Aspheric lens laser interference detection adjustment error separation system and method

Abstract

The invention relates to the technical field of optical detection, in particular to an aspheric lens laser interference detection adjustment error separation system and method, and the method comprises the steps: placing an aspheric lens on a clamping platform, carrying out the multiple laser interference measurement through adjusting different pitch angles and azimuth angles, and collecting a multi-angle interference fringe image; for different installation and adjustment error types, key features of typical error fringe images are extracted and compared with current interference fringe images, and identification of the installation and adjustment error types and levels is achieved; a mathematical model of various installation and adjustment errors is constructed, polynomial parameterization representation is adopted, installation and adjustment compensation parameters are reversely solved according to an identification result, and installation and adjustment are completed; and repeating the measurement and analysis process, and constructing an error adjustment frequency optimization model until the precision requirement is met. The method provided by the invention can improve the adjustment efficiency and precision of the aspherical lens.

Description

Technical Field

[0001] This application relates to the field of optical inspection technology, specifically to a system and method for separating assembly and adjustment errors in laser interferometry inspection of aspherical lenses. Background Technology

[0002] The manufacturing and testing of aspherical lenses is complex and requires extremely high precision in assembly and adjustment. Various assembly and adjustment errors, such as tilting and eccentricity errors, are often introduced during the manufacturing process. These errors affect the optical performance of the aspherical lens and reduce image quality.

[0003] Laser interferometry is a commonly used method for inspecting aspherical lenses. By analyzing interference fringe images, the surface morphology information of the aspherical lens can be obtained, thereby evaluating its processing quality. However, in actual measurement processes, different types of assembly and adjustment errors produce different interference fringe characteristics, and these errors often couple with each other, making them difficult to accurately identify and separate directly from the interference fringe images. This results in the inability to specifically adjust the assembly and adjustment state of the aspherical lens when compensating for assembly and adjustment errors, requiring multiple repeated measurements and adjustments, which is time-consuming, labor-intensive, and inefficient. Therefore, there is an urgent need for a method that can accurately identify and separate assembly and adjustment errors of aspherical lenses to achieve targeted error identification, separation, and compensation, thereby improving the assembly and adjustment efficiency of aspherical lenses.

[0004] In view of this, this application proposes a system and method for separating assembly and adjustment errors in aspherical lenses using laser interferometry detection. Summary of the Invention

[0005] To achieve the above objectives, this application provides a system and method for separating assembly and adjustment errors in aspherical lenses via laser interferometry detection. The specific technical solution is as follows:

[0006] A method for separating assembly and adjustment errors in aspherical lenses using laser interferometry, including:

[0007] The aspherical lens is placed on the clamping platform, and multiple laser interferometry measurements are performed by adjusting different pitch and azimuth angles, and the corresponding interference fringe images are acquired.

[0008] For different types of assembly and adjustment errors, obtain the corresponding typical error stripe images; and extract the key features of each typical error stripe image.

[0009] Image processing and feature extraction are performed on the currently acquired interference fringe image, and the key features are compared with those of typical error fringe images to identify the type and level of assembly and adjustment errors in the current interference fringe image;

[0010] For different types of assembly and adjustment errors, corresponding mathematical models for the errors are constructed and parametrically represented using polynomials.

[0011] Based on the identified types and levels of assembly and adjustment errors, the corresponding assembly and adjustment compensation parameters are calculated inversely using the error mathematical model, and the assembly and adjustment status of the aspherical lens is adjusted accordingly.

[0012] The adjusted aspherical lens is remeasured to obtain a new interference fringe image, which is then analyzed and processed again. An optimization model for the number of error adjustments is constructed, and the minimum number of adjustments is calculated until the required assembly accuracy is achieved.

[0013] Preferably, the aspherical lens is placed on the clamping platform for laser interferometry to form interference fringes;

[0014] By using the clamping platform, the pitch and azimuth angles of the aspherical lens can be changed. The pitch angle represents the angle between the surface normal of the aspherical lens and the optical axis of the interferometer, and the azimuth angle represents the angle of rotation of the aspherical lens around the optical axis.

[0015] By changing the pitch and azimuth angles, interference fringes of the aspherical lens under different laser incident angles were obtained, and interference fringes images were acquired using a CCD camera.

[0016] Preferably, an interference fringe image library of assembly and adjustment errors is established, wherein the types of assembly and adjustment errors include tilt error and eccentricity error;

[0017] Obtain fringe images of different levels of tilt error and eccentricity error, and extract key features of each typical error fringe image, including fringe density and fringe symmetry;

[0018] The convolutional neural network model was trained using the labeled typical error stripe images and their corresponding key features.

[0019] Preferably, key features of the current interference fringe image are extracted, including fringe density and fringe symmetry;

[0020] The gradient change in stripe density is measured by calculating the rate of change in stripe spacing; the asymmetry of stripe distribution is measured by comparing the differences in stripe density in different quadrants.

