An optical lens geometric parameter measurement method and system based on industrial vision

Abstract

The application provides an optical lens geometric parameter measurement method and system based on industrial vision, and relates to the technical field of industrial vision measurement. The technical solution points are: performing gray feature analysis on initial image data to obtain a gray feature analysis result; triggering a polarization modulation instruction to obtain multiple frames of polarization modulation image data; based on the gray feature analysis result, obtaining an overlapping division region, respectively performing feature separation processing on the part of the overlapping division region corresponding to each polarization modulation image data to obtain a first feature set and a second feature set; determining a front surface feature set and a back surface feature set according to the first feature set representation and the second feature set representation; and performing geometric parameter calculation according to the front surface feature set and the back surface feature set to obtain a geometric parameter measurement result. The application aims to identify and process the front and back surface feature overlap of a thin and high-transmittance lens, and improve the accuracy and reliability of optical lens geometric parameter measurement.

Description

Technical Field

[0001] This application relates to the field of industrial vision measurement technology, and more specifically, to a method and system for measuring the geometric parameters of optical lenses based on industrial vision. Background Technology

[0002] In the measurement of optical lenses after processing, non-contact vision measurement systems calculate the center thickness by acquiring images of the central region and extracting reflection features from the front and rear surfaces, thus avoiding contact wear. However, for thin, high-transmittance lenses, the projection distance between the front and rear surfaces of the central region is small, and the reflection intensities are similar. When illuminating paraxially, double-layer images are prone to superposition or proximity. In related technologies, the measurement system struggles to reliably determine the surface attribution of feature points, potentially misidentifying superimposed signals as single-layer or incorrectly paired features. This leads to thickness calculations based on incorrect spatial relationships, resulting in the output center thickness data deviating from the true value and causing a shift in the calculation reference of associated curvature parameters. Summary of the Invention

[0003] The purpose of this application is to provide a method and system for measuring the geometric parameters of optical lenses based on industrial vision, which aims to identify and process the overlapping features of the front and rear surfaces of thin, high-transmittance lenses, thereby improving the accuracy and reliability of measuring the geometric parameters of optical lenses.

[0004] In a first aspect, this application provides a method for measuring the geometric parameters of optical lenses based on industrial vision, including: Initial image data of the central region of the optical lens under test is acquired, and grayscale feature analysis is performed on the initial image data to obtain grayscale feature analysis results; the grayscale feature analysis results include grayscale peak points and grayscale profile morphology; If, based on the grayscale feature analysis results, it is determined that there is an overlap between the front surface features and the rear surface features of the lens, a polarization modulation command is triggered to apply various different polarization modulation conditions to the illumination source of the optical lens measurement system and to acquire images, so as to obtain multi-frame polarization modulation image data corresponding to various different polarization modulation conditions based on the polarization modulation command. Based on the grayscale feature analysis results, overlapping regions are obtained. Feature separation processing is performed on the parts of the overlapping regions corresponding to each polarization modulation image data to obtain a first feature set and a second feature set. The first feature set includes the first center coordinates corresponding to the overlapping regions and the first polarization response features corresponding to the first center coordinates of each polarization modulation image data. The second feature set includes the second center coordinates corresponding to the overlapping regions and the second polarization response features corresponding to the second center coordinates of each polarization modulation image data. The first center coordinates and the second center coordinates correspond to the front surface or the rear surface of the optical lens under test, respectively. Based on the degree of polarization response fluctuation characterized by the first feature set and the degree of polarization response fluctuation characterized by the second feature set, the front surface feature set corresponding to the front surface of the optical lens under test and the rear surface feature set corresponding to the rear surface of the optical lens under test are determined in the first feature set and the second feature set. Geometric parameters are calculated based on the front surface feature set and the rear surface feature set to obtain the geometric parameter measurement results.

[0005] Secondly, this application provides an optical lens geometric parameter measurement system based on industrial vision, comprising: The grayscale analysis module is used to acquire initial image data of the central region of the optical lens under test, perform grayscale feature analysis on the initial image data, and obtain grayscale feature analysis results; the grayscale feature analysis results include grayscale peak points and grayscale profile morphology; The polarization modulation module is used to trigger a polarization modulation command to apply multiple different polarization modulation conditions to the illumination source of the optical lens measurement system and to acquire images when it is determined from the grayscale feature analysis results that there is an overlap between the front surface features and the rear surface features of the lens. The polarization modulation module is used to obtain multi-frame polarization modulation image data corresponding to multiple different polarization modulation conditions based on the polarization modulation command. The feature separation module is used to obtain overlapping regions based on the grayscale feature analysis results, and to perform feature separation processing on the parts of the overlapping regions corresponding to each polarization modulation image data to obtain a first feature set and a second feature set. The first feature set includes a first center coordinate corresponding to the overlapping region and a first polarization response feature corresponding to the first center coordinate of each polarization modulation image data. The second feature set includes a second center coordinate corresponding to the overlapping region and a second polarization response feature corresponding to the second center coordinate of each polarization modulation image data. The first center coordinate and the second center coordinate correspond to the front surface or the rear surface of the optical lens under test, respectively. The surface analysis module is used to determine the front surface feature set of the front surface of the optical lens under test and the rear surface feature set of the optical lens under test based on the polarization response fluctuation degree characterized by the first feature set and the polarization response fluctuation degree characterized by the second feature set. The parameter calculation module is used to calculate geometric parameters based on the front surface feature set and the rear surface feature set to obtain geometric parameter measurement results.

[0006] As can be seen from the above, the optical lens geometric parameter measurement method based on industrial vision provided in this application can identify the overlapping of front and rear surface features through grayscale feature analysis. By applying multiple polarization modulation conditions and acquiring multiple frames of images, combined with feature separation processing, and using the degree of polarization response fluctuation as a criterion, the separated features can be accurately assigned to the front or rear surface, avoiding geometric parameter calculation errors caused by misjudgment, obtaining the true front and rear surface feature points, and thus performing accurate geometric parameter calculations, improving the measurement accuracy and reliability of the center thickness and related geometric parameters of thin, high-transmittance lenses. Attached Figure Description

[0007] Figure 1 This is a flowchart of a method for measuring the geometric parameters of optical lenses based on industrial vision, provided in one embodiment of this application; Figure 2 This is a flowchart of a method for measuring the geometric parameters of optical lenses based on industrial vision, provided in another embodiment of this application; Figure 3 This is a flowchart of a method for measuring the geometric parameters of optical lenses based on industrial vision, provided in another embodiment of this application. Detailed Implementation

[0008] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0009] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0010] like Figure 1 As shown, Figure 1 This is a flowchart of a method for measuring the geometric parameters of an optical lens based on industrial vision, provided in one embodiment of this application. The method may include, but is not limited to, steps S110 to S150.

