A high-precision defect detection method and system for smart glasses

Abstract

This invention discloses a high-precision defect detection method and system for smart glasses, relating to the field of image processing. The method includes: constructing an optical detection system, analyzing imaging parameters and adjusting system configuration to obtain an optically calibrated detection system; capturing the target glasses to the detection area, performing multi-angle surface detection, and acquiring a set of multi-angle glasses surface images; performing denoising and distortion correction preprocessing to obtain a set of multi-angle standard glasses surface images, performing panoramic stitching and 3D reconstruction to obtain a 3D model of the target glasses surface; performing multi-dimensional feature extraction and defect region identification to obtain a set of defect regions on the glasses surface, performing defect feature evaluation, and obtaining the glasses defect feature detection result. This method solves the technical problem of insufficient detection comprehensiveness and accuracy in existing smart glasses defect detection systems, achieving the technical effect of improving the comprehensiveness and accuracy of surface defect detection.

Description

Technical Field

[0001] The present application relates to the field of image processing, and in particular to a high-precision defect detection method and system for smart glasses. Background Art

[0002] As high-precision devices, VR / AR glasses require extremely high quality control during the production process. Any tiny surface defects, such as scratches, stains, and unevenness, may affect the user experience and device performance. In the production process of VR / AR glasses, traditional defect detection methods mainly include manual visual inspection and inspection based on traditional machine vision. Manual visual inspection relies on the experience and subjective judgment of the inspector, which is inefficient and prone to errors. Long-term manual inspection may also cause visual fatigue, further reducing detection accuracy. Although traditional machine vision inspection has a high degree of automation, it is limited by the resolution of the camera and the stability of the light source, making it difficult to achieve high-precision detection of tiny defects on the surface of VR / AR glasses. In addition, traditional machine vision inspection methods are mostly only able to perform two-dimensional plane inspections, unable to fully reflect the three-dimensional shape of the glasses surface, and easily miss some defects hidden in the curved surface or edges.

[0003] Among the current related technologies, smart glasses defect detection has technical problems such as insufficient comprehensiveness and accuracy. Summary of the Invention

[0004] The present application provides a high-precision defect detection method and system for smart glasses. The method adopts the method of building an optical detection system, performing imaging parameter analysis and system adjustment configuration, obtaining a stable and high-precision optical calibration detection system, using a manipulator to accurately control the detection position of the smart glasses, performing multi-angle surface detection on the smart glasses, comprehensively covering all areas of the smart glasses surface, performing pre-processing operations such as denoising and distortion correction on the acquired multi-angle glasses surface image set, obtaining high-quality standard images, constructing a three-dimensional model of the smart glasses surface through panoramic stitching and three-dimensional reconstruction, performing multi-dimensional feature extraction and defect area identification based on the three-dimensional model, accurately finding the defect area on the smart glasses surface, and performing detailed defect feature evaluation, etc., thereby achieving the technical effect of improving the comprehensiveness and accuracy of surface defect detection.

[0005] This application provides a high-precision defect detection method for smart glasses, comprising:

[0006] An optical inspection system is constructed, wherein the optical inspection system includes a stray light elimination lens, a large-target industrial camera, and an illumination surface light source, and imaging parameter analysis and system adjustment configuration are performed on the stray light elimination lens, the large-target industrial camera, and the illumination surface light source to obtain an optical calibration inspection system; according to the optical calibration inspection system, a glasses inspection position area is determined, a robot is controlled to clamp and grasp target glasses to the glasses inspection position area, and multi-angle surface inspection is performed on the target glasses through the optical calibration inspection system to obtain a multi-angle glasses surface image set; denoising and distortion correction preprocessing are performed on the multi-angle glasses surface image set to obtain a multi-angle standard glasses surface image set, and the multi-angle standard glasses surface image set is panoramically stitched and three-dimensionally reconstructed to obtain a target glasses surface three-dimensional model; multi-dimensional feature extraction and defect area identification are performed based on the target glasses surface three-dimensional model to obtain a glasses surface defect area set, and defect feature evaluation is performed on the glasses surface defect area set in sequence to obtain a glasses defect feature detection result.

[0007] In a possible implementation, the optical calibration detection system is obtained by performing the following processing:

[0008] The stray light elimination lens is installed on the interface of the large-target industrial camera to obtain an imaging detection device; characteristic parameter information of the target glasses is obtained, and imaging parameter analysis is performed on the illumination surface light source and the imaging detection device based on the characteristic parameter information to obtain optical imaging detection parameter information; verification and fine-tuning of the illumination surface light source and the imaging detection device are performed based on the optical imaging detection parameter information to determine target imaging detection parameter information; the illumination surface light source and the imaging detection device are adjusted and configured according to the target imaging detection parameter information to obtain the optical calibration detection system.

[0009] In a possible implementation, the optical imaging detection parameter information is obtained by performing the following processing:

[0010] A glasses detection database is obtained through data mining technology, wherein the glasses detection database includes historical glasses characteristic parameters, optical imaging parameters, and corresponding glasses detection imaging data; the glasses detection database is divided and integrated based on the historical glasses characteristic parameters, and the integrated glasses detection database is matched and extracted according to the characteristic parameter information to obtain a glasses detection calibration library; a set of optical imaging evaluation indicators is obtained, and a weighted evaluation is performed on the glasses detection imaging data based on the optical imaging evaluation indicator set to obtain optical imaging detection effect data; and a comparison and optimization is performed within the glasses detection calibration library based on the optical imaging detection effect data to obtain the optical imaging detection parameter information.

[0011] In a possible implementation, the target imaging detection parameter information is determined by performing the following processing:

[0012] Based on the optical imaging detection parameter information, the configuration of the lighting surface light source and the imaging detection device is verified to obtain glasses imaging detection effect information; the glasses imaging detection requirement standard is obtained, and the non-standard parameters of the glasses imaging detection effect information are extracted according to the glasses imaging detection requirement standard to obtain non-standard imaging effect parameters; based on the non-standard imaging effect parameters, associated imaging parameter optimization analysis is performed to obtain an associated imaging parameter optimization step; based on the associated imaging parameter optimization step, the optical imaging detection parameter information is optimized, adjusted, verified and optimized to determine the target imaging detection parameter information.

