Glass cover plate multi-dimensional detection method based on 3D structured light and multi-stage optical reflection

Abstract

The application discloses a glass cover multi-dimensional detection method based on 3D structured light and multi-stage optical reflection, which comprises the following steps: 3D profile and hole site detection and surface micro-defect detection. The 3D profile and hole site detection comprises detection parameter adaptive call, optical component pre-start and calibration, polarization fringe pattern projection and collection, phase data analysis and optimization, 3D point cloud reconstruction and fine extraction and processing on the 3D point cloud model, extraction of first detection data, comparison, obtaining of a detection result, storage of the detection result and corresponding detection data into a database and association of product coding. The surface micro-defect detection comprises reflection angle adaptive adjustment, laser light source start and parameter regulation, laser multi-stage reflection and signal collection, optical signal analysis and defect identification and defect judgment and information recording. The glass cover can be automatically detected in full dimension and integration, the detection precision is high, the detection efficiency is high, the detection data is traceable, and the full life cycle management of product quality is realized.

Description

Technical Field

[0001] This invention relates to the field of glass cover inspection technology, and in particular to a multi-dimensional inspection method for glass covers based on 3D structured light and multi-level optical reflection, for the full-dimensional inspection of 2D / 3D curved edge glass covers. Background Technology

[0002] With the development of the consumer electronics industry, 2D / 3D curved edge glass covers, as core components of smartphones, tablets, and smart wearable devices, face increasingly stringent requirements for surface flatness, curved edge contour accuracy, hole dimensions, and surface micro-defects (scratches, dents, chipping, etc.). Currently, glass cover inspection technologies in the industry are mainly divided into three categories, all of which have significant technical shortcomings: 1. Manual inspection technology: This method involves multi-angle observation under strong light using a high-powered microscope. However, it suffers from high labor intensity, low inspection efficiency, and high rates of missed and false detections. It cannot meet the cycle time requirements of large-scale production, and the inspection results are greatly affected by subjective human factors, resulting in poor consistency. 2. Visual Imaging Inspection Equipment: This equipment uses a multi-camera group to acquire images of the short / long arc edges and hole areas of the glass cover, achieving automated inspection. However, it can only complete the visual recognition of 2D planes and simple 3D contours. It cannot effectively detect micro-contour deviations of 3D arc edges or micro-defects inside the glass cover. Furthermore, the fixed acquisition angle of the camera group creates a visual blind spot for the surface inspection of arc edges, which can easily lead to missed detection of minute defects. 3. Multi-level reflection optical inspection equipment: It amplifies the surface defects of the glass cover plate through multi-level laser reflection, which improves the defect detection sensitivity. However, it can only detect surface defects and cannot complete the quantitative detection of hole position accuracy and arc edge contour size. Moreover, the angle adjustment of the reflector depends on manual operation, the inspection process is cumbersome, and it is difficult to achieve full automation. 4.3D structured light camera system: It can realize high-precision 3D point cloud reconstruction and obtain three-dimensional contour data of objects. However, when used alone, it has poor adaptability to the transparency of glass cover plates. Light reflection on the glass surface can easily cause distortion of point cloud data. In addition, it cannot directly detect micro-defects on the surface and needs to be combined with other detection methods. The integration of the equipment is low.

[0003] In summary, existing technologies suffer from several technical problems, including limited detection dimensions (either only detecting contours or only surface defects), low accuracy in 3D curved surface detection (due to blind spots / data distortion), insufficient automation (some steps rely on manual labor), and low equipment integration (multiple devices are required to complete all-item inspections). These issues make it difficult to meet the demand for full-dimensional, high-precision, and automated inspection of 2D / 3D glass covers. Summary of the Invention

[0004] This application provides a multi-dimensional inspection method for glass covers based on 3D structured light and multi-level optical reflection. It can automatically inspect glass covers in all dimensions with high accuracy and efficiency. At the same time, the inspection data is traceable, realizing full life cycle management of product quality.

[0005] This application provides a multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection, including the following steps: S10, 3D contour and hole position detection; S11: Collect the product specification characteristics of the glass cover and match them with the system's built-in adaptive detection parameter library to automatically retrieve the corresponding specification's detection parameters. S12, the 3D structured light optical engine projects sinusoidal polarized fringe patterns onto the entire surface of the glass cover at a preset frequency, and simultaneously acquires the fringe deformation pattern of the glass cover through an industrial area array camera. S13, the wrapping phase of the stripe deformation pattern is obtained by using the four-step phase shift method, and the wrapping phase is unwrapped to obtain continuous phase distribution data on the surface of the glass cover plate. Then, the phase distribution data is denoised and error compensated to generate phase mapping data. S14. Based on the phase mapping data, a 3D point cloud model of the glass cover plate is generated using the triangulation method, covering the plane, arc edge and inner wall area of ​​the hole of the glass cover plate. S15, perform noise reduction, smoothing and registration processing on the 3D point cloud model, automatically extract the first detection data according to the preset extraction rules, compare the first detection data with the corresponding specification standard data in the standard data threshold library to obtain the detection result, store the detection result and the corresponding detection data in the database, and associate it with the product code; S20, Surface micro-defect detection; S21, After collecting the product specification features of the glass cover in step S11, the reflection angle of the reflected light is automatically adjusted to ensure that the reflected light path completely covers the entire surface of the glass cover. S22 controls the laser beam to repeatedly pass through the surface of the glass cover plate, and the optical receiver synchronously collects the optical signal of the reflected laser. Its collection frequency is synchronized with the collection frame rate of the industrial area array camera. S23, perform brightness analysis, offset analysis and spot analysis on the optical signal collected by the optical receiver to obtain the second detection data, and at the same time, perform magnification and identification processing on the suspected defect area that is close to the judgment threshold. S24. Based on the optical signal analysis results, determine whether there are micro-defects on the surface of the glass cover, store the judgment results and defect information in the database, and associate them with the product code.

