Mobile phone glass cover plate quality detection method based on image sensor

Abstract

The invention relates to the technical field of quality detection of image recognition, and particularly discloses a mobile phone glass cover plate quality detection method based on an image sensor, which comprises the following steps of: performing omnibearing scanning on a mobile phone glass cover plate fixed on a detection platform by using an industrial camera, and determining a crack concentration area according to a crack distribution condition based on acquired image data; the method comprises the following steps: calculating abnormal values of crack distribution characteristics through clustering analysis, identifying areas with abnormal crack distribution, extracting curvature fluctuation characteristics and crack growth rate characteristics of cracks for each crack concentration area, approximately calculating curvature through a finite difference method and obtaining curvature fluctuation characteristic values through Haar wavelet transform, and calculating the crack distribution characteristics according to the curvature fluctuation characteristic values. A local outlier factor algorithm is used for calculating a crack propagation abnormal characteristic value, a comprehensive characteristic vector is constructed to serve as input of a support vector machine model, and the quality score of each mobile phone glass cover plate is predicted, so that the quality grade of each mobile phone glass cover plate is judged, and low-quality products are marked.

Description

Technical Field

[0001] The present invention relates to the technical field of quality detection for image recognition, and particularly to a quality detection method for mobile phone glass covers based on image sensors. Background Art

[0002] With the rapid development of the smartphone market, consumers' quality requirements for mobile phone glass covers are increasing day by day. They not only pay attention to their aesthetic appearance but also attach more importance to durability and anti-breakage ability. Traditional quality detection methods mainly rely on manual visual inspection and simple automated optical detection equipment. Although these methods can identify obvious defects to a certain extent, they have limited ability to recognize tiny cracks, fine scratches, and defects with complex shapes. In addition, traditional methods usually lack the ability to predict the trend of crack propagation and cannot provide a comprehensive quality assessment. Therefore, developing an efficient, accurate, and real-time monitoring automatic detection system has become a key requirement for improving the quality of mobile phone glass covers.

[0003] The existing technologies have the following deficiencies:

[0004] There are many deficiencies in the existing technologies for the quality detection of mobile phone glass covers. First, traditional manual visual inspection is inefficient and easily affected by subjective factors, making it difficult to ensure consistency and accuracy. Second, most of the existing automated optical detection equipment uses fixed threshold segmentation algorithms, which have poor adaptability to images under different lighting conditions, resulting in frequent missed detections or false detections. In addition, existing technologies usually can only identify surface visible defects, lacking in-depth analysis of crack propagation rate and curvature fluctuation characteristics, and unable to effectively predict potential quality hazards. More importantly, the existing systems lack an integrated feedback mechanism to guide the continuous optimization of production processes and are difficult to achieve closed-loop management from detection to improvement. These problems limit production efficiency. Summary of the Invention

[0005] The purpose of the present invention is to provide a quality detection method for mobile phone glass covers based on image sensors to solve the problems in the above background.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A quality detection method for mobile phone glass covers based on image sensors includes the following steps:

[0008] S1: Use an image sensor to perform an omnidirectional scan on the mobile phone glass cover to obtain original image data, and preprocess the original image data;

[0009] The image sensor is an industrial camera;

[0010] S2: Based on the preprocessed original image data, identify the cracks on the mobile phone glass cover plate in the original image, and identify the crack concentration areas according to the crack distribution of the mobile phone glass cover plate;

[0011] S3: Based on the crack concentration areas, for each crack concentration area, extract the curvature fluctuation characteristics and crack propagation rate characteristics of the cracks, analyze the curvature fluctuation and crack propagation rate characteristics of the cracks, and according to the analysis results, construct a comprehensive feature vector as the input of the machine learning model, and judge the quality of each mobile phone glass cover plate according to the model output;

[0012] S4: According to the judgment results, divide the quality of each mobile phone glass cover plate into low quality and high quality, and mark the low-quality mobile phone glass cover plates.

[0013] As a further solution of the present invention: The identification of the cracks on the mobile phone glass cover plate in the original image specifically includes:

[0014] Obtain the original grayscale image of the mobile phone glass cover plate, calculate an initial global threshold, then divide the image into multiple non-overlapping small regions, calculate the local threshold for each small region, and use the calculated local threshold to perform a preliminary binaryzation process on each small region to generate a series of partially binaryzation images; and mark the position and bounding box of each crack in the binaryzation image.