[0021] Preferably, the current interference fringe image and the extracted fringe density and fringe symmetry are combined into a hybrid data matrix, which is used as the input to the convolutional neural network model;

[0022] The trained convolutional neural network model is used to classify and identify the mixed data matrix. The convolutional neural network model extracts the spatial features of the interference fringe image through convolutional and pooling layers, and fuses the extracted fringe density and fringe symmetry features through fully connected layers to output the probability value of the current interference fringe image belonging to tilt error, eccentricity error or surface distortion error.

[0023] The type of assembly and adjustment error with the highest probability value is taken as the identification result, and the error level is divided into three levels: low, medium, and high according to the magnitude of the probability value.

[0024] Preferably, an error mathematical model is constructed, which includes a tilt error mathematical model and an eccentricity error mathematical model;

[0025] For each identified type of assembly and adjustment error, select the corresponding error mathematical model and calculate the initial values ​​of the error mathematical model parameters according to the error level; for tilt error, calculate the range of rotation angle according to the error level; for eccentricity error, calculate the range of translation distance according to the error level.

[0026] The parameters of the error mathematical model are fitted and optimized using the least squares method.

[0027] Preferably, based on the constructed error mathematical model, a mapping relationship is established between the adjustment compensation parameters and the adjustment state of the aspherical lens;

[0028] For tilt error, the inverse transformation of rotation is used to convert the adjustment compensation parameters into the adjustment amount of the pitch and azimuth angles of the aspherical lens on the mounting platform.

[0029] For eccentricity error, the inverse transformation of translation is used to convert the mounting compensation parameters into the lateral position adjustment of the aspherical lens on the mounting platform.

[0030] Preferably, for assembly and adjustment states with different error levels, an optimization model for the number of error adjustments is constructed, and the optimal number of error compensations is adaptively determined based on the error level and compensation accuracy requirements.

[0031] Preferably, the number of error compensation attempts and the convergence conditions are determined based on the error level and the required compensation accuracy.

[0032] After each compensation for the aspherical lens's adjustment error, the aspherical lens is remeasured to obtain the adjusted interference fringe image. The analysis and processing of the adjustment error are repeated to evaluate the compensation effect until the adjustment error meets the requirements.

[0033] An aspherical lens laser interferometric testing and assembly error separation system is used to implement the aspherical lens laser interferometric testing and assembly error separation method, including: a multi-angle interferometric image acquisition module, a typical error feature extraction module, an error identification and classification module, an error modeling parameterization module, an error assembly compensation module, and a feedback optimization and adjustment module;

[0034] The multi-angle interference image acquisition module places the aspherical lens on the clamping platform, performs multiple laser interference measurements by adjusting different pitch and azimuth angles, and acquires the corresponding interference fringe images.

[0035] The typical error feature extraction module acquires corresponding typical error stripe images for different assembly and adjustment error types, and extracts key features for each typical error stripe image.

[0036] The error identification and classification module performs image processing and feature extraction on the currently acquired interference fringe image, compares it with the key features of typical error fringe images, and identifies the type and level of assembly and adjustment errors in the current interference fringe image.

[0037] The error modeling parameterization module constructs corresponding error mathematical models for different types of assembly and adjustment errors, and uses polynomials for parameterization representation.

[0038] The error adjustment compensation module, based on the identified adjustment error type and level, calculates the corresponding adjustment compensation parameters through an error mathematical model and adjusts the adjustment state of the aspherical lens.

[0039] The feedback optimization and adjustment module remeasures the adjusted aspherical lens to obtain a new interference fringe image, analyzes and processes it again, constructs an error adjustment number optimization model, calculates the minimum number of error adjustments, until the assembly accuracy requirements are met.

[0040] The beneficial effects of this application are: by measuring from multiple angles and directions, this application can comprehensively obtain the surface morphology information of aspherical lenses, providing sufficient data support for subsequent error analysis.

[0041] This application establishes a correspondence between assembly and adjustment errors and stripe image features by acquiring typical error stripe images of different assembly and adjustment error types and extracting their key features.

[0042] This application uses image processing and feature extraction to transform the currently acquired interference fringe image into quantifiable feature data; by comparing the features with those of typical error fringe images, the type and severity of assembly and adjustment errors present in the current interference fringe image can be accurately identified.

[0043] This application constructs a mathematical model of assembly and adjustment errors and uses polynomials for parameterization, transforming complex error patterns into a few key parameters, which greatly simplifies the error description and compensation process and improves assembly and adjustment efficiency.

[0044] Based on the identified specific types and levels of assembly and adjustment errors, this application uses an error mathematical model to solve for the corresponding assembly and adjustment compensation parameters, thereby precisely adjusting the assembly and adjustment state of aspherical lenses and eliminating or reducing assembly and adjustment errors.