[0011] Step S110: Obtain initial image data of the central region of the optical lens to be tested, perform grayscale feature analysis on the initial image data, and obtain grayscale feature analysis results; the grayscale feature analysis results include grayscale peak points and grayscale profile morphology. Step S120: If it is determined from the grayscale feature analysis results that there is an overlap between the front surface features and the rear surface features of the lens, trigger a polarization modulation command to apply various different polarization modulation conditions to the illumination source of the optical lens measurement system and to acquire images, so as to obtain multi-frame polarization modulation image data corresponding to various different polarization modulation conditions based on the polarization modulation command. Step S130: Based on the grayscale feature analysis results, the overlapping division region is obtained. Feature separation processing is performed on the part of the overlapping division region corresponding to each polarization modulation image data to obtain a first feature set and a second feature set. The first feature set includes the first center coordinates corresponding to the overlapping division region and the first polarization response feature corresponding to the first center coordinates of each polarization modulation image data. The second feature set includes the second center coordinates corresponding to the overlapping division region and the second polarization response feature corresponding to the second center coordinates of each polarization modulation image data. The first center coordinates and the second center coordinates correspond to the front surface or the rear surface of the optical lens under test, respectively. Step S140: Based on the degree of polarization response fluctuation characterized by the first feature set and the degree of polarization response fluctuation characterized by the second feature set, determine the front surface feature set of the front surface of the optical lens under test and the rear surface feature set of the rear surface of the optical lens under test in the first feature set and the second feature set. Step S150: Calculate the geometric parameters based on the front surface feature set and the rear surface feature set to obtain the geometric parameter measurement results.

[0012] For example, an optical lens refers to a transparent or semi-transparent optical element used to focus, diverge, reflect, or refract light, such as a lens, prism, or filter. Acquiring initial image data of the central region of the optical lens under test can be achieved by photographing the lens under illumination conditions using an industrial camera. For instance, a high-resolution CCD camera can be used to acquire images from a direction perpendicular to the lens surface under uniform white LED illumination, obtaining initial image data. This initial image data refers to the raw image information acquired by the industrial camera under illumination conditions, and can be a grayscale or color image. Subsequently, grayscale feature analysis is performed on the acquired initial image data. Grayscale feature analysis can involve analyzing the distribution of grayscale values ​​in the image to extract edges, peaks, valleys, etc. Grayscale feature analysis can employ various methods; for example, it can identify areas with significant changes in grayscale values ​​by performing one-dimensional or two-dimensional grayscale profile scanning of the image. The grayscale feature analysis results include grayscale peak points and grayscale profile morphology. Grayscale peak points can be local maximum values ​​in the grayscale profile, and grayscale profile morphology can be a grayscale value variation curve along a specific direction. The shape of the grayscale profile morphology can characterize the structural features of the lens surface and is used to describe the grayscale variation trend around the peak points. For example, a grayscale profile can be extracted along the radial direction of the lens center to observe whether there are multiple peaks or broadened peaks, in order to determine whether there is surface overlap.

[0013] When grayscale feature analysis determines that there is overlap between the front and rear surface features of a lens, i.e., when the overlap is determined through the identification and analysis of grayscale peak points and grayscale profile morphology, a polarization modulation command can be triggered. For example, if the grayscale profile shows a broadened single peak or two very close peaks with similar peak intensities, surface overlap can be identified. Polarization modulation can affect the interaction between light and the lens surface by changing the polarization state of the illumination light, thereby acquiring different image information under different polarization states. The corresponding polarization modulation command can be a control signal sent to the illumination source of the optical lens measurement system to adjust the polarization state of the illumination light. Therefore, the polarization modulation command is sent to the illumination source of the optical lens measurement system to apply various different polarization modulation conditions and acquire images. For example, the illumination source can be equipped with an adjustable polarizer. By changing the angle of the polarizer, linearly polarized light with different polarization directions such as 0 degrees, 45 degrees, 90 degrees, and 135 degrees can be generated sequentially, and the camera can acquire images corresponding to each polarization state. This results in a sequence of images acquired under different polarization modulation conditions, i.e., multi-frame polarization modulation image data.

[0014] For example, based on the grayscale feature analysis results, overlapping regions are obtained. These regions can be defined as the image areas corresponding to the broadened peaks or two closely spaced peaks identified in the grayscale profile. Then, feature separation processing is performed on the portions of the overlapping regions corresponding to each polarization-modulated image data to obtain a first feature set and a second feature set. For example, image segmentation algorithms, such as thresholding, region growing, or edge detection, can be used to separate different surface features in the overlapping regions. The first and second feature sets represent the image feature sets corresponding to the front and rear surfaces of the lens after separation processing, respectively. The first feature set includes the first center coordinates of the overlapping region and the first polarization response features corresponding to the first center coordinates of each polarization-modulated image data. The second feature set includes the second center coordinates of the overlapping region and the second polarization response features corresponding to the second center coordinates of each polarization-modulated image data. The first and second center coordinates correspond to the front or rear surface of the optical lens under test, respectively. The polarization response features refer to the response characteristics of the lens surface to polarized light under different polarization modulation conditions, such as reflection intensity and degree of polarization.

[0015] The degree of polarization response fluctuation refers to the amplitude or pattern of change in polarization response characteristics under different polarization modulation conditions. It can be used to distinguish different surfaces. Therefore, based on the degree of polarization response fluctuation characterized by the first feature set and the second feature set, the front surface feature set corresponding to the front surface of the optical lens under test and the rear surface feature set corresponding to the rear surface of the optical lens under test are determined from the first and second feature sets. The variance or standard deviation of each polarization response characteristic in the first and second feature sets can be calculated to quantify the degree of fluctuation. By comparing the degree of fluctuation and combining it with preset discrimination rules, the corresponding front surface feature set and the corresponding rear surface feature set can be distinguished. After determining the front surface feature set and the rear surface feature set, geometric parameters can be calculated based on the feature sets to obtain the geometric parameter measurement results.

[0016] The technical concept of this application lies in effectively identifying whether there is overlap between the front and rear surface features of a lens in the central region by performing grayscale feature analysis on the initial image data. When overlap is detected, the judgment no longer relies on a single grayscale information, but triggers a polarization modulation command to apply various different polarization modulation conditions to the illumination source and acquire multiple frames of polarization modulated image data. Subsequently, based on the overlapping region determined by the grayscale feature analysis results, feature separation processing is performed on the multiple frames of polarization modulated image data. By analyzing the differences in image data under different polarization states, the originally overlapping front and rear surface features can be separated more accurately, resulting in a first feature set and a second feature set. Each feature set includes not only the center coordinates of the corresponding overlapping region, but also the polarization response characteristics of that coordinate in different polarization modulated image data. Furthermore, by analyzing the degree of polarization response fluctuation characterized by the first and second feature sets, it is possible to accurately determine which feature set corresponds to the front surface of the lens and which feature set corresponds to the rear surface, because the front and rear surfaces of the lens have different optical properties, resulting in different response modes and fluctuation degrees to polarized light. Therefore, after determining the front surface feature set and the back surface feature set, geometric parameters can be calculated, thereby obtaining reliable geometric parameter measurement results.

[0017] like Figure 2 As shown, Figure 2 This is a flowchart of a method for measuring the geometric parameters of an optical lens based on industrial vision, provided in another embodiment of this application. Regarding step S130 above, the method may include, but is not limited to, steps S210 to S240.

