A wavefront detection method and device based on single-lens scanning

Abstract

The application discloses a kind of based on single-lens scanning wavefront detection method and device, comprising the following steps: S1, generates scanning response feature set;S2, generates standardization scanning response feature set and aperture coordinate set to be reconstructed;S3, standardization scanning response feature set and aperture coordinate set to be reconstructed are input into improved DeepONet network, generate candidate wavefront phase result;S4, to candidate wavefront phase result Single-lens scanning light path back calculation is executed, and prediction spot response result and spot response residual are generated;S5, spot response residual is back projected to aperture coordinate field, and wavefront residual correction amount is generated;S6, candidate wavefront phase result is corrected using wavefront residual correction amount, and to the wavefront slope field after correction Integrable reconstruction is carried out, and continuous wavefront phase diagram is generated;S7, to continuous wavefront phase diagram Wavefront parameter extraction is carried out, and wavefront detection result is output.The application can reduce the complexity of device, improve wavefront reconstruction accuracy and stability.

Description

Technical Field

[0001] This invention relates to the field of optical detection technology, and in particular to a wavefront detection method and apparatus based on single-lens scanning. Background Technology

[0002] Wavefront detection is an important technique in optical system assembly and adjustment, optical component quality evaluation, adaptive optics control, and laser beam quality analysis. The detection results are typically used to characterize the phase distortion state of the beam under test or the optical system. Existing wavefront detection methods mostly employ interferometric detection or microlens array wavefront sensors. Interferometric detection has high requirements for environmental vibration, optical path stability, and reference light quality, resulting in a complex system structure and limited field adaptability. While microlens array wavefront sensors can obtain local wavefront slopes through a beam array, their application is limited in large-aperture, high dynamic range, or low-cost detection scenarios due to limitations in microlens array fabrication accuracy, sub-aperture size, beam crosstalk, and spatial resolution.

[0003] With the development of computational imaging and deep learning technologies, reconstructing continuous wavefronts using a small amount of sampled spot information has become a new research direction. However, existing learning-based wavefront reconstruction methods usually predict wavefront results directly from sampled data, lacking reverse physical verification of the single-lens scanning optical path, making it difficult to guarantee that the prediction results are consistent with the actual spot response. At the same time, problems such as spot distortion, intensity fluctuations, and local abnormal sampling exist during the scanning process, which can easily lead to non-integrable wavefront slope fields or discontinuous reconstructed phases.

[0004] Therefore, how to provide a wavefront detection method and device based on single-lens scanning is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a wavefront detection method and apparatus based on single-lens scanning. This invention performs scanning acquisition using a single lens relative to the aperture under test, utilizing the spot response characteristics at each scanning position to obtain local wavefront information. This avoids reliance on microlens arrays, reducing the structural complexity and fabrication requirements of the detection device. By improving the DeepONet network to generate candidate wavefront phase results, and using a single-lens scanning-accompanied verification structure to perform optical path back-calculation and residual back-projection, the candidate wavefront results can be verified by measured response in the spot observation domain, improving the consistency between the reconstructed results and the actual scanning optical path. Furthermore, an integrable wavefront reconstruction module uses wavefront residual correction to correct the candidate wavefront phase results, suppressing phase breaks or distortions caused by scanning noise, spot anomalies, and local slope inconsistencies, thereby generating a stable, continuous, and physically constrained wavefront phase map, improving wavefront detection accuracy and engineering applicability.

[0006] A wavefront detection method based on single-lens scanning according to an embodiment of the present invention includes the following steps: S1. Control the single lens to perform scanning acquisition relative to the aperture to be measured, acquire the spot image corresponding to each scanning position, extract the spot response features, and generate a set of scanning response features; S2. Perform position calibration and response normalization on the scan response feature set to generate a standardized scan response feature set. Generate multiple aperture query coordinates according to the aperture boundary and aperture sampling interval of the aperture to be measured, and arrange the multiple aperture query coordinates in spatial order to form a set of aperture coordinates to be reconstructed. S3. Input the standardized scan response feature set and the aperture coordinate set to be reconstructed into the improved DeepONet network. The improved DeepONet network includes a scan wavefront operator generation module, an optical path back projection module, and an integrable wavefront reconstruction module. The scan wavefront operator generation module generates candidate wavefront phase results. S4. The optical path back-calculation and back-projection module is equipped with a single-lens scanning accompanying verification structure. The single-lens scanning accompanying verification structure performs single-lens scanning optical path back-calculation on the candidate wavefront phase result, generates a predicted spot response result, and compares the predicted spot response result with the measured spot response feature in the standardized scanning response feature set to obtain the spot response residual. S5. The single-lens scanning accompanying verification structure back-projects the spot response residual along the accompanying direction of the single-lens scanning optical path to the aperture coordinate domain to generate the wavefront residual correction amount. S6. The integrable wavefront reconstruction module uses the wavefront residual correction amount to correct the candidate wavefront phase result, and performs integrable reconstruction on the corrected wavefront slope field to generate a continuous wavefront phase map. S7. The integrable wavefront reconstruction module performs wavefront parameter extraction on the continuous wavefront phase map and outputs the wavefront detection results.

[0007] Optionally, S1 specifically includes: S11. Read the aperture boundary data of the aperture to be measured, and generate a scanning position sequence covering the aperture to be measured according to the aperture boundary data; S12, drive the single lens to move sequentially to each scanning position in the scanning position sequence relative to the aperture to be measured, and trigger the image acquisition unit to acquire the spot image after the single lens reaches the corresponding scanning position; S13. Add a corresponding scanning position mark to each spot image to form a spot image sequence with position marks; S14. Perform gray-level difference between each spot image in the position-marked spot image sequence and the reference background image to obtain the background subtraction image; perform spot intensity threshold segmentation and connected component filtering on the background subtraction image to extract the effective spot region; S15. Perform centroid localization, boundary extraction and intensity distribution statistics on the effective spot area to obtain spot centroid offset, spot size, spot shape skewness and spot intensity response. S16. Associate the spot centroid offset, spot size, spot shape skew, and spot intensity response with the corresponding scanning position markers to generate a set of scanning response features.

[0008] Optionally, S2 specifically includes: S21. Using the aperture center and aperture boundary of the aperture to be measured as coordinate references, convert the scanning position marks corresponding to each spot response feature in the scanning response feature set into aperture position coordinates. S22. Bind the aperture position coordinates with the corresponding spot centroid offset, spot size, spot shape deviation and spot intensity response to form an aperture calibration scanning response feature set; S23. Perform normalization processing based on the maximum amplitude on the spot centroid offset, spot size, spot shape deviation and spot intensity response in the aperture calibration scanning response feature set to generate a standardized scanning response feature set. S24. Generate multiple aperture query coordinates according to the aperture boundary and aperture sampling interval of the aperture to be measured, remove the aperture query coordinates located outside the aperture boundary, and arrange the remaining aperture query coordinates in spatial order to form a set of aperture coordinates to be reconstructed.

[0009] Optionally, S3 specifically includes: S31. The scanning wavefront operator generation module unpacks the fields of the standardized scanning response feature set to obtain the aperture position coordinates, standardized centroid offset, standardized spot size, standardized morphological skew and standardized intensity response corresponding to each scanning position. S32. Using the aperture position coordinates as the spatial positioning result of the scanning nodes, the standardized centroid offset is decomposed into lateral offset components and longitudinal offset components, and the lateral offset components and longitudinal offset components are configured as the local slope response of the corresponding scanning nodes. The scanning slope node map is constructed according to the spatial adjacency relationship between the scanning nodes. S33. Perform spot state verification on each scanning node in the scanning slope node diagram, and compare the standardized spot size, standardized morphological deviation, and standardized intensity response with the corresponding spot state reference range respectively; when the standardized spot size, standardized morphological deviation, or standardized intensity response of any scanning node exceeds the corresponding spot state reference range, mark the corresponding scanning node as a low confidence scanning node. S34. Along the spatial adjacent edges in the scanning slope node diagram, perform differential processing on the lateral offset components and longitudinal offset components of adjacent scanning nodes, and combine the lateral differential results and longitudinal differential results to form local slope change features. S35. According to the spatial order of the scanning nodes, the local slope change features are spliced ​​and encoded with the spot validity markers of the corresponding scanning nodes, and the local slope change features corresponding to low confidence scanning nodes are suppressed to form a scanning wavefront operator coefficient sequence. S36. The scanning wavefront operator generation module encodes the aperture position of the set of aperture coordinates to be reconstructed, converts each aperture query coordinate into a normalized horizontal coordinate, a normalized vertical coordinate, and a normalized radial coordinate relative to the center of the aperture to be measured, and performs position function expansion on the normalized coordinate components to form an aperture position function sequence. S37. Multiply and accumulate the operator coefficients in the scanning wavefront operator coefficient sequence and the position function vectors in the aperture position function sequence according to the corresponding dimensions, and calculate the candidate wavefront phase value corresponding to each aperture query coordinate one by one. S38. According to the spatial arrangement order in the set of aperture coordinates to be reconstructed, fill the candidate wavefront phase values ​​corresponding to each aperture query coordinate back into the aperture coordinate domain to form the candidate wavefront phase results.