[0013] In a possible implementation, the three-dimensional model of the target eyeglass surface is obtained by performing the following processing:

[0014] A feature point detection algorithm is used to perform multi-dimensional feature point detection and feature point matching on the multi-angle standard glasses surface image set to obtain a matching feature point set; a reference glasses surface image is selected based on the multi-angle standard glasses surface image set; an overlapping area image transformation analysis is performed based on the reference glasses surface image and the matching feature point set to determine the remaining image projection transformation matrix set; the multi-angle standard glasses surface image set is transformed, aligned and spliced ​​based on the remaining image projection transformation matrix set to obtain a panoramic image of the glasses surface; a surface reconstruction algorithm is used to perform depth estimation and point cloud three-dimensional reconstruction on the panoramic image of the glasses surface to obtain a three-dimensional model of the target glasses surface.

[0015] In a possible implementation, the obtaining of the set of defect regions on the surface of the glasses involves performing the following processing:

[0016] Multidimensional feature extraction is performed on the target glasses surface three-dimensional model to obtain a glasses surface multidimensional feature set, wherein the glasses surface multidimensional feature set includes geometric features and texture features; according to the design quality requirements of the target glasses, a glasses surface defect database is constructed, and a support vector machine is used to perform classification training on the glasses surface defect database to obtain a glasses surface defect classifier; based on the glasses surface defect classifier, defect recognition and classification are performed on the glasses surface multidimensional feature set to obtain a glasses surface defect feature type set; based on the glasses surface defect feature type set, anchor frame marking is performed on the target glasses surface three-dimensional model to obtain the glasses surface defect area set.

[0017] In a possible implementation, the glasses defect feature detection result is obtained and the following processing is performed:

[0018] The target glasses surface three-dimensional model is meshed according to a preset size to obtain a glasses surface three-dimensional mesh model; the glasses surface defect area set is mapped and statistically mapped with the glasses surface three-dimensional mesh model to obtain a glasses defect area area set; a quality impact assessment is performed on the glasses surface defect feature type set to determine a surface defect type influencing factor set; and a defect feature weighted assessment is performed on the glasses defect area set based on the surface defect type influencing factor set to obtain the glasses defect feature detection result.

[0019] This application also provides a high-precision defect detection system for smart glasses, comprising:

[0020] An optical calibration detection system building module is used to build an optical detection system, which includes a stray light elimination lens, a large-target industrial camera, and an illumination surface light source. The imaging parameters of the stray light elimination lens, the large-target industrial camera, and the illumination surface light source are analyzed and the system is adjusted and configured to obtain an optical calibration detection system; a glasses multi-angle surface detection module is used to determine the glasses detection position area according to the optical calibration detection system, control the manipulator to clamp and grab the target glasses to the glasses detection position area, and perform calibration on the target glasses through the optical calibration detection system. Multi-angle surface detection, obtaining a multi-angle glasses surface image set; a panoramic stitching module, the panoramic stitching module is used to perform denoising and distortion correction preprocessing on the multi-angle glasses surface image set to obtain a multi-angle standard glasses surface image set, and the multi-angle standard glasses surface image set is panoramically stitched and three-dimensionally reconstructed to obtain a target glasses surface three-dimensional model; a defect recognition module, the defect recognition module is used to perform multi-dimensional feature extraction and defect area recognition based on the target glasses surface three-dimensional model to obtain a glasses surface defect area set, and the defect feature evaluation of the glasses surface defect area set is performed in sequence to obtain glasses defect feature detection results.

[0021] The present application proposes a high-precision defect detection method and system for smart glasses. First, an optical inspection system is established. The optical inspection system includes a stray light elimination lens, a large-target industrial camera, and an illumination surface light source. Imaging parameter analysis and system adjustment configuration are performed on the stray light elimination lens, the large-target industrial camera, and the illumination surface light source to obtain an optical calibration inspection system. Then, based on the optical calibration inspection system, the glasses inspection position area is determined, and the robot is controlled to clamp and grab the target glasses to the glasses inspection position area. The target glasses are subjected to multi-angle surface inspection through the optical calibration inspection system to obtain a multi-angle glasses surface image set. Then, the multi-angle glasses surface image set is denoised and distortion corrected pre-processed to obtain a multi-angle standard glasses surface image set. The multi-angle standard glasses surface image set is panoramically stitched and three-dimensionally reconstructed to obtain a three-dimensional model of the target glasses surface. Finally, multi-dimensional feature extraction and defect area identification are performed based on the three-dimensional model of the target glasses surface to obtain a glasses surface defect area set. The defect features of the glasses surface defect area set are evaluated in sequence to obtain glasses defect feature detection results, thereby achieving the technical effect of improving the comprehensiveness and accuracy of surface defect detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention are briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in precise order. Instead, various steps may be processed in reverse order or simultaneously as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0023] Figure 1 A flowchart of a high-precision defect detection method for smart glasses provided in an embodiment of the present application.

[0024] Figure 2 A schematic structural diagram of a high-precision defect detection system for smart glasses provided in an embodiment of the present application.

[0025] Explanation of the accompanying drawings: optical calibration and detection system building module 10, glasses multi-angle surface detection module 20, panoramic stitching module 30, defect recognition module 40. DETAILED DESCRIPTION

[0026] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0027] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0028] In the following description, reference is made to “some embodiments” which describe a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict. The terms “including” and “having” and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0029] The present application provides a high-precision defect detection method for smart glasses, such as Figure 1 As shown, the method includes:

[0030] Step S100, build an optical detection system, the optical detection system includes a stray light elimination lens, a large-target industrial camera and an illumination surface light source, perform imaging parameter analysis and system adjustment configuration on the stray light elimination lens, the large-target industrial camera and the illumination surface light source, and obtain an optical calibration detection system. Specifically, according to the detection requirements, select a stray light elimination lens with high resolution, low distortion and suitable for industrial applications. The stray light elimination lens is a specially designed lens used to reduce or eliminate light interference from non-detection areas, improve imaging quality, and has low reflection, low scattering and other characteristics. Purchase a large-target industrial camera. The large-target industrial camera is a camera designed for industrial applications with high resolution, high stability, low noise and other characteristics. Its target surface is large and can capture more image details. The illumination surface light source is selected based on factors such as the brightness, uniformity, and color temperature of the light source. The illumination surface light source is a light source device that provides uniform lighting, which is used to ensure that the light in the detection area is sufficient and uniform, thereby improving detection accuracy.