[0018] Step S210: Based on the grayscale feature analysis results, obtain the overlapping regions and create a first initial Gaussian distribution and a second initial Gaussian distribution; Step S220: Analyze the partial data of the corresponding overlapping division region in each polarization modulation image data and the first probability corresponding to the first initial Gaussian distribution and the second probability corresponding to the second initial Gaussian distribution; Step S230: Using the first probability and the second probability, adjust the first initial Gaussian distribution and the second initial Gaussian distribution on each polarization modulation image data respectively to obtain the first independent Gaussian distribution and the second independent Gaussian distribution. Step S240: Extract the mean vectors of the first independent Gaussian distribution and the second independent Gaussian distribution to obtain the first center coordinates and the second center coordinates of the corresponding overlapping division region, and obtain the polarization response features of the first independent Gaussian distribution and the second independent Gaussian distribution on each polarization modulation image data to obtain the first feature set and the second feature set.

[0019] For example, when grayscale feature analysis results indicate that the front and rear surface features of a lens overlap, the overlapping region of the lens's front and rear surface features in the image can be determined based on the grayscale feature analysis results. Within the overlapping region, a first initial Gaussian distribution and a second initial Gaussian distribution are created. These distributions can serve as preliminary estimates of the front and rear surface features. For instance, the mean vector (i.e., center coordinates), covariance matrix (i.e., spot shape and range), and mixing weights can be initialized based on the geometric center and shape of the wide single peak in the grayscale feature analysis results of the initial image data.

[0020] For example, the first initial Gaussian distribution and the second initial Gaussian distribution are used to simulate the pixel intensity distribution of the front surface and the back surface features in the overlapping region, respectively. Therefore, partial data of the corresponding overlapping region in each polarization modulation image data can be analyzed, and the first probability corresponding to the first initial Gaussian distribution and the second probability corresponding to the second initial Gaussian distribution can be calculated to characterize the possibility that the pixel data in the overlapping region belongs to the front surface feature or the back surface feature.

[0021] Using the obtained first and second probabilities, the first and second initial Gaussian distributions can be iteratively adjusted on each polarization-modulated image data to optimize the parameters of the Gaussian distribution, so that the first and second initial Gaussian distributions fit the actual pixel data distribution, thereby more accurately representing the features of the front and back surfaces.

[0022] Mean vectors are extracted from the first and second independent Gaussian distributions to obtain the first and second center coordinates of the corresponding overlapping regions. Simultaneously, polarization response features of the first and second independent Gaussian distributions on various polarization-modulated image data are acquired. These polarization response features may include, but are not limited to, the peak intensity, width, or shape parameters of the Gaussian distribution, and can characterize the response characteristics of the front and rear surfaces to polarized light under different polarization modulation conditions.

[0023] For example, the iterative solution process can be based on the expectation-maximization algorithm. In the expectation step of the expectation-maximization algorithm, using the current Gaussian parameters, the first probability and the second probability that the gray value of each pixel in the overlapping region is generated by the Gaussian distribution of the front surface and the Gaussian distribution of the back surface are calculated. In the maximization step of the expectation-maximization algorithm, using the first probability and the second probability as weights, the mean vector, covariance matrix, and mixing weights of the two Gaussian distributions are recalculated and updated. As the expectation-maximization algorithm iterates continuously, the log-likelihood function gradually converges, and the originally tightly coupled or completely merged broad single peaks are forcibly and reasonably decoupled by the mathematical model into two mutually independent first independent Gaussian distributions and second independent Gaussian distributions.

[0024] Once the decoupling algorithm meets the convergence condition, the system extracts parameters from the fitted first independent Gaussian distribution and the second independent Gaussian distribution. The mean vectors of the first independent Gaussian distribution and the second independent Gaussian distribution represent the sub-pixel level spatial coordinates of the front surface feature points and the back surface feature points on the image plane. At the same time, the peak gray level or integral energy of the first independent Gaussian distribution and the second independent Gaussian distribution under four polarization states are recorded to form two independent polarization response feature sets.

[0025] This application embodiment utilizes the technical concept of Gaussian mixture model. By modeling and iteratively optimizing the pixel data of the overlapping area using Gaussian distribution, it can effectively distinguish and extract the overlapping features of the front and rear surfaces of the optical lens in the image. Even when grayscale features are difficult to distinguish directly, effective separation can be achieved based on the difference in polarization response. This improves the accuracy of locating feature points on the front and rear surfaces in the measurement of complex optical lenses and avoids measurement errors caused by feature overlap.

[0026] In another embodiment of this application, the method for measuring the geometric parameters of optical lenses based on industrial vision may include, but is not limited to, step S330, the above-mentioned step S230.

[0027] Step S330: Using the first probability and the second probability, under the constraint of the preset regularization constraint term based on polarization response difference, adjust the first initial Gaussian distribution and the second initial Gaussian distribution on each polarization modulation image data respectively to obtain the first independent Gaussian distribution and the second independent Gaussian distribution; the regularization constraint term based on polarization response difference is used to constrain the changing trend of the parameters of the first initial Gaussian distribution and the parameters of the second initial Gaussian distribution on each polarization modulation image data.

[0028] It is understandable that, in the process of adjusting the first and second initial Gaussian distributions on each polarization-modulated image data, if there is a lack of effective constraint mechanisms, the adjustment process of the Gaussian distribution may be easily affected by image noise or local feature similarity, resulting in unstable convergence or inability to accurately distinguish overlapping front and rear surface features.

[0029] Based on this, the first and second probabilities can be used to adjust the first and second initial Gaussian distributions on each polarization-modulated image data under the constraint of a preset regularization constraint term based on polarization response differences. The regularization constraint term based on polarization response differences can be a mathematical expression used to ensure that the changing trends of the parameters of the first and second initial Gaussian distributions remain reasonable across different polarization-modulated image data during the iterative adjustment of the Gaussian distribution parameters. That is, when the polarization response differences between the two surface features are large, the Gaussian distribution parameters are allowed a larger independent adjustment space, while when the polarization response differences are small, a stronger correlation constraint is imposed on the parameter changes to avoid overfitting or unstable separation. The regularization constraint term based on polarization response differences can be constructed based on physical models or empirical data. For example, it can be a term inversely proportional to the degree of difference between the polarization response features of the two Gaussian distributions, or a term that penalizes the inconsistency in the changing trends of the two Gaussian distribution parameters.

[0030] The embodiments of this application, through regularization constraints based on polarization response differences, can effectively avoid the problem of unstable parameter adjustment or convergence to local optima caused by lack of constraints during the process of adjusting the first and second initial Gaussian distributions using the first and second probabilities. This allows the parameter adjustment of the Gaussian distribution to better characterize the true optical response characteristics of the front and rear surfaces of the optical lens under different polarization states, thereby improving the accuracy of feature separation.

[0031] In some embodiments, when using the Expectation-Maximization (EM) algorithm for Gaussian Mixture Model (GM) parameter estimation, a regularization constraint term based on polarization response difference can be integrated into the M-step of the EM algorithm. Specifically, when updating the mean, covariance, and weight parameters of the first and second initial Gaussian distributions, in addition to considering the probability contribution of data points, a penalty term can be introduced. This penalty term is a function of the difference in polarization response characteristics of the two Gaussian distributions on the current polarization-modulated image data. For example, it could be an L2 norm term used to minimize the difference in the variation trend of the two Gaussian distribution parameters across different polarization-modulated image data, and is weighted according to a preset polarization response difference threshold. When the polarization response difference between the two Gaussian distributions is less than the threshold, the weight of the penalty term increases, thereby more strictly constraining the parameter variation trend and promoting a closer correlation between the two Gaussian distributions in the parameter space. When the difference is large, the weight of the penalty term decreases, allowing for a larger independent adjustment space.