[0045] This application evaluates the adjustment effect by re-measuring and analyzing the adjusted aspherical lens, and calculates the minimum number of adjustments required to achieve the target accuracy by constructing an error adjustment number optimization model, thus avoiding blind or over-adjustment and improving assembly efficiency. Attached Figure Description

[0046] Figure 1 This is a flowchart of the method for separating assembly and adjustment errors in laser interferometry testing of aspherical lenses in this application;

[0047] Figure 2 Flowchart for constructing the assembly and adjustment error interference fringe image library provided in this application;

[0048] Figure 3 The flowchart for error identification of interference fringe images provided in this application;

[0049] Figure 4 The flowchart for the execution of the error compensation control strategy provided in this application;

[0050] Figure 5 This is a structural diagram of the aspherical lens laser interferometry detection and adjustment error separation system of this application. Detailed Implementation

[0051] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0052] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0053] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of this application. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that excludes other embodiments.

[0054] Example 1:

[0055] Reference Figure 1 This is the first embodiment of the present application, which provides a method for separating assembly and adjustment errors in aspherical lenses using laser interferometry detection.

[0056] Step 1: Place the aspherical lens on the clamping platform, perform multiple laser interferometry measurements by adjusting different pitch and azimuth angles, and acquire the corresponding interference fringe images.

[0057] Place the aspherical lens on the rotatable clamping platform and ensure that the optical axis of the aspherical lens coincides with the rotation axis of the clamping platform. Adjust the position and angle of the clamping platform so that the vertex of the aspherical lens is located on the reference plane of the interferometer and the normal direction of the surface of the aspherical lens is parallel to the reference optical axis of the interferometer.

[0058] Using a HeNe laser as the light source for the interferometer, the laser beam is split, with one beam illuminating a reference plane and the other illuminating the surface of an aspherical mirror. The two beams interfere after being reflected from the surface of the aspherical mirror, forming interference fringes. Using a HeNe laser can obtain a stable, monochromatic, and coherent light source, ensuring the accuracy of the interferometric measurement.

[0059] The pitch angle of the aspherical lens can be changed by rotating the clamping platform. and azimuth ,in , Pitch angle The azimuth angle represents the angle between the surface normal of the aspherical lens and the optical axis of the interferometer. This indicates the angle of rotation of an aspherical lens about its optical axis, achieved by changing... and The surface shape of aspherical lenses is measured at different laser incident angles to obtain more comprehensive measurement data.

[0060] In each The location was determined, and interference fringe images were acquired using a CCD camera. The acquired images were then processed with background noise reduction and image enhancement to improve the signal-to-noise ratio and fringe contrast.

[0061] In different Multiple interference fringe images were acquired at the location, and the acquired interference fringe images were numbered and archived, recording the corresponding data for each image. Location parameters; by establishing the correspondence between images and location parameters, they are used for subsequent data processing and analysis.

[0062] This step allows us to obtain interference fringe images of the aspherical lens at different incident angles, providing ample data support for subsequent assembly and adjustment error analysis. Furthermore, optimizing measurement parameters and the image acquisition process can improve the accuracy and reliability of the measurement.

[0063] Step 2: For different types of assembly and adjustment errors, obtain the corresponding typical error fringe images; and extract the key features of each typical error fringe image, see [link / reference]. Figure 2 The flowchart below shows the construction process for the assembly and adjustment error interference fringe image library in this step.

[0064] An interference fringe image library for assembly and adjustment errors is established. The assembly and adjustment error types include tilt error and eccentricity error, which are two typical error types. The tilt error refers to the fact that the normal direction of the aspherical lens is not parallel to the reference optical axis of the interferometer. The eccentricity error refers to the fact that the vertex of the aspherical lens does not coincide with the reference plane of the interferometer. By establishing the interference fringe image library for assembly and adjustment errors, a reference standard is provided for subsequent error identification.

[0065] Obtain fringe images of different levels of tilt error and eccentricity error. For tilt error, the tilt angle of the aspherical lens is changed. Simulate different levels of error; for example, the tilt angle range is... Step size is For eccentricity error, different levels of error are simulated by changing the offset of the aspherical lens vertex relative to the reference plane. By obtaining typical error fringe images under different error levels, an assembly error interference fringe image library is established.

[0066] Image processing is performed on each typical error fringe image, including Gaussian filtering, adaptive threshold binarization, and morphological operations. For example, Gaussian filtering uses... The convolutional kernel, with a standard deviation of 1.5, is used to suppress high-frequency noise; adaptive thresholding binarization uses local thresholding with a window size of 11×11 to improve stripe contrast; morphological closing operation uses 3×3 rectangular structuring elements to fill stripe discontinuities. Through image processing, a high-quality input image is provided for subsequent feature extraction.

[0067] Key features of each typical error fringe image are extracted, including fringe density and fringe symmetry.

[0068] The gradient change in fringe density is measured by calculating the rate of change of fringe spacing. For eccentricity errors, the image is divided into four quadrants, and the fringe density in each quadrant is calculated separately. The asymmetry of the fringe distribution is measured by comparing the differences in fringe density across different quadrants. By extracting key features, a correspondence between assembly / adjustment error types and fringe features can be established.