[0006] In one possible implementation, in step S11, the detection parameters in the system's built-in adaptive detection parameter library include: structured light projection frequency, camera acquisition frame rate, phase resolution, point cloud reconstruction resolution, and contour / hole position standard threshold range.

[0007] In one possible implementation, before step S12, an anti-reflective polarizing filter is first installed at the projection end of the optical engine, and the anti-reflective polarizing filter is adjusted to a polarization angle that matches the transparency of the glass. At the same time, the optical lens of the industrial area array camera acquisition end is focused to a clear focal length, and the lens aperture is adjusted to F2.8-F4.0.

[0008] In one possible implementation, in step S12, the projection resolution of the 3D structured light optomechanical system is ≥1920x1080, the acquisition frame rate of the industrial area array camera is ≥30fps, and the industrial area array camera continuously acquires multiple frames of valid images.

[0009] In one possible implementation, in step S15, the detection data automatically extracted according to the preset extraction rules includes arc edge contour data and hole position accuracy data. For the arc edge contour data, the radius of curvature, chord length, height and chamfer size of multiple sampling points of the arc edge are extracted, and the sampling point spacing is ≤0.002mm. For the hole position accuracy data, the center coordinates X / Y / Z, hole diameter, hole depth, distance between the hole position and the reference edge, and center distance between multiple holes are extracted.

[0010] In one possible implementation, in step S22, the optical receiver synchronously acquires the optical signals obtained from the reflected laser, including: laser brightness distribution, laser path offset, and scattered spot position.

[0011] In one possible implementation, in step S23, for brightness analysis: the laser brightness distribution data is compared with the standard brightness distribution data to identify areas of abnormal brightness; for offset analysis: the difference between the actual offset of the laser path and the standard offset is calculated, and if the difference exceeds 0.001mm, it is determined to be a micro-defect; for spot analysis: the scattered spot is located and its size is measured to determine the specific location and size of the defect.

[0012] In one possible implementation, in step S24, if a micro-defect exists, the defect type, defect location, and defect size are further identified, wherein the defect type includes scratches, pits, chipped edges, and micro-cracks; the defect location includes X / Y coordinates, accurate to 0.001 mm; and the defect size includes length / depth / width, accurate to 0.001 mm. If no micro-defects are found, the surface is determined to be defect-free.

[0013] In one possible implementation, the multi-dimensional detection method further includes step S30, comprehensive determination; S31, verify data integrity, retrieve the first and second test data and one-way judgment results of the same product. If there is missing data, it is determined to be a product to be re-inspected and a re-inspection signal is triggered; if the data is complete, proceed to the next logical judgment. S32, multi-dimensional logical judgment, judges according to the principle that all items are qualified and any single item is unqualified, including the following situations: Qualified product: The contour / hole position is determined to be qualified in step S10, and the surface is determined to be free of micro-defects in step S20; Outline / Hole position defective product: If the outline / hole position is determined to be defective in step S10 and the surface is determined to be free of micro-defects in step S20, or if the outline / hole position is determined to be defective in step S10 and the surface is determined to have micro-defects in step S20. Surface defect non-conforming product: In step S10, the outline / hole position is determined to be qualified, and in step S20, the surface is determined to have micro-defects; Items awaiting re-inspection: missing test data, abnormal single-item judgment results, or signal fluctuations during equipment testing; S33, Judgment result labeling and signal transmission: Each glass cover is classified and labeled, including product code, comprehensive judgment result, non-conformity items, and detailed defect information. An alarm signal is sent to the products to be re-inspected to remind staff to handle the matter. S34. For all glass covers that are judged to be non-conforming, a standardized defect report is automatically generated. The report includes: basic product information, testing equipment information, testing time, details of non-conforming items, individual judgment results, and comprehensive judgment results. The report is stored in the database in a predetermined format.

