A phase correction method

Abstract

The present application discloses a phase correction method, which obtains a light beam phase distribution diagram collected by a wavefront sensor, extracts an ideal light beam phase distribution diagram from a database, and uses image analysis and processing technology based on computer vision to extract features of the light beam phase distribution diagram and the ideal light beam phase distribution diagram, thereby automatically obtaining the phase difference between the two according to the semantic deviation between the light beam phase distribution characteristics and the ideal light beam phase distribution characteristics, and automatically adjusting the shape of the optical element based on the phase difference. In this way, nonlinear or complex phase distortion can be processed, so as to accurately identify and correct subtle phase changes that are difficult to adjust with traditional mechanical deformation elements, and improve the correction accuracy of high-order modes or multi-modal light beams. At the same time, higher resolution and faster response speed can be provided, so that higher accuracy and faster response can be achieved when adjusting the phase, meeting the needs of real-time correction and realizing an intelligent correction process.

Description

Technical Field

[0001] The present application relates to the field of intelligent correction, and more specifically, to a phase correction method. Background Art

[0002] In recent years, with the development of precision optical technology, precise control of the phase distribution of the light beam has become increasingly important. Whether in the fields of laser processing, optical communication or optical imaging, the phase of the light beam needs to be precisely controlled to achieve the best working effect. Specifically, in high-power lasers, especially when the Fresnel number is greater than 1, the divergence of the laser beam is a key parameter that affects the focusing ability of the laser beam and the power density on the target. Phase correction is a technology that involves adjusting the phase relationship of each part of the laser mode so that the laser beam maintains better coherence during propagation, reduces divergence, and can be more effectively focused on the target.

[0003] However, traditional phase correction measures the phase distribution of the laser beam through a wavefront sensor, and then adjusts the phase through a programmable phase plate or an adaptive optical system to compensate for the wavefront distortion. However, this phase correction method may not be able to achieve high correction accuracy when processing high-order modes or multi-modal beams, especially when dealing with nonlinear or complex phase distortions. These methods often show shortcomings. In addition, traditional correction methods may not be able to respond to rapidly changing phase errors in real time, resulting in poor correction effects. For example, during laser processing, due to the inhomogeneity of the material or other external factors, the phase of the beam may change rapidly in a short period of time. Traditional correction methods often cannot adjust quickly to compensate for these changes, which in turn affects processing quality and efficiency.

[0004] Therefore, an optimized phase correction method is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. An embodiment of the present application provides a phase correction method, which obtains a light beam phase distribution diagram collected by a wavefront sensor, extracts an ideal light beam phase distribution diagram from a database, and uses image analysis and processing technology based on computer vision to extract features of the light beam phase distribution diagram and the ideal light beam phase distribution diagram, thereby automatically obtaining the phase difference between the two according to the semantic deviation between the light beam phase distribution characteristics and the ideal light beam phase distribution characteristics, and automatically adjusting the shape of the optical element based on the phase difference. In this way, nonlinear or complex phase distortion can be processed, so as to accurately identify and correct subtle phase changes that are difficult to adjust with traditional mechanical deformation elements, and improve the correction accuracy of high-order modes or multi-modal light beams. At the same time, higher resolution and faster response speed can be provided, so that higher accuracy and faster response can be achieved when adjusting the phase, meeting the needs of real-time correction and realizing an intelligent correction process.

[0006] According to one aspect of the present application, a phase correction method is provided, which includes: acquiring a beam phase distribution diagram collected by a wavefront sensor; extracting an ideal beam phase distribution diagram from a database; inputting the beam phase distribution diagram and the ideal beam phase distribution diagram into a beam phase distribution feature extractor to obtain a beam phase distribution feature diagram and a beam phase distribution ideal feature diagram; performing a beam distribution detection-beam distribution ideal semantic deviation feature analysis on the beam phase distribution feature diagram and the ideal beam phase distribution feature diagram to obtain a beam distribution detection-beam distribution ideal semantic deviation feature; based on the beam distribution detection-beam distribution ideal semantic deviation feature, obtaining a phase difference value, and based on the phase difference value, generating a phase correction instruction for adjusting the shape of an optical element; wherein, the beam phase distribution feature diagram and the ideal beam phase distribution feature diagram are subjected to a beam distribution detection-beam distribution ideal semantic deviation feature analysis to obtain a beam distribution detection-beam distribution ideal semantic deviation feature. Features include: performing feature fine-grained deconstruction on the beam phase distribution feature map and the beam phase distribution ideal feature map to obtain a set of local feature matrices of the beam phase distribution channel dimension and a set of local feature matrices of the beam phase distribution ideal channel dimension; performing two-way global attention joint processing on the set of local feature matrices of the beam phase distribution channel dimension and the set of local feature matrices of the beam phase distribution ideal channel dimension to obtain a set of local feature matrices of the beam phase distribution channel dimension optimized by one-way global attention and a set of local feature matrices of the beam phase distribution ideal channel dimension optimized by one-way global attention; performing global perception processing on the set of local feature matrices of the beam phase distribution channel dimension optimized by one-way global attention and the set of local feature matrices of the beam phase distribution ideal channel dimension optimized by one-way global attention to obtain a beam distribution detection-beam distribution ideal semantic deviation feature map as the beam distribution detection-beam distribution ideal semantic deviation feature.

