Wavefront distortion information prediction method and device, electronic equipment and storage medium

Abstract

The application provides a wavefront distortion information prediction method, which comprises the following steps: obtaining an original far-field light spot measured by a target far-field light spot measurement system; performing calibration processing on the original far-field light spot by using a first system parameter of the target far-field light spot measurement system and a second system parameter of a standard far-field light spot simulation system, to obtain a standard far-field light spot; inputting the standard far-field light spot into a trained neural network model for prediction processing, to obtain predicted wavefront distortion information, wherein the trained neural network model is trained according to the sample standard far-field light spot after calibration processing and corresponding wavefront distortion true value. According to the application, the original far-field light spots with different parameters measured by different far-field light spot measurement systems are calibrated into standard far-field light spots, and the standard far-field light spots are used as the input of the trained neural network model, so that the wavefront distortion information prediction of various different measurement systems can be applied, and the wavefront distortion information prediction efficiency in different scenes is improved.

Description

Technical Field

[0001] This application relates to the field of optical sensors, and more particularly to a method, apparatus, electronic device, and storage medium for predicting wavefront distortion information. Background Technology

[0002] Wavefront sensing is an optical wavefront aberration measurement technique and a crucial component of adaptive optics, widely used in astronomical imaging, solar imaging, human eye bio-imaging, laser communication, and high-energy laser systems. The emergence of intelligent adaptive optics has significantly reduced the need for expensive wavefront sensors in wavefront detection, as well as the increased demand on beacon laser intensity caused by wavefront sensors. Simultaneously, it enhances detection capabilities in environments with strong turbulence, long transmission distances, faint targets, and high-power conditions.

[0003] Intelligent adaptive optics encompasses numerous sub-technologies, with intelligent wavefront sensing being one of them. Its function is to extract wavefront distortion information from far-field intensity images. Currently, there are three approaches to intelligent wavefront sensing: recovering the slope or wavefront based on a HS measurement array; recovering the wavefront or Zernike aberration based on two intensity images at the focal plane and defocus plane; and recovering the wavefront or Zernike aberration based on a single focal plane image. Existing AI-based wavefront recovery methods based on far-field intensity images generally suffer from the following drawbacks: 1. The number of grid cells and resolution of the input intensity image for the intelligent wavefront sensing model are pre-fixed, making it impossible to flexibly predict wavefront distortion information for intensity images with different grid numbers and resolutions; 2. The intelligent wavefront sensing model is typically learned from simulation or experimental data based on the parameters of a single far-field spot measurement system, making it unsuitable for far-field spot measurement systems with different parameters such as aperture, focal length, wavelength, and camera resolution. For different systems, data must be re-acquired and the neural network model trained again, resulting in wasted resources for data acquisition and model training, and limiting its timeliness.

[0004] Therefore, existing wavefront distortion information prediction methods have the problem of requiring different neural network models to flexibly identify input light intensity images for different scenarios, resulting in low prediction efficiency for wavefront distortion information under different scenarios. Summary of the Invention

[0005] This application provides a wavefront distortion information prediction method, which aims to solve the problem that existing wavefront distortion information prediction methods require training different neural network models to flexibly identify input light intensity images for different scenes, resulting in low wavefront distortion information prediction efficiency in different scenes.

[0006] In a first aspect, embodiments of this application provide a method for predicting wavefront distortion information, characterized in that the method includes the following steps:

[0007] Acquire the original far-field spot measured by the target far-field spot measurement system;

[0008] The original far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot.

[0009] The standard far-field spot is input into the trained neural network model for prediction processing to obtain predicted wavefront distortion information. The trained neural network model is trained based on the calibrated sample standard far-field spot and the corresponding wavefront distortion ground truth value.

[0010] Optionally, the calibration of the original far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot includes:

[0011] Obtain the first system parameters of the far-field spot measurement system, and obtain the second system parameters of the standard far-field spot simulation system;

[0012] The speckle radius of the original far-field spot is determined by the first system parameters, and the speckle radius of the standard far-field spot is determined by the second system parameters.

[0013] The original far-field spot is calibrated using the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot to obtain the standard far-field spot.

