Method for detecting surface roughness of optical glass

Abstract

The present application relates to roughness detection technical field.The present application relates to a kind of optical glass surface roughness detection method.It includes the following steps: S1, using multiple different wavelength light sources respectively with different incident angles to illuminate the optical glass surface to be measured, and the test light spot data scattered by the optical glass to be measured is collected, according to the wavelength coding information of each light source and the light source incident angle calibration relationship, the angle classification of test light spot data is carried out;The present application constructs the hierarchical detection architecture of standard benchmark rapid screening and abnormal area accurate inversion, first pass gray mean square deviation and contour similarity dual-dimension deviation threshold, quickly complete the eligibility determination of whole detection area, only the abnormal area that matches the deviation is sent into subsequent quantitative inversion link, under the premise of guaranteeing detection accuracy, greatly reduce the overall amount of calculation, significantly improve the batch detection efficiency of large aperture optical glass.

Description

Technical Field

[0001] This invention relates to the field of roughness detection technology, and more specifically, to a method for detecting the roughness of optical glass surfaces. Background Technology

[0002] Optical glass is a core component of high-end optical equipment such as photolithography optical systems, high-power laser devices, and precision imaging lenses. Its surface roughness directly determines the core performance indicators of the component, such as light scattering loss, imaging contrast, and laser damage threshold. Therefore, surface roughness detection is a key link in the quality control of optical glass processing and manufacturing.

[0003] Existing optical glass surface roughness testing technologies are mostly used in finished product sampling inspections and laboratory quality verification scenarios after processing. They are usually carried out in a constant temperature and vibration isolation clean laboratory environment. The test objects are mainly small-sized planar optical glass. The test process mostly adopts a single light source and single incident angle optical path architecture. Either the global average roughness value of the measured area is output by integrating the reception, or the surface morphology information is obtained by stitching together point-by-point mechanical scanning. It is mainly used for offline verification of the overall processing level of components and is difficult to adapt to the online, large-area rapid testing needs of mass production lines.

[0004] In actual mass production testing scenarios, existing technologies have several shortcomings: First, traditional integral scattering testing can only obtain the global average roughness and cannot identify local roughness defects such as scratches and insufficient polishing, making it difficult to meet the stringent requirements of high-end optical glass for surface uniformity and directly affecting the final optical performance of the components; Second, the detection method with a single incident angle and a single wavelength is not sensitive enough to surface roughness textures of specific orientations and scales, and is prone to missed detections. Although some solutions achieve angle adjustment by adding a mechanical swing mechanism, the moving parts introduce additional errors and reduce the detection speed and long-term stability of the system. Therefore, a new method for detecting the surface roughness of optical glass is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a method for detecting the surface roughness of optical glass, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, a method for detecting the surface roughness of optical glass is provided, comprising the following steps: S1. Multiple light sources of different wavelengths are used to illuminate the surface of the optical glass under test at different incident angles, and test spot data scattered by the optical glass under test are collected. According to the wavelength encoding information of each light source and the incident angle calibration relationship of the light source, the test spot data is classified by angle. S2. Divide the test spot into multiple detection areas based on the center position of the test spot, and determine the actual glass area corresponding to each detection area on the surface of the optical glass to be tested by combining the imaging calibration relationship. S3. Set the deviation threshold and simultaneously acquire the standard spot data corresponding to the qualified optical glass. Match and compare the test spot data of each detection area with the standard spot data. When the matching deviation between the test spot data and the standard spot data exceeds the deviation threshold, determine the corresponding detection area as an abnormal rough area. Then, summarize the adjacent abnormal rough areas to obtain the abnormal rough distribution area on the surface of the optical glass to be tested. S4. For abnormally rough distribution areas, the average roughness value corresponding to the successfully matched detection area is used as the initial roughness parameter, and an optical scattering model is established. Based on the optical scattering model, simulated spot data is generated for the abnormally rough area. The roughness parameter of the abnormally rough area is adjusted iteratively so that the difference between the simulated spot data and the test spot data meets the deviation threshold, thereby determining the roughness value of the abnormally rough area and completing the detection of the surface roughness of the optical glass.

[0007] As a further improvement to this technical solution, in S1, multiple light sources emitting different wavelengths are set. Each light source is installed above the optical glass to be tested through a fixed bracket with angle adjustment function. Each light source is set with an independent incident angle. All incident angles are based on the normal direction of the surface of the optical glass to be tested. Different wavelengths of light sources correspond to different incident angles, so that the incident beams of each light source converge in the same detection area on the surface of the glass to be tested. An array of photoelectric imaging devices is arranged along the normal direction of the glass to receive backscattered light generated by the glass surface. In a single detection cycle, each light source is lit sequentially according to the wavelength encoding order, and scattered light spot images at the corresponding wavelengths are collected. All single-wavelength light spot images are combined to form the test light spot data.