[0007] Compared with the prior art, the present application provides a phase correction method, which obtains the light beam phase distribution diagram collected by the wavefront sensor, extracts the ideal light beam phase distribution diagram from the database, and uses image analysis and processing technology based on computer vision to extract the features of the light beam phase distribution diagram and the ideal light beam phase distribution diagram, so as to automatically obtain the phase difference between the two according to the semantic deviation between the light beam phase distribution characteristics and the ideal light beam phase distribution characteristics, and automatically adjust the shape of the optical element based on the phase difference. In this way, nonlinear or complex phase distortion can be processed, so as to accurately identify and correct subtle phase changes that are difficult to adjust with traditional mechanical deformation elements, and improve the correction accuracy of high-order modes or multi-modal light beams. At the same time, it can provide higher resolution and faster response speed, so that higher accuracy and faster response can be achieved when adjusting the phase, meeting the needs of real-time correction and realizing an intelligent correction process. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 4 is a flow chart of a phase correction method according to an embodiment of the present application.

[0010] Figure 2 Schematic diagram of data flow of a phase correction method according to an embodiment of the present application.

[0011] Figure 3 4 is a flowchart of sub-step S4 of the phase correction method according to an embodiment of the present application. DETAILED DESCRIPTION

[0012] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0013] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0014] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0015] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0016] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0017] Traditional phase correction measures the phase distribution of the laser beam through a wavefront sensor, and then adjusts the phase through a programmable phase plate or an adaptive optical system to compensate for the wavefront distortion. However, this phase correction method may not be able to achieve high correction accuracy when processing high-order modes or multimodal beams, especially when dealing with nonlinear or complex phase distortions. These methods often show shortcomings. In addition, traditional correction methods may not be able to respond to rapidly changing phase errors in real time, resulting in poor correction effects. For example, during laser processing, due to the inhomogeneity of the material or other external factors, the phase of the beam may change rapidly in a short period of time. Traditional correction methods often cannot adjust quickly to compensate for these changes, which in turn affects processing quality and efficiency. Therefore, an optimized phase correction method is desired.

[0018] In the technical solution of the present application, a phase correction method is proposed. Figure 1 4 is a flow chart of a phase correction method according to an embodiment of the present application. Figure 2 FIG. 2 is a schematic diagram of data flow of a phase correction method according to an embodiment of the present application. Figure 1 and Figure 2 As shown, the phase correction method according to the embodiment of the present application includes the steps of: S1, acquiring a beam phase distribution diagram collected by a wavefront sensor; S2, extracting an ideal beam phase distribution diagram from a database; S3, inputting the beam phase distribution diagram and the ideal beam phase distribution diagram into a beam phase distribution feature extractor to obtain a beam phase distribution feature diagram and a beam phase distribution ideal feature diagram; S4, performing beam distribution detection-beam distribution ideal semantic deviation feature analysis on the beam phase distribution feature diagram and the ideal beam phase distribution feature diagram to obtain beam distribution detection-beam distribution ideal semantic deviation features; S5, obtaining a phase difference value based on the beam distribution detection-beam distribution ideal semantic deviation features, and generating a phase correction instruction for adjusting the shape of the optical element based on the phase difference value.

[0019] In particular, S1 and S2 obtain the light beam phase distribution diagram collected by the wavefront sensor; and extract the ideal light beam phase distribution diagram from the database. The wavefront sensor is an optical device used to measure the wavefront distortion of light waves. Wavefront distortion is caused by the phase distortion that occurs when light waves pass through a medium (such as the atmosphere or an optical element). The working principle of the wavefront sensor is based on interferometry. It combines the light wave to be measured with the reference light wave to form an interference pattern. The shape and intensity of the interference pattern reflect the wavefront distortion of the light wave to be measured.

[0020] In particular, in S3, the beam phase distribution diagram and the ideal beam phase distribution diagram are input into a beam phase distribution feature extractor to obtain a beam phase distribution feature diagram and an ideal beam phase distribution feature diagram. In a specific example of the present application, the beam phase distribution diagram and the ideal beam phase distribution diagram are input into a beam phase distribution feature extractor based on a hole convolutional neural network model to obtain the beam phase distribution feature diagram and the ideal beam phase distribution feature diagram. Considering that the beam phase distribution diagram and the ideal beam phase distribution diagram contain rich detail information, such as phase value, phase gradient, wavefront profile, etc., which are essential for analyzing the behavior and characteristics of the beam. The hole convolution can increase the receptive field without increasing the network depth while keeping the spatial resolution unchanged, which is particularly important for processing the beam phase distribution diagram, because maintaining the resolution helps to retain the detail information in the image. Based on this, in the technical solution of the present application, the beam phase distribution diagram and the ideal beam phase distribution diagram are input into a beam phase distribution feature extractor based on a hole convolutional neural network model to capture and extract key detail feature information, and obtain a beam phase distribution feature diagram and an ideal beam phase distribution feature diagram. It is worth mentioning that the dilated convolutional neural network is a convolutional neural network model that uses dilated convolution operations to expand the receptive field while maintaining spatial resolution. The dilated convolution operation is similar to the standard convolution operation, but it introduces "holes" in the convolution kernel, i.e. zero padding. These holes enable the convolution kernel to skip certain elements in the input feature map, thereby expanding the receptive field.