[0031] Install the stray light elimination lens on the front end of the large-area industrial camera, ensuring that the lens and camera interface are tightly connected and not loose. Secure the large-area industrial camera on a stable bracket or machine table, and adjust the position and angle of the large-area industrial camera to facilitate subsequent inspection operations. Install the illumination surface light source and adjust the position and angle of the illumination surface light source according to the size and shape of the inspection area so that the light can evenly illuminate the surface of the glasses to be inspected.

[0032] Conduct preliminary testing using standard samples or samples with known defects. Take multiple images to analyze image quality, checking parameters such as image clarity, contrast, and color reproduction to ensure that image quality meets inspection requirements. Analyze lens distortion, recording the type and degree of distortion in preparation for correction.

[0033] Based on the imaging parameter analysis results, adjust the focal length, aperture, and other parameters of the stray light elimination lens to achieve optimal imaging results. Adjust the exposure time, gain, and white balance of the large-format industrial camera to ensure moderate image brightness and accurate color. Correct lens distortion using image processing software or the camera's built-in distortion correction function. Adjust the brightness and color temperature of the illumination source to ensure uniform lighting that meets inspection requirements.

[0034] Use a standard calibration plate or calibration software to calibrate the optical inspection system. By photographing the feature points on the calibration plate, the camera's internal and external parameters and distortion coefficients are calculated. Use the calibration results to correct the camera to ensure imaging accuracy and consistency.

[0035] In one possible implementation, the optical calibration and detection system is obtained, and step S100 further includes step S110, in which the stray light elimination lens is installed on the interface of the large-target industrial camera to obtain an imaging detection device. Specifically, the interface specifications of the large-target industrial camera are confirmed to match the installation requirements of the stray light elimination lens, and the stray light elimination lens is aligned and fixed to the interface of the large-target industrial camera using appropriate tools. During the installation process, the stray light elimination lens is ensured to be aligned with the optical axis of the large-target industrial camera to avoid introducing additional imaging errors. Step S120, characteristic parameter information of the target glasses is obtained, and imaging parameter analysis of the illumination surface light source and the imaging detection device is performed based on the characteristic parameter information to obtain optical imaging detection parameter information. Specifically, by measuring or querying official data of the target glasses, characteristic parameter information such as their size, shape, material, and surface treatment is obtained. Based on this characteristic parameter information, optical simulation software or experimental methods are used to perform a preliminary analysis of the type, brightness, intensity, angle, distribution, color temperature, and uniformity of the illumination surface light source, as well as the focus parameters, exposure parameters, camera position, and white balance parameters of the imaging detection device, to predict the imaging effect. In step S130, the illumination surface light source and the imaging detection device are verified and fine-tuned based on the optical imaging detection parameter information to determine the target imaging detection parameter information. Specifically, based on the imaging parameter analysis results, the parameters of the illumination surface light source and the imaging detection device are preliminarily set. Then, test photography is performed using a standard sample or a sample with known defects. By comparing the actual imaging effect with the expected effect, the parameters are fine-tuned. Testing and adjustment are repeated until satisfactory imaging results are achieved. The parameters determined at this point become the target imaging detection parameters. In step S140, the illumination surface light source and the imaging detection device are adjusted and configured according to the target imaging detection parameter information to obtain the optical calibration detection system. Specifically, according to the determined target imaging detection parameters, the final adjustment and configuration of the lighting surface light source and the imaging detection device are carried out, including adjusting the brightness, color temperature and other parameters of the light source, as well as the focal length, exposure time, white balance and other parameters of the camera. After the adjustment is completed, the entire optical detection system is calibrated and calibrated to obtain an optical calibration detection system. The optical calibration detection system is an optical detection system that can stably and accurately obtain the target glasses image after precise adjustment and configuration. This implementation method ensures that each component is configured and adjusted according to the optimal parameters by analyzing the imaging parameters, verifying fine-tuning and further adjusting the configuration of the lighting surface light source and imaging detection device in the constructed optical detection system, reducing imaging errors and interference factors, and achieving the technical effect of improving imaging quality, thereby improving the accuracy and reliability of subsequent defect detection.

[0036] In one possible implementation, the step of obtaining optical imaging detection parameter information in step S120 further includes step S121, whereby a glasses detection database is obtained using data mining techniques. The glasses detection database includes historical glasses characteristic parameters, optical imaging parameters, and corresponding glasses detection imaging data. Specifically, data mining tools and techniques (such as crawlers and API calls) are used to collect the required glasses detection data from data sources such as historical detection records, user feedback, and experimental data. The collected data is cleaned to remove duplicate, erroneous, or irrelevant data to ensure data accuracy and consistency. The cleaned data is stored in the glasses detection database, which includes historical glasses characteristic parameters (such as size, shape, material, and surface treatment), optical imaging parameters (such as focal length, exposure time, and white balance), and corresponding glasses detection imaging data (such as captured glasses images). Step S122 involves partitioning and integrating the glasses detection database based on the historical glasses characteristic parameters, and performing matching extraction on the integrated glasses detection database based on the characteristic parameter information to obtain a glasses detection calibration library. Specifically, the glasses detection database is divided into multiple subsets based on historical glasses characteristic parameters (such as similar size, shape, and material). Each subset is then further integrated to ensure that the data within each subset has similar characteristics. Next, the target glasses' characteristic parameter information is matched within the integrated glasses detection database to find the subset or data set with the highest similarity. From this matched data set, the optical imaging parameters and corresponding glasses detection imaging data are extracted to form a glasses detection calibration library.

[0037] In step S123, a set of optical imaging evaluation indicators is obtained. Based on the set of optical imaging evaluation indicators, a weighted evaluation is performed on the glasses detection imaging data to obtain optical imaging detection effect data. Specifically, a series of indicators for evaluating optical imaging quality, such as clarity, contrast, distortion rate, and noise level, are determined to form a set of optical imaging evaluation indicators. These evaluation indicators are applied to the glasses detection imaging data in the glasses detection calibration library, and weights are assigned based on the importance of each indicator. A weighted formula is used to comprehensively evaluate the glasses detection imaging data to obtain optical imaging detection effect data. The optical imaging detection effect data reflects the quality of imaging under different optical imaging parameters. In step S124, based on the optical imaging detection effect data, an optimal combination of optical imaging parameters is compared and optimized within the glasses detection calibration library to obtain the optical imaging detection parameter information. Specifically, within the glasses detection calibration library, different optical imaging parameter combinations are compared and analyzed based on the optical imaging detection effect data to find the optical imaging parameter combination with the optimal imaging effect (i.e., the highest evaluation score). This optimal optical imaging parameter combination is determined as the optical imaging detection parameter information for the smart glasses to be detected (the target glasses). This implementation method ensures that the obtained glasses detection calibration library is highly similar to the target glasses through data mining and classification integration. It uses a set of weighted evaluation indicators to conduct a comprehensive and objective evaluation of the optical imaging quality, avoiding the one-sidedness brought by a single indicator. Through comparison and optimization, the optimal calibration data is quickly located, achieving the technical effect of improving the accuracy and efficiency of obtaining optical imaging detection parameter information.