[0008] As a further improvement to this technical solution, in step S1, a correspondence between the incident angle of each wavelength light source and the angular distribution of the scattered light spot is established through a calibration experiment as the incident angle calibration relationship of the light source. After obtaining the test spot data, the corresponding incident angle calibration relationship is matched according to the wavelength encoding information carried by the spot data. The test spot data is grouped according to the incident angle dimension, and each group of data corresponds to the scattered light distribution under an incident angle, thus completing the angle classification of the test spot data.

[0009] As a further improvement to this technical solution, in S2, for each group of test spots that has completed angle classification, the gray-scale centroid algorithm is used to calculate the gray-scale centroid coordinates of the spot, and the coordinates are used as the reference for the center position of the test spot. With the center of the light spot as the origin, several rectangular detection areas of equal area are divided in the imaging plane. All detection areas are arranged in a row and column array and completely cover the effective light intensity distribution range of the test light spot. The imaging system is calibrated using a standard calibration plate to establish the imaging calibration relationship between the pixel coordinates of the imaging surface and the physical coordinates of the glass surface under test. Based on this imaging calibration relationship, the pixel coordinate range of each detection area on the imaging surface is converted into the actual physical area coordinates corresponding to the optical glass surface under test. The actual position and actual detection area of ​​the glass surface corresponding to each detection area are determined. The actual glass area is composed of the actual position and actual detection area of ​​the glass surface, thus realizing the corresponding mapping between the imaging detection area and the physical area of ​​the glass surface.

[0010] As a further improvement to this technical solution, in step S3, a deviation threshold is set that includes a light spot grayscale mean square error threshold and a light spot contour similarity threshold. Under the same light source wavelength, incident angle, acquisition distance and imaging conditions as the optical glass under test, spot data is collected on the qualified optical glass. The collected qualified spot data is divided into regions and classified by angle to obtain standard spot data corresponding to different incident angles, different wavelengths and different detection areas. In this process, standard spot data is used as the benchmark data for matching and comparison of each detection area, so that the test spot data of the optical glass under test can be compared with the standard spot data under the same irradiation conditions and the same area position.

[0011] As a further improvement to this technical solution, in S3, the standard spot data obtained by multiple qualified optical glasses under the same testing conditions are statistically analyzed to determine the range of spot fluctuations that can occur in each testing area within the normal roughness range. The upper limit of the matching deviation is set according to the range of spot fluctuations, and the upper limit is used as the deviation threshold. The difference between the test spot characteristics and the standard spot characteristics under the same detection area, the same wavelength and the same incident angle is calculated to obtain the matching deviation of the corresponding detection area. If the matching deviation is greater than the deviation threshold, it is determined that the test spot data of the detection area fails to match the standard spot data, and the monitoring armadillo is identified as an abnormally rough area. If the matching deviation is less than the deviation threshold, the roughness of the area is considered normal.

[0012] As a further improvement to this technical solution, in S3, after the determination of all detection areas is completed, the spatial position of all abnormal rough areas is traversed and identified. When two or more abnormal rough areas have a common adjacent boundary, or the distance between the areas is less than the preset merging distance threshold, these adjacent abnormal rough areas are merged and summarized to finally obtain a continuous sheet of abnormal rough distribution area on the surface of the optical glass to be tested.

[0013] 8. A method for detecting the surface roughness of optical glass according to claim 1, characterized in that: in step S4, a detection area that successfully matches the standard spot data is selected from the optical glass to be tested, and the qualified optical glass roughness value associated with the standard spot data corresponding to the detection area is obtained; The roughness values ​​corresponding to the successfully matched detection areas are averaged to obtain the reference roughness of the optical glass under test in the current detection state. Then, the reference roughness is used as the initial roughness parameter when performing roughness inversion in abnormal roughness distribution areas. An optical scattering model is established. The input parameters of the optical scattering model include the incident light wavelength, incident angle, refractive index of the optical glass, and surface roughness parameters. The output of the optical scattering model is the simulated data of the intensity distribution of the scattered light spot under the corresponding parameter conditions.

[0014] As a further improvement to this technical solution, in step S4, the initial roughness parameters of the abnormally rough distribution area, the wavelength of the light source, the incident angle of the light source, the illumination position of the light source, the image acquisition position, and the position of the optical glass surface to be tested are input into the optical scattering model. The optical scattering model calculates the scattering intensity distribution formed after light shines on an abnormally rough distribution area under the current roughness parameters, and generates simulated spot data corresponding to the test spot data based on the scattering intensity distribution; The difference between the simulated spot data and the test spot data corresponding to the abnormally coarse distribution area is calculated to obtain the simulation deviation. When the simulation deviation exceeds the deviation threshold, the roughness parameter of the abnormally rough area is increased or decreased according to the adjustment step size, and the simulated spot data is regenerated based on the adjusted roughness parameter. Repeat the difference calculation and roughness parameter adjustment until the simulated deviation between the simulated spot data and the test spot data is less than or equal to the deviation threshold. When determining the roughness value of an abnormally rough area, the roughness parameter is iteratively calculated for multiple detection areas within the same abnormally rough distribution area. The roughness parameter corresponding to the final deviation threshold of each detection area is determined as the roughness value of that detection area. Finally, the roughness distribution result of the optical glass surface under test is generated based on the roughness values ​​of multiple detection areas.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This method for detecting the surface roughness of optical glass constructs a hierarchical detection architecture that combines rapid screening of standard benchmarks with precise inversion of abnormal areas. First, it rapidly determines the pass / fail status of the entire detection area using a dual-dimensional deviation threshold based on grayscale mean square error and contour similarity. Only abnormal areas with excessive matching deviations are sent to the subsequent quantitative inversion stage. This significantly reduces the overall computational load while ensuring detection accuracy, thus significantly improving the batch detection efficiency of large-diameter optical glass. Furthermore, the deviation threshold is determined based on the statistical fluctuations of multiple qualified samples, making it adaptable to glass products with different materials and processing techniques. It balances the product pass rate under normal process fluctuations with the detection rate of abnormal defects. Combined with the adjacent abnormal area merging mechanism, it effectively filters out misjudgments caused by dust and imaging noise, improving the anti-interference capability and operational stability of the detection system in industrial environments.