[0014] Optionally, the calibration process of the original far-field spot using the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot to obtain the standard far-field spot includes:

[0015] Determine the first grid number of the original far-field spot, and the first resolution of the original far-field spot based on the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot;

[0016] Determine the second grid number and the second resolution of the standard far-field spot;

[0017] The first grid number and the first resolution of the original far-field spot are interpolated to the second grid number and the second resolution of the standard far-field spot to obtain the standard far-field spot.

[0018] Optionally, the original far-field spot includes a first far-field spot and a second far-field spot, wherein the first far-field spot is a defocused far-field spot, and the second far-field spot is either a defocused far-field spot or a focused far-field spot. The standard far-field spot includes a first standard far-field spot and a second standard far-field spot. The calibration process of the original far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot includes:

[0019] The first far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the first standard far-field spot.

[0020] The second far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the second standard far-field spot.

[0021] Optionally, the step of inputting the standard far-field spot into a trained neural network model for prediction processing to obtain predicted wavefront distortion information includes:

[0022] The first standard far-field spot and the second standard far-field spot are input into the trained neural network model for prediction processing to obtain the predicted wavefront distortion information.

[0023] Optionally, the predicted wavefront distortion information includes Zernike coefficients, and the step of inputting the standard far-field spot into a trained neural network model for prediction processing to obtain the predicted wavefront distortion information includes:

[0024] The standard far-field light spot is input into a trained neural network model for prediction processing, and the Zernike coefficients of a predetermined order are output.

[0025] Optionally, in the step of inputting the standard far-field light spot into a trained neural network model for prediction processing to obtain predicted wavefront distortion information, the method further includes:

[0026] The original far-field spot of the sample containing wavefront distortion and the true value of the wavefront distortion corresponding to the original far-field spot of the sample are measured by a far-field spot measurement system.

[0027] The original far-field spot of the sample is calibrated by the third system parameter of the far-field spot measurement system and the second system parameter of the standard far-field spot simulation system to obtain the standard far-field spot of the sample.

[0028] Based on the standard far-field spot of the sample and the true value of the wavefront distortion corresponding to the original far-field spot of the sample, the neural network model to be trained is trained, and the trained neural network model is obtained after training. The output of the neural network model to be trained and the trained neural network model are both predicted wavefront distortion information.

[0029] Secondly, embodiments of this application also provide a wavefront distortion information prediction device, the wavefront distortion information prediction device comprising:

[0030] The acquisition module is used to acquire the original far-field spot measured by the target far-field spot measurement system;

[0031] The calibration processing module is used to calibrate the original far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain a standard far-field spot.

[0032] The prediction module is used to input the standard far-field spot into the trained neural network model for prediction processing to obtain predicted wavefront distortion information. The trained neural network model is trained based on the calibrated sample standard far-field spot and the corresponding wavefront distortion ground truth value.

[0033] Thirdly, embodiments of this application provide an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the wavefront distortion information prediction method provided in embodiments of this application.

[0034] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in the wavefront distortion information prediction method provided in the embodiments of the invention.

[0035] In this embodiment, the original far-field spot measured by the target far-field spot measurement system is obtained; the original far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain a standard far-field spot; the standard far-field spot is input into a trained neural network model for prediction processing to obtain predicted wavefront distortion information. The trained neural network model is trained based on the calibrated sample standard far-field spot and the corresponding wavefront distortion ground truth. This embodiment, by applying different calibration processes to the original far-field spot, can calibrate original far-field spots with different parameters measured by different far-field spot measurement systems into standard far-field spots. Using the standard far-field spot as the input to the trained neural network model standardizes the input of the neural network model, making it applicable to wavefront distortion information prediction for various measurement systems and improving the prediction efficiency of wavefront distortion information in different scenarios. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart of a wavefront distortion information prediction method provided in an embodiment of this application;

[0038] Figure 2 This is a flowchart of another wavefront distortion information prediction method provided in the embodiments of this application;

[0039] Figure 3 This is a flowchart of another wavefront distortion information prediction method provided in the embodiments of this application;

[0040] Figure 4 This is a schematic diagram of wavefront distortion information prediction provided in an embodiment of this application;

[0041] Figure 5 This is a schematic diagram of the structure of a wavefront distortion information prediction device provided in the embodiments of this application;

[0042] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0043] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0044] like Figure 1 As shown, Figure 1 This is a flowchart of a wavefront distortion information prediction method provided in an embodiment of this application. The wavefront distortion information prediction method includes the following steps:

[0045] 101. Obtain the original far-field spot measured by the target far-field spot measurement system.