[0038] In one possible implementation, determining target imaging detection parameter information, step S130 further includes step S131: verifying the configuration of the illumination surface light source and the imaging detection device based on the optical imaging detection parameter information to obtain eyeglass imaging detection effect information. Specifically, the settings of the illumination surface light source are adjusted based on the optical imaging detection parameter information (such as illumination intensity, illumination angle, and light source type), and the settings of the imaging detection device are adjusted based on the optical imaging detection parameter information (such as focal length, exposure time, white balance, and resolution). Using the configured illumination surface light source and imaging detection device, imaging tests are performed on standard samples or samples with known defects to obtain eyeglass imaging detection effect information. Step S132: obtaining eyeglass imaging detection requirement standards, and extracting non-compliant parameters from the eyeglass imaging detection effect information using the eyeglass imaging detection requirement standards to obtain non-compliant imaging effect parameters. Specifically, specific requirements for eyeglass imaging detection, including thresholds for indicators such as clarity, contrast, distortion rate, and noise level, are obtained from industry standards, customer requirements, or internal specifications. The obtained glasses imaging detection effect information is compared and analyzed with the glasses imaging detection requirement standards to find out the parameters that do not meet the standards. The non-standard parameters found in the comparison analysis are extracted to form a list of non-standard imaging effect parameters. Step S133, based on the non-standard imaging effect parameters, the associated imaging parameters are optimized and analyzed to obtain the optimization step of the associated imaging parameters. Specifically, the correlation between the non-standard imaging effect parameters and the optical imaging detection parameters is analyzed to determine which detection parameters have a significant impact on the imaging effect. According to the correlation analysis results, a targeted optimization strategy is formulated to determine the associated imaging parameters that need to be adjusted and their adjustment direction. Based on the optimization strategy and empirical data, the optimization step of the associated imaging parameters is determined, that is, the specific value or range of the associated imaging parameters each time. Step S134, based on the associated imaging parameter optimization step, the optical imaging detection parameter information is optimized and adjusted and verified to determine the target imaging detection parameter information. Specifically, the associated imaging parameters are adjusted according to the associated imaging parameter optimization step size, the optical imaging detection parameter information is updated, and the imaging detection is re-performed using the updated optical imaging detection parameter information to obtain new glasses imaging detection effect information. The new glasses imaging detection effect information is compared and verified with the glasses imaging detection requirement standards, and the optimization effect is evaluated. If there are still parameters that do not meet the standards, steps S133 and S134 are repeated for further optimization. When all imaging effect parameters meet or exceed the glasses imaging detection requirement standards, the current optical imaging detection parameter information is determined to be the target imaging detection parameter information. This implementation method ensures that the target imaging detection parameter information can meet the glasses imaging detection requirement standards by optimizing, adjusting, verifying and optimizing the optical imaging detection parameter information, thereby achieving the technical effect of improving the accuracy and reliability of the detection results.

[0039] In step S200, a glasses detection location area is determined based on the optical calibration detection system. A manipulator is controlled to grip and grasp the target glasses to the glasses detection location area. The optical calibration detection system then performs multi-angle surface detection on the target glasses to obtain a set of multi-angle glasses surface images. Specifically, a suitable glasses detection location area is calculated and determined based on the imaging characteristics of the optical calibration detection system (such as focal length, field of view, etc.) and the size and shape of the target glasses. This area ensures that the target glasses are completely within the camera's field of view during the detection process and at a suitable focusing distance from the lens to achieve optimal imaging results. A high-precision manipulator or robot is used, and its gripping mechanism is programmed to accurately grasp the target glasses. The manipulator is moved above the pre-set glasses detection location area and precisely placed on the glasses detection location area using visual recognition or mechanical positioning technology.

[0040] Based on the inspection requirements, multiple inspection angles (such as front, side, and top) are set. The robot or optical calibration inspection system (e.g., via a rotating platform or camera arm) is controlled to adjust the position of the target glasses or the camera to achieve multi-angle surface inspection of the target glasses. At each inspection angle, the optical calibration inspection system captures an image of the target glasses' surface and stores it in real time in the image acquisition system. This step is repeated until all preset inspection angles have been captured, forming a multi-angle collection of glasses' surface images. For example, a multi-angle collection of glasses' surface images could include fingerprint images captured by 11 camera positions.

[0041] Step S300 performs denoising and distortion correction preprocessing on the multi-angle eyeglass surface image set to obtain a multi-angle standard eyeglass surface image set. These multi-angle standard eyeglass surface image sets are then panoramically stitched and three-dimensionally reconstructed to obtain a target eyeglass surface three-dimensional model. Specifically, the eyeglass surface images at each angle are denoised using methods such as median filtering, Gaussian filtering, and bilateral filtering to eliminate random noise (e.g., electronic noise, thermal noise, etc.) and fixed pattern noise (e.g., sensor defects, dark current, etc.) in the images. Denoising improves image quality and the accuracy of subsequent processing. Distortion correction is then performed on the denoised images based on the distortion coefficients in the optical calibration and detection system. The correction process includes geometric transformations (e.g., affine transformations, perspective transformations, etc.) and interpolation algorithms (e.g., bilinear interpolation, bicubic interpolation, etc.). Distortion correction eliminates image deformation caused by lens design or installation errors, restoring the image to its original shape.

[0042] The corrected images are standardized, including adjustments to image size, contrast, and brightness, to ensure consistency and comparability between images from different angles in subsequent processing, thereby improving the accuracy and efficiency of panoramic stitching and 3D reconstruction. Image stitching techniques, such as feature point matching algorithms and image content-based optimization algorithms (e.g., graph cut algorithms and energy minimization algorithms), are used to stitch together the multi-angle images of the standard eyeglass surface into a complete panoramic image. Based on the panoramic image and the camera's intrinsic and extrinsic parameters (obtained through calibration in step S100), a 3D reconstruction algorithm is used to restore the 3D shape of the target eyeglass surface. This 3D reconstruction algorithm includes steps such as stereo matching, depth estimation, and surface reconstruction. The resulting 3D model of the target eyeglass surface is used for subsequent defect detection and assessment.