[0015] As a further solution of the present invention: The identification of the crack concentration areas specifically includes:

[0016] Divide the image into multiple non-overlapping small regions, calculate the crack distribution feature outliers in the corresponding regions according to the crack distribution in each region, and judge whether the crack distribution feature outliers in each region are greater than or equal to a preset threshold. If so, record it as a crack concentration area.

[0017] As a further solution of the present invention: The process of obtaining the crack distribution feature outliers is as follows:

[0018] Divide the mobile phone glass cover plate into several non-overlapping regions of the same size, and extract the crack distribution feature values for each region;

[0019] Perform K-means clustering analysis on the crack distribution feature values, and initialize K clustering centers. For each region, find the nearest clustering center according to the Euclidean distance formula and assign it to the corresponding cluster; and calculate and update the position of the new center point of each cluster, repeat the assignment and update process until the clustering center reaches the maximum number of iterations; after obtaining the final clustering result, for each cluster, calculate its average crack distribution feature value, and for each region, calculate the sum of the squares of the differences between the crack distribution feature value and the average feature value of the cluster to which it belongs, and take the square root to obtain the crack distribution feature outliers.

[0020] As a further solution of the present invention: based on the crack concentration area, for each crack concentration area, extract the curvature fluctuation and crack propagation rate characteristics of the crack, and analyze the curvature fluctuation and crack propagation rate characteristics of the crack, specifically including:

[0021] For each crack concentration area, extract the curvature fluctuation characteristics and crack propagation rate characteristics of the crack. According to the degree of curvature fluctuation of the crack, calculate the curvature fluctuation characteristic value. According to the crack propagation rate, calculate the crack propagation abnormal characteristic value. Construct the curvature fluctuation characteristic value and the crack propagation abnormal characteristic value into a comprehensive feature vector. Take the comprehensive feature vectors in all areas as the input of the machine learning model. The output of the model is the quality score of each mobile phone glass cover plate. According to the quality score, judge the quality of each mobile phone glass cover plate.

[0022] As a further solution of the present invention: the obtaining process of the curvature fluctuation characteristic value is as follows:

[0023] For each point on all crack centerlines, approximately calculate the first derivative and the second derivative of the arc length by the finite difference method, and calculate the curvature of each crack through the curvature calculation formula;

[0024] Apply the Haar wavelet transform to the calculated curvature sequence, decompose the curvature sequence into different frequency components, and obtain a series of wavelet coefficients;

[0025] Based on the wavelet coefficients after wavelet transform, calculate the sum of the standard deviations of all wavelet coefficients to obtain the curvature fluctuation characteristic value.

[0026] As a further solution of the present invention: the obtaining process of the crack propagation abnormal characteristic value is as follows:

[0027] Obtain and standardize the crack propagation rate data set, determine the B-distance and B-neighborhood of each standardized data point, and calculate the reachable density of each data point relative to other points; calculate the local outlier factor of each data point according to the reachable density, and calculate the mean value of all local outlier factors to obtain the crack propagation abnormal characteristic value.

[0028] As a further solution of the present invention: the judgment of the quality of each mobile phone glass cover plate specifically includes:

[0029] Judge whether the quality score of each mobile phone glass cover plate is greater than or equal to a preset threshold. If so, mark it as a high-quality mobile phone glass cover plate. If not, mark it as a low-quality mobile phone glass cover plate.

[0030] As a further solution of the present invention: the construction process of the machine learning model is as follows:

[0031] Construct a comprehensive feature vector from the curvature fluctuation eigenvalue and crack propagation anomaly eigenvalue of each crack concentration region;

[0032] Use the comprehensive feature vectors in all regions as the input of a machine learning model, where the machine learning model is a support vector machine model; use the support vector machine model to output the quality score of each mobile phone glass cover plate, and judge the quality of each mobile phone glass cover plate according to the quality score;

[0033] Collect and label a series of mobile phone glass cover plate samples with known quality scores to form a training set, and use the training set to train the support vector machine model; during the training process, use cross-validation technology to adjust the model parameters, and after multiple iterations, obtain the final support vector machine model; input the comprehensive feature vector of each newly input mobile phone glass cover plate into the trained support vector machine model to output the quality score of the mobile phone glass cover plate.