[0032] In another embodiment of this application, the method for measuring the geometric parameters of optical lenses based on industrial vision may include, but is not limited to, steps S410 to S420, regarding step S140.

[0033] Step S410: Analyze the gray-level variance or polarization contrast of multiple first polarization response features to obtain the first polarization response change rate; analyze the gray-level variance or polarization contrast of multiple second polarization response features to obtain the second polarization response change rate. Step S420: Based on the first polarization response change rate, the second polarization response change rate, the preset standard polarization response change rate, the degree of polarization response fluctuation characterized by the first feature set, and the degree of polarization response fluctuation characterized by the second feature set, determine the front surface feature set of the front surface of the optical lens under test and the rear surface feature set of the optical lens under test.

[0034] For example, the first polarization response feature and the second polarization response feature are obtained based on multi-frame polarization-modulated image data. They are used to characterize the response characteristics of the front or rear surface of the optical lens to polarized light under different polarization modulation conditions. Gray-level variance refers to the dispersion of pixel gray-level values ​​within a specific region, which can characterize the degree of drastic changes in texture or brightness in local areas of the image. Polarization contrast refers to the difference in brightness or intensity of the image under different polarization states, which can characterize the intensity change of the polarization response. Therefore, by analyzing the gray-level variance or polarization contrast of multiple first polarization response features, the first polarization response change rate can be obtained, which is used to quantify the response stability or sensitivity of the first feature set under different polarization modulation conditions. By analyzing the gray-level variance or polarization contrast of multiple second polarization response features, the second polarization response change rate can be obtained.

[0035] The preset standard polarization response rate of change is predetermined based on a known standard lens or theoretical model and serves as a benchmark for judging the front and rear surfaces. The nominal parameters of the lens currently being measured can be retrieved via the manufacturing execution system interface, including the theoretical center thickness, theoretical front surface radius of curvature, theoretical rear surface radius of curvature, and the material's center wavelength refractive index. Based on these nominal parameters, a lightweight ray-tracing forward model simulates the complete optical path of a polarized beam emitted from a light source, passing through the air medium, reflecting at the theoretical lens's front surface, refracting into the lens interior, reflecting at the rear surface, and then exiting the front surface again. Combining the Fresnel reflection equation and the empirical birefringence of the lens material, the forward model calculates the expected light intensity change curves and relative attenuation ratios of the theoretically reflected light from the front and rear surfaces after undergoing various polarization modulations under different polarization modulation conditions before reaching the camera sensor, thus obtaining the standard polarization response rate of change.

[0036] Next, for each set of feature points, the gray-level variance or polarization contrast is calculated under various polarization modulation conditions to quantify the response rate of change. The actual response rates of change for the two sets of features are compared with the standard polarization response rates output by the forward model. According to the physical mechanism, the light reflected from the front surface does not enter the interior of the material, and its polarization state is less affected by modulation, resulting in a relatively gentle change in light intensity after passing through the analyzer. In contrast, the light reflected from the rear surface undergoes round-trip penetration into the interior of the material, and is subject to cumulative modulation by birefringence and absorption, resulting in a more drastic change in polarization state. Consequently, the intensity fluctuation after passing through the analyzer is significantly greater than that of the light reflected from the front surface. Therefore, the set of feature points with a larger response rate of change, more drastic fluctuations, and slightly lower overall integrated energy can be identified as the rear surface feature subset; the set of feature points with a smaller response rate of change and relatively stable light intensity can be identified as the front surface feature subset.

[0037] For example, if the response rates of two sets of features are extremely similar and cannot form a significant distinction, or if the response curves of both sets deviate significantly from the theoretical expectations, it indicates that there may be strong external stray light interference, or that there are serious local defects within the lens material that disrupt the normal polarization transmission law. In this case, a compensation parameter can be automatically calculated based on the direction of the deviation and fed back to the light source controller and camera. For example, the light source can be instructed to increase the overall brightness by 20%, or the camera can be instructed to extend the exposure time by a specified number of microseconds. Then, the polarization modulation image acquisition and decoupling separation process can be retried. If the attribution determination still fails after three consecutive retries, the algorithm attempt on the lens can be stopped, and it can be marked as a defective product with an abnormal measurement state. The pneumatic sorting mechanism can be directly driven through the control interface to remove it, and an alarm log can be output to the operation interface, thereby preventing uncertain erroneous data from flowing into the final solution model.

[0038] This application embodiment analyzes the grayscale variance or polarization contrast of polarization response characteristics and calculates the rate of change of polarization response, which can construct a more comprehensive discrimination model, effectively distinguish the front surface and the back surface, improve the recognition accuracy in complex optical environments or when facing special lens materials, and enhance the robustness of the entire measurement method to noise and environmental changes.

[0039] In another embodiment of this application, the method for measuring the geometric parameters of an optical lens based on industrial vision is provided. When the grayscale feature analysis results characterize multiple overlapping regions, the first feature set includes multiple first center coordinates corresponding to the multiple overlapping regions, and the second feature set includes multiple second center coordinates corresponding to the multiple overlapping regions. Regarding the above step S150, the method may include, but is not limited to, steps S510 to S540.

[0040] Step S510: Obtain multiple front surface coordinate points based on the front surface feature set, and obtain multiple rear surface coordinate points based on the rear surface feature set. Step S520: Based on the preset initial adjacent distance pairing condition, determine the target rear surface coordinate points corresponding to each front surface coordinate point and pair them to obtain multiple feature pairs to be verified. Step S530: Calculate the Euclidean distance vector of each feature pair to be verified. Confirm that the feature pair to be verified is a dual-surface feature pair if the magnitude of the Euclidean distance vector is within the preset tolerance boundary and the direction of the Euclidean distance vector is consistent with the eccentricity direction of the optical lens to be tested. Step S540: Calculate the geometric parameters based on the dual-surface feature pair to obtain the geometric parameter measurement results.

[0041] For example, when the grayscale feature analysis results indicate that there are multiple overlapping regions, the multiple front surface coordinate points and multiple back surface coordinate points extracted from the first feature set and the second feature set correspond to different overlapping regions. The multiple front surface coordinate points and multiple back surface coordinate points can be the center coordinates obtained through feature separation processing, or they can be surface feature points after refinement processing.