[0043] In one possible implementation, obtaining the target eyeglass surface 3D model, step S300 further includes step S310, where a feature point detection algorithm is used to perform multi-dimensional feature point detection and feature point matching on the multi-angle standard eyeglass surface image set to obtain a set of matched feature points. Specifically, a feature point detection algorithm, such as SIFT (Scale Invariant Feature Transform), SURF (Speeded Up Robust Features), or ORB (Oriented FAST and Rotated BRIEF), is used to detect key points (feature points) on each standard eyeglass surface image, identifying unique and repeatable feature points within the image. These feature points are robust to image scale, rotation, and illumination variations. For each detected feature point, a descriptor is generated that contains texture information about the area surrounding the feature point. Feature descriptors are then used to match feature points, using a brute-force matcher or FLANN (Fast Nearest Neighbor Search Library) matcher, to identify corresponding identical feature points in standard eyeglass surface images from different angles. Step S320, a reference eyeglass surface image is selected based on the multi-angle standard eyeglass surface image set. Specifically, the quality of each image in the multi-angle standard eyeglass surface image set, such as clarity and illumination uniformity, is analyzed. The image with the closest viewing angle or containing the most feature information is selected as the baseline eyeglass surface image for reference in subsequent transformation and stitching. Step S330: Based on the baseline eyeglass surface image and the set of matched feature points, image transformation analysis is performed in the overlapping region to determine a set of projection transformation matrices for the remaining images. Specifically, a transformation model (e.g., affine transformation or perspective transformation) is used to describe the geometric relationship between the images. Using the matched feature point pairs, an optimization algorithm (e.g., RANSAC or least squares method) is used to calculate the projection transformation matrix of each image in the multi-angle standard eyeglass surface image set relative to the baseline eyeglass surface image. Step S340: Based on the set of remaining image projection transformation matrices, the multi-angle standard eyeglass surface image set is transformed, aligned, and stitched to obtain a panoramic eyeglass surface image. Specifically, based on the calculated projection transformation matrix for the remaining images, each image in the multi-angle standard eyeglass surface image set is transformed into the same coordinate system as the baseline eyeglass surface image. The transformed images are then fused and stitched in the overlapping region to generate a complete panoramic eyeglass surface image. Step S350 uses a surface reconstruction algorithm to perform depth estimation and point cloud 3D reconstruction on the panoramic image of the eyeglass surface to obtain a 3D model of the target eyeglass surface. Specifically, the depth information of each pixel in the panoramic image of the eyeglass surface is estimated using techniques such as stereo vision, structured light, or optical flow. Based on the estimated depth information, the 2D image pixels are converted into points in 3D space to form point cloud data. The point cloud data is then used to construct a 3D model of the target eyeglass surface through techniques such as triangulation and mesh optimization.This implementation method improves the accuracy and robustness of 3D reconstruction through feature point detection and matching, image alignment and stitching, depth information estimation and surface reconstruction, achieving the technical effect of enabling the final 3D model to more realistically reflect the geometric shape and surface details of the target glasses.

[0044] Step S400 involves performing multi-dimensional feature extraction and defect area identification based on the target 3D surface model of the eyeglasses to obtain a set of defect areas on the eyeglasses surface. Defect features are then sequentially evaluated on the set of defect areas to obtain an eyeglass defect feature detection result. Specifically, the target 3D surface model of the eyeglasses is meticulously analyzed, employing image processing techniques (e.g., edge detection and texture analysis), machine learning algorithms (e.g., feature learning and feature selection), or specialized detection algorithms (e.g., surface defect detection algorithms) to extract features in multiple dimensions, including shape features (e.g., curvature and edge profile), texture features (e.g., surface roughness and gloss), and color features (e.g., color difference and spots). Using these extracted multi-dimensional features and combining them with pre-set defect recognition rules or models, the target 3D surface model of the eyeglasses is scanned and identified. During the identification process, actual features are compared with standard features or features in a defect feature library to determine whether any deviations or anomalies exist, thereby identifying potential defect areas. Defect areas can manifest in various forms, such as scratches, cracks, pits, color difference, and stains. All identified defect areas are organized into a set, with each defect area clearly identified and its location within the set.

[0045] Using a defect feature assessment algorithm, a detailed feature evaluation is performed on each defect in the set of defect areas on the eyewear surface, including defect type, size, shape, depth (if possible), and location. Based on the evaluation results, defects can be classified, graded, or ranked. The results of the defect feature assessment are compiled into a detailed inspection report containing information such as the image, location, type, feature description, and assessment results of the defect area, for subsequent quality control, product traceability, and process improvement. The embodiments of the present application can be used to detect fingerprints and / or dirt, scratches, etc. on the surface of VR / AR glasses. By building an optical detection system, performing imaging parameter analysis and system adjustment configuration, a stable and high-precision optical calibration detection system is obtained. A manipulator is used to accurately control the detection position of the smart glasses, and multi-angle surface detection is performed on the smart glasses to comprehensively cover all areas of the smart glasses surface. The acquired multi-angle glasses surface image set is pre-processed by denoising, correcting distortion, and other pre-processing operations to obtain high-quality standard images. Through panoramic stitching and three-dimensional reconstruction, a three-dimensional model of the smart glasses surface is constructed. Based on the three-dimensional model, multi-dimensional feature extraction and defect area identification are performed to accurately find the defect area on the smart glasses surface, and detailed defect feature evaluation is performed. These technical means achieve precise positioning and identification of surface defects of the smart glasses, achieving the technical effect of improving the comprehensiveness and accuracy of surface defect detection.