[0014] In one possible implementation, the multi-dimensional detection method further includes step S40, data storage and tracing; S41, Data Standardization and Integration: Integrate the first inspection data, second inspection data, and standardized defect reports for the same product code, and process them according to the preset data format to ensure that the data fields, units, and accuracy are uniform. The length unit is mm, the time unit is s, and the accuracy is 0.001mm. S42 features multi-dimensional database storage. The integrated standardized data is stored in an industrial-grade database, employing a two-dimensional association method. The product dimension is categorized and stored by product code, production batch, and product specification, enabling direct retrieval of the entire process testing data for a single glass cover by product code. The testing dimension is categorized and stored by testing time, testing station, and equipment number, allowing retrieval of all testing data for a specific time period / station / equipment by time / station / equipment. The industrial-grade database features automatic backup functionality, performing local backups hourly and cloud backups daily. S43 allows users to retrieve corresponding test data and defect reports by inputting any keyword from product code, production batch, or inspection time into the operating terminal. Data can be exported in Excel, PDF, Flash, JPG, and print formats. The integrated standardized data is then integrated with the enterprise's MES and QMS systems. S44, Perform statistical analysis on the test data according to a preset cycle, and generate quality analysis reports, including: Pass rate statistics: overall pass rate, pass rate of products of each specification, and pass rate of each testing station; Non-conformance statistics: main types of contour / hole position non-conformance, main types of surface defects, and the percentage of each non-conformance item; Defect location statistics: the concentrated locations of defects in the same production batch; S45, Data Lifecycle Management, sets the data storage lifecycle, archives expired data to ensure long-term data traceability, and marks and isolates abnormal data.

[0015] Beneficial effects: Compared with the prior art, the multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection provided in this application integrates 3D structured light contour detection and multi-level reflection micro-defect detection. It can complete the full-item detection of the arc edge contour size, hole position accuracy, and surface or near-surface micro-defects of the glass cover plate in one loading. It can solve the problems of single detection dimension and multiple equipment required by the prior art, and reduce the equipment integration cost. The point cloud data has high accuracy, while the contour detection accuracy is less than 0.005mm and the defect detection sensitivity can reach 0.001mm; 3D contour inspection and surface defect inspection are performed simultaneously, with a single product inspection time of ≤5s, resulting in high work efficiency and adapting to the high-cycle requirements of large-scale production; overlapping inspection stations can avoid errors caused by secondary positioning and further improve inspection accuracy. The detection parameters can be adjusted adaptively without manual intervention, which can also significantly reduce labor intensity; The test data is stored in conjunction with the product code. Through defect reports, it can provide data support for production process optimization and realize full life cycle management of product quality.

[0016] These and other objects, features and advantages of the present invention will become fully apparent from the following detailed description. Attached Figure Description

[0017] Figure 1 The flowchart of the multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection of this application is shown. Detailed Implementation

[0018] The following description is intended to disclose the present invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art. The basic principles of the invention defined in the following description can be applied to other embodiments, modifications, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the invention.

[0019] Those skilled in the art should understand that, in the disclosure of this specification, the terms "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the above terms should not be construed as limiting the present invention.

[0020] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. It will also be understood that terms, such as those defined in commonly used dictionaries, shall be interpreted as having a meaning consistent with their meaning in the context of the relevant art and shall not be interpreted as having an idealized or overly formal meaning unless expressly so defined herein.

[0021] It is understood that the term "a" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple, and the term "a" should not be understood as a limitation on the number.