[0042] Pairing is performed based on a preset initial adjacent distance pairing condition, which can initially screen feature points that may belong to the same pair of front and rear surfaces. The preset initial adjacent distance pairing condition can be that if the distance between a coordinate point on a front surface and a coordinate point on a rear surface is less than a preset threshold, then they may constitute a feature pair to be verified. Further, feature pairs to be verified that have a magnitude of the Euclidean distance vector within a preset tolerance boundary and whose direction is consistent with the eccentricity direction of the optical lens under test are identified as dual-surface feature pairs. Here, the magnitude of the Euclidean distance vector represents the actual distance between the front and rear surfaces, the preset tolerance boundary is used to exclude pairings that do not meet the physical size constraints, and the direction of the Euclidean distance vector being consistent with the eccentricity direction of the optical lens further ensures the correctness of the pairing, because the direction of the line connecting the feature points on the front and rear surfaces should be consistent with the geometric axis or eccentricity direction of the lens.

[0043] This application embodiment uses a feature pairing and verification mechanism under multiple overlapping regions to accurately identify dual-surface feature pairs, greatly improving the accuracy of front and rear surface feature pairing. This enables the accurate identification of the corresponding front and rear surfaces when facing lenses with multi-layer optical structures or complex surface morphologies, thereby avoiding geometric parameter measurement errors caused by incorrect pairing.

[0044] In another embodiment of this application, the preset tolerance boundary is obtained through the following method: Obtain the range of geometric parameter distribution of the optical lens under test during the manufacturing process; Based on the distribution range of geometric parameters and preset calibration parameters, the threshold range of the projection distance of feature points on the front and back surfaces of the optical lens under test is analyzed to obtain the tolerance boundary.

[0045] For example, during the manufacturing process of optical lenses, various factors such as material properties, processing technology, and equipment precision cause fluctuations and distributions in the geometric parameters of the lens, such as thickness, radius of curvature, and eccentricity. This represents the range of geometric parameter distribution of the optical lens under test during manufacturing. Based on the range of geometric parameter distribution and preset calibration parameters, combined with the preset calibration parameters of the optical measurement system itself, such as magnification, pixel size, and distortion correction parameters, the projection distances of feature points on the front and rear surfaces of the lens onto the image can be theoretically analyzed and calculated. By mapping the range of geometric parameter distribution to the image space, the maximum and minimum possible ranges of variation in the projection distances of feature points on the front and rear surfaces onto the image under different combinations of geometric parameters can be predicted. This represents the threshold range of the projection distances of feature points on the front and rear surfaces, i.e., the tolerance boundary.

[0046] For example, the preset tolerance boundary can be calculated again using the nominal parameters of the lens, combined with the lens thickness manufacturing tolerance, curvature processing tolerance, and the small tilt attitude tolerance allowed on the conveyor belt, through geometric optics derivation, to obtain the maximum allowable upper limit and minimum allowable lower limit of the theoretical projection distance of an incident ray parallel to the optical axis at the front surface reflection point and the rear surface reflection point on the image sensor plane under the most extreme tolerance combination.

[0047] This application embodiment establishes a tolerance boundary that highly conforms to the actual production situation by combining the geometric parameter distribution range of the optical lens under test during the manufacturing process with the preset calibration parameters of the optical measurement system. This ensures the rationality and accuracy of the tolerance boundary, improves the accuracy of double-surface feature pair identification, reduces errors introduced by inaccurate feature pair matching, and makes the measurement method more adaptable to lenses of different batches and different tolerance ranges, thereby improving the reliability of the entire measurement system.

[0048] In another embodiment of this application, the method for measuring the geometric parameters of optical lenses based on industrial vision may include, but is not limited to, step S610, the above step S420.

[0049] Step S610: If the deviation between the first polarization response change rate and the standard polarization response change rate exceeds a preset deviation threshold, obtain the grayscale vector of the first center coordinates in each polarization modulation image data, calculate the cross product integral of the grayscale Laplacian operator result of the first center coordinates within the neighborhood radius of the corresponding pixel in the initial image data and the grayscale vector of the first center coordinates in each polarization modulation image data to obtain the first gradient tensor, extract the first eccentricity of the covariance matrix of the first independent Gaussian distribution corresponding to the first polarization response change rate, and fuse the first eccentricity with the first gradient tensor to obtain the first fusion coefficient. When the first fusion coefficient is greater than the preset stress distortion constant, the non-overlapping region adjacent to the overlapping region corresponding to the first center coordinate is determined. The polarization response curves corresponding to the determined front and rear surface coordinates of the non-overlapping region are extracted as local references. The front surface feature set of the front surface of the optical lens under test and the rear surface feature set of the rear surface of the optical lens under test are determined by polynomial extrapolation. If the deviation between the second polarization response change rate and the standard polarization response change rate exceeds a preset deviation threshold, the grayscale vectors of the second center coordinates in each polarization-modulated image data are obtained. The cross product integral of the grayscale Laplacian operator result within the neighborhood radius of the corresponding pixel in the initial image data and the grayscale vectors of the second center coordinates in each polarization-modulated image data is calculated to obtain the second gradient tensor. The second eccentricity of the covariance matrix of the second independent Gaussian distribution corresponding to the second polarization response change rate is extracted. The second eccentricity and the second gradient tensor are fused to obtain the second fusion coefficient. When the second fusion coefficient is greater than the preset stress distortion constant, the non-overlapping region adjacent to the overlapping region corresponding to the second center coordinate is determined. The polarization response curves corresponding to the determined front and rear surface coordinates of the non-overlapping region are extracted as local references. The front surface feature set of the front surface of the optical lens under test and the rear surface feature set of the optical lens under test are determined by polynomial extrapolation.

[0050] Understandably, in practical applications, when optical lenses have internal stress or surface distortion, the polarization response characteristics may undergo complex changes, making it difficult to accurately distinguish the front and back surface features through simple polarization response fluctuation or rate of change analysis, and even leading to misjudgment.

[0051] For example, when the deviation of the first polarization response change rate or the second polarization response change rate from the preset standard polarization response change rate exceeds the preset deviation threshold, it indicates that the polarization response of the current overlapping area may be affected by abnormal factors, resulting in a difference between the polarization characteristics and the ideal situation. At this time, the grayscale vector of the first center coordinate or the second center coordinate on each polarization modulation image data is obtained, that is, the grayscale value sequence of the center coordinate and its surrounding pixels is extracted to characterize the brightness response of the point under different polarization states.

[0052] The gradient tensor is obtained by calculating the cross product integral of the grayscale Laplacian operator result within the neighborhood radius of the pixel corresponding to the center coordinate in the initial image data and the grayscale vector of the center coordinate in each polarization-modulated image data. This gradient tensor quantifies the local gradient information and directionality of the image grayscale change at the center coordinate. The Laplacian operator is used to detect image edges and textures, while the cross product integral with the grayscale vector comprehensively considers the dynamic changes in polarization response and local structural features. The eccentricity of the covariance matrix of the first independent Gaussian distribution corresponding to the first polarization response change rate or the second independent Gaussian distribution corresponding to the second polarization response change rate is extracted. This eccentricity characterizes the shape characteristics of the Gaussian distribution, i.e., the direction and extent of its extension in image space, representing the geometric shape or distortion of surface features. The eccentricity is fused with the gradient tensor to obtain the fusion coefficient, which comprehensively considers the geometric shape of surface features and local grayscale changes, thereby more comprehensively assessing the existence of stress distortion or other anomalies.