[0046] In one possible implementation, obtaining a set of eyeglass surface defect regions in step S400 further includes step S410, performing multidimensional feature extraction on the target eyeglass surface three-dimensional model to obtain a set of multidimensional eyeglass surface features. The multidimensional eyeglass surface feature set includes geometric and textural features. Specifically, geometric analysis algorithms (such as curvature analysis and shape descriptors) are used to extract geometric features of the eyeglass surface, such as curvature changes, edge contours, and concave and convex areas, from the target eyeglass surface three-dimensional model. Texture analysis techniques (such as gray-level co-occurrence matrix and local binary pattern) are used to extract textural features of the eyeglass surface, such as color, texture pattern, and texture direction, from the texture map of the target eyeglass surface three-dimensional model. The extracted geometric and textural features are integrated into a set of multidimensional eyeglass surface features to provide data support for subsequent classification and recognition. Step S420, based on the design quality requirements of the target eyeglasses, a database of eyeglass surface defects is constructed, and classification training is performed on the database using a support vector machine to obtain an eyeglass surface defect classifier. Specifically, based on the design quality requirements of the target glasses, various types of surface defect samples, including scratches, dents, and chromatic aberration, are collected. These samples are manually labeled to clearly identify the defect type of each sample. These labeled samples are then stored in a database, forming a glasses surface defect database. A support vector machine (SVM) classification algorithm is used, and the SVM is trained using the glasses surface defect database to obtain a glasses surface defect classifier. In step S430, the glasses surface defect classifier performs defect recognition and classification on the multidimensional surface feature set of the glasses, obtaining a set of glasses surface defect feature types. Specifically, the extracted multidimensional surface feature set is input into the trained glasses surface defect classifier. The glasses surface defect classifier performs defect recognition and classification on the glasses surface based on the input multidimensional surface feature set, outputting the defect type for each region, and integrating the classification results into a set of glasses surface defect feature types. In step S440, anchor frames are labeled on the target glasses surface 3D model based on the set of glasses surface defect feature types, obtaining a set of glasses surface defect regions. Specifically, based on the information in the set of eyeglass surface defect feature types, the system identifies specific defect areas on the target eyeglass surface 3D model. Anchor boxes (i.e., bounding boxes) are drawn around the identified defect areas to clearly identify the specific location and scope of the defects. All marked defect areas are then combined into a set, which is output as the final set of eyeglass surface defect areas. This implementation method, through multidimensional feature extraction and the use of an SVM classifier, accurately identifies various defect types on the eyeglass surface and pinpoints the specific defect areas, achieving the technical effect of improving the efficiency and accuracy of defect area identification.

[0047] In one possible implementation, obtaining the eyeglass defect feature detection results in step S400 further includes step S450, meshing the target eyeglass surface 3D model according to a preset size to obtain a 3D eyeglass surface mesh model. Specifically, based on detection accuracy requirements and computational resource constraints, a preset mesh size (e.g., mesh side length, area, etc.) is determined. A meshing algorithm (e.g., Delaunay triangulation, Voronoi diagram, etc.) is used to mesh the target eyeglass surface 3D model to generate a 3D eyeglass surface mesh model. Each mesh cell (e.g., triangle, quadrilateral, etc.) in the 3D eyeglass surface mesh model meets the preset size requirements. The generated 3D eyeglass surface mesh model is optimized to eliminate possible overlaps, gaps, or irregular shapes to ensure mesh model quality and accuracy. In step S460, the set of eyeglass surface defect regions is mapped and statistically analyzed with the 3D eyeglass surface mesh model to obtain a set of eyeglass defect region areas. Specifically, based on the information in the set of defect areas on the surface of the glasses, each defect area is located on the three-dimensional grid model of the surface of the glasses, and each defect area is mapped to the corresponding grid unit in the three-dimensional grid model of the surface of the glasses. The number of grid units covered by each defect area is counted, and the area of ​​each defect area is calculated based on the size of the grid unit and the number of grid cells covered by the defect area, and the areas are summarized into a set of areas of defect areas of the glasses.

[0048] Step S470: Perform a quality impact assessment on the set of surface defect feature types and determine a set of surface defect type influencing factors. Specifically, analyze each defect type in the set of surface defect feature types to determine its specific impact on the quality of the glasses (e.g., impact on aesthetics, reduced durability, etc.). Based on the degree of impact of each defect type on the quality of the glasses, develop quality impact assessment criteria (e.g., a scoring system, weighting distribution, etc.). Based on the assessment criteria, determine one or more quality impact factors for each defect type and quantify the degree of impact (e.g., the depth of the scratch, the area of ​​the dent, etc.). Step S480: Perform a weighted defect feature assessment on the set of defect area areas of the glasses based on the set of surface defect type influencing factors to obtain the defect feature detection results of the glasses. Specifically, perform a weighted calculation on each defect area in the set of defect area areas based on the set of surface defect type influencing factors. The weighted calculation results are summarized to obtain a weighted assessment value for each defect area. This weighted assessment value is then used to evaluate the defect feature detection results of the entire target glasses. The defect feature detection results are then compiled into a report containing information such as the location, area, type, and impact of the defect area for reference in subsequent processing and quality control. This implementation method accurately calculates the area of ​​the defect region through grid division and mapping statistics, performs quality impact assessment and area-weighted assessment on the defect feature type, comprehensively reflects the degree of influence of the defect on the quality of the glasses, and achieves the technical effect of accurately evaluating the impact of the surface defect characteristics of the glasses on the quality.

[0049] In the above, refer to Figure 1 A high-precision defect detection method for smart glasses according to an embodiment of the present invention is described in detail. Figure 2 A high-precision defect detection system for smart glasses according to an embodiment of the present invention is described.

[0050] A high-precision defect detection system for smart glasses, according to an embodiment of the present invention, is designed to address the technical issues of insufficient comprehensiveness and accuracy in existing smart glasses defect detection, thereby improving the comprehensiveness and accuracy of surface defect detection. The high-precision defect detection system for smart glasses includes an optical calibration and detection system construction module 10, a glasses multi-angle surface detection module 20, a panoramic stitching module 30, and a defect recognition module 40.

[0051] The optical calibration and detection system construction module 10 is used to build an optical detection system, which includes a stray light elimination lens, a large-target industrial camera, and an illumination surface light source. The imaging parameters of the stray light elimination lens, the large-target industrial camera, and the illumination surface light source are analyzed and the system adjustment configuration is performed to obtain an optical calibration and detection system; the glasses multi-angle surface detection module 20 is used to determine the glasses detection position area according to the optical calibration and detection system, control the manipulator to clamp and grab the target glasses to the glasses detection position area, and perform multi-angle surface detection on the target glasses through the optical calibration and detection system to obtain a multi-angle glasses surface image set; the panoramic stitching module 30 is used to perform denoising and distortion correction preprocessing on the multi-angle glasses surface image set to obtain a multi-angle standard glasses surface image set, perform panoramic stitching and three-dimensional reconstruction on the multi-angle standard glasses surface image set to obtain a target glasses surface three-dimensional model; the defect recognition module 40 is used to perform multi-dimensional feature extraction and defect area recognition based on the target glasses surface three-dimensional model to obtain a glasses surface defect area set, and perform defect feature evaluation on the glasses surface defect area set in turn to obtain a glasses defect feature detection result.