[0016] 2. In this method for detecting the surface roughness of optical glass, a multi-wavelength and multi-incident angle bound coding optical path design is adopted. Multi-dimensional backscattered spot data can be obtained without mechanical angle adjustment mechanism, which effectively improves the detection sensitivity of surface roughness textures of different scales and orientations, and reduces the risk of missed detection of scratches and micro-defects in specific directions. At the same time, the spot partitioning benchmark is determined by the gray-scale centroid algorithm, and the precise coordinate mapping between the detection area and the glass surface is realized by imaging calibration. This breaks through the limitation of traditional scattering methods that can only output global average roughness. It can accurately locate the position and distribution range of local roughness defects, and can comprehensively reflect the surface processing uniformity of optical glass, which is more in line with the quality evaluation needs of high-end optical components. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart of a method for detecting the surface roughness of optical glass according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 As shown, the purpose of this embodiment is to provide a method for detecting the surface roughness of optical glass, including the following steps: S1. Multiple light sources of different wavelengths are used to illuminate the surface of the optical glass under test at different incident angles, and test spot data scattered by the optical glass under test are collected. According to the wavelength encoding information of each light source and the incident angle calibration relationship of the light source, the test spot data is classified by angle. In S1, multiple light sources emitting different wavelengths are set up. Each light source is mounted above the optical glass under test through a fixed bracket with angle adjustment function. Each light source is set with an independent incident angle. All incident angles are based on the normal direction of the surface of the optical glass under test. Select at least three sets of narrowband monochromatic light sources with non-overlapping center wavelengths. The light source type is narrowband LED or laser diode, and the half-width of the spectrum is controlled within 5nm. Install each light source on a fixed bracket with an angle scale and a locking knob. All brackets are evenly arranged around the upper side of the optical glass to be tested, and the light outlets are all facing the preset detection center point on the glass surface. Using the normal direction of the glass surface as the angular reference, the pitch angle of the bracket is finely adjusted one by one so that the incident beam of each light source forms a different angle with the normal, that is, a different incident angle. Simultaneously adjust the focusing state and horizontal position of each light source to ensure that all incident beams completely overlap at the detection center point on the glass surface, forming a confocal detection area. This ensures that the same glass area can be irradiated by multiple sets of beams with different angles and wavelengths. After the angle and position are calibrated, lock the adjustment mechanism of all brackets to fix the relative spatial position of the optical path.

[0020] Different wavelengths of light sources correspond to different incident angles, so that the incident beams of each light source converge in the same detection area on the surface of the glass to be tested. An array of photoelectric imaging devices is arranged along the normal direction of the glass to receive backscattered light generated by the glass surface. In a single detection cycle, each light source is lit sequentially according to the wavelength encoding order, and scattered light spot images at the corresponding wavelengths are collected. All single-wavelength light spot images are combined to form the test light spot data.

[0021] The area array photoelectric imaging device is installed facing the detection center along the normal direction of the glass under test, and the photosensitive plane of the imaging device is kept strictly parallel to the surface of the glass under test, so as to receive the backscattered light signal generated by the glass surface. Each light source is assigned a unique digital code as a wavelength code identifier. The wavelength code is matched one-to-one with the lighting control sequence of the light source to form a timing control logic. Within a single detection cycle, each light source is lit up in sequence according to the preset coding order, and only one light source is kept lit at any given time. Each time a light source is lit, the area array imaging device synchronously triggers exposure acquisition to obtain a frame of scattered light spot image at the corresponding wavelength. Each frame of image records the light intensity gray value corresponding to each pixel coordinate in the form of a two-dimensional matrix. After all light sources of all wavelengths have completed acquisition, all single-wavelength light spot images are integrated and stored in the encoding order to jointly form the test light spot dataset of the current detection area of ​​the glass under test.