[0053] A preset stress distortion constant is used to determine whether the fusion coefficient is sufficient to indicate the presence of detailed stress distortion. When the fusion coefficient is greater than the preset stress distortion constant, the surface features can be considered to be affected by stress distortion. Therefore, by identifying the non-overlapping regions adjacent to the overlapping regions corresponding to the center coordinates, areas with well-defined polarization response regularities where the front and rear surface features are clearly separated can be found in the image. Subsequently, the polarization response curves corresponding to the determined front and rear surface coordinates of the non-overlapping regions are extracted as local benchmarks. Using the polarization response characteristics of reliable regions as a reference, the front surface feature set of the corresponding front surface and the rear surface feature set of the corresponding rear surface of the optical lens under test are determined by polynomial extrapolation. The regularity can be extended to the overlapping regions affected by stress distortion using the polarization response curves of the local benchmarks through mathematical models, thereby correcting or re-estimating the front and rear surface feature sets and improving the accuracy of identification.

[0054] This application embodiment, through multidimensional feature analysis and adaptive correction mechanism, can more sensitively and comprehensively detect surface feature anomalies caused by stress when facing stress distortion or complex surface defects of optical lenses. It avoids misjudgment that may be caused by relying solely on the degree of polarization response fluctuation. When stress distortion is detected, the polarization response curves of adjacent non-overlapping regions are used as local references for polynomial extrapolation, which can correct the front and back surface feature sets of the overlapping regions affected by distortion, thereby improving the accuracy and reliability of optical lens geometric parameter measurement under complex working conditions.

[0055] In another embodiment of the optical lens geometry parameter measurement method based on industrial vision provided in this application, the polarization modulation command is specifically used for: Obtain the refractive index range of the material and the optical response parameters of the surface coating of the optical lens under test; Based on the material's refractive index range and the optical response parameters of the surface coating, the driving voltage sequence of the illumination source of the optical lens measurement system is adjusted to adjust the polarization state of the emitted illumination light, and the exposure time and gain settings of the camera sensor are adjusted to instruct the camera to acquire multiple frames of polarization-modulated image data corresponding to different polarization states.

[0056] For example, the refractive index range of the material and the optical response parameters of the surface coating of the optical lens under test can be obtained by consulting the design specifications, material data sheets, or conducting pre-calibration experiments of the optical lens before measurement. This will allow you to obtain the refractive index range of the material used in the optical lens under test and the optical response characteristics of its surface coating, such as reflection, transmission, or absorption of different polarized light.

[0057] Based on the material's refractive index range and the optical response parameters of the surface coating, the driving voltage sequence of the illumination source of the optical lens measurement system is adjusted to adjust the polarization state of the emitted illumination light. The driving voltage sequence usually works in conjunction with a polarization control device to precisely control the polarization state of the emitted illumination light. For specific lens materials and coatings, it is possible to maximize the response differences between the front and rear surface features under different polarization states, thereby optimizing the feature separation effect.

[0058] The camera sensor's exposure time and gain settings are adjusted to instruct the camera to acquire multiple frames of polarization-modulated image data corresponding to different polarization states. After each polarization state adjustment, the camera sensor's exposure time and gain are adjusted based on the intensity of light reflected or transmitted by the optical lens under the current polarization state, as well as the attenuation or enhancement effect of the lens material and coating on the light. This ensures that the image data acquired under different polarization modulation conditions has optimal brightness, contrast, and signal-to-noise ratio, avoiding overexposure or underexposure of the image, thereby providing high-quality raw data for feature separation processing.

[0059] In some embodiments, the optical lens under test is made of a high-refractive-index glass material, with an anti-reflective coating on the front surface and a reflective coating on the rear surface. When executing polarization modulation commands, the refractive index range of the glass material and the optical response parameters of the two coatings are first acquired. The driving voltage sequence of the illumination source is then adjusted; for example, a liquid crystal tunable polarizer is controlled to sequentially output linearly polarized light at 0 degrees, 45 degrees, 90 degrees, 135 degrees, and circularly polarized light. Simultaneously, due to the difference in response of the anti-reflective and reflective coatings to different polarizations, it can be predicted that reflected light will be stronger in some polarization states and weaker in others. Therefore, when acquiring images of polarization states with strong reflected light, the camera's exposure time is appropriately shortened and the gain reduced to prevent overexposure; conversely, when acquiring images of polarization states with weak reflected light, the exposure time is appropriately extended and the gain increased to ensure sufficient brightness and signal-to-noise ratio.

[0060] like Figure 3 As shown, Figure 3 This is a flowchart of a method for measuring the geometric parameters of an optical lens based on industrial vision, provided in another embodiment of this application. Regarding step S130 above, the method may include, but is not limited to, steps S710 to S750.

[0061] Step S710: Based on the grayscale feature analysis results, obtain the overlapping regions; Step S720: Extract texture features from the overlapping regions corresponding to each polarization modulation image data to obtain the texture complexity distribution information of the overlapping regions. Step S730: Based on the texture complexity distribution information, divide the overlapping regions into regions with similar texture complexity within a preset range to obtain multiple overlapping sub-regions; Step S740: For the texture complexity corresponding to each overlapping sub-region, set the parameter configuration of the preset separation algorithm and execute it to obtain the first feature information and the second feature information of each overlapping sub-region. Step S750: Integrate the first feature information and the second feature information corresponding to each overlapping sub-region to obtain the first feature set and the second feature set.

[0062] For example, based on the gray-level feature analysis results, after obtaining the overlapping regions, texture features can be extracted from the overlapping regions corresponding to each polarization modulation image data. Texture feature extraction refers to analyzing the gray-level changes and spatial arrangement patterns of pixels within the overlapping regions using image processing techniques, such as gray-level co-occurrence matrix (GLCM), local binary mode (LBP), or wavelet transform, thereby quantifying the texture characteristics of the image and obtaining the texture complexity distribution information of the overlapping regions. The texture complexity distribution information serves as a statistic or descriptor describing the texture features of local image regions, used to characterize the richness and variation patterns of texture details within the region, such as texture roughness, contrast, and directionality.

[0063] Based on the texture complexity distribution information, regions with similar texture complexity within a preset range are divided into overlapping sub-regions, resulting in multiple overlapping sub-regions. Regions with similar texture characteristics can be grouped together, thus providing a more targeted processing method for feature separation. For example, for regions with higher texture complexity, more refined separation algorithm parameters are used; while for regions with lower texture complexity, relatively simple parameter configurations are used.

[0064] Furthermore, based on the texture complexity of each overlapping sub-region, the parameters of a preset separation algorithm can be configured and executed. This preset separation algorithm can be based on machine learning algorithms such as Gaussian mixture models, K-means clustering, and support vector machines, or on image processing algorithms such as image segmentation and edge detection. Adjusting the parameter configuration allows the separation algorithm to better adapt to overlapping sub-regions with different texture complexities, thereby improving the accuracy of feature separation. By executing the preset separation algorithm, the first and second feature information of each overlapping sub-region can be obtained. Finally, the first and second feature information corresponding to each overlapping sub-region are integrated to obtain the final first and second feature sets.