[0034] Advantages of the present invention:

[0035] (1) By adopting a high-resolution industrial camera and advanced image processing technologies, such as local threshold segmentation, curvature fluctuation feature extraction, and crack propagation rate analysis, etc., the present invention can achieve accurate identification and quantitative evaluation of cracks and other subtle defects on mobile phone glass cover plates. Specifically, the local threshold segmentation technology enables accurate capture of crack edge information under different lighting conditions, while the curvature fluctuation feature extraction accurately calculates the morphological features of each crack through the finite difference method and Haar wavelet transform. Further combined with the local outlier factor algorithm based on reachable density to analyze the crack propagation rate, so as to comprehensively describe the development trend and potential risks of cracks. In particular, these feature vectors are input into an optimized support vector machine model for quality score prediction, which not only significantly improves the accuracy of single defect judgment, but also effectively identifies abnormal regions with complex patterns, greatly improving the overall detection accuracy and efficiency. In addition, this automated detection scheme greatly reduces the need for manual inspection and significantly shortens the detection cycle, especially suitable for real-time quality monitoring on large-scale production lines, ensuring that the quality of each mobile phone glass cover plate meets high standards, and thus providing strong technical support for improving production efficiency and reducing the defective rate. Through this systematic method, the present invention not only realizes efficient and reliable defect detection, but also provides a data-driven basis for continuous optimization of the production process, helping enterprises maintain a leading position in the fierce market competition.

[0036] (2) By precisely marking low-quality mobile phone glass covers and arranging for further manual or advanced automated equipment review, the present invention can detail the specific defect information of each defective product, including multi-dimensional data such as the location, morphological characteristics, and propagation rate of cracks. These detailed defect analysis results not only provide valuable data support for subsequent process improvement but also enable the production team to quickly locate weak links in the production process, such as improper parameter settings or equipment aging in specific processes. Based on these data, production process engineers can adjust key process parameters and optimize the production line configuration accordingly, effectively reducing the occurrence frequency of similar defects and significantly improving the overall product quality. In addition, the real-time feedback mechanism and continuous monitoring function provided by the present invention enable production enterprises to continuously optimize their production processes during daily operations, identify potential quality risks through data analysis, and take preventive measures. In the long run, this systematic quality management strategy helps to significantly reduce the defective rate, improve resource utilization, reduce waste, enhance the market competitiveness of products, and ensure that enterprises occupy a favorable position in the fierce market competition. Through this comprehensive quality inspection and management solution, the present invention not only achieves a fine assessment of individual products but also provides enterprises with a path for sustainable improvement, driving the entire production system towards higher quality standards. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The present invention will be further described below in conjunction with the accompanying drawings.

[0038] Figure 1 It is a specific step flow block diagram of the quality inspection method for mobile phone glass covers based on an image sensor according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0040] Please refer to Figure 1 As shown, the present invention is a quality inspection method for mobile phone glass covers based on an image sensor, including the following steps:

[0041] S1: Use an image sensor to perform an all-round scan of the mobile phone glass cover, obtain the original image data, and preprocess the original image data;

[0042] The image sensor is an industrial camera;

[0043] S2: Based on the preprocessed original image data, identify the cracks on the mobile phone glass cover plate in the original image, and identify the crack concentration areas according to the crack distribution of the mobile phone glass cover plate;

[0044] S3: Based on the crack concentration areas, for each crack concentration area, extract the curvature fluctuation characteristics and crack propagation rate characteristics of the cracks, analyze the curvature fluctuation and crack propagation rate characteristics of the cracks, and according to the analysis results, construct a comprehensive feature vector as the input of the machine learning model, and judge the quality of each mobile phone glass cover plate according to the model output;

[0045] S4: According to the judgment results, divide the quality of each mobile phone glass cover plate into low quality and high quality, and mark the low-quality mobile phone glass cover plates.

[0046] In S1, use an image sensor to perform an omnidirectional scan of the mobile phone glass cover plate to obtain the original image data, and preprocess the original image data. Specifically, it includes:

[0047] Use a high-resolution industrial camera to perform an omnidirectional scan of the mobile phone glass cover plate. First, fix the glass cover plate on the detection platform, and adjust its position and angle through an automated mechanical device to ensure that the sensor can cover the entire surface area. Configure an appropriate light source system (such as a ring-shaped LED) to reduce reflection interference and enhance surface texture features. The image sensor acquires data row by row or region by region at a preset resolution and frame rate to generate the original image data containing the surface information of the glass cover plate.