[0069] The extracted stripe features are trained and classified using a convolutional neural network (CNN) model. For example, a CNN model containing two convolutional layers, two pooling layers, and two fully connected layers is constructed, with an input image size of [size missing]. The kernel size is The pooling window size is The activation function used is ReLU, the optimization algorithm used is Adam, and the loss function used is cross-entropy.

[0070] Tilting error and eccentricity error are treated as two categories. Key features of typical error fringe images are integrated with the images, and the typical error fringe images are labeled. The labeled typical error fringe images and their corresponding key features are used as training samples to train a CNN model. By training the CNN model, image features are learned and used to identify the type of assembly error in interference fringe images.

[0071] The trained CNN model is tested and evaluated using a 5-fold cross-validation method. For example, the image library of interference fringes due to alignment errors is randomly divided into 5 subsets. Four subsets are selected as the training set each time, and the remaining subset is used as the test set. This experiment is repeated 5 times, and the average of the 5 results is taken as the final performance metric. The accuracy, precision, recall, and F1 score of the CNN model on the test set are calculated. For example, the accuracy for identifying tilt and eccentric errors is ensured to be no less than 95%, and the precision and recall no less than 93%. Cross-validation is used to evaluate the generalization ability and robustness of the CNN model, avoiding overfitting and underfitting.

[0072] This step constructs a database of interference fringe images for assembly and adjustment errors, and extracts key features from each typical error fringe image. Based on a Convolutional Neural Network (CNN) model, an error classification model is built to automatically identify tilt and eccentric errors in the interference fringe images. Compared to traditional machine learning algorithms, the CNN model can directly learn fringe features from images, avoiding the tediousness and subjectivity of manual feature design, and improving the richness and discriminative power of feature representation. Simultaneously, cross-validation is used to comprehensively evaluate the performance of the CNN model, ensuring its accuracy and reliability. This provides crucial input information for subsequent quantitative error analysis and compensation, significantly improving the efficiency and accuracy of assembly and adjustment error separation.

[0073] Step 3: Perform image processing and feature extraction on the currently acquired interference fringe image, compare it with the key features of typical error fringe images, and identify the type and level of assembly and adjustment errors in the current interference fringe image. (See [link / reference]). Figure 3 This is a flowchart of the interference fringe image error identification process for this step.

[0074] Image processing is performed on the currently acquired interference fringe image, including Gaussian filtering, adaptive threshold binarization, and morphological operations. Key features of the current interference fringe image are extracted, including fringe density and fringe symmetry.

[0075] The gradient change in fringe density is measured by calculating the rate of change of fringe spacing. Let the interference fringe image be... The parameters of the fringe lines obtained by Hough transform are: ,in For the first The distance from a straight line to the origin. For the first straight lines and The included angle of the axis, Number the detected striped lines. The total number of fringe lines; the rate of change of fringe density gradient is defined. for:

[0076]

[0077] in, For two adjacent straight lines in The spacing in a direction can be expressed using linear parameters. and Calculated; The larger the value, the more drastic the change in fringe density along the tilt direction, and the more severe the tilt error.

[0078] For eccentricity error, the image is divided into four quadrants, and the fringe density in each quadrant is calculated. The asymmetry of the fringe distribution is measured by comparing the differences in fringe density between different quadrants; let the ... The number of stripe pixels in each quadrant is The quadrant area is , Define the fringe density asymmetry index. for:

[0079]

[0080] in, To find the minimum value function, The range of values ​​is The closer the value is to 1, the greater the difference in fringe density in different quadrants, and the more severe the eccentricity error.

[0081] Current interference fringe image and extracted features , The data is combined into a mixed dataset and used as input to a convolutional neural network (CNN) model. Specifically, the mixed data is... Normalization to Within the range, a normalized image is obtained. ;Will , and Concatenate them into a three-channel hybrid data matrix As input to the CNN model, that is: .

[0082] Using a trained CNN model, on the mixed data matrix Classification and recognition are performed; the CNN model extracts spatial features of the interference fringe image through convolutional and pooling layers, and fuses the extracted features through fully connected layers. and Output the probability value of whether the current interference fringe image belongs to tilt error, eccentricity error or surface distortion error; take the assembly error type with the highest probability value as the recognition result, and divide the error level into three levels: low, medium and high according to the magnitude of the probability value.

[0083] By combining the original image and extracted features into hybrid data, the error characteristics of interference fringe images can be more comprehensively identified, improving the classification performance of CNN models. Simultaneously, normalizing the features to the same range as the image balances the scale differences between different types of data, promoting the convergence and generalization of CNN models.