[0065] This application embodiment, through texture feature extraction and region division based on texture complexity, can process image data of overlapping areas of features on the front and rear surfaces of optical lenses more precisely, thereby more accurately identifying and separating feature information belonging to the front and rear surfaces, realizing refined adaptive processing of overlapping areas, and improving the accuracy of feature separation.

[0066] This application also provides an embodiment of an optical lens geometric parameter measurement system based on industrial vision, including: The grayscale analysis module is used to acquire initial image data of the central region of the optical lens under test, perform grayscale feature analysis on the initial image data, and obtain grayscale feature analysis results; the grayscale feature analysis results include grayscale peak points and grayscale profile morphology; The polarization modulation module is used to trigger polarization modulation commands to apply various polarization modulation conditions to the illumination source of the optical lens measurement system and to acquire images when it is determined from the grayscale feature analysis results that there is an overlap between the front surface features and the rear surface features of the lens. This allows for the acquisition of multi-frame polarization modulation image data corresponding to various polarization modulation conditions based on the polarization modulation commands. The feature separation module is used to obtain overlapping regions based on grayscale feature analysis results, and to perform feature separation processing on the parts of the overlapping regions corresponding to each polarization modulation image data to obtain a first feature set and a second feature set. The first feature set includes the first center coordinates corresponding to the overlapping regions and the first polarization response features corresponding to the first center coordinates of each polarization modulation image data. The second feature set includes the second center coordinates corresponding to the overlapping regions and the second polarization response features corresponding to the second center coordinates of each polarization modulation image data. The first center coordinates and the second center coordinates correspond to the front surface or the rear surface of the optical lens under test, respectively. The surface analysis module is used to determine the front surface feature set of the front surface of the optical lens under test and the rear surface feature set of the optical lens under test based on the polarization response fluctuation degree characterized by the first feature set and the polarization response fluctuation degree characterized by the second feature set. The parameter calculation module is used to calculate geometric parameters based on the front surface feature set and the rear surface feature set, and obtain the geometric parameter measurement results.

[0067] For example, the grayscale analysis module can be an image processing unit running image processing algorithms, capable of receiving initial image data from an industrial camera and performing grayscale feature analysis. The polarization modulation module can include an illumination source controller and an image acquisition controller. The illumination source controller can be a programmable power supply used to drive an LED or laser source equipped with an adjustable polarizer to apply various polarization modulation conditions. The image acquisition controller is used to synchronously control the industrial camera to acquire images, obtaining multi-frame polarization-modulated image data corresponding to different polarization modulation conditions. The feature separation module can be an image processing unit, such as an FPGA or GPU accelerator, running image segmentation and feature extraction algorithms, capable of receiving grayscale feature analysis results and multi-frame polarization-modulated image data. The surface analysis module can be a data analysis unit, capable of receiving a first feature set and a second feature set, and analyzing the degree of polarization response fluctuation represented therein. The parameter calculation module can be a calculation unit, capable of receiving a front surface feature set and a back surface feature set, and combining them with preset calibration parameters to perform geometric parameter calculations.

[0068] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for measuring the geometric parameters of optical lenses based on industrial vision, characterized in that, include: Acquire initial image data of the central region of the optical lens under test, and perform grayscale feature analysis on the initial image data to obtain... Gray-scale feature analysis results; the gray-scale feature analysis results include gray-scale peak points and gray-scale profile morphology; If, based on the grayscale feature analysis results, it is determined that there is an overlap between the front surface features and the rear surface features of the lens, a polarization modulation command is triggered to apply various different polarization modulation conditions to the illumination source of the optical lens measurement system and to acquire images, so as to obtain multi-frame polarization modulation image data corresponding to various different polarization modulation conditions based on the polarization modulation command. Based on the grayscale feature analysis results, overlapping regions are obtained. Feature separation processing is performed on the parts of the overlapping regions corresponding to each polarization modulation image data to obtain a first feature set and a second feature set. The first feature set includes the first center coordinates corresponding to the overlapping regions and the first polarization response features corresponding to the first center coordinates of each polarization modulation image data. The second feature set includes the second center coordinates corresponding to the overlapping regions and the second polarization response features corresponding to the second center coordinates of each polarization modulation image data. The first center coordinates and the second center coordinates correspond to the front surface or the rear surface of the optical lens under test, respectively. Based on the degree of polarization response fluctuation characterized by the first feature set and the degree of polarization response fluctuation characterized by the second feature set, the front surface feature set corresponding to the front surface of the optical lens under test and the rear surface feature set corresponding to the rear surface of the optical lens under test are determined in the first feature set and the second feature set. Geometric parameters are calculated based on the front surface feature set and the rear surface feature set to obtain the geometric parameter measurement results.

2. The method for measuring the geometric parameters of optical lenses based on industrial vision according to claim 1, characterized in that, Based on the grayscale feature analysis results, overlapping regions are obtained. Feature separation processing is then performed on the corresponding overlapping regions of each polarization modulation image data to obtain a first feature set and a second feature set, including: Based on the grayscale feature analysis results, overlapping regions are obtained, and a first initial Gaussian distribution and a second initial Gaussian distribution are created. Analyze the partial data of the corresponding overlapping regions in each polarization modulation image data and their corresponding first probabilities and second probabilities with the first initial Gaussian distribution; Using the first probability and the second probability, the first initial Gaussian distribution and the second initial Gaussian distribution are adjusted on each of the polarization modulation image data respectively to obtain the first independent Gaussian distribution and the second independent Gaussian distribution; The mean vectors of the first independent Gaussian distribution and the second independent Gaussian distribution are extracted to obtain the first center coordinates and the second center coordinates of the corresponding overlapping regions. The polarization response features of the first independent Gaussian distribution and the second independent Gaussian distribution on each polarization modulation image data are obtained to obtain the first feature set and the second feature set.

3. The method for measuring the geometric parameters of optical lenses based on industrial vision according to claim 2, characterized in that, The step of adjusting the first initial Gaussian distribution and the second initial Gaussian distribution on each of the polarization modulation image data using the first probability and the second probability respectively to obtain the first independent Gaussian distribution and the second independent Gaussian distribution includes: Using the first probability and the second probability, under the constraint of a preset regularization constraint term based on polarization response difference, the first initial Gaussian distribution and the second initial Gaussian distribution are adjusted on each of the polarization modulated image data to obtain the first independent Gaussian distribution and the second independent Gaussian distribution; the regularization constraint term based on polarization response difference is used to constrain the changing trend of the parameters of the first initial Gaussian distribution and the parameters of the second initial Gaussian distribution on each of the polarization modulated image data.

4. The method for measuring the geometric parameters of optical lenses based on industrial vision according to claim 2, characterized in that, The step of determining the front surface feature set corresponding to the front surface of the optical lens under test and the rear surface feature set corresponding to the rear surface of the optical lens under test based on the polarization response fluctuation degree characterized by the first feature set and the second feature set includes: The gray-level variance or polarization contrast of multiple first polarization response features is analyzed to obtain the first polarization response change rate. The gray-level variance or polarization contrast of multiple second polarization response features is analyzed to obtain the second polarization response change rate. Based on the first polarization response change rate, the second polarization response change rate, the preset standard polarization response change rate, the degree of polarization response fluctuation characterized by the first feature set, and the degree of polarization response fluctuation characterized by the second feature set, the front surface feature set of the corresponding front surface of the optical lens under test and the rear surface feature set of the corresponding rear surface of the optical lens under test are determined.