[0052] The specific configuration of the optical calibration detection system building module 10 will be described in detail below. As described above, an optical calibration detection system is obtained, and the optical calibration detection system building module 10 may further include: an imaging detection device acquisition unit for mounting the stray light elimination lens on the interface of the large-target industrial camera to obtain an imaging detection device; an imaging parameter analysis unit for acquiring characteristic parameter information of the target glasses, performing imaging parameter analysis on the illumination surface light source and the imaging detection device based on the characteristic parameter information to obtain optical imaging detection parameter information; a verification and fine-tuning unit for verifying and fine-tuning the illumination surface light source and the imaging detection device based on the optical imaging detection parameter information to determine the target imaging detection parameter information; and an adjustment and configuration unit for adjusting and configuring the illumination surface light source and the imaging detection device according to the target imaging detection parameter information to obtain the optical calibration detection system.

[0053] Among them, the imaging parameter analysis unit for obtaining optical imaging detection parameter information may further include: a glasses detection database acquisition subunit for acquiring a glasses detection database through data mining technology, wherein the glasses detection database includes historical glasses characteristic parameters, optical imaging parameters and corresponding glasses detection imaging data; a glasses detection calibration library acquisition subunit for dividing and integrating the glasses detection database based on the historical glasses characteristic parameters, and matching and extracting the integrated glasses detection database according to the characteristic parameter information to obtain a glasses detection calibration library; an optical imaging detection effect data acquisition subunit for acquiring an optical imaging evaluation index set, and performing weighted evaluation on the glasses detection imaging data based on the optical imaging evaluation index set to obtain optical imaging detection effect data; an optical imaging detection parameter information acquisition subunit for performing comparison and optimization in the glasses detection calibration library based on the optical imaging detection effect data to obtain the optical imaging detection parameter information.

[0054] Among them, the target imaging detection parameter information is determined, and the verification and fine-tuning unit can further include: a glasses imaging detection effect information acquisition subunit is used to verify the configuration of the lighting surface light source and the imaging detection device based on the optical imaging detection parameter information, and obtain the glasses imaging detection effect information; a substandard parameter extraction subunit is used to obtain the glasses imaging detection requirement standard, and extract substandard parameters of the glasses imaging detection effect information according to the glasses imaging detection requirement standard to obtain substandard imaging effect parameters; an associated imaging parameter optimization analysis subunit is used to perform associated imaging parameter optimization analysis based on the substandard imaging effect parameters to obtain the associated imaging parameter optimization step; a verification and optimization subunit is used to optimize and adjust the optical imaging detection parameter information and verify and optimize it based on the associated imaging parameter optimization step to determine the target imaging detection parameter information.

[0055] The specific configuration of the panoramic stitching module 30 will be described in detail below. As described above, to obtain a target eyeglass surface three-dimensional model, the panoramic stitching module 30 may further include: a feature point matching unit for performing multi-dimensional feature point detection and feature point matching on the multi-angle standard eyeglass surface image set using a feature point detection algorithm to obtain a matching feature point set; a reference eyeglass surface image selection unit for selecting a reference eyeglass surface image based on the multi-angle standard eyeglass surface image set; an overlapping area image transformation analysis unit for performing overlapping area image transformation analysis based on the reference eyeglass surface image and the matching feature point set to determine a set of remaining image projection transformation matrices; a transformation alignment and stitching unit for performing transformation alignment and stitching on the multi-angle standard eyeglass surface image set based on the remaining image projection transformation matrix set to obtain a panoramic image of the eyeglass surface; and a point cloud three-dimensional reconstruction unit for performing depth estimation and point cloud three-dimensional reconstruction on the panoramic image of the eyeglass surface using a surface reconstruction algorithm to obtain the target eyeglass surface three-dimensional model.

[0056] The specific configuration of the defect recognition module 40 will be described in detail below. As described above, to obtain a set of defect regions on the surface of the glasses, the defect recognition module 40 may further include: a multidimensional feature extraction unit for performing multidimensional feature extraction on the target glasses surface three-dimensional model to obtain a glasses surface multidimensional feature set, wherein the glasses surface multidimensional feature set includes geometric features and texture features; a glasses surface defect classifier training unit for constructing a glasses surface defect database based on the design quality requirements of the target glasses, and using a support vector machine to perform classification training on the glasses surface defect database to obtain a glasses surface defect classifier; a defect recognition and classification unit for performing defect recognition and classification on the glasses surface multidimensional feature set based on the glasses surface defect classifier to obtain a glasses surface defect feature type set; and an anchor frame labeling unit for performing anchor frame labeling on the target glasses surface three-dimensional model based on the glasses surface defect feature type set to obtain the glasses surface defect region set.

[0057] Among them, the defect recognition module 40 for obtaining the glasses defect feature detection result may further include: a grid division unit for gridding the target glasses surface three-dimensional model according to a preset size to obtain the glasses surface three-dimensional grid model; a mapping statistics unit for mapping statistics between the glasses surface defect area set and the glasses surface three-dimensional grid model to obtain the glasses defect area area set; a quality impact assessment unit for performing a quality impact assessment on the glasses surface defect feature type set to determine a surface defect type influencing factor set; a defect feature weighted assessment unit for performing a defect feature weighted assessment on the glasses defect area area set based on the surface defect type influencing factor set to obtain the glasses defect feature detection result.