[0021] In particular, in S4, the beam phase distribution feature map and the ideal beam phase distribution feature map are subjected to beam distribution detection-beam distribution ideal semantic deviation feature analysis to obtain beam distribution detection-beam distribution ideal semantic deviation features. Considering that the beam phase distribution feature map and the ideal beam phase distribution feature map contain some tiny difference features, such as edges, textures and other tiny differences. And these tiny, tiny differences are crucial for the subsequent generation of phase differences. Based on this, in order to identify and separate different phase change details in a more fine-grained manner, so that the features can be understood and processed more accurately, so as to reflect the specific differences between the actual detected phase distribution and the ideal phase distribution, in the technical solution of the present application, the beam phase distribution feature map and the ideal beam phase distribution feature map are subjected to beam distribution detection-beam distribution ideal semantic deviation feature analysis to obtain a beam distribution detection-beam distribution ideal semantic deviation feature map as a beam distribution detection-beam distribution ideal semantic deviation feature. It is worth mentioning that the semantic deviation feature analysis models the comprehensive recognition and perception capabilities of image semantics through fine-grained feature deconstruction, global attention interaction, feature coupling and differential calculation. In a specific example of the present application, such as Figure 3 As shown, the S4 includes: S41, performing feature fine-grained deconstruction on the beam phase distribution feature map and the beam phase distribution ideal feature map to obtain a set of local feature matrices of the beam phase distribution channel dimension and a set of local feature matrices of the beam phase distribution ideal channel dimension; S42, performing two-way global attention joint processing on the set of local feature matrices of the beam phase distribution channel dimension and the set of local feature matrices of the beam phase distribution ideal channel dimension to obtain a set of local feature matrices of the beam phase distribution channel dimension optimized by one-way global attention and a set of local feature matrices of the beam phase distribution ideal channel dimension optimized by one-way global attention; S43, performing global perception processing on the set of local feature matrices of the beam phase distribution channel dimension optimized by one-way global attention and the set of local feature matrices of the beam phase distribution ideal channel dimension optimized by one-way global attention to obtain a beam distribution detection-beam distribution ideal semantic deviation feature map as the beam distribution detection-beam distribution ideal semantic deviation feature.

[0022] Specifically, the S41 performs feature fine-grained deconstruction on the beam phase distribution feature graph and the beam phase distribution ideal feature graph to obtain a set of local feature matrices of the beam phase distribution channel dimension and a set of local feature matrices of the beam phase distribution ideal channel dimension. That is, feature fine-grained segmentation is performed along the channel dimension to decompose the complex feature graph into smaller and more specific feature sets, so as to capture finer-grained beam phase features, and obtain a set of local feature matrices of the beam phase distribution channel dimension and a set of local feature matrices of the beam phase distribution ideal channel dimension.

[0023] Specifically, the S42 performs a two-way global attention joint processing on the set of local feature matrices of the channel dimension of the beam phase distribution and the set of local feature matrices of the ideal channel dimension of the beam phase distribution to obtain a set of local feature matrices of the channel dimension of the beam phase distribution optimized by one-way global attention and a set of local feature matrices of the ideal channel dimension of the beam phase distribution optimized by one-way global attention. Considering that there are mutual correlations and interactive relationships between the beam phase distribution characteristics and the ideal characteristics of the beam phase distribution, such as the phase deviation between the two, the matching degree of the interference fringes, the consistency of the wavefront shape, etc., these correlations reflect the difference between the actual detection beam phase distribution and the ideal state, which is crucial for accurately evaluating the beam quality. Considering that there are mutual correlations and interactive relationships between the beam phase distribution characteristics and the ideal characteristics of the beam phase distribution, such as the phase deviation between the two, the matching degree of the interference fringes, the consistency of the wavefront shape, etc., these correlations reflect the difference between the actual detection beam phase distribution and the ideal state, which is crucial for accurately evaluating the beam quality. Furthermore, each local feature matrix of the beam phase distribution channel dimension in the set of local feature matrices of the beam phase distribution channel dimension is used as a query feature matrix, and the set of local feature matrices of the ideal channel dimension of the beam phase distribution is used as a set of key feature matrices, and the query feature matrix and the set of key feature matrices are input into a one-way global attention interaction module based on the first converter structure to obtain a set of local feature matrices of the beam phase distribution channel dimension that are optimized by one-way global attention. At the same time, each local feature matrix of the ideal channel dimension of the beam phase distribution in the set of local feature matrices of the ideal channel dimension of the beam phase distribution is used as a query feature matrix, and the set of local feature matrices of the beam phase distribution channel dimension is used as a set of key feature matrices, and the set of query feature matrices and the set of key feature matrices are input into a one-way global attention interaction module based on the second converter structure to obtain a set of local feature matrices of the ideal channel dimension of the beam phase distribution that are optimized by one-way global attention. It should be understood that after the feature map has been deconstructed in a fine-grained manner, a Transformer-based unidirectional global attention module is used to evaluate the correlation between different feature matrices through a multi-head attention mechanism, thereby strengthening the connection between them. This method enables the model to focus more on areas with rich and important feature information. Specifically, in the present technical solution, each local feature matrix in the beam phase distribution feature map can have in-depth semantic communication with all local feature matrices in the ideal feature map of the beam phase distribution in a unidirectional direction, so as to more comprehensively obtain the global semantic connection between each local feature matrix in the beam phase distribution feature map and the overall semantic information of the ideal feature map of the beam phase distribution. Similarly, the global semantic interaction between each local feature matrix in the ideal feature map of the beam phase distribution and all local matrices of the beam phase distribution feature map is also the same.In this way, the two-way global semantic interaction can ensure that the relationship between the two feature maps is fully utilized during the processing, thereby achieving a more comprehensive understanding of the actual detection and ideal beam phase distribution.