5. The method for measuring the geometric parameters of optical lenses based on industrial vision according to claim 1, characterized in that, When the grayscale feature analysis result represents multiple overlapping regions, the first feature set includes multiple first center coordinates corresponding to the multiple overlapping regions, and the second feature set includes multiple second center coordinates corresponding to the multiple overlapping regions. The step of calculating geometric parameters based on the front surface feature set and the rear surface feature set to obtain geometric parameter measurement results includes: Based on the front surface feature set, multiple front surface coordinate points are obtained; based on the rear surface feature set, multiple rear surface coordinate points are obtained. Based on the preset initial adjacent distance pairing condition, the target rear surface coordinate points corresponding to each of the front surface coordinate points are determined and paired to obtain multiple feature pairs to be verified. Calculate the Euclidean distance vector of each of the feature pairs to be verified, and confirm that the feature pairs to be verified are dual-surface feature pairs if the magnitude of the Euclidean distance vector is within the preset tolerance boundary and the direction of the Euclidean distance vector is consistent with the eccentricity direction of the optical lens to be tested. Based on the dual-surface features, geometric parameters are calculated to obtain geometric parameter measurement results.

6. The method for measuring the geometric parameters of optical lenses based on industrial vision according to claim 5, characterized in that, The preset tolerance boundary is obtained in the following way: Obtain the range of geometric parameter distribution of the optical lens under test during the manufacturing process; Based on the distribution range of the geometric parameters and the preset calibration parameters, the threshold range of the projection distance of feature points on the front and back surfaces of the optical lens under test is analyzed to obtain the tolerance boundary.

7. The method for measuring the geometric parameters of optical lenses based on industrial vision according to claim 4, characterized in that, The step of determining the front surface feature set of the front surface of the optical lens under test and the rear surface feature set of the rear surface of the optical lens under test based on the first polarization response change rate, the second polarization response change rate, and the preset standard polarization response change rate includes: If the deviation between the first polarization response change rate and the standard polarization response change rate exceeds a preset deviation threshold, the grayscale vector of the first center coordinates on each of the polarization-modulated image data is obtained. The cross product integral of the grayscale Laplacian operator result within the neighborhood radius of the corresponding pixel in the initial image data and the grayscale vector of the first center coordinates on each of the polarization-modulated image data is calculated to obtain the first gradient tensor. The first eccentricity of the covariance matrix of the first independent Gaussian distribution corresponding to the first polarization response change rate is extracted. The first eccentricity is fused with the first gradient tensor to obtain the first fusion coefficient. When the first fusion coefficient is greater than a preset stress distortion constant, a non-overlapping region adjacent to the overlapping region corresponding to the first center coordinate is determined. The polarization response curves corresponding to the determined front and rear surface coordinates of the non-overlapping region are extracted as local references. The front surface feature set of the front surface of the optical lens under test and the rear surface feature set of the optical lens under test are determined by polynomial extrapolation. If the deviation between the second polarization response change rate and the standard polarization response change rate exceeds a preset deviation threshold, the grayscale vector of the second center coordinates on each of the polarization-modulated image data is obtained. The cross product integral of the grayscale Laplacian operator result within the neighborhood radius of the corresponding pixel in the initial image data and the grayscale vector of the second center coordinates on each of the polarization-modulated image data is calculated to obtain the second gradient tensor. The second eccentricity of the covariance matrix of the second independent Gaussian distribution corresponding to the second polarization response change rate is extracted. The second eccentricity and the second gradient tensor are fused to obtain the second fusion coefficient. When the second fusion coefficient is greater than the preset stress distortion constant, a non-overlapping region adjacent to the overlapping region corresponding to the second center coordinate is determined. The polarization response curves corresponding to the determined front and rear surface coordinates of the non-overlapping region are extracted as local references. The front surface feature set of the front surface of the optical lens under test and the rear surface feature set of the rear surface of the optical lens under test are determined by polynomial extrapolation.

8. The method for measuring the geometric parameters of optical lenses based on industrial vision according to claim 1, characterized in that, The polarization modulation command is specifically used for: Obtain the material refractive index range and the optical response parameters of the surface coating of the optical lens under test; Based on the refractive index range of the material and the optical response parameters of the surface coating, the driving voltage sequence of the illumination source of the optical lens measurement system is adjusted to adjust the polarization state of the emitted illumination light, and the exposure time and gain setting of the camera sensor are adjusted to instruct the camera to acquire multiple frames of polarization modulation image data corresponding to different polarization states.

9. The method for measuring the geometric parameters of optical lenses based on industrial vision according to claim 1, characterized in that, The step of performing feature separation processing on the overlapping regions corresponding to each of the polarization modulation image data to obtain a first feature set and a second feature set includes: Based on the grayscale feature analysis results, the overlapping regions are obtained; Texture features are extracted from the overlapping regions corresponding to each polarization modulation image data to obtain the texture complexity distribution information of the overlapping regions. Based on the texture complexity distribution information, the overlapping regions are divided into regions with similar texture complexity within a preset range to obtain multiple overlapping sub-regions; For each overlapping sub-region, the parameters of a preset separation algorithm are configured and executed to obtain the first feature information and the second feature information of each overlapping sub-region. The first feature information and the second feature information corresponding to each of the overlapping sub-regions are integrated to obtain the first feature set and the second feature set.

10. A system for measuring the geometric parameters of optical lenses based on industrial vision, characterized in that, include: The grayscale analysis module is used to acquire initial image data of the central region of the optical lens under test, and to perform grayscale feature analysis on the initial image data to obtain... Gray-scale feature analysis results; the gray-scale feature analysis results include gray-scale peak points and gray-scale profile morphology; The polarization modulation module is used to trigger a polarization modulation command to apply multiple different polarization modulation conditions to the illumination source of the optical lens measurement system and to acquire images when it is determined from the grayscale feature analysis results that there is an overlap between the front surface features and the rear surface features of the lens. The polarization modulation module is used to obtain multi-frame polarization modulation image data corresponding to multiple different polarization modulation conditions based on the polarization modulation command. The feature separation module is used to obtain overlapping regions based on the grayscale feature analysis results, and to perform feature separation processing on the parts of the overlapping regions corresponding to each polarization modulation image data to obtain a first feature set and a second feature set. The first feature set includes a first center coordinate corresponding to the overlapping region and a first polarization response feature corresponding to the first center coordinate of each polarization modulation image data. The second feature set includes a second center coordinate corresponding to the overlapping region and a second polarization response feature corresponding to the second center coordinate of each polarization modulation image data. The first center coordinate and the second center coordinate correspond to the front surface or the rear surface of the optical lens under test, respectively. The surface analysis module is used to determine the front surface feature set of the front surface of the optical lens under test and the rear surface feature set of the optical lens under test based on the polarization response fluctuation degree characterized by the first feature set and the polarization response fluctuation degree characterized by the second feature set. The parameter calculation module is used to calculate geometric parameters based on the front surface feature set and the rear surface feature set to obtain geometric parameter measurement results.

Images