[0022] refer to Figure 1 This application provides a multi-dimensional detection method for glass covers based on 3D structured light and multi-level optical reflection, including the following steps: S10, 3D contour and hole position detection, generates high-precision 3D point cloud data based on polarized structured light phase analysis method, completes quantitative detection of arc edge contour size, hole position coordinates and accuracy, detection accuracy ≤0.005mm, and is started and processed in parallel with subsequent surface micro-defect detection. S11, Adaptive Detection Parameter Retrieval: After receiving the signal that the product has arrived, the central controller collects the product specification features of the glass cover through the vision recognition module, including features such as size, curvature, number / distribution of holes, etc., and matches them with the system's built-in adaptive detection parameter library to automatically retrieve the corresponding specification detection parameters. The detection parameters in the system's built-in adaptive detection parameter library include: structured light projection frequency, camera acquisition frame rate, phase resolution, point cloud reconstruction resolution, contour / hole position standard threshold range, etc. S12, Optical Component Pre-start and Calibration: First, the 3D structured light optical engine and industrial area array camera are pre-started to complete the optical path calibration. Then, the anti-reflective polarizing filter at the projection end of the 3D structured light optical engine is automatically adjusted to the corresponding polarization angle to adapt to the transparency of the glass cover and suppress surface specular reflection. At the same time, the high-transmittance optical lens at the camera acquisition end is automatically focused to a clear focal length, and the lens aperture is adjusted to F2.8-F4.0 to ensure that the acquired fringe pattern is not blurred or overexposed. Polarization fringe pattern projection and acquisition: The 3D structured light optomechanical system projects a sinusoidal polarization fringe pattern onto the entire surface of the glass cover plate (including the arc edge and hole area) at a preset frequency. The projection resolution is preferably ≥1920×1080. In addition, while the optomechanical system is projecting, an industrial area array camera simultaneously acquires the fringe deformation pattern of the glass cover plate. Preferably, the industrial area array camera acquires the fringe deformation pattern of the cover plate surface at a frame rate of ≥30fps, and continuously acquires 5 valid images to remove blurry and noisy frames to ensure the validity of the image data. S13, Phase data analysis and optimization: The acquired fringe deformation map is sent to the phase analysis unit. The phase analysis unit uses a four-step phase shift method to obtain the wrapped phase of the fringe deformation map and performs unwrapping processing on the wrapped phase to obtain continuous phase distribution data on the surface of the glass cover. Then, the phase distribution data is subjected to noise reduction processing and error compensation to eliminate the phase deviation caused by optical path distortion and glass refraction, and generate accurate phase height mapping data. Noise reduction processing is performed by Gaussian filtering. S14, 3D point cloud reconstruction and refined extraction: Based on phase mapping data, a high-precision 3D point cloud model of the glass cover is generated using the triangulation method. The preferred point cloud density is ≥1000 points / mm², covering the plane, arc edge and inner wall area of ​​the hole of the glass cover. S15, perform noise reduction, smoothing and registration processing on the 3D point cloud model, automatically extract the first detection data according to the preset extraction rules, compare the first detection data with the corresponding specification standard data in the standard data threshold library to obtain the detection result, store the detection result and the corresponding detection data in the database, and associate it with the product code; S20, surface micro-defect detection, based on the principle of laser multi-level reflection amplification, to achieve accurate detection of surface / near-surface micro-defects (scratches, pits, chipping, micro-cracks, etc., with a detection sensitivity of 0.001mm) on the glass cover plate. It is performed synchronously with step S10, with the optical path and 3D structured light optical path coaxially linked, and the detection stations completely overlapping. S21, Adaptive adjustment of reflection angle: After collecting the product specification features of the glass cover in step S11, based on the principle of multi-level laser reflection amplification, the central controller sends a synchronous start signal to the multi-level reflection defect detection module while retrieving the detection parameters of the glass cover. According to the product specifications (such as arc curvature and cover thickness), the reflection angle of the laser reflected light is automatically adjusted with an adjustment accuracy of ≤0.1° to ensure that the reflected light path completely covers the entire surface of the glass cover, including the curved surface of the arc edge, the edge of the hole and other areas that are prone to visual blind spots. The laser reflection times are preferably 3-5 times, which can be adapted according to the thickness of the glass cover. S22, Laser source startup and parameter control: The laser source generator starts up and emits a high-power monochromatic laser beam (wavelength 650nm, power adjustable, adaptable to glass cover plates of different thicknesses). After being collimated by a collimating lens, the laser beam is projected onto the first reflector in the plate-shaped reflector group. The central controller adjusts the laser power in real time according to the transparency of the glass cover plate to ensure that the optical receiver can receive a stable laser signal after the laser beam passes through the glass cover plate, without being too weak or too strong. Laser multi-level reflection and signal acquisition: After multi-level reflection by the plate-shaped reflector group, the laser beam is repeatedly transmitted through the surface / near-surface area of ​​the glass cover plate. If there are micro-defects in the cover plate, the refractive index of the glass at the defect will change, causing the laser beam to produce phenomena such as brightness attenuation, path offset, and scattering. The optical receiver (high-sensitivity photoelectric sensor + area array imager) synchronously acquires the optical signals of the reflected laser, including laser brightness distribution, laser path offset, and scattered spot position. The acquisition frequency is synchronized with the acquisition frame rate of the industrial area array camera in step S12 to ensure matching detection time. S23, Optical Signal Analysis and Defect Identification: The central controller performs brightness analysis, offset analysis, and spot analysis on the optical signals collected by the optical receiver to obtain the second detection data. For brightness analysis: the laser brightness distribution data is compared with the standard brightness distribution data to identify areas of abnormal brightness; the brightness at the defect location will show a local decrease or increase. For offset analysis: the difference between the actual offset of the laser path and the standard offset is calculated; if the difference exceeds 0.001mm, it is judged as a micro-defect. For spot analysis: the scattered spot is located and its size is measured to determine the specific location and size of the defect. Suspected defect areas approaching the judgment threshold are magnified for identification processing to eliminate interference factors such as dust and water stains on the glass cover surface (through laser focusing detection, dust / water stains are surface adhesions, while defects are damage to the glass body, and the optical path signal characteristics are different). S24, Defect Judgment and Information Recording: The image analysis unit determines whether there are micro-defects on the surface / near surface of the glass cover plate based on the optical signal analysis results, stores the judgment results and defect information (if there are no defects, it is recorded as none) in the database, and associates them with the product code. If micro-defects exist, the defect type, defect location, and defect size are further identified. The defect type includes scratches, dents, chipping, and micro-cracks; the defect location includes X / Y coordinates, accurate to 0.001mm; the defect size includes length / depth / width, accurate to 0.001mm; if there are no micro-defects, the surface is determined to be defect-free.