[0022] In S1, the correspondence between the incident angle of each wavelength light source and the angular distribution of the scattered light spot is established through calibration experiments as the incident angle calibration relationship of the light source. After obtaining the test spot data, the corresponding incident angle calibration relationship is matched according to the wavelength encoding information carried by the spot data. The test spot data is grouped according to the incident angle dimension. Each group of data corresponds to the scattered light distribution under an incident angle, thus completing the angle classification of the test spot data.

[0023] A standard Lambert diffuse reflector plate calibrated by metrology is used as the calibration reference. It is placed horizontally on the testing station of the glass to be tested. The height and horizontal position of the diffuse reflector plate are adjusted so that its surface is in the same spatial position as the testing surface of the glass to be tested, ensuring that the calibration optical path is completely consistent with the formal testing optical path. Following the same lighting sequence and acquisition parameters as the formal test, light sources of each wavelength were lit sequentially, and scattered light spot images corresponding to the standard diffuse reflector were acquired. Based on the ideal Lambertian scattering characteristics of the standard diffuse reflector, the angle distribution of each calibration light spot image was analyzed, and the actual scattering angle corresponding to each pixel position on the imaging surface was calculated. Then, the corresponding mapping relationship between pixel coordinates and scattering angle under the incident angle of that wavelength was established. This mapping relationship was bound and stored with the corresponding wavelength code as the incident angle calibration relationship of the light source of that wavelength. After obtaining the test spot dataset of the glass under test, the wavelength code identifier corresponding to each frame of spot image is first extracted. The incident angle calibration relationship of the light source corresponding to the wavelength is retrieved according to the code matching. All spot images are grouped and organized according to different incident angles, with the incident angle as the grouping dimension. Each group of data corresponds to a certain incident angle and contains the complete two-dimensional scattered light intensity distribution at that angle.

[0024] S2. Divide the test spot into multiple detection areas based on the center position of the test spot, and determine the actual glass area corresponding to each detection area on the surface of the optical glass to be tested by combining the imaging calibration relationship. In S2, for each group of test spots that has completed angle classification, the gray-scale centroid algorithm is used to calculate the gray-scale centroid coordinates of the spot, and the coordinates are used as the reference for the center position of the test spot. For each group of scattered light spot images that have completed angle classification, background threshold segmentation is first performed. Dark noise in the imaging background is filtered out by a preset grayscale threshold, the effective light intensity pixel area of ​​the light spot is extracted, and the interference of invalid background pixels on the center calculation is eliminated.

[0025] Within the effective light spot area, the gray value of each pixel is used as a weight to calculate the weighted average of the horizontal and vertical coordinates of all pixels, thus obtaining the gray-level centroid coordinates of the light spot. Compared to the geometric center, the gray-level centroid is based on the distribution of light intensity energy, which can more accurately reflect the energy center position of the scattered light spot. It can adapt to situations with irregular light spot contours and uneven light intensity distribution, effectively avoiding reference misalignment caused by slight light spot offset or light intensity asymmetry.

[0026] The calculated gray-scale centroid coordinates are used as the reference for the center position of the group of light spots. All subsequent area divisions are based on this point as the origin of the coordinates, ensuring that the light spot partitioning reference is consistent for different incident angles and different detection batches.

[0027] With the center of the light spot as the origin, several rectangular detection areas of equal area are divided in the imaging plane. All detection areas are arranged in a row and column array and completely cover the effective light intensity distribution range of the test light spot. Using the gray-scale centroid coordinates of the light spot as the origin, the imaging surface is divided at equal intervals along the horizontal and vertical orthogonal directions according to the preset pixel step size, generating several rectangular detection areas of the same size. All detection areas are arranged in a regular row and column array, with adjacent areas closely connected, without overlap or gaps. During the division process, the effective light intensity boundary of the light spot is used as the limit, and the area is expanded outward to the outermost detection area to completely cover all effective light intensity pixels, ensuring that all effective scattered signals within the light spot are included in the detection range without any signal omission. The step size for dividing the detection area can be flexibly adjusted according to the detection resolution requirements. The smaller the step size, the smaller the physical area corresponding to a single detection area, and the stronger the spatial resolution capability.

[0028] The imaging system is calibrated using a standard calibration plate to establish an imaging calibration relationship between the pixel coordinates of the imaging surface and the physical coordinates of the glass surface under test. Based on this imaging calibration relationship, the pixel coordinate range of each detection area on the imaging surface is converted into the actual physical area coordinates corresponding to the optical glass surface under test. The actual position and actual detection area of ​​the glass surface corresponding to each detection area are determined. The actual glass area is composed of the actual position and actual detection area of ​​the glass surface, realizing the corresponding mapping between the imaging detection area and the physical area of ​​the glass surface. The steps are as follows: A calibration plate calibrated by metrology is used as the calibration reference. The surface of the calibration plate is engraved with a high-precision grid scale with known physical spacing. The scale deviation is controlled within the micrometer level. The calibration plate is placed horizontally on the testing station of the optical glass to be tested. The height and horizontal orientation of the calibration plate are finely adjusted so that the scale working surface of the calibration plate and the glass testing surface are in the same spatial plane, ensuring that the object distance of the calibration optical path and the formal testing optical path are completely consistent.