[0024] In an embodiment of the present application, a set of local feature matrices of the beam phase distribution channel dimension and a set of local feature matrices of the ideal channel dimension of the beam phase distribution are subjected to bidirectional global attention joint processing to obtain a set of local feature matrices of the beam phase distribution channel dimension that are optimized by one-way global attention and a set of local feature matrices of the ideal channel dimension of the beam phase distribution that are optimized by one-way global attention, including: taking each local feature matrix of the beam phase distribution channel dimension in the set of local feature matrices of the beam phase distribution channel dimension as a query feature matrix, taking the set of local feature matrices of the ideal channel dimension of the beam phase distribution as a key feature matrix, and combining the query feature matrix and the set of local feature matrices of the ideal channel dimension of the beam phase distribution. The set of key feature matrices is input into the one-way global attention interaction module based on the first converter structure to obtain the set of local feature matrices of the one-way global attention optimized beam phase distribution channel dimension; each local feature matrix of the ideal channel dimension of the beam phase distribution in the set of local feature matrices of the ideal channel dimension of the beam phase distribution is used as a query feature matrix, and the set of local feature matrices of the beam phase distribution channel dimension is used as a set of key feature matrices, and the set of the query feature matrix and the key feature matrix is ​​input into the one-way global attention interaction module based on the second converter structure to obtain the set of local feature matrices of the ideal channel dimension of the one-way global attention optimized beam phase distribution.

[0025] Among them, each local feature matrix of the beam phase distribution channel dimension in the set of local feature matrices of the beam phase distribution channel dimension is used as a query feature matrix, and the set of local feature matrices of the ideal channel dimension of the beam phase distribution is used as a set of key feature matrices, and the query feature matrix and the set of the key feature matrix are input into the one-way global attention interaction module based on the first converter structure to obtain the set of local feature matrices of the one-way global attention optimized beam phase distribution channel dimension. The process includes: matrix multiplication of the local feature matrix of the beam phase distribution channel dimension with the transposed matrix of each local feature matrix of the ideal channel dimension of the beam phase distribution in the set of local feature matrices of the ideal channel dimension of the beam phase distribution to obtain a set of beam phase distribution detection-ideal correlation feature matrices; calculating the beam phase distribution detection The set of beam phase distribution detection-ideal correlation feature matrices is obtained by positionally dividing each beam phase distribution detection-ideal correlation feature matrix in the set of beam phase distribution detection-ideal correlation feature matrices with the square root of the scale of the beam phase distribution detection-ideal correlation feature matrix; the set of the beam phase distribution detection-ideal correlation modulation feature matrices is input into the softmax function for activation processing, and the obtained set of beam phase distribution detection-ideal correlation activation feature matrices is matrix-multiplied with the beam phase distribution channel dimension local feature matrix to obtain a set of beam phase distribution detection-ideal correlation attention interaction feature matrices; the set of the beam phase distribution detection-ideal correlation attention interaction feature matrices is multiplied by position point to obtain the unidirectional global attention optimized beam phase distribution channel dimension local feature matrix.More specifically, each local feature matrix of the ideal channel dimension of the beam phase distribution in the set of local feature matrices of the ideal channel dimension of the beam phase distribution is used as a query feature matrix, and the set of local feature matrices of the ideal channel dimension of the beam phase distribution is used as a set of key feature matrices. The query feature matrix and the set of key feature matrices are input into a one-way global attention interaction module based on a second converter structure to obtain the set of local feature matrices of the ideal channel dimension of the beam phase distribution optimized by the one-way global attention. The process includes: matrix multiplication of the local feature matrix of the ideal channel dimension of the beam phase distribution with the transposed matrix of each local feature matrix of the channel dimension of the beam phase distribution in the set of local feature matrices of the ideal channel dimension of the beam phase distribution to obtain a set of ideal-detection correlation feature matrices of the beam phase distribution; calculating the ideal channel dimension of the beam phase distribution. The ideal-detection associated feature matrix of each beam phase distribution in the set of the ideal-detection associated feature matrix is ​​divided by the square root of the scale of the ideal-detection associated feature matrix of the beam phase distribution to obtain a set of ideal-detection associated modulation feature matrices of the beam phase distribution; the set of the ideal-detection associated modulation feature matrices of the beam phase distribution is input into the softmax function for activation processing, and the obtained set of ideal-detection associated activation feature matrices of the beam phase distribution is matrix-multiplied with the local feature matrix of the ideal channel dimension of the beam phase distribution to obtain a set of ideal-detection associated attention interaction feature matrices of the beam phase distribution; the set of the ideal-detection associated attention interaction feature matrices of the beam phase distribution is multiplied by the position point to obtain the one-way global attention optimized beam phase distribution ideal channel dimension local feature matrix.

[0026] Specifically, the S43 performs global perception processing on the set of local feature matrices of the channel dimension of the one-way global attention optimized beam phase distribution and the set of local feature matrices of the ideal channel dimension of the one-way global attention optimized beam phase distribution to obtain a beam distribution detection-beam distribution ideal semantic deviation feature map as the beam distribution detection-beam distribution ideal semantic deviation feature. That is, each optimized local feature matrix is ​​coupled along the channel dimension to obtain a one-way global interactive optimized beam phase distribution feature map and a one-way global interactive optimized beam phase distribution ideal feature map, so as to more comprehensively and comprehensively integrate the overall semantic expression of the actual detection and ideal beam phase distribution. Finally, by calculating the direct weighted difference between the two to obtain the beam distribution detection-beam distribution ideal semantic deviation feature map, the difference between the actual detection beam phase distribution feature and the ideal beam phase distribution feature can be effectively highlighted, so that the system can clearly point out which parts deviate from the ideal state, thereby helping the system to more accurately identify and locate the position and degree of phase deviation.