[0084] For example, the specific criteria for classifying error levels are as follows: When the maximum probability value output by the CNN model is greater than or equal to 0.85, it indicates that the model has a high degree of confidence in this type of assembly error, and the error feature is obvious and strong in the interference fringe image, therefore it is classified as a "high-level error" and should be prioritized; when the probability value is within... When the probability value is between [ ], it indicates that the model's identification of this assembly error type is relatively reliable, but it may still be affected by other features, and is therefore judged as a "medium-level error." Further confirmation using other criteria is recommended; when the probability value is between [ ], it indicates that the model's identification of this assembly error type is relatively reliable, but it may still be affected by other features, and is therefore judged as a "medium-level error." When the error level is between 0 and 1, it indicates that the error characteristics are not significant enough or overlap with other types, resulting in a low confidence level for identification. It is classified as a "low-level error" and should be handled with caution. The numerical values ​​for classifying error levels can be set according to requirements.

[0085] For example, if the maximum probability value output by the CNN model is greater than a preset threshold (e.g., 0.85), the current interference fringe image is marked as an unrecognizable error type and classified as surface distortion error. Surface distortion error refers to a large deviation between the actual shape and the designed shape of an aspherical lens, resulting in irregular distortion and deformation of the interference fringe image, which is difficult to describe using tilt error or eccentricity error. By introducing the concept of surface distortion error, the completeness and accuracy of error identification are improved, avoiding the incorrect classification of special error types as tilt error or eccentricity error.

[0086] This step utilizes image preprocessing, feature extraction, and CNN classification techniques to automatically process and identify errors in the acquired interference fringe images; it also extracts the fringe density gradient change rate. and stripe density asymmetry index As a key feature, the severity of tilt and eccentricity errors is quantified; a CNN model is used to classify the features and output probability values ​​for error type and level; the concept of surface distortion error is also introduced to improve the completeness of error identification.

[0087] Step 4: For different types of assembly and adjustment errors, construct corresponding error mathematical models and use polynomials for parameterization.

[0088] Construct mathematical models for the errors, including mathematical models for tilt error and eccentricity error; construct the mathematical model for tilt error, assuming the surface shape equation of the aspherical lens is... , It is a function describing the surface shape of an aspherical lens, treating the tilt error as a function of the surface shape equation around... shaft and Rotational transformations of the axes, with rotation angles of respectively and Then the surface shape equation under tilt error It can be represented as:

[0089]

[0090] in, and The angle parameter of the tilt error reflects the severity of the tilt error. By establishing a mathematical model of the tilt error, the error is correlated with the surface shape equation of the aspherical lens, providing a theoretical basis for subsequent error compensation.

[0091] Construct a mathematical model for eccentricity error, and assume the surface shape equation of the aspherical lens is: The eccentricity error is considered as the surface equation in direction and Translation transformations in the direction, with translation distances of respectively and Then the surface shape equation under eccentricity error Represented as: ;in and The distance parameter represents the eccentricity error, reflecting its severity. By establishing a mathematical model of the eccentricity error, the error can be correlated with the surface shape equation of the aspherical lens, providing a theoretical basis for subsequent error compensation.

[0092] For each identified assembly and adjustment error type, the corresponding error mathematical model is selected, and the initial values ​​of the error mathematical model parameters are calculated according to the error level; for tilt errors, the rotation angle is calculated according to the error level. and The range; for eccentricity error, calculate the translation distance according to the error level. and The range.

[0093] The least squares method is used to fit and optimize the parameters of the error mathematical model. For example, taking tilt error as an example, the optimal rotation angle is solved by minimizing the sum of squared residuals between the actual measured interference fringe image and the interference fringe image generated by the tilt error mathematical model. and ,Right now:

[0094]

[0095] in, This is the actual measured interference fringe image. and The parameter represents the index value; Interference fringe image generated for the mathematical model of tilt error. and The number of rows and columns of the image is given; similarly, the parameters of the mathematical models for eccentricity error and surface distortion error are optimized and solved; through parameter optimization, the mathematical model that best fits the actual error can be obtained, providing quantitative indicators for subsequent error compensation.

[0096] This step constructs corresponding mathematical models for different types of assembly and adjustment errors and represents them using a polynomial parameterization method, thereby achieving a quantitative description and modeling of assembly and adjustment errors. Through the estimation, fitting optimization, and verification of the error mathematical model parameters, an accurate and reliable error mathematical model is obtained, providing a theoretical basis and quantitative indicators for subsequent assembly and adjustment error compensation.

[0097] Step 5: Based on the identified type and level of assembly and adjustment error, the corresponding assembly and adjustment compensation parameters are calculated inversely using the error mathematical model, and the assembly and adjustment status of the aspherical lens is adjusted.

[0098] Based on the identified assembly and adjustment error types and levels, and the constructed error mathematical model, the error parameters that need to be compensated are calculated. For tilt error, the parameter that needs to be compensated is the rotation angle. and For eccentricity error, the parameter that needs to be compensated is the translation distance. and By specifying the compensation parameters, the assembly and adjustment errors can be corrected and compensated.

[0099] Based on the constructed error mathematical model, a mapping relationship is established between the adjustment compensation parameters and the adjustment state of the aspherical lens; for tilt error, the adjustment compensation parameters are transformed through the inverse transformation of rotation. and This translates to the adjustment of the pitch and azimuth angles of the aspherical lens on the mounting platform. and ,Right now: .