[0058] A high-precision defect detection system for smart glasses provided by an embodiment of the present invention can execute a high-precision defect detection method for smart glasses provided by any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0059] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0060] The above specific embodiments do not constitute a limitation to the scope of protection of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of this application should be included in the scope of protection of this application. In some cases, the actions or steps recorded in this application can be performed in an order different from that in the embodiments and can still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A high-precision defect detection method for smart glasses, characterized in that: The method comprises: Building an optical detection system, wherein the optical detection system includes a stray light elimination lens, a large target area industrial camera, and an illumination surface light source, performing imaging parameter analysis and system adjustment configuration on the stray light elimination lens, the large target area industrial camera, and the illumination surface light source, and obtaining an optical calibration detection system; According to the optical calibration detection system, determine the glasses detection position area, control the manipulator to clamp and grab the target glasses to the glasses detection position area, and perform multi-angle surface detection on the target glasses through the optical calibration detection system to obtain a multi-angle glasses surface image set; Performing denoising and distortion correction preprocessing on the multi-angle glasses surface image set to obtain a multi-angle standard glasses surface image set, and performing panoramic stitching and three-dimensional reconstruction on the multi-angle standard glasses surface image set to obtain a target glasses surface three-dimensional model; Perform multi-dimensional feature extraction and defect area recognition based on the target glasses surface three-dimensional model to obtain a glasses surface defect area set, and sequentially perform defect feature evaluation on the glasses surface defect area set to obtain glasses defect feature detection results; The method of obtaining an optical calibration detection system comprises: The stray light elimination lens is mounted on the interface of the large-target-area industrial camera to obtain an imaging detection device; Acquiring characteristic parameter information of the target glasses, and performing imaging parameter analysis on the illumination surface light source and the imaging detection device based on the characteristic parameter information to obtain optical imaging detection parameter information; Based on the optical imaging detection parameter information, verify and fine-tune the illumination surface light source and the imaging detection device to determine the target imaging detection parameter information; Adjust and configure the illumination surface light source and the imaging detection device according to the target imaging detection parameter information to obtain the optical calibration detection system; The obtaining of optical imaging detection parameter information includes: Acquire a glasses detection database by using data mining technology, wherein the glasses detection database includes historical glasses characteristic parameters, optical imaging parameters, and corresponding glasses detection imaging data; The glasses detection database is divided and integrated based on the historical glasses characteristic parameters, and matching and extracting the integrated glasses detection database according to the characteristic parameter information to obtain a glasses detection calibration library; Acquire an optical imaging evaluation index set, and perform weighted evaluation on the glasses detection imaging data based on the optical imaging evaluation index set to obtain optical imaging detection effect data; Based on the optical imaging detection effect data, comparison and optimization are performed in the glasses detection calibration library to obtain the optical imaging detection parameter information; The determining of target imaging detection parameter information includes: Based on the optical imaging detection parameter information, configuration verification is performed on the illumination surface light source and the imaging detection device to obtain glasses imaging detection effect information; Acquire glasses imaging detection requirement standards, extract substandard parameters of the glasses imaging detection effect information according to the glasses imaging detection requirement standards, and obtain substandard imaging effect parameters; Performing an optimization analysis of associated imaging parameters based on the imaging effect parameters that do not meet the standards, and obtaining an optimization step length of the associated imaging parameters; The optical imaging detection parameter information is optimized, adjusted and verified based on the associated imaging parameter optimization step size to determine the target imaging detection parameter information.

2. A high-precision defect detection method for smart glasses as claimed in claim 1, characterized in that: The step of obtaining the target glasses surface three-dimensional model comprises: Using a feature point detection algorithm to perform multi-dimensional feature point detection and feature point matching on the multi-angle standard glasses surface image set to obtain a matching feature point set; Selecting a reference glasses surface image according to the multi-angle standard glasses surface image set; Performing image transformation analysis of overlapping areas based on the reference glasses surface image and the matching feature point set to determine a set of remaining image projection transformation matrices; Based on the remaining image projection transformation matrix set, the multi-angle standard glasses surface image set is transformed, aligned and spliced ​​to obtain a panoramic image of the glasses surface; A surface reconstruction algorithm is used to perform depth estimation and point cloud three-dimensional reconstruction on the panoramic image of the glasses surface to obtain a three-dimensional model of the target glasses surface.

3. The high-precision defect detection method for smart glasses according to claim 1, characterized in that: The step of obtaining a set of defective regions on the surface of glasses comprises: Performing multi-dimensional feature extraction on the target glasses surface three-dimensional model to obtain a glasses surface multi-dimensional feature set, wherein the glasses surface multi-dimensional feature set includes geometric features and texture features; According to the design quality requirements of the target glasses, a glasses surface defect database is constructed, and a support vector machine is used to perform classification training on the glasses surface defect database to obtain a glasses surface defect classifier; Based on the glasses surface defect classifier, defect recognition and classification are performed on the glasses surface multidimensional feature set to obtain a glasses surface defect feature type set; The target glasses surface three-dimensional model is anchored based on the glasses surface defect feature type set to obtain the glasses surface defect area set.

4. A high-precision defect detection method for smart glasses as claimed in claim 3, characterized in that: The obtaining of the glasses defect feature detection result comprises: Meshing the target glasses surface three-dimensional model according to a preset size to obtain a glasses surface three-dimensional mesh model; Mapping and counting the defect area set of the glasses surface and the three-dimensional mesh model of the glasses surface to obtain an area set of the glasses defect areas; Performing a quality impact assessment on the set of surface defect feature types of the glasses to determine a set of surface defect type influencing factors; Based on the surface defect type influencing factor set, a weighted defect feature evaluation is performed on the eyeglass defect region area set to obtain the eyeglass defect feature detection result.

5. A high-precision defect detection system for smart glasses, characterized in that: The system is used to implement the high-precision defect detection method for smart glasses according to any one of claims 1 to 4, and the system comprises: An optical calibration detection system building module, which is used to build an optical detection system, wherein the optical detection system includes a stray light elimination lens, a large target area industrial camera, and an illumination surface light source, and performs imaging parameter analysis and system adjustment configuration on the stray light elimination lens, the large target area industrial camera, and the illumination surface light source to obtain an optical calibration detection system; A glasses multi-angle surface detection module, the glasses multi-angle surface detection module is used to determine the glasses detection position area according to the optical calibration detection system, control the manipulator to clamp and grab the target glasses to the glasses detection position area, and perform multi-angle surface detection on the target glasses through the optical calibration detection system to obtain a multi-angle glasses surface image set; A panoramic stitching module, which is used to perform denoising and distortion correction preprocessing on the multi-angle glasses surface image set to obtain a multi-angle standard glasses surface image set, and perform panoramic stitching and three-dimensional reconstruction on the multi-angle standard glasses surface image set to obtain a target glasses surface three-dimensional model; A defect recognition module is used to perform multi-dimensional feature extraction and defect area recognition based on the three-dimensional model of the target glasses surface, obtain a set of defect areas on the glasses surface, and sequentially perform defect feature evaluation on the set of defect areas on the glasses surface to obtain glasses defect feature detection results.

Images