[0048] After obtaining the original image data, perform preliminary preprocessing on it to improve the accuracy of subsequent analysis. The preprocessing steps include image denoising (such as using Gaussian filtering), adjusting brightness and contrast to eliminate the influence of uneven illumination, and image registration to correct the distortion caused by changes in the scanning angle or position. Finally, obtain clear and standardized image data, laying a foundation for subsequent crack detection and feature extraction.

[0049] In S2, based on the preprocessed original image data, identify the cracks on the mobile phone glass cover plate in the original image, and identify the crack concentration areas according to the crack distribution of the mobile phone glass cover plate. Specifically, it includes:

[0050] Obtain the original grayscale image of the mobile phone glass cover plate, calculate an initial global threshold, and then divide the image into multiple non-overlapping small regions. Calculate the local threshold for each small region. The calculation expression is: T i = m(x,y) + A*(x,y)*(1 + α*ΔC);

[0051] Wherein, A is a preset sensitivity value, α is a preset weight factor, ΔC represents the contrast difference factor between adjacent regions, m(x, y) represents the mean value of pixels within a region, (x, y) represents the region coordinates, Ti represents the local threshold within the i-th region, and i represents the number of non-overlapping small regions into which the image is segmented; the calculated local threshold is used to perform preliminary binarization processing on each small region to generate a series of partially binarized images; and the position and bounding box of each crack are marked in the binarized image.

[0052] The image is segmented into a plurality of non-overlapping small regions. According to the crack distribution in each region, the outlier value of the crack distribution characteristics in the corresponding region is calculated, and it is determined whether the outlier value of the crack distribution characteristics in each region is greater than or equal to a preset threshold. If so, it is recorded as a crack concentration region.

[0053] The process for obtaining the outlier value of the crack distribution characteristics is as follows:

[0054] The mobile phone glass cover plate is divided into a number of non-overlapping regions of the same size, and the crack distribution characteristic value F i =[L i , C i , D i ;

[0055] Wherein, L i represents the total length of all cracks within the i-th region, C i represents the number of all cracks within the i-th region, D i represents the crack density within the i-th region, and F i represents the crack distribution characteristic value within the i-th region;

[0056] Perform K-means clustering analysis on the crack distribution characteristic values, and initialize K clustering centers. For each region, find the nearest clustering center according to the Euclidean distance formula and assign it to the corresponding cluster; and calculate and update the position of the new center point of each cluster, repeat the assignment and update process until the clustering center reaches the maximum number of iterations; after obtaining the final clustering result, for each cluster, calculate its average crack distribution characteristic value, and for each region, calculate the sum of the squares of the differences between the crack distribution characteristic value and the average characteristic value of the cluster to which it belongs, and take the square root to obtain the outlier value of the crack distribution characteristics.

[0057] It should be noted that: through the multi-dimensional comprehensive analysis of the crack distribution characteristic values, regions with abnormal crack distributions can be identified more accurately. Combining with the K-means clustering algorithm, it can automatically complete the crack distribution pattern classification and outlier value calculation without manual intervention, which is suitable for real-time detection on large-scale production lines. The identified abnormal regions can be used as an important basis for improving the production process, helping to reduce the defective rate and improve product quality.

[0058] In S3, based on the crack concentration regions, for each crack concentration region, extract the curvature fluctuation characteristics and crack propagation rate characteristics of the cracks, analyze the curvature fluctuation and crack propagation rate characteristics of the cracks, and according to the analysis results, construct a comprehensive feature vector as the input of the machine learning model. According to the model output, judge the quality of each mobile phone glass cover plate, specifically including:

[0059] For each crack concentration region, extract the curvature fluctuation characteristics and crack propagation rate characteristics of the cracks. Calculate the curvature fluctuation eigenvalue according to the degree of curvature fluctuation of the cracks, and calculate the crack propagation anomaly eigenvalue according to the crack propagation rate. Construct the curvature fluctuation eigenvalue and the crack propagation anomaly eigenvalue into a comprehensive feature vector. Use the comprehensive feature vectors in all regions as the input of the machine learning model. The output of the model is the quality score of each mobile phone glass cover plate. Judge the quality of each mobile phone glass cover plate according to the quality score.