[0027] In an embodiment of the present application, a set of local feature matrices of the channel dimension of the one-way global attention optimized beam phase distribution and a set of local feature matrices of the ideal channel dimension of the one-way global attention optimized beam phase distribution are globally perceived to obtain a beam distribution detection-beam distribution ideal semantic deviation feature map, including: respectively coupling the set of local feature matrices of the channel dimension of the one-way global attention optimized beam phase distribution and the set of local feature matrices of the ideal channel dimension of the one-way global attention optimized beam phase distribution along the channel dimension to obtain a one-way global interactive optimized beam phase distribution feature map and a one-way global interactive optimized beam phase distribution ideal feature map; calculating the position-weighted difference between the one-way global interactive optimized beam phase distribution feature map and the one-way global interactive optimized beam phase distribution ideal feature map to obtain the beam distribution detection-beam distribution ideal semantic deviation feature map.

[0028] Among them, the process of calculating the position-weighted difference between the one-way global interactive optimization beam phase distribution feature map and the one-way global interactive optimization beam phase distribution ideal feature map to obtain the beam distribution detection-beam distribution ideal semantic deviation feature map includes: multiplying the one-way global interactive optimization beam phase distribution feature map and the one-way global interactive optimization beam phase distribution ideal feature map with the first weighted hyperparameter and the second weighted hyperparameter respectively by position, and then subtracting the obtained one-way global interactive optimization beam phase distribution weighted feature map and the one-way global interactive optimization beam phase distribution ideal weighted feature map by position to obtain the beam distribution detection-beam distribution ideal semantic deviation feature map.

[0029] In summary, in the above embodiment, the beam phase distribution feature diagram and the beam phase distribution ideal feature diagram are subjected to beam distribution detection-beam distribution ideal semantic deviation feature analysis to obtain beam distribution detection-beam distribution ideal semantic deviation features, including: the beam phase distribution feature diagram and the beam phase distribution ideal feature diagram are subjected to beam distribution detection-beam distribution ideal semantic deviation feature analysis according to the following formula to obtain the beam distribution detection-beam distribution ideal semantic deviation feature diagram; wherein the formula is: ;in, and are the light beam phase distribution characteristic diagram and the light beam phase distribution ideal characteristic diagram, respectively. It is a feature fine-grained deconstruction operation. are the first, second, ..., and third local characteristic matrices of the beam phase distribution channel dimension in the set. ,..., The local characteristic matrix of the beam phase distribution channel dimension, are the first, second, ..., and third local characteristic matrices of the ideal channel dimension of the beam phase distribution. ,..., The ideal channel dimension local characteristic matrix of the beam phase distribution is: is the number of characteristic matrices in the set of local characteristic matrices of the beam phase distribution channel dimension and the set of local characteristic matrices of the beam phase distribution ideal channel dimension, and They are and The transposed matrix of represents matrix multiplication, and are the scales of the beam phase distribution detection-ideal correlation characteristic matrix and the beam phase distribution ideal-detection correlation characteristic matrix, respectively. and function, The first local feature matrix in the set of channel dimensions for optimizing the beam phase distribution for unidirectional global attention A one-way global attention optimizes the beam phase distribution channel dimension local feature matrix, The ideal channel dimension local feature matrix for optimizing the phase distribution of the beam for one-way global attention is the first One-way global attention optimizes the beam phase distribution ideal channel dimension local feature matrix, To couple the features of each feature matrix along the channel dimension, and They are respectively the one-way global interactive optimization beam phase distribution characteristic diagram and the one-way global interactive optimization beam phase distribution ideal characteristic diagram, and is the corresponding weighted hyperparameter, The beam distribution detection-beam distribution ideal semantic deviation feature map.

[0030] It is worth mentioning that in other specific examples of the present application, the beam phase distribution feature map and the beam phase distribution ideal feature map can also be subjected to beam distribution detection-beam distribution ideal semantic deviation feature analysis in other ways to obtain beam distribution detection-beam distribution ideal semantic deviation features, for example: input the beam phase distribution feature map and the beam phase distribution ideal feature map; for each pixel in the feature map, calculate the pixel-level difference between the beam phase distribution feature map and the beam phase distribution ideal feature map; convert the pixel-level difference into a semantic deviation map. Each pixel value in the semantic deviation map represents the degree of semantic deviation of the pixel in the beam phase distribution; use a convolutional neural network (CNN) or other machine learning algorithms to extract semantic deviation features from the semantic deviation map.

[0031] In particular, the S5 obtains a phase difference value based on the beam distribution detection-beam distribution ideal semantic deviation feature, and generates a phase correction instruction for adjusting the shape of the optical element based on the phase difference value. In a specific example of the present application, the beam distribution detection-beam distribution ideal semantic deviation feature map is input into a phase difference demodulation module to obtain a phase difference decoding value; and based on the phase difference decoding value, the phase correction instruction is generated, and the phase correction instruction is used to adjust the shape of the optical element. Specifically, the beam distribution detection-beam distribution ideal semantic deviation feature map is input into a phase difference demodulation module to obtain a phase difference decoding value, including: inputting the beam distribution detection-beam distribution ideal semantic deviation feature map into a decoder-based phase difference demodulation module to obtain the phase difference decoding value. That is, the beam distribution detection-beam distribution ideal semantic deviation feature obtained by performing semantic deviation feature analysis using the beam phase distribution feature map and the beam phase distribution ideal feature map is decoded to automatically obtain the phase difference value between the two, and automatically adjust the shape of the optical element based on the phase difference value. In this way, nonlinear or complex phase distortions can be processed, thereby accurately identifying and correcting subtle phase changes that are difficult to adjust with traditional mechanical deformation elements, and improving the correction accuracy of high-order modes or multi-modal beams. At the same time, higher resolution and faster response speed can be provided, so that higher accuracy and faster response can be achieved when adjusting the phase, meeting the needs of real-time correction and realizing an intelligent correction process. In one example, the measured phase difference is decoded into a surface height value, which involves using a known phase-height relationship and then generating a mathematical model of the shape of the optical element based on the decoded surface height value. This can be a parameterized model, such as a Zernike polynomial or a spline curve.