[0010] Optionally, S4 specifically includes: S41. Index the candidate wavefront phase results in the aperture domain, determine the local cropping center according to the aperture position coordinates corresponding to each scanning position in the standardized scanning response feature set, and use the effective light transmission range of a single lens as the cropping window. Crop the local phase segment corresponding to the scanning position in the candidate wavefront phase results, and mark the aperture area covered by the cropping window as the scanning sub-aperture area marker. S42. Calculate the adjacent phase difference along the transverse aperture coordinate direction within the local phase segment to form a local transverse phase change sequence; calculate the adjacent phase difference along the longitudinal aperture coordinate direction to form a local longitudinal phase change sequence; perform region convergence on the local transverse phase change sequence and the local longitudinal phase change sequence respectively to obtain the local transverse phase slope and the local longitudinal phase slope, and perform coordinate projection on the local transverse phase slope and the local longitudinal phase slope according to the direction reference of the single lens scanning coordinate system to form the scanning direction slope component; S43. Substitute the slope component of the scanning direction into the slope-spot offset transfer relationship of the single lens scanning optical path, calculate the predicted lateral spot offset and the predicted longitudinal spot offset at the corresponding scanning position, and combine the predicted lateral spot offset and the predicted longitudinal spot offset into the predicted spot centroid offset. S44. Perform phase amplitude statistics and phase change direction statistics on local phase segments to obtain phase fluctuation amplitude and main change direction; determine the predicted spot size and predicted intensity response based on phase fluctuation amplitude, determine the predicted shape skew based on main change direction, and combine the predicted spot size, predicted shape skew, and predicted intensity response into a predicted spot auxiliary response. S45. Using the scan position mark as an index, select the predicted spot response features and measured spot response features corresponding to the same scan position, and pair them item by item according to the centroid offset, spot size, shape deviation and intensity response component type. S46. Perform the difference operation between the measured value and the predicted value for each pair of similar response components to form centroid offset residual, size residual, shape skew residual and intensity response residual, and combine the size residual, shape skew residual and intensity response residual into auxiliary response residual. S47. Encapsulate the centroid offset residual, auxiliary response residual, scanning sub-aperture region marker and the spot validity marker of the corresponding scanning node into a spot residual element. Each spot residual element carries the content of residual value, residual action area and residual confidence status. S48. Arrange the spot residual elements corresponding to each scanning position in sequence according to the scanning position sequence to form a spot response residual composed of multiple spot residual elements and used for accompanying back projection.

[0011] Optionally, S5 specifically includes: S51. Perform residual element analysis on the spot residual elements corresponding to each scanning position to separate the centroid offset residual, auxiliary response residual, scanning sub-aperture region marker and spot validity marker. S52. Based on the scanning sub-aperture region marking, locate the corresponding scanning position in the aperture coordinate domain, and determine the aperture query coordinates in the scanning sub-aperture region as the residual back projection coordinates. S53. The centroid offset residual is divided into the transverse centroid residual and the longitudinal centroid residual. According to the accompanying transmission direction of the single lens scanning optical path, the transverse centroid residual is expanded into a transverse slope correction segment and the longitudinal centroid residual is expanded into a longitudinal slope correction segment. S54. Perform residual amplitude statistics on the auxiliary response residuals to form the auxiliary residual intensity of the corresponding scanning sub-aperture region, and generate residual modulation coefficients by combining the spot validity markers; wherein, the residual modulation coefficients corresponding to low confidence scanning nodes are smaller than the residual modulation coefficients corresponding to normal scanning nodes. S55. Modulate the transverse slope correction segment and the longitudinal slope correction segment respectively using the residual modulation coefficient to form the aperture domain residual segment corresponding to the scanning position. S56. Write the aperture domain residual segments corresponding to each scanning position into the aperture coordinate domain according to the scanning sub-aperture region; when the same aperture query coordinate is covered by multiple aperture domain residual segments, perform weighted synthesis according to the residual modulation coefficients corresponding to each aperture domain residual segment. S57. According to the spatial order of the aperture coordinate set to be reconstructed, the coordinates of the weighted composite aperture domain residual fragments are rearranged to generate wavefront residual correction quantities containing transverse slope correction components and longitudinal slope correction components.

[0012] Optionally, S6 specifically includes: S61. The integrable wavefront reconstruction module reads the candidate wavefront phase results and wavefront residual corrections, and establishes the positional correspondence between the candidate wavefront phase results and wavefront residual corrections according to the spatial order of the aperture coordinate set to be reconstructed. S62. Perform phase change extraction between adjacent aperture query coordinates on the candidate wavefront phase results to obtain the candidate transverse wavefront slope field and the candidate longitudinal wavefront slope field. S63. Extract the transverse slope correction component and the longitudinal slope correction component from the wavefront residual correction. S64. The transverse slope correction component is superimposed onto the candidate transverse wavefront slope field, and the longitudinal slope correction component is superimposed onto the candidate longitudinal wavefront slope field to form the corrected wavefront slope field. S65. Using the candidate wavefront phase results as the initial phase value, perform an integrability consistency check on the corrected wavefront slope field to identify non-integrable quantities in the corrected wavefront slope field that cannot form a continuous phase surface. S66. Perform reduction processing on non-integrable quantities and retain integrable quantities in the modified wavefront slope field to form an integrable wavefront slope field. S67. Using the candidate wavefront phase results as the initial phase value, perform phase accumulation reconstruction on the integrable wavefront slope field along the coordinates of adjacent apertures in the set of aperture coordinates to be reconstructed, and generate a continuous wavefront phase map.

[0013] Optionally, S7 specifically includes: S71. The integrable wavefront reconstruction module reads the continuous wavefront phase map and extracts the effective region within the aperture according to the aperture boundary of the aperture to be measured, thereby obtaining the effective wavefront phase region. S72. Perform reference plane subtraction processing on the phase values ​​in the effective wavefront phase region to obtain the reference-free wavefront phase region. S73. Read the phase values ​​corresponding to the aperture query coordinates in the phase region of the reference wavefront, and perform maximum value extraction, minimum value extraction and mean square statistics on the phase values ​​to generate wavefront peak and valley values ​​and wavefront root mean square values. S74. According to the aperture coordinate system of the aperture to be measured, project the phase value in the phase region of the reference wavefront to the set of aberration basis functions to obtain the corresponding aberration components. S75. Combine the continuous wavefront phase map, the reference wavefront phase region, the wavefront peak and valley values, the wavefront root mean square value, and the aberration components to form the wavefront detection result.

[0014] A wavefront detection device based on single-lens scanning according to an embodiment of the present invention includes the following modules: The scanning acquisition module is used to control the single lens to perform scanning acquisition relative to the aperture to be measured, acquire the spot image corresponding to each scanning position, extract the spot response features, and generate a set of scanning response features; The scanning response standardization module is used to perform position calibration and response normalization processing on the scanning response feature set, generate a standardized scanning response feature set, and generate multiple aperture query coordinates according to the aperture boundary and aperture sampling interval of the aperture to be measured. The multiple aperture query coordinates are arranged in spatial order to form a set of aperture coordinates to be reconstructed. The candidate wavefront generation module is used to input the standardized scan response feature set and the aperture coordinate set to be reconstructed into the improved DeepONet network. The improved DeepONet network includes a scan wavefront operator generation module, an optical path back projection module, and an integrable wavefront reconstruction module. The candidate wavefront phase results are generated by the scan wavefront operator generation module. The optical path back calculation and verification module is used to perform single-lens scanning optical path back calculation on the candidate wavefront phase result through the single-lens scanning accompanying verification structure set in the optical path back calculation and back projection module, generate the predicted spot response result, and compare the predicted spot response result with the measured spot response feature in the standardized scanning response feature set to obtain the spot response residual. The residual-accompanied back-projection module is used to back-project the spot response residual along the accompanying direction of the single-lens scanning optical path to the aperture coordinate domain through the single-lens scanning accompanying verification structure, thereby generating the wavefront residual correction amount. The integrable wavefront reconstruction module is used to correct the candidate wavefront phase results using the wavefront residual correction amount, and to perform integrable reconstruction on the corrected wavefront slope field to generate a continuous wavefront phase map. The wavefront detection output module is used to extract wavefront parameters from continuous wavefront phase maps and output wavefront detection results.

[0015] The beneficial effects of this invention are: 1. This invention performs scanning acquisition with a single lens relative to the aperture to be measured, and obtains local wavefront information by utilizing the light spot response characteristics at different scanning positions. This avoids relying on microlens arrays for parallel sampling, reduces the structural complexity and device fabrication requirements of the wavefront detection device, and is suitable for large aperture, low cost and on-site detection scenarios.

[0016] 2. This invention sets up an optical path back-projection module and a single-lens scanning accompanying verification structure in the improved DeepONet network. It performs single-lens scanning optical path back-projection on the candidate wavefront phase results, compares the predicted spot response results with the measured spot response characteristics using residuals, and then back-projects the spot response residuals to the aperture coordinate domain. This allows the wavefront reconstruction results to undergo physical verification in the spot observation domain, thereby improving the reliability of the detection results.