[0100] For eccentricity error, the assembly and adjustment compensation parameters are obtained through the inverse transformation of the translation transformation. and This is converted into the lateral position adjustment of the aspherical lens on the mounting platform, i.e.: , and These are the adjustments made to the horizontal and vertical positions of the lens after the inverse transformation.

[0101] By establishing a mapping relationship between assembly and adjustment compensation parameters and assembly and adjustment status, the error mathematical model is linked to the actual assembly and adjustment operation, guiding the precise adjustment and optimization of the assembly and adjustment process.

[0102] Based on the mapping relationship between the adjustment compensation parameters and the adjustment state, the adjustment state of the aspherical lens is adjusted. For tilt and eccentricity errors, the rotation and translation errors are compensated by adjusting the angle and position of the aspherical lens on the clamping platform; specifically, the pitch and azimuth angles of the lens are adjusted respectively. and Adjust the lateral position of the lenses respectively and By precisely adjusting the mounting and adjustment settings, the mounting and adjustment errors of aspherical lenses can be effectively compensated and corrected.

[0103] This step, based on the identified assembly and adjustment error types and the constructed error mathematical model, establishes a mapping relationship between the assembly and adjustment compensation parameters and the assembly and adjustment state by reverse calculation. Based on this, the assembly and adjustment state of the aspherical lens is precisely adjusted and optimized, achieving effective compensation and correction of assembly and adjustment errors. Furthermore, through closed-loop compensation evaluation and optimization, as well as rigorous final inspection and acceptance, the processing accuracy and performance indicators of the aspherical lens can be continuously improved to meet the application requirements of high-precision optical systems.

[0104] Step 6: Remeasure the adjusted aspherical lens to obtain a new interference fringe image, analyze and process it again, construct an optimization model for the number of error adjustments, calculate the minimum number of adjustments until the required assembly accuracy is achieved. (See [link to documentation]). Figure 4 This is the flowchart for the error compensation control strategy execution in this step.

[0105] Based on the error level and compensation accuracy requirements, determine the assembly and adjustment status, and adjust the number of error compensation cycles and convergence conditions. For low-level errors, adjustments can be made in one step without iteration; for high-level errors, multiple iterations are required, with each adjustment gradually decreasing until the compensation accuracy requirements are met. By setting reasonable error compensation cycles and convergence conditions, assembly and adjustment efficiency can be improved while ensuring compensation accuracy, avoiding the waste of time and costs caused by over-adjustment.

[0106] Specifically, for assembly and adjustment of different error levels, an optimization model for the number of error adjustments is constructed. Based on the error level and compensation accuracy requirements, the optimal number of error compensations is adaptively determined to improve assembly and adjustment efficiency while ensuring compensation accuracy.

[0107] Define error level quantification indicators This is used to measure the severity of assembly and adjustment errors; Calculations are performed based on the identified error characteristics: .

[0108] in, and These are the rate of change of the fringe density gradient and the fringe density asymmetry index for the current error, respectively. and The corresponding threshold; and These are weighting coefficients used to balance the contributions of different error characteristics.

[0109] Quantitative indicators based on error level Establish the number of error compensation cycles The optimization model; Should follow The value increases with the increase of , but the rate of increase should gradually slow down to avoid too many invalid iterations; an exponential decay function model is constructed to describe it. and Relationship: ;in To determine the maximum number of error compensation attempts, This is the attenuation coefficient, used to control the rate at which the number of error compensation cycles increases.

[0110] Introducing a compensation accuracy factor Number of error compensations Perform adaptive adjustments: ,in, Number of error compensations The correction value after adaptive adjustment.

[0111] For tilt error, let the compensation accuracy factor be... , Using the rate of change of the fringe density gradient of the current error With target value Expressed as a ratio: ;in, The target value for the stripe density gradient change rate can be set according to actual needs and processing capabilities. When When the current tilt error does not meet the compensation accuracy requirements, it indicates that the iteration needs to continue; when When the current tilt error meets the compensation accuracy requirements, the iteration can be terminated.

[0112] For eccentricity error, let the compensation accuracy factor be... , Using the current error as an index of fringe density asymmetry With target value Expressed as a ratio: ;in, The target value for the stripe density asymmetry index can be set according to actual needs and processing capabilities. When the error does not reach the required compensation accuracy, it indicates that the current eccentricity error has not met the compensation accuracy requirements and further iteration is needed; when When the current eccentricity error meets the compensation accuracy requirements, the iteration can be terminated.

[0113] Set the iterative convergence condition, when continuous... The compensation accuracy factor of the next iteration When all values ​​are less than or equal to 1, the iteration process is considered to have converged and the iteration can be terminated. The value can be set according to actual needs and experience, usually 3 to 5.