[0029] Under the same lens focal length, aperture, and imaging distance parameters as the formal test, a clear grid image of the calibration board is acquired, and the pixel coordinates of all grid intersections in the image are identified by the corner point extraction algorithm. Combined with the actual physical coordinates of the grid intersections, the transformation relationship from the pixel coordinates of the imaging surface to the physical coordinates of the glass surface is established by fitting calculation. At the same time, the radial distortion and tangential distortion correction coefficients of the lens are solved, and finally an imaging calibration relationship including distortion correction is formed, realizing a precise mapping from the pixel scale to the physical scale. The established imaging calibration relationship is invoked, and all detection areas are traversed one by one. The pixel coordinates of the four vertices of each detection area are substituted into the calibration transformation relationship in turn, and after distortion correction, they are converted into the physical coordinates corresponding to the glass surface. Based on the transformed physical coordinate boundaries, the actual center position, actual physical size and actual coverage area of ​​each detection area on the glass surface are calculated. Then, the actual position and actual area together constitute the actual glass area corresponding to the detection area, thus completing the mapping from the virtual detection unit of the imaging plane to the physical area of ​​the glass surface. Once the mapping is complete, all subsequent roughness determination results for the detection area can be directly mapped to the specific physical location on the glass surface, achieving precise spatial positioning of defects.

[0030] S3. Set the deviation threshold and simultaneously acquire the standard spot data corresponding to the qualified optical glass. Match and compare the test spot data of each detection area with the standard spot data. When the matching deviation between the test spot data and the standard spot data exceeds the deviation threshold, determine the corresponding detection area as an abnormal rough area. Then, summarize the adjacent abnormal rough areas to obtain the abnormal rough distribution area on the surface of the optical glass to be tested. In S3, under the same light source wavelength, incident angle, acquisition distance and imaging conditions as the optical glass under test, spot data is acquired on the qualified optical glass. The acquired qualified spot data is divided into regions and classified by angle to obtain standard spot data corresponding to different incident angles, different wavelengths and different detection areas. Several pieces of optical glass that have passed high-precision metrological calibration are selected as standard samples. The surface roughness of the samples is within the acceptable tolerance range and covers the normal fluctuation range of normal production process. The standard samples are fixed at the testing station to ensure that all testing conditions such as light source wavelength, incident angle, imaging distance, and exposure parameters are completely consistent with the testing conditions of the glass to be tested. Backscattered light spot images of the standard samples are collected one by one. For the spot data of all standard samples, perform angle classification and detection area division operations that are completely consistent with the data to be tested to obtain multiple sets of standard spot data corresponding to different incident angles and different detection area positions, and construct a standard spot benchmark library. In this process, standard spot data is used as the benchmark data for matching and comparison of each detection area, so that the test spot data of the optical glass under test can be compared with the standard spot data under the same irradiation conditions and the same area position.

[0031] In S3, the standard spot data obtained by multiple qualified optical glasses under the same testing conditions are statistically analyzed to determine the range of spot fluctuations that can occur in each testing area within the normal roughness range. The upper limit of the matching deviation is set according to the range of spot fluctuations, and this upper limit is used as the deviation threshold. Statistical analysis was performed on all standard sample data from the same detection area and under the same incident conditions in the standard spot reference library. The inherent fluctuation range of grayscale distribution and the inherent fluctuation range of contour morphology were calculated respectively to obtain the natural fluctuation range of spot characteristics under normal roughness.

[0032] Based on the upper limit of the normal fluctuation range, and after adding a reasonable safety margin, the gray-level mean square error threshold and the contour similarity threshold are determined respectively. The two thresholds together form the complete deviation threshold.

[0033] Among them, the grayscale mean square error threshold is used to limit the maximum allowable deviation of the light intensity value. The larger the value, the greater the allowable difference in light intensity. The contour similarity threshold is used to limit the maximum allowable distortion of the spot shape. The smaller the value, the greater the allowable shape difference. The formula is as follows: ; in, This is the grayscale mean square error threshold. This represents the average of the grayscale values ​​of all standard samples in this region, where n is a safety margin coefficient, typically set to 3 to cover 99.7% of normal fluctuations. The standard deviation of the gray level for all samples; ; in, The value ranges from 0 to 1, with values ​​closer to 1 indicating a higher requirement for contour consistency. The standard deviation of the contour area. This is the average area of ​​the region's outline across all standard samples; The difference between the test spot characteristics and the standard spot characteristics under the same detection area, the same wavelength and the same incident angle is calculated to obtain the matching deviation of the corresponding detection area. For each detection area of ​​the glass to be tested, standard spot data of the corresponding position and conditions are retrieved in sequence according to the incident angle, and matching comparison is carried out under the same conditions. The gray mean square deviation and contour similarity coefficient of the spot to be tested and the standard spot are calculated respectively, and the two calculation results are verified with the corresponding deviation thresholds. If the matching deviation is greater than the deviation threshold, it is determined that the test spot data of the detection area fails to match the standard spot data, and the monitoring armadillo is identified as an abnormally rough area. If the matching deviation is less than the deviation threshold, the roughness of the region is determined to be normal. For the k-th incident angle and the (i, j)-th detection region, the formula is as follows: ; in, Where is the grayscale mean square deviation, and P is the total number of valid pixels in the region. Let (i,j) be the pixel range of the (i,j)th detection region. Let be the pixel grayscale matrix of the light spot to be measured. This represents the pixel grayscale matrix of a standard light spot. Binarize the two light spots separately to extract their contour regions: ; in, The binary region of the light spot to be measured. This is a standard binary region of light spot. The contour similarity coefficient; when > ,or < The region was determined to be an abnormally rough region.