[0017] 3. This invention uses an integrable wavefront reconstruction module to correct the candidate wavefront phase result using the wavefront residual correction amount, and performs integrable reconstruction on the corrected wavefront slope field. This can suppress phase breaks or distortions caused by scanning noise, spot abnormalities and local slope inconsistencies, and generate a more stable, continuous wavefront phase map that conforms to physical constraints. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is an overall flowchart of a wavefront detection method and device based on single-lens scanning proposed in this invention. Figure 2 This is a schematic diagram of the improved DeepONet network structure in the wavefront detection method and device based on single-lens scanning proposed in this invention. Figure 3 This is a schematic diagram of the device structure of a wavefront detection method and apparatus based on single-lens scanning proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figure 1 and Figure 2 A wavefront detection method based on single-lens scanning includes the following steps: S1. Control the single lens to perform scanning acquisition relative to the aperture to be measured, acquire the spot image corresponding to each scanning position, extract the spot response features, and generate a set of scanning response features; S2. Perform position calibration and response normalization on the scan response feature set to generate a standardized scan response feature set. Generate multiple aperture query coordinates according to the aperture boundary and aperture sampling interval of the aperture to be measured, and arrange the multiple aperture query coordinates in spatial order to form a set of aperture coordinates to be reconstructed. S3. Input the standardized scan response feature set and the aperture coordinate set to be reconstructed into the improved DeepONet network. The improved DeepONet network includes a scan wavefront operator generation module, an optical path back projection module, and an integrable wavefront reconstruction module. The scan wavefront operator generation module generates candidate wavefront phase results. S4. In the optical path back calculation and back projection module, a single-lens scanning accompanying verification structure is set up. The single-lens scanning accompanying verification structure performs single-lens scanning optical path back calculation on the candidate wavefront phase result, generates the predicted spot response result, and compares the predicted spot response result with the measured spot response characteristics in the standardized scanning response feature set to obtain the spot response residual. S5. The single-lens scanning coupled verification structure back-projects the spot response residual along the coupled direction of the single-lens scanning optical path to the aperture coordinate domain, generating the wavefront residual correction amount. S6. The integrable wavefront reconstruction module uses the wavefront residual correction amount to correct the candidate wavefront phase results and performs integrable reconstruction on the corrected wavefront slope field to generate a continuous wavefront phase map. S7, the integrable wavefront reconstruction module performs wavefront parameter extraction on the continuous wavefront phase map and outputs the wavefront detection results.

[0021] In this embodiment, S1 specifically includes: S11. Read the aperture boundary data of the aperture to be measured, and generate a scanning position sequence covering the aperture to be measured according to the aperture boundary data; S12, drive the single lens to move sequentially to each scanning position in the scanning position sequence relative to the aperture to be measured, and trigger the image acquisition unit to acquire the spot image after the single lens reaches the corresponding scanning position; S13. Add a corresponding scanning position mark to each spot image to form a spot image sequence with position marks; S14. Perform gray-level difference between each spot image in the position-marked spot image sequence and the reference background image to obtain the background subtraction image; perform spot intensity threshold segmentation and connected component filtering on the background subtraction image to extract the effective spot region; S15. Perform centroid localization, boundary extraction and intensity distribution statistics on the effective spot area to obtain spot centroid offset, spot size, spot shape skewness and spot intensity response. S16. Associate the spot centroid offset, spot size, spot shape skew, and spot intensity response with the corresponding scanning position markers to generate a set of scanning response features.

[0022] In the specific implementation process, the aperture boundary data of the aperture to be measured is first read. The aperture boundary data includes the aperture center coordinates, aperture boundary contour, aperture radius or equivalent aperture size, effective aperture area mask, and aperture coordinate system orientation reference. The single-lens scanning coverage area is defined by the aperture boundary data, and then a scanning position sequence is generated by combining the effective light transmission range of the single lens and the aperture sampling requirements. The scanning interval can be determined by the effective light transmission diameter of the single lens and the target sampling density; when it is necessary to ensure continuous coverage of adjacent scanning areas, the scanning interval is no greater than the effective light transmission diameter of the single lens; when it is necessary to improve reconstruction stability, the scanning interval is less than the effective light transmission diameter of the single lens, so that adjacent scanning sub-apertures form overlapping coverage. The scanning position sequence can be arranged in a row-by-row, column-by-column, or serpentine scanning manner, and the position outside the aperture boundary is not included in the scanning position sequence.

[0023] After generating the scan position sequence, the single lens is driven to move sequentially to each scan position relative to the aperture to be measured. Once the single lens reaches the current scan position, it triggers the image acquisition unit to acquire the spot image corresponding to that scan position and adds a scan position marker to the spot image, ensuring that the subsequently extracted spot response features remain consistent with the corresponding scan position. A reference background image is acquired under conditions of no effective spot incidence, maintaining the same exposure time, gain, and detector operating parameters as the spot image. Gray-level difference is performed between each spot image and the reference background image to obtain a background subtraction image. Then, spot intensity thresholding is performed on the background subtraction image. The spot intensity threshold can be determined by the mean background gray level and the background gray level fluctuation range of the reference background image, for example, by using an integer multiple of the mean background gray level plus the standard deviation of the background gray level as the threshold. Pixels that meet the spot intensity threshold requirements are selected as candidate spot pixels. Isolated noise regions, regions with areas smaller than the lower limit of the noise area, and regions with areas larger than the upper limit of the reasonable spot area are eliminated through connected component filtering, thereby obtaining the effective spot region.

[0024] After obtaining the effective spot area, spot features that characterize the local wavefront response are extracted from the effective spot area. During centroid localization, the gray value of each pixel within the effective spot area is used as the weight of that pixel's coordinates. The gray-weighted sum of the abscissas of all pixels is then divided by the sum of the gray values ​​of all pixels within the effective spot area to obtain the abscissa of the spot centroid. Similarly, the gray-weighted sum of the ordinates of all pixels is then divided by the sum of the gray values ​​of all pixels within the effective spot area to obtain the ordinate of the spot centroid. The position of the spot centroid is formed by the abscissa and ordinate of the spot centroid, and the coordinate difference between the spot centroid position and the reference spot center position is calculated to obtain the spot centroid offset. The spot size can be determined by the outer boundary size, equivalent area diameter, or width in the horizontal and vertical directions of the effective spot area. The spot morphology skew can be determined by the principal axis direction, horizontal distribution width, and vertical distribution width of the effective spot area. The spot intensity response can be determined by the gray-level peak value, gray-level mean value, or gray-level cumulative value of the effective spot area. Finally, the spot centroid offset, spot size, spot morphology deviation, and spot intensity response are correlated with the scanning position markers to generate a set of scanning response features.

[0025] In this embodiment, S2 specifically includes: S21. Using the aperture center and aperture boundary of the aperture to be measured as coordinate references, convert the scanning position marks corresponding to each spot response feature in the scanning response feature set into aperture position coordinates. S22. Bind the aperture position coordinates with the corresponding spot centroid offset, spot size, spot shape deviation and spot intensity response to form an aperture calibration scanning response feature set; S23. Perform normalization processing based on the maximum amplitude on the spot centroid offset, spot size, spot shape deviation and spot intensity response in the aperture calibration scanning response feature set to generate a standardized scanning response feature set. S24. Generate multiple aperture query coordinates according to the aperture boundary and aperture sampling interval of the aperture to be measured, remove the aperture query coordinates located outside the aperture boundary, and arrange the remaining aperture query coordinates in spatial order to form a set of aperture coordinates to be reconstructed.

[0026] In the specific implementation process, the center of the aperture to be measured is used as the origin of the coordinate system, and the direction reference of the aperture coordinate system is used as the coordinate axis direction. The scanning position mark is converted into aperture position coordinates. The aperture position coordinates are bound to the spot centroid offset, spot size, spot morphology deviation, and spot intensity response at the same scanning position to form a set of aperture calibration scanning response features. In order to reduce the impact of different dimensions and different amplitude ranges on the stability of network input, normalization processing based on the maximum amplitude is performed on each type of spot response feature. Specifically, the maximum amplitude of spot centroid offset, spot size, spot morphology deviation, and spot intensity response within the current scanning batch is calculated, and the maximum amplitude of the same type of feature is used as the normalization reference. The feature value of the same type at each scanning position is divided by the corresponding normalization reference to obtain the standardized response value. When the maximum amplitude of a certain type of feature is zero or less than the lower limit of system noise, the standardized response value of that type is set to zero or replaced by the calibration default value to avoid invalid normalization.

[0027] The set of aperture coordinates to be reconstructed is used to define the output positions of subsequent candidate wavefront phase values ​​and continuous wavefront phase maps. When generating the set of aperture coordinates to be reconstructed, candidate aperture query coordinates are generated in the aperture coordinate system according to the aperture boundary and aperture sampling interval. Coordinate points outside the aperture boundary are removed using an effective aperture area mask, retaining only the effective query coordinates within the aperture boundary. The retained aperture query coordinates can be arranged in ascending order of horizontal coordinates followed by ascending vertical coordinates within the same horizontal coordinate, or in a serpentine order consistent with the scanning position sequence. The sorting method remains consistent throughout the same detection task, ensuring that subsequent candidate wavefront phase values, wavefront residual corrections, and continuous wavefront phase maps have a unified coordinate index.