[0060] The process of obtaining the curvature fluctuation eigenvalue is as follows:

[0061] For each point on all crack centerlines, approximately calculate the first derivative and second derivative of the arc length by the finite difference method, and calculate the curvature of each crack through the curvature calculation formula. The calculation expression is: where a represents the number of cracks, x′ a and y′ a represent the first derivative of the image coordinates, x″ a and y″ a represent the second derivative of the image coordinates, and k a represents the curvature of the a-th crack;

[0062] Apply the Haar wavelet transform to the calculated curvature sequence, decompose the curvature sequence into different frequency components, and obtain a series of wavelet coefficients. The Haar wavelet transform is represented as a matrix operation: W = H·K, where H represents the Haar wavelet transform matrix, K represents the vector composed of the curvature sequence, and W represents the transformed coefficient vector;

[0063] Based on the wavelet coefficients after wavelet transform, calculate the sum of the standard deviations of all wavelet coefficients to obtain the curvature fluctuation eigenvalue.

[0064] The process of obtaining the crack propagation anomaly eigenvalue is as follows:

[0065] Collect the crack propagation rate measurement values at multiple time points or spatial positions, standardize the collected crack propagation rates. For each standardized data point, find its distance to the B-th nearest neighbor as the distance of the current point; define the data point set containing the current point and its first B nearest neighbors as its B-neighborhood;

[0066] For each standardized data point and another data point, calculate the average distance from all points within the B-neighborhood of the standardized data point to the other data point; divide the number of points within the B-neighborhood of the other data point by this average distance to obtain the reachability density of the other data point relative to the standardized data point. For each standardized data point, calculate the average value of the ratio of the reachability density of all points within its B-neighborhood relative to another data point to the reachability density of the other data point relative to these points to obtain the local outlier factor. Calculate the mean value of all local outlier factors to obtain the crack propagation anomaly eigenvalue.

[0067] It should be noted that: by accurately extracting and analyzing the curvature fluctuation characteristics and crack propagation rate characteristics of the cracks in the mobile phone glass cover plate, the intelligent evaluation of the quality of the glass cover plate is realized. The core technical effect lies in accurately calculating the curvature fluctuation eigenvalue of the crack using the finite difference method and Haar wavelet transform, and evaluating the crack propagation anomaly eigenvalue through the local outlier factor algorithm, so as to construct a comprehensive feature vector as the input of the machine learning model to predict the quality score of each glass cover plate. This method not only improves the detection accuracy but also can effectively identify the micro defects and crack propagation anomalies that are difficult to discover by traditional methods. The innovation points are mainly reflected in: combining mathematical analysis (finite difference method, wavelet transform) with statistical methods (local outlier factor), providing a new multi-dimensional feature extraction and analysis framework for the intelligent evaluation of the quality of material surface cracks. This method has higher accuracy and reliability than traditional single-parameter evaluation, especially suitable for the refined analysis of complex crack morphologies.

[0068] The construction process of the machine learning model is as follows:

[0069] Construct a comprehensive feature vector from the curvature fluctuation eigenvalue and crack propagation anomaly eigenvalue of each crack concentration area;

[0070] Use the comprehensive feature vectors in all regions as the input of the machine learning model, where the machine learning model is a support vector machine model; use the support vector machine model to output the quality score of each mobile phone glass cover plate, and judge the quality of each mobile phone glass cover plate according to the quality score.

[0071] The construction process of the support vector machine model is as follows: Select a linear kernel according to the characteristics of the data to divide samples of different categories in the high-dimensional space; Set the key parameters of the support vector machine, including the regularization parameter and the selection of the kernel function for the RBF kernel, and optimize the key parameters through the method of cross-validation; Collect and label a series of mobile phone glass cover plate samples with known quality scores to form a training set, and use the training set to train the support vector machine model. By minimizing the structural risk, find the optimal hyperplane that can maximize the separation of samples of different quality levels; During the training process, use the cross-validation technique to adjust the model parameters to ensure that the model has good generalization ability; After multiple iterations, obtain the final support vector machine model; For the comprehensive feature vector of each newly input mobile phone glass cover plate, input it into the trained support vector machine model to output the quality score of the mobile phone glass cover plate.