[0023] In some embodiments, in step S15, the detection data automatically extracted according to the preset extraction rules includes arc edge contour data and hole position accuracy data. For the arc edge contour data, the radius of curvature, chord length, height and chamfer size of multiple sampling points of the arc edge are extracted, and the sampling point spacing is ≤0.002mm. For the hole position accuracy data, the center coordinates X / Y / Z, hole diameter, hole depth, distance between the hole position and the reference edge, and center distance between multiple holes are extracted.

[0024] In some embodiments, prior to step S10, the multi-dimensional detection method further includes automatic feeding; Material receiving and initial positioning: The receiving module automatically moves to the material receiving position on the production line according to the production line rhythm. It receives the glass cover plate for inspection through a flexible receiving pad (anti-scratch). The receiving sensor detects whether the cover plate is in place. If it is not in place, it sends a material shortage signal to the central controller and the receiving module resets to wait. If it is in place, it triggers the initial positioning gripper to gently clamp the edge of the glass cover plate, limiting the left and right displacement of the cover plate. Precise positioning and attitude calibration: After initial positioning, the glass cover plate moves to the positioning seat station along with the receiving module. The short-side positioning cylinder and the long-side positioning cylinder on the positioning seat work together in the order of "long side positioning first, short side calibration later". The polyurethane buffer head at the cylinder end contacts the edge of the glass cover plate to avoid hard contact that could cause the cover plate to chip. At the same time, the visual alignment sensor on the positioning seat collects data on the edge position of the glass cover plate and feeds the data back to the central controller. The extension and retraction of the cylinder is adjusted in real time to complete the calibration of the horizontal attitude of the glass cover plate, ensuring that the center of the peeled cover plate coincides with the center of the inspection station, with a positioning error of ≤0.01mm. Vacuum Adsorption and Station Transfer: After positioning calibration, the vacuum suction cup in the positioning seat is activated, generating a vacuum pressure of -0.08~-0.1MPa to stably adsorb the glass cover plate (preventing slippage during transfer); the robotic arm of the vacuum suction cup loading component moves along a preset trajectory to transfer the adsorbed glass cover plate to the inspection station. The transfer process is divided into 3 stages (fast movement - slow movement - fine stop), and the positioning error compensation during the fine stop stage is ≤0.005mm; Position Detection and Signal Feedback: After the glass cover is transferred to the inspection station, the vacuum adsorption device at the inspection station is activated to take over adsorption of the cover. At the same time, the laser position sensor at the station detects whether the glass cover is fully in place and without tilting from the X, Y, and Z directions. After confirmation of position, the robotic arm of the vacuum suction cup loading component resets, and the automated loading and unloading positioning module sends a high-level position signal to the central control processing module to trigger the subsequent inspection process. If tilting or offset of the cover is detected, an abnormal signal is sent, and the module controls the robotic arm to re-grip and calibrate. If the calibration fails three times, an alarm is triggered and the cover is sorted to the re-inspection area.

[0025] In some embodiments, the multi-dimensional detection method further includes step S30, comprehensive determination; S31, verify data integrity, retrieve the first and second test data and one-way judgment results of the same product. If there is missing data, it is determined to be a product to be re-inspected and a re-inspection signal is triggered; if the data is complete, proceed to the next logical judgment. S32, multi-dimensional logical judgment, judges according to the principle that all items are qualified and any single item is unqualified, including the following situations: Qualified product: The contour / hole position is determined to be qualified in step S10, and the surface is determined to be free of micro-defects in step S20; Outline / Hole position defective product: If the outline / hole position is determined to be defective in step S10 and the surface is determined to be free of micro-defects in step S20, or if the outline / hole position is determined to be defective in step S10 and the surface is determined to have micro-defects in step S20 (the main defect is marked as outline / hole position defect). Surface defect non-conforming product: In step S10, the outline / hole position is determined to be qualified, and in step S20, the surface is determined to have micro-defects; Items awaiting re-inspection: missing test data, abnormal results for a single item (such as data exceeding the reasonable range), or signal fluctuations during equipment testing; S33, Judgment result labeling and signal transmission: Each glass cover is classified and labeled, including product code, comprehensive judgment result, non-conformance items (if any), and detailed defect information (if any). An alarm signal is sent separately for the product to be re-inspected to remind the staff to handle it. S34 automatically generates a standardized defect report for all glass covers that are deemed non-conforming. The report includes: basic product information (including specifications, product code, and production batch), testing equipment information (including equipment number and testing station), testing time, details of non-conforming items (including defect type, location, size, and deviation value), individual judgment results, and comprehensive judgment results. The report is stored in a database in a predetermined format (such as PDF, Excel, JPG, etc.) and can be retrieved at any time.