[0028] In this embodiment, S3 specifically includes: S31. The scanning wavefront operator generation module unpacks the fields of the standardized scanning response feature set to obtain the aperture position coordinates, standardized centroid offset, standardized spot size, standardized morphological skew and standardized intensity response corresponding to each scanning position. S32. Using the aperture position coordinates as the spatial positioning result of the scanning nodes, the standardized centroid offset is decomposed into lateral offset components and longitudinal offset components, and the lateral offset components and longitudinal offset components are configured as the local slope response of the corresponding scanning nodes. The scanning slope node map is constructed according to the spatial adjacency relationship between the scanning nodes. S33. Perform spot state verification on each scanning node in the scanning slope node diagram, and compare the standardized spot size, standardized morphological deviation, and standardized intensity response with the corresponding spot state reference range respectively; when the standardized spot size, standardized morphological deviation, or standardized intensity response of any scanning node exceeds the corresponding spot state reference range, mark the corresponding scanning node as a low confidence scanning node. S34. Along the spatial adjacent edges in the scanning slope node diagram, perform differential processing on the lateral offset components and longitudinal offset components of adjacent scanning nodes, and combine the lateral differential results and longitudinal differential results to form local slope change features. S35. According to the spatial order of the scanning nodes, the local slope change features are spliced ​​and encoded with the spot validity markers of the corresponding scanning nodes, and the local slope change features corresponding to low confidence scanning nodes are suppressed to form a scanning wavefront operator coefficient sequence. S36. The scanning wavefront operator generation module encodes the aperture position of the set of aperture coordinates to be reconstructed, converts each aperture query coordinate into normalized horizontal coordinates, normalized vertical coordinates and normalized radial coordinates relative to the center of the aperture to be measured, and performs position function expansion on the normalized coordinate components to form an aperture position function sequence. S37. Multiply and accumulate the operator coefficients in the scanning wavefront operator coefficient sequence and the position function vectors in the aperture position function sequence according to the corresponding dimensions, and calculate the candidate wavefront phase value corresponding to each aperture query coordinate one by one. S38. According to the spatial arrangement order in the set of aperture coordinates to be reconstructed, fill the candidate wavefront phase values ​​corresponding to each aperture query coordinate back into the aperture coordinate domain to form the candidate wavefront phase results.

[0029] In practical implementation, the improved DeepONet network's scanning wavefront operator generation module is used to complete the operator mapping from the single-lens scanning response to the candidate wavefront phase. The original DeepONet network typically includes a branch network and a backbone network. The branch network encodes the input function sample values, and the backbone network encodes the query coordinates. The function value at the query location is then output through the operator mapping between the two. When this structure is directly used for single-lens scanning wavefront detection, it is difficult to explicitly distinguish the physical meaning of spot centroid shift, spot morphology anomalies, spatial adjacency relationships of scanning nodes, and the aperture query location. This easily allows abnormal spots or local noise to participate in wavefront phase generation. Therefore, the scanning wavefront operator generation module includes a scanning node construction unit, a spot validity marking unit, a local slope encoding unit, an aperture position encoding unit, and a phase operator assembly unit.

[0030] The scanning node construction unit unpacks the standardized scanning response feature set, uses the aperture position coordinates as the spatial position of the scanning node, and decomposes the standardized centroid offset into lateral and longitudinal offset components, writing them as local slope responses into the corresponding scanning nodes. This transforms the unstructured input sample values ​​in a standard DeepONet into a scanning slope node map with aperture spatial relationships. If the scanning positions are regularly arranged, adjacent scanning nodes in the horizontal, vertical, and diagonal directions are connected as spatial adjacency edges; if the scanning positions are irregularly arranged, adjacent nodes are selected according to the spatial distance between them, and spatial adjacency edges are established. The spot validity marking unit compares the standardized spot size, standardized morphological skew, and standardized intensity response with the corresponding spot state reference range, which can be obtained statistically from the scanning samples of an aberration-free reference beam or calibration beam. When the spot size, morphological skew, or intensity response exceeds the corresponding reference range, the corresponding scanning node is marked as a low-confidence scanning node, and the remaining scanning nodes are marked as normal scanning nodes.

[0031] The local slope coding unit performs differential processing on the lateral and longitudinal offset components of adjacent scan nodes along the spatial adjacency edges of the scan slope node graph to form local slope change features. For any scan node, the lateral offset component of the current scan node is subtracted from the lateral offset component of the adjacent scan nodes to obtain the lateral slope change; the longitudinal offset component of the current scan node is subtracted from the longitudinal offset component of the adjacent scan nodes to obtain the longitudinal slope change. When a scan node has multiple adjacent nodes, these are calculated separately according to the adjacency direction and arranged or converged to form the local slope change features of that scan node. During concatenation coding, the local slope change features are used as the main feature components, and the spot validity marker is used as the confidence component, concatenating them to form the node slope coding vector. For normal scan nodes, local slope variation features are encoded using their original values. For low-confidence scan nodes, a low-confidence suppression coefficient is used to reduce the lateral and longitudinal slope variations. The low-confidence suppression coefficient is smaller than the retention coefficient for normal scan nodes and can be obtained statistically from calibration samples or determined jointly by the degree of spot size anomaly, morphological skewness anomaly, and intensity response anomaly. All node slope encoding vectors are arranged in spatial order of the scan nodes, forming a scan wavefront operator coefficient sequence.

[0032] The aperture position encoding unit uses the center of the aperture to be measured as the origin of the coordinate system. It converts each aperture query coordinate into normalized lateral, longitudinal, and radial coordinates. The normalized lateral and longitudinal coordinates are obtained by dividing the corresponding coordinate value by the aperture radius, while the normalized radial coordinate is obtained by dividing the distance from the aperture query coordinate to the aperture center by the aperture radius. Subsequently, the aperture position encoding unit constructs position function vectors in a fixed order: constant term, normalized lateral coordinate term, normalized longitudinal coordinate term, normalized radial coordinate term, lateral coordinate square term, longitudinal coordinate square term, lateral and longitudinal coordinate product term, and radial coordinate square term. The position function vectors corresponding to each aperture query coordinate are arranged according to the spatial order of the aperture coordinate set to be reconstructed, forming an aperture position function sequence. The phase operator assembly unit converts the scanning wavefront operator coefficient sequence into an operator coefficient vector that matches the dimension of the position function vector. Then, it multiplies and sums the operator coefficient vector with the position function vector of each aperture query coordinate according to the corresponding dimension to calculate the candidate wavefront phase value for the corresponding aperture query coordinate. After completing the calculation of all aperture query coordinates, the candidate wavefront phase values ​​are backfilled into the aperture coordinate domain according to the spatial order of the set of aperture coordinates to be reconstructed, forming the candidate wavefront phase result.

[0033] In this embodiment, S4 specifically includes: S41. Index the candidate wavefront phase results in the aperture domain, determine the local cropping center according to the aperture position coordinates corresponding to each scanning position in the standardized scanning response feature set, and use the effective light transmission range of a single lens as the cropping window. Crop the local phase segment corresponding to the scanning position in the candidate wavefront phase results, and mark the aperture area covered by the cropping window as the scanning sub-aperture area marker. S42. Calculate the adjacent phase difference along the transverse aperture coordinate direction within the local phase segment to form a local transverse phase change sequence; calculate the adjacent phase difference along the longitudinal aperture coordinate direction to form a local longitudinal phase change sequence; perform region convergence on the local transverse phase change sequence and the local longitudinal phase change sequence respectively to obtain the local transverse phase slope and the local longitudinal phase slope, and perform coordinate projection on the local transverse phase slope and the local longitudinal phase slope according to the direction reference of the single lens scanning coordinate system to form the scanning direction slope component; S43. Substitute the slope component of the scanning direction into the slope-spot offset transfer relationship of the single lens scanning optical path, calculate the predicted lateral spot offset and the predicted longitudinal spot offset at the corresponding scanning position, and combine the predicted lateral spot offset and the predicted longitudinal spot offset into the predicted spot centroid offset. S44. Perform phase amplitude statistics and phase change direction statistics on local phase segments to obtain phase fluctuation amplitude and main change direction; determine the predicted spot size and predicted intensity response based on phase fluctuation amplitude, determine the predicted shape skew based on main change direction, and combine the predicted spot size, predicted shape skew, and predicted intensity response into a predicted spot auxiliary response. S45. Using the scan position mark as an index, select the predicted spot response features and measured spot response features corresponding to the same scan position, and pair them item by item according to the centroid offset, spot size, shape deviation and intensity response component type. S46. Perform the difference operation between the measured value and the predicted value for each pair of similar response components to form centroid offset residual, size residual, shape skew residual and intensity response residual, and combine the size residual, shape skew residual and intensity response residual into auxiliary response residual. S47. Encapsulate the centroid offset residual, auxiliary response residual, scanning sub-aperture region marker and the spot validity marker of the corresponding scanning node into a spot residual element. Each spot residual element carries the content of residual value, residual action area and residual confidence status. S48. Arrange the spot residual elements corresponding to each scanning position in sequence according to the scanning position sequence to form a spot response residual composed of multiple spot residual elements and used for accompanying back projection.