[0114] In each iteration, based on the current compensation precision factor The compensation parameter adjustment amount for the next iteration is dynamically adjusted, for example:

[0115]

[0116]

[0117] in, This represents the current number of error compensation attempts. , , and For the first The adjustment amount of the assembly and adjustment compensation parameters in the next iteration. , , and For the first The adjustment amount of the assembly and adjustment compensation parameters in the next iteration; by adaptively adjusting the compensation parameters, the iteration convergence speed can be accelerated and the assembly and adjustment efficiency can be improved.

[0118] By constructing an optimization model for the number of error adjustments, the number of error compensations and convergence conditions are adaptively determined based on the error level and compensation accuracy factor. The assembly and adjustment compensation parameters are dynamically adjusted, which can minimize the number of assembly and adjustment error compensations while ensuring compensation accuracy and improving assembly and adjustment efficiency.

[0119] After each adjustment of the assembly state, the aspherical lens is remeasured to obtain the adjusted interference fringe image. The analysis and processing of assembly errors are repeated to evaluate the compensation effect. If the type and level of the assembly error change, or the error amplitude decreases significantly, the compensation adjustment is effective. If the type and level of the assembly error remain unchanged, or the error amplitude does not improve significantly, the assembly compensation parameters and adjustment strategy need to be further optimized until a satisfactory compensation effect is achieved.

[0120] This step precisely compensates for tilt and eccentricity errors by adjusting the angle and position of the aspherical lens on the clamping platform, thereby improving processing accuracy and surface quality. By combining the error level and compensation accuracy factor, the number of error compensation cycles is adaptively adjusted to avoid redundant iterations and improve assembly efficiency. Furthermore, by setting convergence conditions and dynamic compensation strategies, efficient and stable correction of assembly errors is achieved. The compensation effect is evaluated in a timely manner through a feedback closed-loop mechanism, realizing intelligent and automated assembly error compensation.

[0121] Example 2:

[0122] Reference Figure 5 This is the second embodiment of the present application, which provides an aspherical lens laser interferometry detection and adjustment error separation system.

[0123] The system includes: a multi-angle interferometric image acquisition module, a typical error feature extraction module, an error identification and classification module, an error modeling and parameterization module, an error assembly and adjustment compensation module, and a feedback optimization and adjustment module.

[0124] The multi-angle interference image acquisition module places the aspherical lens on the clamping platform, performs multiple laser interference measurements by adjusting different pitch and azimuth angles, and acquires the corresponding interference fringe images.

[0125] The typical error feature extraction module acquires corresponding typical error stripe images for different assembly and adjustment error types, and extracts key features for each typical error stripe image.

[0126] The error identification and classification module performs image processing and feature extraction on the currently acquired interference fringe image, compares it with the key features of typical error fringe images, and identifies the type and level of assembly and adjustment errors in the current interference fringe image.

[0127] The error modeling parameterization module constructs corresponding error mathematical models for different types of assembly and adjustment errors, and uses polynomials for parameterization.

[0128] The error adjustment compensation module, based on the identified adjustment error type and level, uses an error mathematical model to inversely calculate the corresponding adjustment compensation parameters and adjusts the adjustment state of the aspherical lens.

[0129] The feedback optimization and adjustment module remeasures the adjusted aspherical lens to obtain a new interference fringe image, analyzes and processes it again, constructs an error adjustment number optimization model, calculates the minimum number of error adjustments, until the assembly accuracy requirements are met.

[0130] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings or direct couplings or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0131] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of this application without departing from the spirit and scope of protection of the claims. All of these variations are within the protection scope of this application.

Claims

1. A method for separating assembly and adjustment errors in aspherical lenses using laser interferometry, characterized in that, include: The aspherical lens is placed on the clamping platform, and multiple laser interferometry measurements are performed by adjusting different pitch and azimuth angles, and the corresponding interference fringe images are acquired. For different types of assembly and adjustment errors, obtain the corresponding typical error stripe images; and extract the key features of each typical error stripe image. Image processing and feature extraction are performed on the currently acquired interference fringe image, and the key features are compared with those of typical error fringe images to identify the type and level of assembly and adjustment errors in the current interference fringe image; For different types of assembly and adjustment errors, corresponding mathematical models for the errors are constructed and parametrically represented using polynomials. Based on the identified types and levels of assembly and adjustment errors, the corresponding assembly and adjustment compensation parameters are calculated inversely using the error mathematical model, and the assembly and adjustment status of the aspherical lens is adjusted accordingly. The adjusted aspherical lens is remeasured to obtain a new interference fringe image, which is then analyzed and processed again. An optimization model for the number of error adjustments is constructed, and the minimum number of adjustments is calculated until the required assembly accuracy is achieved.

2. The method for separating assembly and adjustment errors in aspherical lenses using laser interferometry detection according to claim 1, characterized in that, An aspherical lens is placed on a clamping platform for laser interferometry to form interference fringes. By using the clamping platform, the pitch and azimuth angles of the aspherical lens can be changed. The pitch angle represents the angle between the surface normal of the aspherical lens and the optical axis of the interferometer, and the azimuth angle represents the angle of rotation of the aspherical lens around the optical axis. By changing the pitch and azimuth angles, interference fringes of the aspherical lens under different laser incident angles were obtained, and interference fringes images were acquired using a CCD camera.