[0046] In this embodiment, the aforementioned target far-field spot measurement system is a far-field spot measurement system that requires wavefront distortion information prediction, and the aforementioned original far-field spot is the spot measured by the target far-field spot measurement system. The original far-field spot of the wavefront-distorted beam is measured by the far-field spot measurement system. This original far-field spot has a corresponding grid number and grid resolution, which are pre-fixed in the intelligent wavefront sensor. The grid number can be set to M×N, and the grid resolution to (Δx0, Δy0). That is, the original far-field spot has a grid number of M×N and a grid resolution of (Δx0, Δy0).

[0047] 102. The original far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot.

[0048] In this embodiment of the application, the first system parameters may include the beam wavelength, focal length, and aperture of the target far-field spot measurement system, and the second system parameters may include the beam wavelength, focal length, and aperture of the standard far-field spot simulation system.

[0049] The original far-field spot can be calibrated by mapping the correlation between the first system parameters and the second system parameters to obtain a standard far-field spot.

[0050] The standard far-field spot is the spot measured by the standard far-field spot simulation system. The standard far-field spot has a corresponding number of grids and grid resolution; the number of grids for the standard far-field spot is m×m, and the grid resolution is (Δxy1, Δxy1).

[0051] By mapping the correlation between the first system parameters and the second system parameters, the original far-field spot (M,N,Δx′0,Δy′0) is calibrated to obtain the standard far-field spot (m,m,Δxy1,Δxy1).

[0052] 103. Input the standard far-field light spot into the trained neural network model for prediction processing to obtain the predicted wavefront distortion information.

[0053] In this embodiment, the trained neural network model is obtained by training based on the calibrated sample standard far-field spot and the corresponding wavefront distortion ground truth value. The sample standard far-field spot has a grid size of m×m and a grid resolution of (Δxy1, Δxy1). The wavefront distortion ground truth value is the actual wavefront distortion value corresponding to the sample standard far-field spot. A training dataset can be constructed using the sample standard far-field spot and the corresponding wavefront distortion ground truth value. A usable neural network model is obtained by iteratively training the training dataset. This usable neural network model is the trained neural network model.

[0054] After obtaining the standard far-field spot of the original far-field spot, the standard far-field spot is input into the trained neural network model. The trained neural network model performs prediction processing on the standard far-field spot and outputs the predicted wavefront distortion information.

[0055] In this embodiment, the original far-field spot measured by the target far-field spot measurement system is obtained; the original far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain a standard far-field spot; the standard far-field spot is input into a trained neural network model for prediction processing to obtain predicted wavefront distortion information. The trained neural network model is trained based on the calibrated sample standard far-field spot and the corresponding wavefront distortion ground truth. By performing different calibration processes on the original far-field spot, original far-field spots with different parameters measured by different far-field spot measurement systems can be calibrated into standard far-field spots. Using the standard far-field spot as the input to the trained neural network model standardizes the input of the neural network model, making it applicable to wavefront distortion information prediction for various measurement systems and improving the prediction efficiency of wavefront distortion information in different scenarios.

[0056] When applied to different far-field spot measurement systems, the embodiments of this application do not require the re-collection of a large amount of simulation or experimental data, nor do they require repeated training of the neural network. This greatly reduces the resource consumption of data acquisition and network training, and improves the timeliness of system application.

[0057] Please refer to Figure 2 , Figure 2 This is a flowchart of another wavefront distortion information prediction method provided in the embodiments of this application, such as... Figure 2As shown, the original far-field spot (2) of the beam containing wavefront distortion is measured by the far-field spot measurement system (1). The original far-field spot (2) has a grid number of M×N and a grid resolution of (Δx0, Δy0). The original far-field spot (2) is input into the calibration model (3) to output the standard far-field spot (4). The standard far-field spot (4) has a grid number of m×m and a grid resolution of Δxy1. During the training phase, the standard far-field spot (4) and the true value of wavefront distortion are used as training datasets, and a usable neural network model (5) is obtained through iterative training. During the usage phase, the standard far-field spot (4) is input into the trained neural network model (5), and the neural network model (5) predicts the wavefront distortion information (6) from the standard far-field spot (4). The embodiments of this application can be applied to intelligent wavefront sensing of different far-field spot measurement systems, and improve the prediction efficiency of wavefront distortion information under different far-field spot measurement systems.