[0034] In the specific implementation, the single-lens scanning accompanying verification structure is used to verify the optical path consistency of the candidate wavefront phase results. Unlike the previous DeepONet which directly output the candidate wavefront phase results, this implementation re-introduces the candidate wavefront phase results into the single-lens scanning imaging process to determine whether the candidate wavefront phase results can produce a spot shift and spot shape consistent with the measured spot response at each scanning position. When performing local phase segment truncation, the aperture position coordinates corresponding to the current scanning position are used as the truncation center, and the aperture range corresponding to the effective light transmission diameter of the single lens is used as the truncation window. The set of aperture query coordinates covered by the truncation window is the scanning sub-aperture region corresponding to that scanning position, and the scanning sub-aperture region also serves as the area of ​​application for subsequent residual back projection. If the truncation window intersects with the aperture boundary, only the aperture query coordinates falling within the aperture boundary are retained.

[0035] When extracting the phase slope within a local phase segment, the phase values ​​of adjacent aperture query coordinates are compared along the lateral aperture coordinate direction to obtain the lateral phase change sequence; similarly, the phase values ​​of adjacent aperture query coordinates are compared along the longitudinal aperture coordinate direction to obtain the longitudinal phase change sequence. During region convergence, average convergence or weighted convergence is performed on the lateral and longitudinal phase change sequences within the same scanning sub-aperture region, respectively, to obtain the local lateral phase slope and local longitudinal phase slope. The weight of the weighted convergence can be determined according to the distance from the aperture query coordinate to the cut-off center; the closer the aperture query coordinate is to the cut-off center, the higher its weight, thereby reducing the impact of incomplete edge cutting. After obtaining the local lateral and longitudinal phase slopes, coordinate projection is performed based on the directional correspondence between the single-lens scanning coordinate system and the aperture coordinate system, converting the slope component in the aperture coordinate system into the scanning direction slope component in the single-lens scanning coordinate system.

[0036] The slope component of the scanning direction is then substituted into the slope-spot offset transfer relationship of the single-lens scanning optical path to obtain the predicted lateral spot offset and the predicted longitudinal spot offset. The slope-spot offset transfer relationship is jointly determined by the single-lens focal length, detector pixel size, imaging distance, and system calibration coefficients, or it can be obtained through reference wavefront scanning calibration. The predicted spot auxiliary response is determined based on the phase distribution state of the local phase segment. The phase fluctuation amplitude is characterized by the maximum phase value, minimum phase value, or phase dispersion within the local phase segment. When the phase fluctuation amplitude is large, the corresponding predicted spot size increases or the predicted intensity response decreases. The phase change direction is determined by the main phase gradient direction within the local phase segment and is used to determine the predicted shape skew. After completing the predicted spot response calculation, using the scanning position mark as an index, the predicted spot response characteristics at the same scanning position are paired with the measured spot response characteristics for similar components. The centroid offset component is used to calculate the centroid offset residual, and the spot size, shape skew, and intensity response components are used to calculate the auxiliary response residual. The residual calculation uses the method of subtracting the predicted value from the measured value, so that the residual direction can represent the deviation direction of the candidate wavefront phase result relative to the true scanning response.

[0037] To facilitate subsequent backprojection processing, the residuals corresponding to each scanning position are organized into spot residual elements. These spot residual elements include centroid offset residuals, auxiliary response residuals, scanning sub-aperture region markers, and spot validity markers. The centroid offset residual serves as the primary correction factor, used to generate subsequent lateral and longitudinal slope correction segments; the auxiliary response residual is used to adjust the residual correction intensity; the scanning sub-aperture region marker defines the effective range of the residuals when backprojected back to the aperture coordinate domain; and the spot validity marker reduces the impact of low-confidence scanning node residuals on wavefront correction. Through this structured residual representation, the single-lens scanning accompanying verification structure not only calculates the difference between predicted and measured values ​​but also organizes this difference into an aperture domain correction basis that can be directly used in subsequent accompanying backprojection.

[0038] In this embodiment, S5 specifically includes: S51. Perform residual element analysis on the spot residual elements corresponding to each scanning position to separate the centroid offset residual, auxiliary response residual, scanning sub-aperture region marker and spot validity marker. S52. Based on the scanning sub-aperture region marking, locate the corresponding scanning position in the aperture coordinate domain, and determine the aperture query coordinates in the scanning sub-aperture region as the residual back projection coordinates. S53. The centroid offset residual is divided into the transverse centroid residual and the longitudinal centroid residual. According to the accompanying transmission direction of the single lens scanning optical path, the transverse centroid residual is expanded into a transverse slope correction segment and the longitudinal centroid residual is expanded into a longitudinal slope correction segment. S54. Perform residual amplitude statistics on the auxiliary response residuals to form the auxiliary residual intensity of the corresponding scanning sub-aperture region, and generate residual modulation coefficients by combining the spot validity markers; wherein, the residual modulation coefficients corresponding to low confidence scanning nodes are smaller than the residual modulation coefficients corresponding to normal scanning nodes. S55. Modulate the transverse slope correction segment and the longitudinal slope correction segment respectively using the residual modulation coefficient to form the aperture domain residual segment corresponding to the scanning position. S56. Write the aperture domain residual segments corresponding to each scanning position into the aperture coordinate domain according to the scanning sub-aperture region; when the same aperture query coordinate is covered by multiple aperture domain residual segments, perform weighted synthesis according to the residual modulation coefficients corresponding to each aperture domain residual segment. S57. According to the spatial order of the aperture coordinate set to be reconstructed, the coordinates of the weighted composite aperture domain residual fragments are rearranged to generate wavefront residual correction quantities containing transverse slope correction components and longitudinal slope correction components.

[0039] In the specific implementation process, after the single-lens scanning accompanying verification structure completes the construction of the spot residual element, it inversely converts the residual in the spot observation domain into slope correction information in the aperture wavefront domain. In each spot residual element, the centroid offset residual includes the transverse centroid residual and the longitudinal centroid residual, which are used to characterize the offset direction and offset magnitude of the predicted spot centroid relative to the measured spot centroid; the auxiliary response residual is obtained by combining the size residual, shape skew residual, and intensity response residual, which are used to characterize the comprehensive deviation of spot diffusion, stretching, and intensity change at the scanning position; the scanning sub-aperture region marker is used to determine the effective range of the residual written back to the aperture coordinate domain; and the spot validity marker is used to determine whether the residual element comes from a low-confidence scanning node.

[0040] During the back-projection process, the scanned sub-aperture region is first used as an index to locate the corresponding scanned sub-aperture region in the aperture coordinate domain. The aperture lookup coordinates within this region are then used as the coordinates for residual back-projection. For centroid offset residuals, the lateral centroid residuals are expanded into lateral slope correction segments along the lateral accompanying propagation direction of the single-lens scanning optical path, and the longitudinal centroid residuals are expanded into longitudinal slope correction segments along the longitudinal accompanying propagation direction. During expansion, the aperture lookup coordinates within the same scanned sub-aperture region share the residual direction of that scanned position, and spatial attenuation is set according to the distance from the aperture lookup coordinates to the center of the scanned sub-aperture. This ensures that coordinates closer to the center of the sub-aperture receive a larger correction contribution, while coordinates closer to the edge of the sub-aperture receive a smaller correction contribution.

[0041] The auxiliary response residual is not directly written back as a phase value, but is used to adjust the participation level of the slope correction segment. Specifically, amplitude statistics are performed on the size residual, shape skew residual, and intensity response residual to obtain the auxiliary residual intensity; this is then combined with the spot validity marker to generate the residual modulation coefficient. When the spot validity marker represents a normal scan node, the residual modulation coefficient maintains a high value, allowing the corresponding residual segment to fully participate in the correction; when the spot validity marker represents a low-confidence scan node, the residual modulation coefficient decreases, allowing the residual segment corresponding to that scan node to participate in the correction only at a small proportion. Subsequently, the single-lens scanning accompanying verification structure uses the residual modulation coefficient to modulate the transverse slope correction segment and the longitudinal slope correction segment respectively, forming the aperture domain residual segment at the corresponding scan position.

[0042] The aperture domain residual segments include lateral slope correction and longitudinal slope correction components, and are written into the aperture coordinate domain according to the scanned sub-aperture regions. Since the effective light transmission range of adjacent scan positions may overlap during single-lens scanning, multiple aperture domain residual segments may be received for the same aperture query coordinate. In this case, the residual modulation coefficients corresponding to each aperture domain residual segment are used as the synthesis weights to weight and synthesize multiple lateral slope corrections and multiple longitudinal slope corrections at the same aperture query coordinate. After writing the aperture domain residual segments for all scan positions, the synthesized results are rearranged according to the spatial order of the aperture coordinate set to be reconstructed, generating the wavefront residual correction. This wavefront residual correction includes lateral slope correction and longitudinal slope correction components, which can be directly superimposed onto the candidate wavefront slope field in the subsequent integrable wavefront reconstruction module.

[0043] In this embodiment, S6 specifically includes: S61. The integrable wavefront reconstruction module reads the candidate wavefront phase results and wavefront residual corrections, and establishes the positional correspondence between the candidate wavefront phase results and wavefront residual corrections according to the spatial order of the aperture coordinate set to be reconstructed. S62. Perform phase change extraction between adjacent aperture query coordinates on the candidate wavefront phase results to obtain the candidate transverse wavefront slope field and the candidate longitudinal wavefront slope field. S63. Extract the transverse slope correction component and the longitudinal slope correction component from the wavefront residual correction. S64. The transverse slope correction component is superimposed onto the candidate transverse wavefront slope field, and the longitudinal slope correction component is superimposed onto the candidate longitudinal wavefront slope field to form the corrected wavefront slope field. S65. Using the candidate wavefront phase results as the initial phase value, perform an integrability consistency check on the corrected wavefront slope field to identify non-integrable quantities in the corrected wavefront slope field that cannot form a continuous phase surface. S66. Perform reduction processing on non-integrable quantities and retain integrable quantities in the modified wavefront slope field to form an integrable wavefront slope field. S67. Using the candidate wavefront phase results as the initial phase value, perform phase accumulation reconstruction on the integrable wavefront slope field along the coordinates of adjacent apertures in the set of aperture coordinates to be reconstructed, and generate a continuous wavefront phase map.