[0034] In S3, after determining all detection areas, the spatial location of all abnormal rough areas is identified. When two or more abnormal rough areas have a common adjacent boundary, or the distance between the areas is less than the preset merging distance threshold, these adjacent abnormal rough areas are merged and summarized to finally obtain a continuous area of ​​abnormal rough distribution on the surface of the optical glass to be tested.

[0035] After completing the anomaly determination of all detection areas, the spatial coordinates of all abnormal rough areas are traversed one by one to establish the region adjacency relationship judgment logic. First, it is determined whether two abnormal areas have a common adjacent boundary, that is, whether they are rectangular units that are directly adjacent in the top, bottom, left and right. For non-adjacent abnormal regions, calculate the minimum spatial distance between their boundaries. If the distance is less than the preset merging spacing threshold, they are determined to be adjacent abnormal regions that can be merged. All abnormally rough areas that meet the conditions of direct adjacency or proximity are merged and summarized, while isolated single-point anomalies without any adjacent areas are filtered out to eliminate misjudgments caused by imaging noise and tiny dust particles. Finally, a continuous distribution area of ​​abnormally rough areas on the surface of the optical glass under test is obtained. The merged result is more in line with the actual characteristics of the continuous distribution of optical glass processing defects.

[0036] S4. For abnormally rough distribution areas, the average roughness value corresponding to the successfully matched detection area is used as the initial roughness parameter, and an optical scattering model is established. Based on the optical scattering model, simulated spot data is generated for the abnormally rough area. The roughness parameter of the abnormally rough area is adjusted iteratively so that the difference between the simulated spot data and the test spot data meets the deviation threshold, thereby determining the roughness value of the abnormally rough area and completing the detection of the surface roughness of the optical glass.

[0037] In S4, a detection area that successfully matches the standard spot data is selected from the optical glass to be tested, and the qualified optical glass roughness value associated with the standard spot data corresponding to the detection area is obtained. From all the test areas of the glass to be tested, the test areas that are qualified after matching and comparison and successfully match the standard spot data are selected. The qualified test areas adjacent to the abnormal roughness distribution area are selected as reference samples. At the same time, the standard roughness value corresponding to each reference qualified area is extracted. This value is associated with the standard spot data and comes from the calibration results of qualified glass samples.

[0038] The roughness values ​​corresponding to the successfully matched detection areas are averaged to obtain the reference roughness of the optical glass under test in the current detection state. Then, the reference roughness is used as the initial roughness parameter when performing roughness inversion in abnormal roughness distribution areas. The roughness values ​​of all reference areas are calculated by arithmetic mean to obtain the reference roughness value under the current detection conditions. This value is used as the initial iteration parameter for roughness inverse inversion in abnormal roughness distribution areas. An optical scattering model is established. The input parameters of the optical scattering model include the incident light wavelength, incident angle, refractive index of the optical glass, and surface roughness parameters. The output of the optical scattering model is the simulated data of the intensity distribution of the scattered light spot under the corresponding parameter conditions.

[0039] An optical scattering model is constructed by combining Kirchhoff's scattering approximation theory with a two-way reflection distribution function. This model is suitable for the scattering characteristics of ultra-smooth optical glass surfaces and has high computational accuracy in the sub-nanometer to nanometer roughness range.

[0040] The model's input parameters include incident light wavelength, incident angle of the light source, refractive index of the optical glass under test, surface roughness parameters, as well as geometric layout parameters such as the spatial position of the light source, the image acquisition position, and the position of the glass surface under test.

[0041] The model assumes a Gaussian random rough surface and calculates the coherent superposition of scattering from surface micro-elements to obtain the light intensity distribution at different scattering angles. Then, combining this with the geometric projection relationship of the imaging system, it generates two-dimensional simulated light spot intensity distribution data that is completely consistent with the measured light spot pixel scale and coordinate system. The formula is as follows: ; in, It is a two-way reflection distribution function. Angle of incidence The scattering angle is... The azimuth angle of the scattering. The incident light wavelength, For Fresnel reflection coefficient, The root mean square roughness of the surface. Let be the surface power spectral density function. For spatial wave vector difference; ; in, Let be the simulated light intensity at pixel (x,y) within the (i,j)th detection region. For the incident light intensity, The solid angle corresponding to a single pixel in the imaging system. and The pixel coordinates are obtained by conversion using imaging calibration relationships.