[0026] In some embodiments, after step S30, the multi-dimensional detection method further includes automatic sorting and unloading, that is, the judgment result in step S33 is sent to the automatic sorting and unloading module in real time to trigger the sorting process. Based on the comprehensive judgment signal of the central controller, the glass cover plate is automatically transferred to the workstation, accurately sorted, classified and unloaded, and reminded of material shortage / fullness. The whole process is automated and there is no human intervention. Release and transfer at the inspection station: After inspection is completed, the vacuum adsorption device at the inspection station is turned off, releasing the glass cover plate; the vacuum suction cup of the unloading and conveying mechanism is activated, adsorbing the cover plate and transferring it to the sorting station according to the preset trajectory. During the transfer, the visual recognition sensor at the sorting station is triggered to re-check the product code of the glass cover plate to ensure that the code matches the judgment signal and avoid sorting errors. Adaptive sorting path retrieval: After receiving the comprehensive judgment signal, the automatic sorting and unloading module automatically retrieves the corresponding sorting path. For qualified products: retrieve the qualified product unloading channel path; for products with unqualified contours / holes: retrieve the contour / hole defective product channel path; for products with unqualified surface defects: retrieve the surface defective product channel path; for products awaiting re-inspection: retrieve the re-inspection area path. Precise sorting and unloading: The robotic arm of the vacuum suction cup sorting component moves according to the retrieved path. The vacuum suction cup (with anti-scratch pad) at the end of the robotic arm stably adsorbs the cover plate and moves it to the unloading position of the corresponding channel / area. The material carrier platform (with flexible buffer layer) at the unloading position receives the cover plate. The counting sensor in the carrier platform records the unloaded quantity. At the same time, the vacuum suction cup closes, releases the glass cover plate, and the robotic arm resets to wait for the next sorting instruction. During the sorting process, if the cover plate adsorption fails (insufficient vacuum pressure), a re-adsorption instruction is triggered. If the re-adsorption fails three times, an alarm is triggered, and the glass cover plate is sorted to the waiting area. Full / Short Material Signal Feedback: Each material carrier platform in each discharge channel / area is equipped with a full material sensor and a short material sensor. When the number of covers in the carrier platform reaches the preset full material threshold, the full material sensor sends a full material signal to the central control processing module, which triggers an audible and visual alarm to remind staff to remove materials in time. When there is no continuous discharge from the qualified product channel, the short material sensor sends a short material signal, which is fed back to the production line by the module, reminding the production line to check the material loading process. Sorting data recording: The automatic sorting and discharging module stores data such as sorting time, sorting channel, and discharging quantity for each cover plate to the central controller, associates it with the product code, and realizes full data traceability of the sorting process.

[0027] In some embodiments, the multi-dimensional detection method further includes step S40, data storage and traceability, which standardizes the storage, manages the structure, and traces the data of the entire detection process in a structured manner, providing accurate data support for process optimization at the production end; S41, Data Standardization and Integration: Integrate the first inspection data, second inspection data, and standardized defect reports for the same product code, and process them according to the preset data format to ensure that the data fields, units, and accuracy are uniform. The length unit is mm, the time unit is s, and the accuracy is 0.001mm. S42 features multi-dimensional database storage. The integrated, standardized data is stored in an industrial-grade database (supporting massive data storage and high-speed retrieval). The storage method employs a two-dimensional association: Product dimension: categorized by product code, production batch, and product specification, enabling direct retrieval of the entire process testing data for a single glass cover via product code; Testing dimension: categorized by testing time, testing station, and equipment number, allowing retrieval of all testing data for a specific time period / station / equipment via time / station / equipment. The industrial-grade database features automatic backup, performing hourly local backups and daily cloud backups to prevent data loss. S43 allows users to retrieve corresponding test data and defect reports by inputting any keyword from product code, production batch, or test time into the operating terminal. Data can be exported in Excel, PDF, Flash, JPG, and print formats. The integrated standardized data is interfaced with the enterprise's MES (Manufacturing Execution System) and QMS (Quality Management System), enabling the production and quality ends to directly retrieve test data from the enterprise management system, thus achieving data sharing. S44, statistically analyzes the test data according to a preset cycle (hour / day / week / month) and generates a quality analysis report, including: Pass rate statistics: overall pass rate, pass rate of products of each specification, and pass rate of each testing station; Non-conformance statistics: main types of contour / hole position non-conformance (such as curvature deviation, hole position offset), main types of surface defects (such as scratches, chipping), and the percentage of each non-conformance item; Defect location statistics: The concentrated locations of defects in the same production batch are analyzed to determine whether they are related to the production process (such as forming, polishing, drilling). The quality analysis report provides accurate data support for process optimization at the production end. For example, if the curvature deviation of the arc edge of a certain batch of cover plates is high, the mold parameters for glass hot bending can be optimized at the production end; if surface scratches are concentrated on the long edge, the fixture design for the polishing process can be optimized. S45, data lifecycle management, sets the data storage lifecycle (e.g., 1 year / 3 years / 5 years), archives expired data (compressed storage, not deleted), ensuring long-term data traceability; it also supports the labeling and isolation of abnormal data (e.g., detection deviation data caused by equipment failure), preventing abnormal data from affecting process optimization analysis.