3. The method for separating assembly and adjustment errors in aspherical lenses using laser interferometry detection according to claim 2, characterized in that, Establish an image library of assembly and adjustment errors, wherein the types of assembly and adjustment errors include tilt error and eccentricity error; Obtain fringe images of different levels of tilt error and eccentricity error, and extract key features of each typical error fringe image, including fringe density and fringe symmetry; The convolutional neural network model was trained using the labeled typical error stripe images and their corresponding key features.

4. The method for separating assembly and adjustment errors in aspherical lenses using laser interferometry detection according to claim 1, characterized in that, Extract key features from the current interference fringe image, including fringe density and fringe symmetry; The gradient change in stripe density is measured by calculating the rate of change in stripe spacing; the asymmetry of stripe distribution is measured by comparing the differences in stripe density in different quadrants.

5. The method for separating assembly and adjustment errors in aspherical lenses using laser interferometry according to claim 4, characterized in that, The current interference fringe image and the extracted fringe density and fringe symmetry are combined into a hybrid data matrix, which is then used as input to the convolutional neural network model. The trained convolutional neural network model is used to classify and identify the mixed data matrix. The convolutional neural network model extracts the spatial features of the interference fringe image through convolutional and pooling layers, and fuses the extracted fringe density and fringe symmetry features through fully connected layers to output the probability value of the current interference fringe image belonging to tilt error, eccentricity error or surface distortion error. The type of assembly and adjustment error with the highest probability value is taken as the identification result, and the error level is divided into three levels: low, medium, and high according to the magnitude of the probability value.

6. The method for separating assembly and adjustment errors in aspherical lenses using laser interferometry detection according to claim 5, characterized in that, Construct an error mathematical model, which includes a tilt error mathematical model and an eccentricity error mathematical model; For the identified assembly and adjustment error types, select the corresponding error mathematical model and calculate the initial values ​​of the error mathematical model parameters according to the error level; For tilt error, calculate the range of rotation angle based on the error level; For eccentricity error, calculate the range of translation distance based on the error level; The parameters of the error mathematical model are fitted and optimized using the least squares method.

7. The method for separating assembly and adjustment errors in aspherical lenses using laser interferometry according to claim 6, characterized in that, Based on the constructed error mathematical model, a mapping relationship between the adjustment compensation parameters and the adjustment state of the aspherical lens is established. For tilt error, the inverse transformation of rotation is used to convert the adjustment compensation parameters into the adjustment amount of the pitch and azimuth angles of the aspherical lens on the mounting platform. For eccentricity error, the inverse transformation of translation is used to convert the mounting compensation parameters into the lateral position adjustment of the aspherical lens on the mounting platform.

8. The method for separating assembly and adjustment errors in aspherical lenses using laser interferometry according to claim 7, characterized in that, For assembly and adjustment of different error levels, an optimization model for the number of error adjustments is constructed, and the optimal number of error compensations is adaptively determined based on the error level and compensation accuracy requirements.

9. The method for separating assembly and adjustment errors in aspherical lenses using laser interferometry detection according to claim 8, characterized in that, Based on the error level and compensation accuracy requirements, determine the number of error compensation attempts and the convergence conditions; After each compensation for the aspherical lens's adjustment error, the aspherical lens is remeasured to obtain the adjusted interference fringe image. The analysis and processing of the adjustment error are repeated to evaluate the compensation effect until the adjustment error meets the requirements.

10. An aspherical lens laser interferometry testing and assembly error separation system, used to implement the aspherical lens laser interferometry testing and assembly error separation method according to any one of claims 1 to 9, characterized in that, include: The system includes a multi-angle interferometric image acquisition module, a typical error feature extraction module, an error identification and classification module, an error modeling and parameterization module, an error assembly and adjustment compensation module, and a feedback optimization and adjustment module. The multi-angle interference image acquisition module places the aspherical lens on the clamping platform, performs multiple laser interference measurements by adjusting different pitch and azimuth angles, and acquires the corresponding interference fringe images. The typical error feature extraction module acquires corresponding typical error stripe images for different assembly and adjustment error types, and extracts key features for each typical error stripe image. The error identification and classification module performs image processing and feature extraction on the currently acquired interference fringe image, compares it with the key features of typical error fringe images, and identifies the type and level of assembly and adjustment errors in the current interference fringe image. The error modeling parameterization module constructs corresponding error mathematical models for different types of assembly and adjustment errors, and uses polynomials for parameterization representation. The error adjustment compensation module, based on the identified adjustment error type and level, calculates the corresponding adjustment compensation parameters through an error mathematical model and adjusts the adjustment state of the aspherical lens. The feedback optimization and adjustment module remeasures the adjusted aspherical lens to obtain a new interference fringe image, analyzes and processes it again, constructs an error adjustment number optimization model, calculates the minimum number of error adjustments, until the assembly accuracy requirements are met.

Images