[0058] Optionally, in the step of calibrating the original far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot, the first system parameters of the spot measurement system and the second system parameters of the standard far-field spot simulation system can be obtained; the speckle radius of the original far-field spot can be determined using the first system parameters, and the speckle radius of the standard far-field spot can be determined using the second system parameters; the original far-field spot can be calibrated using the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot to obtain the standard far-field spot.

[0059] In this embodiment of the application, the first system parameters may include the beam wavelength, focal length, and aperture of the target far-field spot measurement system, and the second system parameters may include the beam wavelength, focal length, and aperture of the standard far-field spot simulation system.

[0060] The speckle radius of the original far-field spot and the speckle radius of the standard far-field spot can be determined by looking up tables. A mapping table is pre-established between the first system parameters and the speckle radius of the original far-field spot; by looking up the first system parameters in the mapping table, the speckle radius of the original far-field spot can be obtained. Similarly, a mapping table is pre-established between the second system parameters and the speckle radius of the standard far-field spot; by looking up the second system parameters in the mapping table, the speckle radius of the standard far-field spot can be obtained. Alternatively, besides looking up tables, the speckle radii of the original and standard far-field spots can also be calculated using the following formula:

[0061] R0=1.22λ0f0 / D0

[0062] R1=1.22λ1f1 / D1

[0063] Where R1 is the original far-field spot speckle radius, R0 is the standard far-field spot speckle radius, λ1, f1, D1 are the beam wavelength, focal length, and aperture of the target far-field spot measurement system, respectively, and λ0, f0, D0 are the beam wavelength, focal length, and aperture of the standard far-field spot simulation system, respectively.

[0064] The original far-field spot can be calibrated by comparing its speckle radius with that of the standard far-field spot to obtain a standard far-field spot. Using both the original and standard speckle radii for calibration yields a more accurate standard far-field spot, thus improving prediction accuracy.

[0065] Optionally, in the step of calibrating the original far-field spot using the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot to obtain the standard far-field spot, the first grid number of the original far-field spot and the first resolution of the original far-field spot based on the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot can be determined. The second grid number and the second resolution of the standard far-field spot are then determined. The first grid number and the first resolution of the original far-field spot are interpolated to the second grid number and the second resolution of the standard far-field spot to obtain the standard far-field spot.

[0066] In this embodiment, the first grid number of the original far-field spot is M×N, and the first resolution of the original far-field spot is (Δx′0, Δy′0). The second grid number of the standard far-field spot is m×m, and the second resolution of the standard far-field spot is (Δxy1, Δxy1).

[0067] Specifically, the first resolution of the original far-field spot can be calculated using the following formula:

[0068]

[0069] Using the above formula, the grid resolution of the original far-field spot can be calibrated from (Δx0,Δy0) to the first resolution (Δx′0,Δy′0), and the speckle radius calibrated far-field spot (M,N,Δx′0,Δy′0) can be obtained.

[0070] The speckle radius calibration of the far-field spot (M,N,Δx′0,Δy′0) can be interpolated to (m,m,Δxy1,Δxy1) using the centroid of the original far-field spot as the grid center, to obtain the standard far-field spot (m,m,Δxy1,Δxy1).

[0071] By interpolating the first grid number and first resolution of the original far-field spot to the second grid number and second resolution of the standard far-field spot, a more accurate standard far-field spot can be obtained, thereby improving the accuracy of prediction.

[0072] Optionally, the original far-field spot includes a first far-field spot and a second far-field spot. The first far-field spot is a defocused far-field spot, and the second far-field spot is either a defocused far-field spot or a focused far-field spot. The standard far-field spot includes a first standard far-field spot and a second standard far-field spot. In the step of calibrating the original far-field spot to obtain the standard far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system, the first far-field spot can be calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the first standard far-field spot; the second far-field spot can be calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the second standard far-field spot.