[0044] In practical implementation, the integrable wavefront reconstruction module is used to convert the wavefront residual correction obtained from the accompanying back projection into a continuous wavefront phase map. The previous DeepONet network typically outputs the candidate wavefront phase results directly after generation, lacking a check for the integrability of the wavefront slope field. When there is scanning noise, low-confidence scan node residuals, or inconsistent overlap of multiple scan sub-aperture residuals, the candidate wavefront phase results are prone to problems such as local phase abrupt changes, inconsistent accumulation paths, or discontinuous phase surfaces. This implementation uses the integrable wavefront reconstruction module to first convert the candidate wavefront phase results into a slope field, then uses the wavefront residual correction to perform slope correction, and finally performs integrability processing on the corrected slope field.

[0045] The integrable wavefront reconstruction module first aligns the candidate wavefront phase results with the wavefront residual corrections according to the spatial order of the aperture coordinate set to be reconstructed, ensuring that each aperture query coordinate corresponds simultaneously to the candidate phase value, the transverse slope correction component, and the longitudinal slope correction component. Then, it extracts the transverse and longitudinal phase differences of the candidate phase values ​​between adjacent coordinates in the aperture coordinate set to be reconstructed, forming candidate transverse wavefront slope fields and candidate longitudinal wavefront slope fields. After completing the slope field representation, the integrable wavefront reconstruction module superimposes the transverse slope correction components onto the candidate transverse wavefront slope field and the longitudinal slope correction components onto the candidate longitudinal wavefront slope field, forming the corrected wavefront slope field.

[0046] Subsequently, an integrability consistency check is performed. Specifically, a local closed cell composed of adjacent aperture query coordinates is selected in the aperture coordinate domain. Phase increment accumulation is performed on the corrected transverse slope component and longitudinal slope component along different accumulation paths of the local closed cell. When the difference in phase increment obtained from different paths exceeds the consistency limit, the corresponding slope component within the local closed cell is marked as a non-integrable quantity. The consistency limit can be determined by the aberration-free reference beam, the calibration beam scan sample, or the phase reconstruction error allowed by the system. For the identified non-integrable quantities, the integrable wavefront reconstruction module performs a reduction process. The reduction process does not directly delete the corresponding slope component, but sets a reduction coefficient according to the magnitude of the path difference corresponding to the non-integrable quantity, and proportionally reduces the transverse slope component and longitudinal slope component corresponding to the non-integrable quantity; the larger the path difference, the smaller the reduction coefficient. For slope components not marked as non-integrable quantities, they are retained with their original values. After the reduction process, the retained slope components form the integrable wavefront slope field.

[0047] In the phase accumulation and reconstruction stage, the integrable wavefront reconstruction module uses the reference aperture query coordinate phase value from the candidate wavefront phase results as the initial phase value, and gradually accumulates the integrable wavefront slope field along the spatial adjacency relationship in the set of aperture coordinates to be reconstructed. For adjacent aperture query coordinates, according to the lateral or longitudinal adjacency direction between them, the corresponding slope component is accumulated onto the phase value of the previous aperture query coordinate to obtain the phase value of the next aperture query coordinate. When the same aperture query coordinate can be reached by multiple accumulation paths, consistent synthesis is performed on the phase values ​​obtained from multiple paths. The synthesis method can be average synthesis or weighted synthesis according to path confidence. After completing the phase accumulation of all aperture query coordinates, a continuous wavefront phase map is obtained.

[0048] In this embodiment, S7 specifically includes: S71. The integrable wavefront reconstruction module reads the continuous wavefront phase map and extracts the effective region within the aperture according to the aperture boundary of the aperture to be measured, thereby obtaining the effective wavefront phase region. S72. Perform reference plane subtraction processing on the phase values ​​in the effective wavefront phase region to obtain the reference-free wavefront phase region. S73. Read the phase values ​​corresponding to the aperture query coordinates in the phase region of the reference wavefront, and perform maximum value extraction, minimum value extraction and mean square statistics on the phase values ​​to generate wavefront peak and valley values ​​and wavefront root mean square values. S74. According to the aperture coordinate system of the aperture to be measured, project the phase value in the phase region of the reference wavefront to the set of aberration basis functions to obtain the corresponding aberration components. S75. Combine the continuous wavefront phase map, the reference wavefront phase region, the wavefront peak and valley values, the wavefront root mean square value, and the aberration components to form the wavefront detection result.

[0049] In the specific implementation process, after the continuous wavefront phase map is generated, the effective region within the aperture is first extracted according to the aperture boundary of the aperture to be measured, and invalid phase data outside the aperture boundary is removed to obtain the effective wavefront phase region. To avoid the influence of overall tilt, installation deviation, or reference surface offset on the wavefront evaluation results, a reference surface subtraction process is performed on the effective wavefront phase region. The reference surface can be determined by the average phase value in the effective wavefront phase region, the least squares fitting plane, or the reference calibration surface; by subtracting the corresponding reference surface from the phase values ​​in the effective wavefront phase region, the reference-deferred wavefront phase region is obtained.

[0050] In the dereferenced wavefront phase region, the maximum and minimum phase values ​​are extracted, and the difference between them is used as the wavefront peak-valley value. Simultaneously, mean square statistics are performed on the phase values ​​within the dereferenced wavefront phase region to generate the wavefront root mean square value. The wavefront peak-valley value characterizes the overall fluctuation range of the wavefront phase, while the wavefront root mean square value characterizes the degree of wavefront phase deviation. Further, according to the aperture coordinate system of the aperture under test, the phase values ​​in the dereferenced wavefront phase region are projected onto an aberration basis function set. This set can include basis functions corresponding to defocus, astigmatism, coma, spherical aberration, and higher-order aberrations. During projection, the phase values ​​at each query coordinate within the aperture are matched and calculated with the corresponding aberration basis functions to obtain the coefficients of each aberration component, thereby converting the continuous wavefront phase map into an aberration representation that facilitates optical evaluation and system adjustment. Finally, the continuous wavefront phase map, the dereferenced wavefront phase region, the wavefront peak-valley value, the wavefront root mean square value, and the aberration components are combined to form the wavefront detection result.

[0051] refer to Figure 3 A wavefront detection device based on single-lens scanning includes the following modules: The scanning acquisition module is used to control the single lens to perform scanning acquisition relative to the aperture to be measured, acquire the spot image corresponding to each scanning position, extract the spot response features, and generate a set of scanning response features; The scanning response standardization module is used to perform position calibration and response normalization processing on the scanning response feature set, generate a standardized scanning response feature set, and generate multiple aperture query coordinates according to the aperture boundary and aperture sampling interval of the aperture to be measured. The multiple aperture query coordinates are arranged in spatial order to form a set of aperture coordinates to be reconstructed. The candidate wavefront generation module is used to input the standardized scan response feature set and the aperture coordinate set to be reconstructed into the improved DeepONet network. The improved DeepONet network includes a scan wavefront operator generation module, an optical path back projection module, and an integrable wavefront reconstruction module. The candidate wavefront phase results are generated by the scan wavefront operator generation module. The optical path back calculation and verification module is used to perform single-lens scanning optical path back calculation on the candidate wavefront phase result through the single-lens scanning accompanying verification structure set in the optical path back calculation and back projection module, generate the predicted spot response result, and compare the predicted spot response result with the measured spot response feature in the standardized scanning response feature set to obtain the spot response residual. The residual-accompanied back-projection module is used to back-project the spot response residual along the accompanying direction of the single-lens scanning optical path to the aperture coordinate domain through the single-lens scanning accompanying verification structure, thereby generating the wavefront residual correction amount. The integrable wavefront reconstruction module is used to correct the candidate wavefront phase results using the wavefront residual correction amount, and to perform integrable reconstruction on the corrected wavefront slope field to generate a continuous wavefront phase map. The wavefront detection output module is used to extract wavefront parameters from continuous wavefront phase maps and output wavefront detection results.

[0052] Example 1: To verify the feasibility of this invention in practice, it was applied to a laser beam quality inspection platform. This platform is used to detect the wavefront distortion state of a aligned laser beam after passing through an optical window and beam expander. In field testing, traditional microlens array wavefront sensors suffer from limitations in the number of sub-apertures, beam crosstalk, and high device costs in large-aperture detection scenarios. While ordinary learning-based wavefront reconstruction methods can output wavefront results from sampled data, they typically lack reverse verification of the actual beam response, making them prone to phase breaks and reconstruction errors when local beam anomalies or scanning noise are present.