[0028] Taking a certain model of 3D rounded edge mobile phone glass cover (size 150×70mm, arc curvature R2.5, including 2 camera holes) as an example, the following tests were conducted: 1. The automated loading and unloading positioning module receives the glass cover plate, and the short / long side positioning cylinders accurately position it with a positioning accuracy of 0.008mm. The vacuum suction cup loading component then transfers it to the inspection station.

[0029] The optomechanical control unit of the 2.3D structured light contour detection module projects a polarization fringe pattern. After the camera control unit acquires the fringe pattern, the point cloud reconstruction unit generates high-precision 3D point cloud data. The point cloud data processing unit extracts the arc edge contour dimension as R2.502mm and the hole position coordinate deviation as 0.003mm, both of which are within the standard range, and the contour / hole position is judged to be qualified.

[0030] 3. Simultaneously, the angle adjustment drive unit of the multi-level reflection defect detection module adjusts the reflective component group to the optimal reflection angle. The laser source generator emits laser light, which covers the surface of the glass cover after multi-level reflection. The laser brightness received by the optical receiver is consistent within the standard threshold range with no offset, and it is determined that there are no micro-defects on the surface.

[0031] 4. The test result determination unit determines that the glass cover is a qualified product, and the automatic sorting and unloading module transfers it to the qualified product unloading channel.

[0032] 5. The central control processing module binds and stores the test data of the glass cover with the product code to complete the test.

[0033] If the arc edge contour dimension of a glass cover is detected to be R2.510mm (exceeding the standard range), it is judged as a contour defective product, sorted to the contour / hole defective product channel, and a defect report is generated, indicating a contour deviation of 0.010mm; if a scratch of 0.002mm is detected on the surface of the glass cover, and the laser brightness shows a local shift, it is judged as a surface defect defective product, sorted to the surface defective product channel, and the defect report indicates the location and size of the scratch.

[0034] It should be noted that the terms "first" and "second" used in this application are for descriptive purposes only and do not indicate any order. They should not be construed as indicating or implying relative importance, and can be interpreted as names.

[0035] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are merely examples and do not limit the invention. The advantages of the present invention have been fully and effectively realized. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments; any variations or modifications can be made to the implementation of the present invention without departing from these principles.

Claims

1. A multi-dimensional detection method for glass covers based on 3D structured light and multi-level optical reflection, characterized in that, Includes the following steps: S10, 3D contour and hole position detection; S11: Collect the product specification characteristics of the glass cover and match them with the system's built-in adaptive detection parameter library to automatically retrieve the corresponding specification's detection parameters. S12, the 3D structured light optical engine projects sinusoidal polarized fringe patterns onto the entire surface of the glass cover at a preset frequency, and simultaneously acquires the fringe deformation pattern of the glass cover through an industrial area array camera. S13, the wrapping phase of the stripe deformation pattern is obtained by using the four-step phase shift method, and the wrapping phase is unwrapped to obtain continuous phase distribution data on the surface of the glass cover plate. Then, the phase distribution data is denoised and error compensated to generate phase mapping data. S14. Based on the phase mapping data, a 3D point cloud model of the glass cover plate is generated using the triangulation method, covering the plane, arc edge and inner wall area of ​​the hole of the glass cover plate. S15, perform noise reduction, smoothing and registration processing on the 3D point cloud model, automatically extract the first detection data according to the preset extraction rules, compare the first detection data with the corresponding specification standard data in the standard data threshold library to obtain the detection result, store the detection result and the corresponding detection data in the database, and associate it with the product code; S20, Surface micro-defect detection; S21, After collecting the product specification features of the glass cover in step S11, the reflection angle of the reflected light is automatically adjusted to ensure that the reflected light path completely covers the entire surface of the glass cover. S22, control the laser beam to repeatedly pass through the surface of the glass cover plate, and the optical receiver synchronously collects the optical signal of the reflected laser. Its collection frequency is synchronized with the collection frame rate of the industrial area array camera in step S12. S23, perform brightness analysis, offset analysis and spot analysis on the optical signal collected by the optical receiver to obtain the second detection data, and at the same time, perform magnification and identification processing on the suspected defect area that is close to the judgment threshold. S24. Based on the optical signal analysis results, determine whether there are micro-defects on the surface of the glass cover, store the judgment results and defect information in the database, and associate them with the product code.

2. The multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection as described in claim 1, characterized in that, In step S11, the detection parameters in the system's built-in adaptive detection parameter library include: structured light projection frequency, camera acquisition frame rate, phase resolution, point cloud reconstruction resolution, and contour / hole position standard threshold range.

3. The multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection as described in claim 1, characterized in that, Before step S12, an anti-reflective polarizing filter is first installed at the projection end of the optical engine. The anti-reflective polarizing filter is adjusted to a polarization angle that matches the transparency of the glass. At the same time, the optical lens of the industrial area array camera acquisition end is focused to a clear focal length, and the lens aperture is adjusted to F2.8-F4.

0.