[0073] In this embodiment, an off-focus far-field spot refers to a spot where the focus is not on the target point, while an in-focus far-field spot refers to a spot where the focus is on the target point. The original far-field spot can be either an in-focus far-field spot or an off-focus far-field spot, measured simultaneously by the target far-field spot measurement system, or two off-focus far-field spots. Since the original far-field spot consists of two spots, two standard far-field spots are obtained by simultaneously calibrating the two original far-field spots using a calibration model. The neural network model input consists of two standard far-field spots, which can be composed of one in-focus far-field spot and one off-focus far-field spot, or two off-focus far-field spots.

[0074] Please refer to Figure 3 The figure is a flowchart of another wavefront distortion information prediction method provided in an embodiment of this application, such as... Figure 3 As shown, the target far-field spot measurement system (1) simultaneously measures the in-focus far-field spot and the off-focus far-field spot, or two off-focus far-field spots. The original far-field spot (2) consists of two spots. The calibration model (3) simultaneously calibrates the two original far-field spots to obtain two standard far-field spots (4). The neural network model input consists of two standard far-field spots, which are composed of one in-focus far-field spot and one off-focus far-field spot, or two off-focus far-field spots.

[0075] When one of the two far-field spots in the original far-field spot is a defocused far-field spot, by calibrating the two far-field spots of the original far-field spot, two standard far-field spots can be obtained, which can improve the richness of the input information and thus improve the prediction accuracy of the neural network model.

[0076] Optionally, in the step of inputting the standard far-field spot into the trained neural network model for prediction processing to obtain the predicted wavefront distortion information, the first standard far-field spot and the second standard far-field spot can be input into the trained neural network model for prediction processing to obtain the predicted wavefront distortion information.

[0077] In the embodiments of this application, the first standard far-field spot and the second standard far-field spot are obtained by calibrating two original far-field spots. The two original far-field spots consist of one in-focus far-field spot and one out-of-focus far-field spot, or two out-of-focus far-field spots.

[0078] When one of the two far-field spots in the original far-field spot is a defocused far-field spot, by calibrating the two far-field spots of the original far-field spot, two standard far-field spots can be obtained, which can improve the richness of the input information and thus improve the prediction accuracy of the neural network model.

[0079] It should be noted that the standard far-field light spots in the trained neural network model are also two standard far-field light spots, and the two standard far-field light spots correspond to one wavefront distortion ground truth value.

[0080] Optionally, the predicted wavefront distortion information includes Zernike coefficients. In the step of inputting the standard far-field spot into the trained neural network model for prediction processing to obtain the predicted wavefront distortion information, the standard far-field spot can be input into the trained neural network model for prediction processing, and the Zernike coefficients of a predetermined order can be output.

[0081] In this embodiment, the output wavefront distortion information of the trained neural network model is a two-dimensional wavefront, a Zernike aberration, or an OPD aberration. This embodiment uses Zernike aberration as an example for illustration.

[0082] Please refer to Figure 4 , Figure 4 This is a schematic diagram illustrating wavefront distortion information prediction provided in an embodiment of this application. For example... Figure 4 As shown, the target far-field spot measurement system (1) measures the... Figure 4 The original light spot shown has a grid number of 109×109 and a grid resolution of 4.56μm; the original light spot is calibrated using calibration model (3) to... Figure 4The standard spot shown has a grid number of 128×128 and a grid resolution of 0.1 times the speckle radius. The standard spot is input into the trained neural network model (5) to predict the Zernike coefficients of the 3rd to 9th order. The residual variance rms of the Zernike coefficients of typical frames are 0.13 rad and 0.05 rad, and the corresponding compensation far-field spot Steller ratio SR is 0.64 and 0.09.

[0083] Optionally, before inputting the standard far-field spot into the trained neural network model for prediction processing to obtain the predicted wavefront distortion information, the original far-field spot of the sample beam containing wavefront distortion and the corresponding true value of wavefront distortion can be measured using a far-field spot measurement system. The original far-field spot of the sample is calibrated using the third system parameters of the far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot of the sample. Based on the standard far-field spot of the sample and the corresponding true value of wavefront distortion of the original far-field spot of the sample, the neural network model to be trained is trained. After training, a trained neural network model is obtained. The outputs of both the neural network model to be trained and the trained neural network model are the predicted wavefront distortion information.