[0053] During testing, the beam to be tested is introduced into the wavefront detection platform, and a single lens performs a two-dimensional scan relative to the aperture to be tested. The effective light-passing diameter of the single lens is set to 2 mm, the scanning interval is set to 1 mm, and the image acquisition unit uses an industrial camera to acquire spot images corresponding to each scanning position. First, the aperture boundary data of the aperture to be tested is read to generate a sequence of scanning positions covering the aperture to be tested; then, the single lens is driven to move sequentially to each scanning position and acquire the corresponding spot image. After background subtraction, spot intensity threshold segmentation, and connected component filtering of the spot images, the spot centroid offset, spot size, spot morphology skew, and spot intensity response are extracted to generate a set of scanning response features.

[0054] After obtaining the scan response feature set, aperture position calibration and normalization based on the maximum amplitude are performed on each spot response feature to generate a standardized scan response feature set. Simultaneously, a set of coordinates for the aperture to be reconstructed is generated according to the aperture boundary and aperture sampling interval of the aperture to be measured. Then, the standardized scan response feature set and the set of coordinates for the aperture to be reconstructed are input into the improved DeepONet network. The scan wavefront operator generation module in the improved DeepONet network maps the scan nodes, uses standardized centroid offset to form a local slope response, and combines spot validity markers and local slope change features to generate a scan wavefront operator coefficient sequence. Finally, through the assembly of the aperture position function sequence and the phase operator, candidate wavefront phase results are generated.

[0055] After the candidate wavefront phase results are generated, the single-lens scanning adjoint verification structure in the optical path back-projection module performs single-lens scanning optical path back-calculation on the candidate wavefront phase results. Specifically, a local phase segment is cropped for each scanning position, the local transverse phase slope and local longitudinal phase slope are extracted, and a predicted spot response result is generated based on the slope-spot offset transfer relationship. Then, the predicted spot response result and the measured spot response characteristics are compared with the residual to calculate the residual, construct the spot residual element, and back-project the spot response residual along the adjoint direction of the single-lens scanning optical path to the aperture coordinate domain to generate the wavefront residual correction amount. Finally, the integrable wavefront reconstruction module uses the wavefront residual correction amount to correct the candidate wavefront slope field, and performs integrable consistency check and phase accumulation reconstruction on the corrected wavefront slope field to generate a continuous wavefront phase map, outputting the wavefront peak and valley values, wavefront root mean square value, and aberration components.

[0056] To verify the detection effect of this invention, 120 sets of wavefront samples were selected for testing, including those with slight defocus, astigmatism, coma, local thermal distortion, and complex aberrations. Using the interferometer measurement results as the reference wavefront, the traditional microlens array wavefront sensing method, the ordinary DeepONet wavefront reconstruction method, and the method of this invention were compared. The statistical results are shown in Table 1 below.

[0057] Table 1 Comparison of Wavefront Detection Results

[0058] As shown in Table 1 above, the method of this invention outperforms both the traditional microlens array wavefront sensing method and the ordinary DeepONet wavefront reconstruction method in terms of RMS error, PV error, and mean residual of the spot response. The traditional microlens array wavefront sensing method is affected by sub-aperture sampling density and spot crosstalk, resulting in insufficient ability to represent local wavefront changes. The ordinary DeepONet wavefront reconstruction method can improve the continuous wavefront reconstruction capability, but due to the lack of back-verification of the spot observation domain, reconstruction bias still exists under conditions of spot distortion or local anomalous sampling. The method of this invention obtains the local spot response through single-lens scanning and introduces a single-lens scanning-accompanied verification structure into the improved DeepONet network. This allows the candidate wavefront phase results to undergo optical path back-calculation, residual construction, and accompanying back-projection correction, thereby making the final wavefront phase map more consistent with the actual spot response.

[0059] This embodiment demonstrates that the present invention can complete continuous wavefront phase reconstruction using the spot response characteristics acquired by single-lens scanning without relying on a microlens array. Through the coordinated processing of the scanning wavefront operator generation module, the optical path back-projection module, and the integrable wavefront reconstruction module, the present invention can suppress the influence of spot anomalies, scanning noise, and local slope inconsistencies on the wavefront reconstruction results, improve the stability and physical consistency of the wavefront detection results, and is suitable for applications such as laser beam quality detection, optical system assembly and adjustment, and large-aperture wavefront detection.

[0060] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A wavefront detection method based on single-lens scanning, characterized in that, Includes the following steps: S1. Control the single lens to perform scanning acquisition relative to the aperture to be measured, acquire the spot image corresponding to each scanning position, extract the spot response features, and generate a set of scanning response features; S2. Perform position calibration and response normalization on the scan response feature set to generate a standardized scan response feature set. Generate multiple aperture query coordinates according to the aperture boundary and aperture sampling interval of the aperture to be measured, and arrange the multiple aperture query coordinates in spatial order to form a set of aperture coordinates to be reconstructed. S3. Input the standardized scan response feature set and the aperture coordinate set to be reconstructed into the improved DeepONet network. The improved DeepONet network includes a scan wavefront operator generation module, an optical path back projection module, and an integrable wavefront reconstruction module. The scan wavefront operator generation module generates candidate wavefront phase results. S4. The optical path back-calculation and back-projection module is equipped with a single-lens scanning accompanying verification structure. The single-lens scanning accompanying verification structure performs single-lens scanning optical path back-calculation on the candidate wavefront phase result, generates a predicted spot response result, and compares the predicted spot response result with the measured spot response feature in the standardized scanning response feature set to obtain the spot response residual. S5. The single-lens scanning accompanying verification structure back-projects the spot response residual along the accompanying direction of the single-lens scanning optical path to the aperture coordinate domain to generate the wavefront residual correction amount. S6. The integrable wavefront reconstruction module uses the wavefront residual correction amount to correct the candidate wavefront phase result, and performs integrable reconstruction on the corrected wavefront slope field to generate a continuous wavefront phase map. S7. The integrable wavefront reconstruction module performs wavefront parameter extraction on the continuous wavefront phase map and outputs the wavefront detection results.

2. The wavefront detection method based on single-lens scanning according to claim 1, characterized in that, S1 specifically includes: S11. Read the aperture boundary data of the aperture to be measured, and generate a scanning position sequence covering the aperture to be measured according to the aperture boundary data; S12, drive the single lens to move sequentially to each scanning position in the scanning position sequence relative to the aperture to be measured, and trigger the image acquisition unit to acquire the spot image after the single lens reaches the corresponding scanning position; S13. Add a corresponding scanning position mark to each spot image to form a spot image sequence with position marks; S14. Perform gray-level difference between each spot image in the position-marked spot image sequence and the reference background image to obtain the background subtraction image; perform spot intensity threshold segmentation and connected component filtering on the background subtraction image to extract the effective spot region; S15. Perform centroid localization, boundary extraction and intensity distribution statistics on the effective spot area to obtain spot centroid offset, spot size, spot shape skewness and spot intensity response. S16. Associate the spot centroid offset, spot size, spot shape skew, and spot intensity response with the corresponding scanning position markers to generate a set of scanning response features.

3. The wavefront detection method based on single-lens scanning according to claim 1, characterized in that, S2 specifically includes: S21. Using the aperture center and aperture boundary of the aperture to be measured as coordinate references, convert the scanning position marks corresponding to each spot response feature in the scanning response feature set into aperture position coordinates. S22. Bind the aperture position coordinates with the corresponding spot centroid offset, spot size, spot shape deviation and spot intensity response to form an aperture calibration scanning response feature set; S23. Perform normalization processing based on the maximum amplitude on the spot centroid offset, spot size, spot shape deviation and spot intensity response in the aperture calibration scanning response feature set to generate a standardized scanning response feature set. S24. Generate multiple aperture query coordinates according to the aperture boundary and aperture sampling interval of the aperture to be measured, remove the aperture query coordinates located outside the aperture boundary, and arrange the remaining aperture query coordinates in spatial order to form a set of aperture coordinates to be reconstructed.

4. The wavefront detection method based on single-lens scanning according to claim 1, characterized in that, S3 specifically includes: S31. The scanning wavefront operator generation module unpacks the fields of the standardized scanning response feature set to obtain the aperture position coordinates, standardized centroid offset, standardized spot size, standardized morphological skew and standardized intensity response corresponding to each scanning position. S32. Using the aperture position coordinates as the spatial positioning result of the scanning nodes, the standardized centroid offset is decomposed into lateral offset components and longitudinal offset components, and the lateral offset components and longitudinal offset components are configured as the local slope response of the corresponding scanning nodes. The scanning slope node map is constructed according to the spatial adjacency relationship between the scanning nodes. S33. Perform spot state verification on each scanning node in the scanning slope node diagram, and compare the standardized spot size, standardized morphological deviation, and standardized intensity response with the corresponding spot state reference range respectively; when the standardized spot size, standardized morphological deviation, or standardized intensity response of any scanning node exceeds the corresponding spot state reference range, mark the corresponding scanning node as a low confidence scanning node. S34. Along the spatial adjacent edges in the scanning slope node diagram, perform differential processing on the lateral offset components and longitudinal offset components of adjacent scanning nodes, and combine the lateral differential results and longitudinal differential results to form local slope change features. S35. According to the spatial order of the scanning nodes, the local slope change features are spliced ​​and encoded with the spot validity markers of the corresponding scanning nodes, and the local slope change features corresponding to low confidence scanning nodes are suppressed to form a scanning wavefront operator coefficient sequence. S36. The scanning wavefront operator generation module encodes the aperture position of the set of aperture coordinates to be reconstructed, converts each aperture query coordinate into a normalized horizontal coordinate, a normalized vertical coordinate, and a normalized radial coordinate relative to the center of the aperture to be measured, and performs position function expansion on the normalized coordinate components to form an aperture position function sequence. S37. Multiply and accumulate the operator coefficients in the scanning wavefront operator coefficient sequence and the position function vectors in the aperture position function sequence according to the corresponding dimensions, and calculate the candidate wavefront phase value corresponding to each aperture query coordinate one by one. S38. According to the spatial arrangement order in the set of aperture coordinates to be reconstructed, fill the candidate wavefront phase values ​​corresponding to each aperture query coordinate back into the aperture coordinate domain to form the candidate wavefront phase results.