4. The multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection as described in claim 3, characterized in that, In step S12, the projection resolution of the 3D structured light optical engine is ≥1920x1080, the acquisition frame rate of the industrial area scan camera is ≥30fps, and the industrial area scan camera continuously acquires multiple frames of valid images.

5. The multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection as described in claim 4, characterized in that, In step S15, the detection data automatically extracted according to the preset extraction rules includes arc edge contour data and hole position accuracy data. For the arc edge contour data, the radius of curvature, chord length, height and chamfer size of multiple sampling points of the arc edge are extracted, and the sampling point spacing is ≤0.002mm. For the hole position accuracy data, the center coordinates X / Y / Z, hole diameter, hole depth, distance between the hole position and the reference edge and the center distance between multiple holes are extracted.

6. The multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection as described in claim 1, characterized in that, In step S22, the optical receiver synchronously acquires the optical signals obtained from the reflected laser, including: laser brightness distribution, laser path offset, and scattered spot position.

7. The multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection as described in claim 1, characterized in that, In step S23, for brightness analysis: the laser brightness distribution data is compared with the standard brightness distribution data to identify areas of abnormal brightness; for offset analysis: the difference between the actual offset of the laser path and the standard offset is calculated. If the difference exceeds 0.001mm, it is determined to be a micro-defect; for spot analysis: the scattered spot is located and its size is measured to determine the specific location and size of the defect.

8. The multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection as described in claim 7, characterized in that, In step S24, if micro-defects exist, the defect type, defect location, and defect size are further identified. The defect type includes scratches, pits, chipped edges, and micro-cracks. The defect location includes X / Y coordinates, accurate to 0.001 mm. The defect size includes length / depth / width, accurate to 0.001 mm. If no micro-defects are found, the surface is determined to be defect-free.

9. The multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection as described in claim 1, characterized in that, It also includes step S30, comprehensive determination; S31, verify the integrity of the data, retrieve the first and second test data and the one-way judgment result of the same product. If there is missing data, it is determined to be a product to be re-inspected and a re-inspection signal is triggered. If the data is complete, proceed to the next logical judgment; S32, multi-dimensional logical judgment, judges according to the principle that all items are qualified and any single item is unqualified, including the following situations: Qualified product: The contour / hole position is determined to be qualified in step S10, and the surface is determined to be free of micro-defects in step S20; Outline / Hole position defective product: If the outline / hole position is determined to be defective in step S10 and the surface is determined to be free of micro-defects in step S20, or if the outline / hole position is determined to be defective in step S10 and the surface is determined to have micro-defects in step S20. Surface defect non-conforming product: In step S10, the outline / hole position is determined to be qualified, and in step S20, the surface is determined to have micro-defects; Items awaiting re-inspection: missing test data, abnormal single-item judgment results, or signal fluctuations during equipment testing; S33, Judgment result labeling and signal transmission: Each glass cover is classified and labeled, including product code, comprehensive judgment result, non-conformity items, and detailed defect information. An alarm signal is sent to the products to be re-inspected to remind staff to handle the matter. S34. For all glass covers that are judged to be non-conforming, a standardized defect report is automatically generated. The report includes: basic product information, testing equipment information, testing time, details of non-conforming items, individual judgment results, and comprehensive judgment results. The report is stored in the database in a predetermined format.

10. The multi-dimensional detection method for glass cover plates based on 3D structured light and multi-level optical reflection as described in claim 1, characterized in that, It also includes step S40, data storage and traceability; S41, Data Standardization and Integration: Integrate the first inspection data, second inspection data, and standardized defect reports for the same product code, and process them according to the preset data format to ensure that the data fields, units, and accuracy are uniform. The length unit is mm, the time unit is s, and the accuracy is 0.001mm. S42 features multi-dimensional database storage. The integrated standardized data is stored in an industrial-grade database, employing a two-dimensional association method. The product dimension is categorized and stored by product code, production batch, and product specification, enabling direct retrieval of the entire process testing data for a single glass cover by product code. The testing dimension is categorized and stored by testing time, testing station, and equipment number, allowing retrieval of all testing data for a specific time period / station / equipment by time / station / equipment. The industrial-grade database features automatic backup functionality, performing local backups hourly and cloud backups daily. S43 allows users to retrieve corresponding test data and defect reports by inputting any keyword from product code, production batch, or inspection time into the operating terminal. Data can be exported in Excel, PDF, Flash, JPG, and print formats. The integrated standardized data is then integrated with the enterprise's MES and QMS systems. S44, Perform statistical analysis on the test data according to a preset cycle, and generate quality analysis reports, including: Pass rate statistics: overall pass rate, pass rate of products of each specification, and pass rate of each testing station; Non-conformance statistics: main types of contour / hole position non-conformance, main types of surface defects, and the percentage of each non-conformance item; Defect location statistics: the concentrated locations of defects in the same production batch; S45, Data Lifecycle Management, sets the data storage lifecycle, archives expired data to ensure long-term data traceability, and marks and isolates abnormal data.

Images