[0032] In a preferred embodiment, inputting the beam distribution detection-beam distribution ideal semantic deviation feature map into a decoder-based phase difference demodulation module to obtain a phase difference decoding value includes: calculating the sum of the absolute values ​​of each eigenvalue of the beam distribution detection-beam distribution ideal semantic deviation feature map to obtain a first beam distribution detection-beam distribution ideal semantic deviation and modulation value, and calculating the square root of the sum of the squares of each eigenvalue of the beam distribution detection-beam distribution ideal semantic deviation feature map to obtain a second beam distribution detection-beam distribution ideal semantic deviation and modulation value; performing dot subtraction between the beam distribution detection-beam distribution ideal semantic deviation feature map and the second beam distribution detection-beam distribution ideal semantic deviation and modulation value, and performing dot multiplication with the number of eigenvalues ​​of the beam distribution detection-beam distribution ideal semantic deviation feature map and the inverse of the first beam distribution detection-beam distribution ideal semantic deviation and modulation value, and taking the inverse of each eigenvalue to obtain the first beam distribution detection-beam distribution ideal semantic deviation and modulation value. An ideal semantic deviation phase conversion feature map is obtained; after dot-wise subtraction of the beam distribution detection-beam distribution ideal semantic deviation feature map from the first beam distribution detection-beam distribution ideal semantic deviation and modulation value, dot-wise multiplication is performed with the square root of the number of eigenvalues ​​of the beam distribution detection-beam distribution ideal semantic deviation feature map and the inverse of the second beam distribution detection-beam distribution ideal semantic deviation and modulation value, and the inverse of each eigenvalue is taken to obtain a second beam distribution detection-beam distribution ideal semantic deviation phase conversion feature map; and dot-wise subtraction of the first beam distribution detection-beam distribution ideal semantic deviation phase conversion feature map from the second beam distribution detection-beam distribution ideal semantic deviation phase conversion feature map and the dot product vector of the weighted hyperparameter to obtain an optimized beam distribution detection-beam distribution ideal semantic deviation feature map; and inputting the optimized beam distribution detection-beam distribution ideal semantic deviation feature map into a decoder-based phase difference demodulation module to obtain a phase difference decoding value.

[0033] Here, the beam distribution detection-beam distribution ideal semantic deviation feature map is recorded as The optimization expression is: ;in, is the beam distribution detection-beam distribution ideal semantic deviation feature map, To take the inverse of each eigenvalue in the feature map, are the eigenvalues ​​of the beam distribution detection-beam distribution ideal semantic deviation feature map, is the first beam profile detection - beam profile ideal semantic deviation and modulation value, is the second beam profile detection - beam profile ideal semantic deviation and modulation value, is the number of eigenvalues ​​of the beam distribution detection-beam distribution ideal semantic deviation feature map, is the first beam distribution detection-beam distribution ideal semantic deviation phase conversion characteristic diagram, is the second beam distribution detection-beam distribution ideal semantic deviation phase conversion characteristic diagram, is a weighted hyperparameter, It's a point reduction. is the dot product, It is the optimized beam distribution detection-beam distribution ideal semantic deviation feature map.

[0034] That is, the applicant of the present application takes into account that the beam phase distribution feature map and the beam phase distribution ideal feature map represent the image semantic features of the beam phase distribution map and the beam phase ideal distribution map respectively, and when they are passed through a bidirectional global attention joint perception module based on fine-grained deconstruction, the spatial fine-grained image semantic feature distribution difference based on the source image semantic distribution will also have the global attention difference of local spatial distribution deconstruction. Therefore, it is expected to enhance the detail semantic joint perception aggregation expression effect of the beam distribution detection-beam distribution ideal semantic deviation feature map based on the local-global bidirectional image semantic feature spatial distribution difference.

[0035] Therefore, in the preferred example, the difference of the characteristic value of the beam distribution detection-beam distribution ideal semantic deviation feature map relative to the overall feature set of the beam distribution detection-beam distribution ideal semantic deviation feature map and the difference of the modulation representation are used as the semantic change intensity information, and the phase-like conversion corresponding to the position-based intensity modulation is performed through the different and modulation representation forms, so that the aggregation enhancement of the semantic change phase perception can improve the axial aggregation receptive field along the feature aggregation direction by performing the spatial translation operation based on alternating stacking under the set scale balance of the beam distribution detection-beam distribution ideal semantic deviation feature map, thereby improving the perception effect of the aggregated semantics of the beam distribution detection-beam distribution ideal semantic deviation feature map for the detailed semantic changes, so as to improve the expression effect of the beam distribution detection-beam distribution ideal semantic deviation feature map, and improve the accuracy of the phase difference decoding value obtained by inputting it into the phase difference demodulation module based on the decoder. In this way, nonlinear or complex phase distortion can be processed, so as to accurately identify and correct the subtle phase changes that are difficult to adjust with traditional mechanical deformation elements, and improve the correction accuracy of high-order modes or multi-modal beams. At the same time, it can provide higher resolution and faster response speed, so that higher accuracy and faster response can be achieved when adjusting the phase, meeting the needs of real-time correction and realizing an intelligent correction process.