[0084] In this embodiment of the application, the original far-field spot of the sample beam containing wavefront distortion measured by each far-field spot measurement system and the true value of the wavefront distortion corresponding to the original far-field spot of the sample can be collected, and each original far-field spot of the sample corresponds to a true value of wavefront distortion.

[0085] All original far-field spots are calibrated to obtain standard far-field spots. Each original far-field spot yields a standard far-field spot. Then, based on the correspondence between the original far-field spots and the ground truth wavefront distortion, the standard far-field spots are associated with the ground truth wavefront distortion to obtain the training dataset. It should be noted that when the original far-field spots consist of an in-focus far-field spot and an out-of-focus far-field spot, or two out-of-focus far-field spots, there are also two standard far-field spots, and each of the two standard far-field spots corresponds to one ground truth wavefront distortion.

[0086] During the training process, supervised training can be used. Specifically, the standard far-field light spot of the sample is input into the neural network model to be trained to obtain the prediction result corresponding to the standard far-field light spot. The loss is calculated by comparing the prediction result corresponding to the standard far-field light spot with the true value of the wavefront distortion corresponding to the standard far-field light spot to obtain the loss value between the prediction result corresponding to the standard far-field light spot and the true value of the wavefront distortion corresponding to the standard far-field light spot. The parameters of the neural network model to be trained are updated through backpropagation of the loss value, and the optimization is iterated with the minimum loss value as the optimization objective. After the iteration is completed, the trained neural network model is obtained.

[0087] A general neural network model can be trained using the standard far-field spot and its corresponding wavefront distortion ground truth. When using it, you only need to calibrate the original far-field spot to the standard far-field spot and input it into the trained neural network model to output the corresponding predicted wavefront distortion information. There is no need to train other neural network models, which improves the prediction efficiency of wavefront distortion information.

[0088] like Figure 5 As shown in the embodiment of this application, a wavefront distortion information prediction device is also provided, characterized in that it includes:

[0089] The acquisition module 501 is used to acquire the original far-field spot measured by the target far-field spot measurement system;

[0090] The calibration processing module 502 is used to calibrate the original far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain a standard far-field spot.

[0091] The prediction module 503 is used to input the standard far-field spot into the trained neural network model for prediction processing to obtain predicted wavefront distortion information. The trained neural network model is trained based on the calibrated sample standard far-field spot and the corresponding wavefront distortion ground truth value.

[0092] Optionally, the calibration processing module 502 is further configured to acquire the first system parameters of the far-field spot measurement system and the second system parameters of the standard far-field spot simulation system; determine the speckle radius of the original far-field spot using the first system parameters and determine the speckle radius of the standard far-field spot using the second system parameters; and perform calibration processing on the original far-field spot using the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot to obtain the standard far-field spot.

[0093] Optionally, the calibration processing module 502 is further configured to determine the first grid number of the original far-field spot, and the first resolution of the original far-field spot based on the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot; determine the second grid number and the second resolution of the standard far-field spot; interpolate the first grid number and the first resolution of the original far-field spot to the second grid number and the second resolution of the standard far-field spot; and obtain the standard far-field spot.

[0094] Optionally, the original far-field spot includes a first far-field spot and a second far-field spot, wherein the first far-field spot is a defocused far-field spot, and the second far-field spot is either a defocused far-field spot or a focused far-field spot. The standard far-field spot includes a first standard far-field spot and a second standard far-field spot. The calibration processing module 502 is further configured to calibrate the first far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the first standard far-field spot; and to calibrate the second far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the second standard far-field spot.

[0095] Optionally, the prediction module 503 is further configured to input the first standard far-field spot and the second standard far-field spot into a trained neural network model for prediction processing to obtain predicted wavefront distortion information.

[0096] Optionally, the predicted wavefront distortion information includes Zernike coefficients, and the prediction module 503 is further used to input the standard far-field spot into a trained neural network model for prediction processing and output Zernike coefficients of a predetermined order.