[0042] In S4, the initial roughness parameters of the abnormally rough distribution area, the wavelength of the light source, the incident angle of the light source, the illumination position of the light source, the image acquisition position, and the position of the optical glass surface to be tested are input into the optical scattering model. The initial roughness parameters of the abnormal region, the wavelength of the light source at the corresponding position, the incident angle and the geometric layout parameters are input into the optical scattering model, and the simulated light spot data of the region is obtained by forward calculation. The optical scattering model calculates the scattering intensity distribution formed after light shines on an abnormally rough distribution area under the current roughness parameters, and generates simulated spot data corresponding to the test spot data based on the scattering intensity distribution; The difference between the simulated spot data and the test spot data corresponding to the abnormally coarse distribution area is calculated to obtain the simulation deviation. When the simulation deviation exceeds the deviation threshold, the roughness parameter of the abnormally rough area is increased or decreased according to the adjustment step size, and the simulated spot data is regenerated based on the adjusted roughness parameter. In the early stages of iteration, a large step size is used for rapid approximation, and when the threshold is approached, a small step size is used for fine adjustment. Repeat the difference calculation and roughness parameter adjustment until the simulated deviation between the simulated spot data and the test spot data is less than or equal to the deviation threshold. When determining the roughness value of an abnormally rough area, the roughness parameter is iteratively calculated for multiple detection areas within the same abnormally rough distribution area. The roughness parameter corresponding to the final deviation threshold of each detection area is determined as the roughness value of that detection area. Finally, the roughness distribution result of the optical glass surface under test is generated based on the roughness values ​​of multiple detection areas.

[0043] For all independent detection areas contained within the same abnormally rough distribution area, the above iterative inversion calculation is performed one by one to obtain the accurate roughness value corresponding to each detection unit.

[0044] For test areas that are determined to have acceptable roughness, their corresponding standard roughness values ​​are directly used as the test results.

[0045] The roughness values ​​of all the detection areas are arranged in an orderly manner according to their spatial row and column positions to construct a two-dimensional roughness distribution matrix of the optical glass surface to be tested.

[0046] Based on this distribution matrix, a roughness distribution map can be further generated, which can intuitively present the differences in roughness at different locations on the glass surface and the distribution pattern of abnormal areas.

[0047] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for detecting the surface roughness of optical glass, characterized in that: Includes the following steps: S1. Multiple light sources of different wavelengths are used to illuminate the surface of the optical glass under test at different incident angles, and test spot data scattered by the optical glass under test are collected. According to the wavelength encoding information of each light source and the incident angle calibration relationship of the light source, the test spot data is classified by angle. S2. Divide the test spot into multiple detection areas based on the center position of the test spot, and determine the actual glass area corresponding to each detection area on the surface of the optical glass to be tested by combining the imaging calibration relationship. S3. Set the deviation threshold and simultaneously acquire the standard spot data corresponding to the qualified optical glass. Match and compare the test spot data of each detection area with the standard spot data. When the matching deviation between the test spot data and the standard spot data exceeds the deviation threshold, determine the corresponding detection area as an abnormal rough area. Then, summarize the adjacent abnormal rough areas to obtain the abnormal rough distribution area on the surface of the optical glass to be tested. S4. For abnormally rough distribution areas, the average roughness value corresponding to the successfully matched detection area is used as the initial roughness parameter, and an optical scattering model is established. Based on the optical scattering model, simulated spot data is generated for the abnormally rough area. The roughness parameter of the abnormally rough area is adjusted iteratively so that the difference between the simulated spot data and the test spot data meets the deviation threshold, thereby determining the roughness value of the abnormally rough area and completing the detection of the surface roughness of the optical glass.

2. The method for detecting the surface roughness of optical glass according to claim 1, characterized in that: In step S1, multiple light sources emitting different wavelengths are set. Each light source is mounted above the optical glass to be tested via a fixed bracket with angle adjustment function. Each light source is set with an independent incident angle. All incident angles are based on the normal direction of the surface of the optical glass to be tested. Different wavelengths of light sources correspond to different incident angles, so that the incident beams of each light source converge in the same detection area on the surface of the glass to be tested. An array of photoelectric imaging devices is arranged along the normal direction of the glass to receive backscattered light generated by the glass surface. In a single detection cycle, each light source is lit sequentially according to the wavelength encoding order, and scattered light spot images at the corresponding wavelengths are collected. All single-wavelength light spot images are combined to form the test light spot data.

3. A method for detecting the surface roughness of optical glass according to claim 1, characterized in that: In step S1, a calibration experiment is conducted to establish the correspondence between the incident angle of each wavelength light source and the angular distribution of the scattered light spot as the incident angle calibration relationship of the light source. After obtaining the test spot data, the corresponding incident angle calibration relationship is matched according to the wavelength encoding information carried by the spot data. The test spot data is grouped according to the incident angle dimension, and each group of data corresponds to the scattered light distribution under an incident angle, thus completing the angle classification of the test spot data.