5. The wavefront detection method based on single-lens scanning according to claim 1, characterized in that, S4 specifically includes: S41. Index the candidate wavefront phase results in the aperture domain, determine the local cropping center according to the aperture position coordinates corresponding to each scanning position in the standardized scanning response feature set, and use the effective light transmission range of a single lens as the cropping window. Crop the local phase segment corresponding to the scanning position in the candidate wavefront phase results, and mark the aperture area covered by the cropping window as the scanning sub-aperture area marker. S42. Calculate the adjacent phase difference along the transverse aperture coordinate direction within the local phase segment to form a local transverse phase change sequence; calculate the adjacent phase difference along the longitudinal aperture coordinate direction to form a local longitudinal phase change sequence; perform region convergence on the local transverse phase change sequence and the local longitudinal phase change sequence respectively to obtain the local transverse phase slope and the local longitudinal phase slope, and perform coordinate projection on the local transverse phase slope and the local longitudinal phase slope according to the direction reference of the single lens scanning coordinate system to form the scanning direction slope component; S43. Substitute the slope component of the scanning direction into the slope-spot offset transfer relationship of the single lens scanning optical path, calculate the predicted lateral spot offset and the predicted longitudinal spot offset at the corresponding scanning position, and combine the predicted lateral spot offset and the predicted longitudinal spot offset into the predicted spot centroid offset. S44. Perform phase amplitude statistics and phase change direction statistics on local phase segments to obtain phase fluctuation amplitude and main change direction; determine the predicted spot size and predicted intensity response based on phase fluctuation amplitude, determine the predicted shape skew based on main change direction, and combine the predicted spot size, predicted shape skew, and predicted intensity response into a predicted spot auxiliary response. S45. Using the scan position mark as an index, select the predicted spot response features and measured spot response features corresponding to the same scan position, and pair them item by item according to the centroid offset, spot size, shape deviation and intensity response component type. S46. Perform the difference operation between the measured value and the predicted value for each pair of similar response components to form centroid offset residual, size residual, shape skew residual and intensity response residual, and combine the size residual, shape skew residual and intensity response residual into auxiliary response residual. S47. Encapsulate the centroid offset residual, auxiliary response residual, scanning sub-aperture region marker and the spot validity marker of the corresponding scanning node into a spot residual element. Each spot residual element carries the content of residual value, residual action area and residual confidence status. S48. Arrange the spot residual elements corresponding to each scanning position in sequence according to the scanning position sequence to form a spot response residual composed of multiple spot residual elements and used for accompanying back projection.

6. The wavefront detection method based on single-lens scanning according to claim 1, characterized in that, S5 specifically includes: S51. Perform residual element analysis on the spot residual elements corresponding to each scanning position to separate the centroid offset residual, auxiliary response residual, scanning sub-aperture region marker and spot validity marker. S52. Based on the scanning sub-aperture region marking, locate the corresponding scanning position in the aperture coordinate domain, and determine the aperture query coordinates in the scanning sub-aperture region as the residual back projection coordinates. S53. The centroid offset residual is divided into the transverse centroid residual and the longitudinal centroid residual. According to the accompanying transmission direction of the single lens scanning optical path, the transverse centroid residual is expanded into a transverse slope correction segment and the longitudinal centroid residual is expanded into a longitudinal slope correction segment. S54. Perform residual amplitude statistics on the auxiliary response residuals to form the auxiliary residual intensity of the corresponding scanning sub-aperture region, and generate residual modulation coefficients by combining the spot validity markers; wherein, the residual modulation coefficients corresponding to low confidence scanning nodes are smaller than the residual modulation coefficients corresponding to normal scanning nodes. S55. Modulate the transverse slope correction segment and the longitudinal slope correction segment respectively using the residual modulation coefficient to form the aperture domain residual segment corresponding to the scanning position. S56. Write the aperture domain residual segments corresponding to each scanning position into the aperture coordinate domain according to the scanning sub-aperture region; when the same aperture query coordinate is covered by multiple aperture domain residual segments, perform weighted synthesis according to the residual modulation coefficients corresponding to each aperture domain residual segment. S57. According to the spatial order of the aperture coordinate set to be reconstructed, the coordinates of the weighted composite aperture domain residual fragments are rearranged to generate wavefront residual correction quantities containing transverse slope correction components and longitudinal slope correction components.

7. The wavefront detection method based on single-lens scanning according to claim 1, characterized in that, S6 specifically includes: S61. The integrable wavefront reconstruction module reads the candidate wavefront phase results and wavefront residual corrections, and establishes the positional correspondence between the candidate wavefront phase results and wavefront residual corrections according to the spatial order of the aperture coordinate set to be reconstructed. S62. Perform phase change extraction between adjacent aperture query coordinates on the candidate wavefront phase results to obtain the candidate transverse wavefront slope field and the candidate longitudinal wavefront slope field. S63. Extract the transverse slope correction component and the longitudinal slope correction component from the wavefront residual correction. S64. The transverse slope correction component is superimposed onto the candidate transverse wavefront slope field, and the longitudinal slope correction component is superimposed onto the candidate longitudinal wavefront slope field to form the corrected wavefront slope field. S65. Using the candidate wavefront phase results as the initial phase value, perform an integrability consistency check on the corrected wavefront slope field to identify non-integrable quantities in the corrected wavefront slope field that cannot form a continuous phase surface. S66. Perform reduction processing on non-integrable quantities and retain integrable quantities in the modified wavefront slope field to form an integrable wavefront slope field. S67. Using the candidate wavefront phase results as the initial phase value, perform phase accumulation reconstruction on the integrable wavefront slope field along the coordinates of adjacent apertures in the set of aperture coordinates to be reconstructed, and generate a continuous wavefront phase map.

8. The wavefront detection method based on single-lens scanning according to claim 1, characterized in that, Specifically, S7 includes: S71. The integrable wavefront reconstruction module reads the continuous wavefront phase map and extracts the effective region within the aperture according to the aperture boundary of the aperture to be measured, thereby obtaining the effective wavefront phase region. S72. Perform reference plane subtraction processing on the phase values ​​in the effective wavefront phase region to obtain the reference-free wavefront phase region. S73. Read the phase values ​​corresponding to the aperture query coordinates in the phase region of the reference wavefront, and perform maximum value extraction, minimum value extraction and mean square statistics on the phase values ​​to generate wavefront peak and valley values ​​and wavefront root mean square values. S74. According to the aperture coordinate system of the aperture to be measured, project the phase value in the phase region of the reference wavefront to the set of aberration basis functions to obtain the corresponding aberration components. S75. Combine the continuous wavefront phase map, the reference wavefront phase region, the wavefront peak and valley values, the wavefront root mean square value, and the aberration components to form the wavefront detection result.

9. A wavefront detection device based on single-lens scanning, characterized in that, A wavefront detection method based on single-lens scanning according to any one of claims 1 to 8 includes the following modules: The scanning acquisition module is used to control the single lens to perform scanning acquisition relative to the aperture to be measured, acquire the spot image corresponding to each scanning position, extract the spot response features, and generate a set of scanning response features; The scanning response standardization module is used to perform position calibration and response normalization processing on the scanning response feature set, generate a standardized scanning response feature set, and generate multiple aperture query coordinates according to the aperture boundary and aperture sampling interval of the aperture to be measured. The multiple aperture query coordinates are arranged in spatial order to form a set of aperture coordinates to be reconstructed. The candidate wavefront generation module is used to input the standardized scan response feature set and the aperture coordinate set to be reconstructed into the improved DeepONet network. The improved DeepONet network includes a scan wavefront operator generation module, an optical path back projection module, and an integrable wavefront reconstruction module. The candidate wavefront phase results are generated by the scan wavefront operator generation module. The optical path back calculation and verification module is used to perform single-lens scanning optical path back calculation on the candidate wavefront phase result through the single-lens scanning accompanying verification structure set in the optical path back calculation and back projection module, generate the predicted spot response result, and compare the predicted spot response result with the measured spot response feature in the standardized scanning response feature set to obtain the spot response residual. The residual-accompanied back-projection module is used to back-project the spot response residual along the accompanying direction of the single-lens scanning optical path to the aperture coordinate domain through the single-lens scanning accompanying verification structure, thereby generating the wavefront residual correction amount. The integrable wavefront reconstruction module is used to correct the candidate wavefront phase results using the wavefront residual correction amount, and to perform integrable reconstruction on the corrected wavefront slope field to generate a continuous wavefront phase map. The wavefront detection output module is used to extract wavefront parameters from continuous wavefront phase maps and output wavefront detection results.

Images