[0097] Optionally, the wavefront distortion information prediction device further includes:

[0098] The sample acquisition module is used to measure the original far-field spot of the sample containing the wavefront distortion beam and the true value of the wavefront distortion corresponding to the original far-field spot of the sample through the far-field spot measurement system.

[0099] The sample calibration processing module is used to calibrate the original far-field spot of the sample using the third system parameters of the far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot of the sample.

[0100] The model training module is used to train the neural network model to be trained based on the sample standard far-field spot and the wavefront distortion ground value corresponding to the original far-field spot of the sample. After training, a trained neural network model is obtained. The outputs of the neural network model to be trained and the trained neural network model are both predicted wavefront distortion information.

[0101] like Figure 6 As shown in the embodiments of this application, an electronic device is also provided, characterized in that it includes a processor, which can execute any of the above-described wavefront distortion information prediction methods.

[0102] Specifically, it includes a processor 601 and a memory 602, as well as a computer program stored in the memory 602 and capable of running on the processor 601, which executes a wavefront distortion information prediction method, wherein:

[0103] The processor 601 executes the calculator program for the wavefront distortion information prediction method stored in the memory 602, performing the following steps:

[0104] Acquire the original far-field spot measured by the target far-field spot measurement system;

[0105] The original far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot.

[0106] The standard far-field spot is input into the trained neural network model for prediction processing to obtain predicted wavefront distortion information. The trained neural network model is trained based on the calibrated sample standard far-field spot and the corresponding wavefront distortion ground truth value.

[0107] Optionally, the processor 601 performs calibration processing on the original far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain a standard far-field spot, including:

[0108] Obtain the first system parameters of the far-field spot measurement system, and obtain the second system parameters of the standard far-field spot simulation system;

[0109] The speckle radius of the original far-field spot is determined by the first system parameters, and the speckle radius of the standard far-field spot is determined by the second system parameters.

[0110] The original far-field spot is calibrated using the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot to obtain the standard far-field spot.

[0111] Optionally, the calibration process performed by the processor 601, which uses the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot to obtain a standard far-field spot, includes:

[0112] Determine the first grid number of the original far-field spot, and the first resolution of the original far-field spot based on the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot;

[0113] Determine the second grid number and the second resolution of the standard far-field spot;

[0114] The first grid number and the first resolution of the original far-field spot are interpolated to the second grid number and the second resolution of the standard far-field spot to obtain the standard far-field spot.

[0115] Optionally, the original far-field spot includes a first far-field spot and a second far-field spot, wherein the first far-field spot is a defocused far-field spot, and the second far-field spot is either a defocused far-field spot or a focused far-field spot. The standard far-field spot includes a first standard far-field spot and a second standard far-field spot. The calibration process performed by the processor 601 on the original far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot includes:

[0116] The first far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the first standard far-field spot.

[0117] The second far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the second standard far-field spot.

[0118] Optionally, the step of processor 601 inputting the standard far-field light spot into a trained neural network model for prediction processing to obtain predicted wavefront distortion information includes:

[0119] The first standard far-field spot and the second standard far-field spot are input into the trained neural network model for prediction processing to obtain the predicted wavefront distortion information.

[0120] Optionally, the predicted wavefront distortion information includes Zernike coefficients. The processor 601 executes the step of inputting the standard far-field spot into a trained neural network model for prediction processing to obtain the predicted wavefront distortion information, including:

[0121] The standard far-field light spot is input into a trained neural network model for prediction processing, and the Zernike coefficients of a predetermined order are output.

[0122] Optionally, in the step of inputting the standard far-field light spot into a trained neural network model for prediction processing to obtain predicted wavefront distortion information, the method executed by the processor 601 further includes:

[0123] The original far-field spot of the sample containing wavefront distortion and the true value of the wavefront distortion corresponding to the original far-field spot of the sample are measured by a far-field spot measurement system.

[0124] The original far-field spot of the sample is calibrated by the third system parameter of the far-field spot measurement system and the second system parameter of the standard far-field spot simulation system to obtain the standard far-field spot of the sample.

[0125] Based on the standard far-field spot of the sample and the true value of the wavefront distortion corresponding to the original far-field spot of the sample, the neural network model to be trained is trained, and the trained neural network model is obtained after training. The output of the neural network model to be trained and the trained neural network model are both predicted wavefront distortion information.