4. A method for detecting the surface roughness of optical glass according to claim 1, characterized in that: In S2, for each group of test spots that has completed angle classification, the gray-scale centroid algorithm is used to calculate the gray-scale centroid coordinates of the spot, and the coordinates are used as the reference for the center position of the test spot. With the center of the light spot as the origin, several rectangular detection areas of equal area are divided in the imaging plane. All detection areas are arranged in a row and column array and completely cover the effective light intensity distribution range of the test light spot. The imaging system is calibrated using a standard calibration plate to establish an imaging calibration relationship between the pixel coordinates of the imaging surface and the physical coordinates of the glass surface under test. Based on this imaging calibration relationship, the pixel coordinate range of each detection area on the imaging surface is converted into the actual physical area coordinates of the optical glass surface under test. The actual position and actual detection area of ​​the glass surface corresponding to each detection area are determined. The actual glass area is composed of the actual position and actual detection area of ​​the glass surface, thus realizing the corresponding mapping between the imaging detection area and the physical area of ​​the glass surface.

5. A method for detecting the surface roughness of optical glass according to claim 1, characterized in that: In step S3, under the same light source wavelength, incident angle, acquisition distance, and imaging conditions as the optical glass to be tested, spot data is acquired on the qualified optical glass. The acquired qualified spot data is divided into regions and classified by angle to obtain standard spot data corresponding to different incident angles, different wavelengths, and different detection regions. In this process, standard spot data is used as the benchmark data for matching and comparison of each detection area, so that the test spot data of the optical glass under test can be compared with the standard spot data under the same irradiation conditions and the same area position.

6. A method for detecting the surface roughness of optical glass according to claim 1, characterized in that: In step S3, statistical analysis is performed on the standard spot data obtained by multiple qualified optical glasses under the same testing conditions to determine the allowable spot fluctuation range in each testing area within the normal roughness range. An upper limit value of the matching deviation is set according to the spot fluctuation range, and this upper limit value is used as the deviation threshold. The difference between the test spot characteristics and the standard spot characteristics under the same detection area, the same wavelength and the same incident angle is calculated to obtain the matching deviation of the corresponding detection area. If the matching deviation is greater than the deviation threshold, it is determined that the test spot data of the detection area fails to match the standard spot data, and the monitoring armadillo is identified as an abnormally rough area. If the matching deviation is less than the deviation threshold, the roughness of the area is considered normal.

7. A method for detecting the surface roughness of optical glass according to claim 1, characterized in that: In step S3, after determining all detection areas, the spatial location of all abnormal rough areas is identified. When two or more abnormal rough areas have a common adjacent boundary, or the distance between the areas is less than the preset merging distance threshold, these adjacent abnormal rough areas are merged and summarized to finally obtain a continuous abnormal rough distribution area on the surface of the optical glass to be tested.

8. A method for detecting the surface roughness of optical glass according to claim 1, characterized in that: In step S4, a detection area that successfully matches the standard spot data is selected from the optical glass to be tested, and the qualified optical glass roughness value associated with the standard spot data corresponding to the detection area is obtained. The roughness values ​​corresponding to the successfully matched detection areas are averaged to obtain the reference roughness of the optical glass under test in the current detection state. Then, the reference roughness is used as the initial roughness parameter when performing roughness inversion in abnormal roughness distribution areas. An optical scattering model is established. The input parameters of the optical scattering model include the incident light wavelength, incident angle, refractive index of the optical glass, and surface roughness parameters. The output of the optical scattering model is the simulated data of the intensity distribution of the scattered light spot under the corresponding parameter conditions.

9. A method for detecting the surface roughness of optical glass according to claim 1, characterized in that: In step S4, the initial roughness parameters of the abnormally rough distribution area, the wavelength of the light source, the incident angle of the light source, the illumination position of the light source, the image acquisition position, and the position of the optical glass surface to be tested are input into the optical scattering model. The optical scattering model calculates the scattering intensity distribution formed after light shines on an abnormally rough distribution area under the current roughness parameters, and generates simulated spot data corresponding to the test spot data based on the scattering intensity distribution; The difference between the simulated spot data and the test spot data corresponding to the abnormally coarse distribution area is calculated to obtain the simulation deviation. When the simulation deviation exceeds the deviation threshold, the roughness parameter of the abnormally rough area is increased or decreased according to the adjustment step size, and the simulated spot data is regenerated based on the adjusted roughness parameter. Repeat the difference calculation and roughness parameter adjustment until the simulated deviation between the simulated spot data and the test spot data is less than or equal to the deviation threshold. When determining the roughness value of an abnormally rough area, the roughness parameter is iteratively calculated for multiple detection areas within the same abnormally rough distribution area. The roughness parameter corresponding to the final deviation threshold of each detection area is determined as the roughness value of that detection area. Finally, the roughness distribution result of the optical glass surface under test is generated based on the roughness values ​​of multiple detection areas.

Images