[0036] In summary, the phase correction method according to the embodiment of the present application is explained, which obtains the light beam phase distribution diagram collected by the wavefront sensor, extracts the ideal light beam phase distribution diagram from the database, and uses image analysis and processing technology based on computer vision to extract the features of the light beam phase distribution diagram and the ideal light beam phase distribution diagram, so as to automatically obtain the phase difference between the two according to the semantic deviation between the light beam phase distribution characteristics and the ideal light beam phase distribution characteristics, and automatically adjust the shape of the optical element based on the phase difference. In this way, nonlinear or complex phase distortion can be processed, so as to accurately identify and correct subtle phase changes that are difficult to adjust with traditional mechanical deformation elements, and improve the correction accuracy of high-order modes or multi-modal light beams. At the same time, higher resolution and faster response speed can be provided, so that higher accuracy and faster response can be achieved when adjusting the phase, meeting the needs of real-time correction and realizing an intelligent correction process.

[0037] In most cases using stable resonators, the obstacle to producing a "pure" single-pass mode is the generation of competing multi-pass modes.

[0038] The transverse structure of these modes is identical to the classical single-pass, azimuthally polarized modes, with the spot diameter determined by the “equivalent length” which is equal to the resonator length multiplied by the number of round trips at paraxial resonance.

[0039] The energy of the "pure" higher order modes is determined by the mode spot diameter (for the same resonator length) compared to the fundamental TEM00 mode in single pass mode. In the case of generating separate multi-pass modes, the spot diameter is even larger, as it is determined by the "equivalent length".

[0040] Compared to multimode generation of TEMpq with a completely uncertain amplitude-phase field distribution, the emission quality of "pure" higher-order modes is significantly better. However, there is also a fundamental problem here. With increasing mode order, despite the increase in power, the power density at distant objects decreases due to the increased divergence. Correcting the phase in adjacent rings significantly improves the situation. It turns out that the "in-phase ring" type radiation evolves into the lowest-order ring A-TEM01* mode. This approach to improving the radiation quality of high-power lasers is based on fundamental knowledge and is very feasible at the current state of the art.

[0041] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A phase correction method, characterized in that: The method comprises: obtaining a beam phase distribution diagram collected by a wavefront sensor; extracting an ideal beam phase distribution diagram from a database; inputting the beam phase distribution diagram and the ideal beam phase distribution diagram into a beam phase distribution feature extractor to obtain a beam phase distribution feature diagram and an ideal beam phase distribution feature diagram; performing a beam distribution detection-beam distribution ideal semantic deviation feature analysis on the beam phase distribution feature diagram and the ideal beam phase distribution feature diagram to obtain a beam distribution detection-beam distribution ideal semantic deviation feature; obtaining a phase difference value based on the beam distribution detection-beam distribution ideal semantic deviation feature, and generating a phase correction instruction for adjusting the shape of an optical element based on the phase difference value; wherein, performing a beam distribution detection-beam distribution ideal semantic deviation feature analysis on the beam phase distribution feature diagram and the ideal beam phase distribution feature diagram to obtain a beam distribution detection-beam distribution ideal semantic deviation feature, comprising: The phase distribution feature map and the ideal feature map of the beam phase distribution are subjected to feature fine-grained deconstruction to obtain a set of local feature matrices of the beam phase distribution channel dimension and a set of local feature matrices of the beam phase distribution ideal channel dimension; the set of local feature matrices of the beam phase distribution channel dimension and the set of local feature matrices of the beam phase distribution ideal channel dimension are subjected to bidirectional global attention joint processing to obtain a set of local feature matrices of the beam phase distribution channel dimension optimized by one-way global attention and a set of local feature matrices of the beam phase distribution ideal channel dimension optimized by one-way global attention; the set of local feature matrices of the beam phase distribution channel dimension optimized by one-way global attention and the set of local feature matrices of the beam phase distribution ideal channel dimension are subjected to global perception processing to obtain a beam distribution detection-beam distribution ideal semantic deviation feature map as the beam distribution detection-beam distribution ideal semantic deviation feature.

2. The phase correction method according to claim 1, characterized in that: The beam phase distribution diagram and the ideal beam phase distribution diagram are input into a beam phase distribution feature extractor to obtain a beam phase distribution feature diagram and an ideal beam phase distribution feature diagram, including: inputting the beam phase distribution diagram and the ideal beam phase distribution diagram into a beam phase distribution feature extractor based on a hole convolutional neural network model to obtain the beam phase distribution feature diagram and the ideal beam phase distribution feature diagram.

3. The phase correction method according to claim 2, characterized in that: The set of local feature matrices of the beam phase distribution channel dimension and the set of local feature matrices of the ideal beam phase distribution channel dimension are subjected to two-way global attention joint processing to obtain a set of local feature matrices of the beam phase distribution channel dimension that are optimized by one-way global attention and a set of local feature matrices of the ideal beam phase distribution channel dimension that are optimized by one-way global attention, including: taking each local feature matrix of the beam phase distribution channel dimension in the set of local feature matrices of the beam phase distribution channel dimension as a query feature matrix, taking the set of local feature matrices of the ideal beam phase distribution channel dimension as a key feature matrix, and combining the query feature matrix and the key feature matrix. The set of arrays is input into the one-way global attention interaction module based on the first converter structure to obtain the set of local feature matrices of the one-way global attention optimized beam phase distribution channel dimension; each local feature matrix of the ideal channel dimension of the beam phase distribution in the set of local feature matrices of the ideal channel dimension of the beam phase distribution is used as a query feature matrix, and the set of local feature matrices of the beam phase distribution channel dimension is used as a set of key feature matrices, and the set of the query feature matrix and the key feature matrix are input into the one-way global attention interaction module based on the second converter structure to obtain the set of local feature matrices of the ideal channel dimension of the one-way global attention optimized beam phase distribution.