[0126] This application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the wavefront distortion information prediction method provided in this application and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0127] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0128] The above-disclosed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Therefore, any equivalent variations made in accordance with the claims of this application shall still fall within the scope of this application.

Claims

1. A method for predicting wavefront distortion information, characterized in that, The method includes the following steps: Acquire the original far-field spot measured by the target far-field spot measurement system; The original far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot. The standard far-field spot is input into the trained neural network model for prediction processing to obtain the predicted wavefront distortion information. The trained neural network model is trained based on the calibrated sample standard far-field spot and the corresponding wavefront distortion ground truth. The calibration process of the original far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the standard far-field spot includes: Obtain the first system parameters of the far-field spot measurement system, and obtain the second system parameters of the standard far-field spot simulation system; The speckle radius of the original far-field spot is determined by the first system parameters, and the speckle radius of the standard far-field spot is determined by the second system parameters. The original far-field spot is calibrated using the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot to obtain the standard far-field spot. The calibration process, which uses the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot to obtain the standard far-field spot, includes: Determine the first grid number of the original far-field spot, and the first resolution of the original far-field spot based on the speckle radius of the original far-field spot and the speckle radius of the standard far-field spot; Determine the second grid number and the second resolution of the standard far-field spot; The first grid number and the first resolution of the original far-field spot are interpolated to the second grid number and the second resolution of the standard far-field spot to obtain the standard far-field spot.

2. The wavefront distortion information prediction method as described in claim 1, characterized in that, The original far-field spot includes a first far-field spot and a second far-field spot. The first far-field spot is a defocused far-field spot, and the second far-field spot is either a defocused far-field spot or a focused far-field spot. The standard far-field spot includes a first standard far-field spot and a second standard far-field spot. The standard far-field spot is obtained by calibrating the original far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system. This includes: The first far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the first standard far-field spot. The second far-field spot is calibrated using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain the second standard far-field spot.

3. The wavefront distortion information prediction method as described in claim 2, characterized in that, The step of inputting the standard far-field light spot into a trained neural network model for prediction processing to obtain predicted wavefront distortion information includes: The first standard far-field spot and the second standard far-field spot are input into the trained neural network model for prediction processing to obtain the predicted wavefront distortion information.

4. The wavefront distortion information prediction method as described in claim 1, characterized in that, The predicted wavefront distortion information includes the Zernike coefficient. The step of inputting the standard far-field spot into a trained neural network model for prediction processing to obtain the predicted wavefront distortion information includes: The standard far-field light spot is input into a trained neural network model for prediction processing, and the Zernike coefficients of a predetermined order are output.

5. The wavefront distortion information prediction method as described in claim 1, characterized in that, The method further includes inputting the standard far-field light spot into a trained neural network model for prediction processing to obtain predicted wavefront distortion information. The original far-field spot of the sample containing wavefront distortion and the true value of the wavefront distortion corresponding to the original far-field spot of the sample are measured by a far-field spot measurement system. The original far-field spot of the sample is calibrated by the third system parameter of the far-field spot measurement system and the second system parameter of the standard far-field spot simulation system to obtain the standard far-field spot of the sample. Based on the standard far-field spot of the sample and the true value of the wavefront distortion corresponding to the original far-field spot of the sample, the neural network model to be trained is trained, and the trained neural network model is obtained after training. The output of the neural network model to be trained and the trained neural network model are both predicted wavefront distortion information.

6. A wavefront distortion information prediction device, employing the wavefront distortion information prediction method as described in any one of claims 1-5, characterized in that, The wavefront distortion information prediction device includes: The acquisition module is used to acquire the original far-field spot measured by the target far-field spot measurement system; The calibration processing module is used to calibrate the original far-field spot using the first system parameters of the target far-field spot measurement system and the second system parameters of the standard far-field spot simulation system to obtain a standard far-field spot. The prediction module is used to input the standard far-field spot into the trained neural network model for prediction processing to obtain predicted wavefront distortion information. The trained neural network model is trained based on the calibrated sample standard far-field spot and the corresponding wavefront distortion ground truth value.

7. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps in the wavefront distortion information prediction method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps in the wavefront distortion information prediction method as described in any one of claims 1 to 5.

Images