted position of a specific ID and an observed spot in the current frame. The matching cost is a weighted combination of motion consistency cost and appearance similarity cost. Step S140-3: Input the above cost matrix into the global optimal allocation algorithm to solve for the ID-observation pairing scheme that minimizes the overall matching cost, and output the matching results, including: successfully matched ID-observation pairs, the set of unmatched historical trajectory IDs, and the set of unmatched observation spots; Step S150: Based on the matching results of steps S140-3, for successfully matched observed spots, the state tracker with the corresponding ID is updated using its centroid position as the measurement value, and its unmatched count is reset; for unmatched historical trajectories, its unmatched count is incremented by 1; when the number of consecutive unmatched frames reaches a preset threshold, it is determined that the spot corresponding to the unmatched trajectory has disappeared, its tracker is terminated, and the historical spot trajectory corresponding to the disappeared spot is deleted from the system; for unmatched observed spots, they are directly filtered out as noise, and no new trajectory is created. Step S160: For each successfully matched ID-observation pair, calculate the displacement vector between its current centroid position and the initial reference centroid position corresponding to it in step S110 for the target spot. Step S170: Based on the displacement vector, a slope vector is formed, and the complete wavefront phase to be measured is reconstructed using a wavefront restoration algorithm.

2. The ultra-large dynamic range Shake-Hartmann wavefront reconstruction method according to claim 1, characterized in that, In step S130, the method for performing global spot detection is either the global threshold method or the connected component analysis method.

3. The ultra-large dynamic range Shake-Hartmann wavefront reconstruction method according to claim 1, characterized in that, Step S130 further includes: extracting the appearance features of each observation spot, wherein the appearance features of each observation spot include the integrated light intensity, point spread function shape parameter, and spot size.

4. The ultra-large dynamic range Shake-Hartmann wavefront reconstruction method according to claim 1, characterized in that, In step S140-1, the state tracker uses a tracking algorithm based on Bayesian filtering; the state vector of the state tracker contains the position and velocity information of the observed light spot, which is used to predict the motion of each light spot trajectory.

5. The ultra-large dynamic range Shack-Hartmann wavefront reconstruction method according to claim 4, characterized in that, Tracking algorithms based on Bayesian filtering employ any one of the following: Kalman filter, particle filter, or interactive multi-model filter.

6. The ultra-large dynamic range Shake-Hartmann wavefront reconstruction method according to claim 1, characterized in that, In step S140-3, the global optimal allocation algorithm adopts an assignment algorithm based on bipartite graph matching, which can be the Hungarian algorithm, auction algorithm, or Jonker-Volgenant algorithm.

7. The ultra-large dynamic range Shack-Hartmann wavefront reconstruction method according to claim 1, characterized in that, In step S140-2, the construction of the association cost matrix is ​​based on the distance between the predicted location and the observed location, represented by Mahalanobis distance or Euclidean distance.

8. A computing device, characterized in that, include: At least one processor and a memory storing program instructions; When the program instructions are read and executed by the processor, the computing device performs the ultra-large dynamic range Shake-Hartmann wavefront reconstruction method as described in any one of claims 1-7.

9. A readable storage medium storing program instructions, characterized in that, When the program instructions are read and executed by the computing device, the computing device performs the ultra-large dynamic range Shake-Hartmann wavefront reconstruction method as described in any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the ultra-large dynamic range Shake-Hartmann wavefront reconstruction method as described in any one of claims 1-7.

Images