4. The phase correction method according to claim 3, characterized in that: Each local feature matrix of the beam phase distribution channel dimension in the set of local feature matrices of the beam phase distribution channel dimension is used as a query feature matrix, and the set of local feature matrices of the ideal channel dimension of the beam phase distribution is used as a set of key feature matrices, and the query feature matrix and the set of key feature matrices are input into a one-way global attention interaction module based on a first converter structure to obtain the set of local feature matrices of the one-way global attention optimized beam phase distribution channel dimension, including: matrix multiplication of the local feature matrix of the beam phase distribution channel dimension with the transposed matrix of each local feature matrix of the ideal channel dimension of the beam phase distribution in the set of local feature matrices of the ideal channel dimension of the beam phase distribution to obtain a set of beam phase distribution detection-ideal correlation feature matrices; calculating the beam phase distribution detection- The positional division between each beam phase distribution detection-ideal correlation feature matrix in the set of ideal correlation feature matrices and the square root of the scale of the beam phase distribution detection-ideal correlation feature matrix is ​​performed to obtain a set of beam phase distribution detection-ideal correlation modulation feature matrices; the set of the beam phase distribution detection-ideal correlation modulation feature matrices is input into the softmax function for activation processing, and the obtained set of beam phase distribution detection-ideal correlation activation feature matrices is matrix multiplied with the beam phase distribution channel dimension local feature matrix to obtain a set of beam phase distribution detection-ideal correlation attention interaction feature matrices; the set of the beam phase distribution detection-ideal correlation attention interaction feature matrices is multiplied by position point to obtain the unidirectional global attention optimized beam phase distribution channel dimension local feature matrix.

5. The phase correction method according to claim 4, characterized in that: Each local feature matrix of the ideal channel dimension of the beam phase distribution in the set of local feature matrices of the ideal channel dimension of the beam phase distribution is used as a query feature matrix, and the set of local feature matrices of the ideal channel dimension of the beam phase distribution is used as a set of key feature matrices, and the query feature matrix and the set of key feature matrices are input into a one-way global attention interaction module based on a second converter structure to obtain the set of local feature matrices of the ideal channel dimension of the beam phase distribution optimized by the one-way global attention, including: matrix multiplication of the local feature matrix of the ideal channel dimension of the beam phase distribution with the transposed matrix of each local feature matrix of the channel dimension of the beam phase distribution in the set of local feature matrices of the ideal channel dimension of the beam phase distribution to obtain a set of ideal-detection correlation feature matrices of the beam phase distribution; calculating the ideal-detection correlation feature matrix of the beam phase distribution. The ideal-detection correlation feature matrix of each beam phase distribution in the set of the measurement correlation feature matrix is ​​divided by the square root of the scale of the ideal-detection correlation feature matrix of the beam phase distribution to obtain a set of ideal-detection correlation modulation feature matrices of the beam phase distribution; the set of the ideal-detection correlation modulation feature matrices of the beam phase distribution is input into the softmax function for activation processing, and the obtained set of ideal-detection correlation activation feature matrices of the beam phase distribution is matrix-multiplied with the local feature matrix of the ideal channel dimension of the beam phase distribution to obtain a set of ideal-detection correlation attention interaction feature matrices of the beam phase distribution; the set of the ideal-detection correlation attention interaction feature matrices of the beam phase distribution is multiplied by the position point to obtain the one-way global attention optimized beam phase distribution ideal channel dimension local feature matrix.

6. The phase correction method according to claim 5, characterized in that: The set of local feature matrices of the channel dimension of the one-way global attention optimized beam phase distribution and the set of local feature matrices of the ideal channel dimension of the one-way global attention optimized beam phase distribution are globally perceived to obtain a beam distribution detection-beam distribution ideal semantic deviation feature map, including: respectively coupling the set of local feature matrices of the channel dimension of the one-way global attention optimized beam phase distribution and the set of local feature matrices of the ideal channel dimension of the one-way global attention optimized beam phase distribution along the channel dimension to obtain a one-way global interactive optimized beam phase distribution feature map and a one-way global interactive optimized beam phase distribution ideal feature map; calculating the position-weighted difference between the one-way global interactive optimized beam phase distribution feature map and the one-way global interactive optimized beam phase distribution ideal feature map to obtain the beam distribution detection-beam distribution ideal semantic deviation feature map.

7. The phase correction method according to claim 6, characterized in that: The position-weighted difference between the one-way global interactive optimization beam phase distribution feature map and the one-way global interactive optimization beam phase distribution ideal feature map is calculated to obtain the beam distribution detection-beam distribution ideal semantic deviation feature map, including: multiplying the one-way global interactive optimization beam phase distribution feature map and the one-way global interactive optimization beam phase distribution ideal feature map with the first weighted hyperparameter and the second weighted hyperparameter respectively by position, and then subtracting the obtained one-way global interactive optimization beam phase distribution weighted feature map and the one-way global interactive optimization beam phase distribution ideal weighted feature map by position to obtain the beam distribution detection-beam distribution ideal semantic deviation feature map.

8. The phase correction method according to claim 7, characterized in that: Based on the beam distribution detection-beam distribution ideal semantic deviation characteristics, a phase difference value is obtained, and based on the phase difference value, a phase correction instruction for adjusting the shape of the optical element is generated, including: inputting the beam distribution detection-beam distribution ideal semantic deviation feature map into a phase difference demodulation module to obtain a phase difference decoding value; based on the phase difference decoding value, generating the phase correction instruction, and the phase correction instruction is used to adjust the shape of the optical element.

9. The phase correction method according to claim 8, characterized in that: The beam distribution detection-beam distribution ideal semantic deviation feature map is input into a phase difference demodulation module to obtain a phase difference decoding value, including: inputting the beam distribution detection-beam distribution ideal semantic deviation feature map into a decoder-based phase difference demodulation module to obtain the phase difference decoding value.

Images