[0072] It should be noted that: By constructing a comprehensive feature vector and using the support vector machine model to score the quality of each mobile phone glass cover plate, it covers the complete process from feature vector construction to model training and then to outputting the quality score, ensuring the systematicness and operability of the technical solution. This method can not only improve the accuracy of quality detection, but also provide strong support for automated production.

[0073] In S4, according to the judgment result, divide the quality of each mobile phone glass cover plate into low quality and high quality, and mark the low-quality mobile phone glass cover plates, specifically including:

[0074] Judge whether the quality score of each mobile phone glass cover plate is greater than or equal to the preset threshold. If so, mark it as a high-quality mobile phone glass cover plate; if not, mark it as a low-quality mobile phone glass cover plate.

[0075] It should be noted that: The low-quality glass cover plates will undergo further manual inspection or recheck by advanced automated detection equipment to confirm the specific situation of the defects. According to the recheck results, these cover plates will be repaired, or directly scrapped, and the relevant data will be recorded for improving the production process and quality control process. This process not only ensures the high-quality standards of the products leaving the factory, but also provides data support for continuous production optimization.

[0076] Working principle of the present invention: A high-resolution industrial camera is used to perform an all-round scan of the mobile phone glass cover plate fixed on the detection platform. By configuring an appropriate light source system (such as a ring-shaped LED), reflection interference is reduced and surface texture features are enhanced. The original image data is collected and preprocessed to ensure the accuracy of subsequent analysis. Based on the preprocessed image data, local threshold segmentation technology is used to identify cracks, and the crack concentration area is determined according to the crack distribution. The local threshold in each small area is calculated for preliminary binarization processing, the crack position and bounding box are marked, and further, the outlier of the crack distribution feature is calculated through clustering analysis to identify the area with abnormal crack distribution. For each crack concentration area, the curvature fluctuation feature and crack propagation rate feature of the crack are extracted. The curvature fluctuation feature value is obtained by approximating the curvature through the finite difference method and its Haar wavelet transform, and the crack propagation abnormal feature value is calculated using the local outlier factor algorithm. A comprehensive feature vector is constructed as the input of the support vector machine model. This model predicts the quality score of each mobile phone glass cover plate based on the knowledge learned from the training set, thereby judging its quality grade. According to the quality score, the mobile phone glass cover plates are divided into two categories: low quality and high quality. The low-quality products are marked, and further manual or advanced automated equipment is arranged for recheck. After confirming the specific situation of the defects, decisions on repair or scrapping are made, and relevant data is recorded for process improvement. The whole process not only improves the detection efficiency and accuracy but also provides a real-time feedback mechanism for the production line, which helps to continuously optimize the production process and improve the product quality.

[0077] The above has described a detailed implementation example of the present invention, but the content described is only the preferred implementation example of the present invention and cannot be considered as used to limit the implementation scope of the present invention. All equivalent changes and improvements made according to the scope of the present invention application should still fall within the patent coverage scope of the present invention.

Claims

1. A method for detecting the quality of a mobile phone glass cover plate based on an image sensor, characterized in that, Including the following steps: S1: Use an image sensor to perform an omni-directional scan on the mobile phone glass cover plate, obtain the original image data, and preprocess the original image data; The image sensor is an industrial camera; S2: Based on the preprocessed original image data, identify the cracks in the mobile phone glass cover plate in the original image, and identify the crack concentration areas according to the crack distribution of the mobile phone glass cover plate; S3: Based on the crack concentration areas, for each crack concentration area, extract the curvature fluctuation characteristics and crack propagation rate characteristics of the cracks, analyze the curvature fluctuation and crack propagation rate characteristics of the cracks, and according to the analysis results, construct a comprehensive feature vector as the input of the machine learning model, and judge the quality of each mobile phone glass cover plate according to the model output; S4: According to the judgment results, divide the quality of each mobile phone glass cover plate into low quality and high quality, and mark the low-quality mobile phone glass cover plates.

2. The method for detecting the quality of a mobile phone glass cover plate based on an image sensor according to claim 1, wherein, The identification of the cracks in the mobile phone glass cover plate in the original image specifically includes: Obtain the original grayscale image of the mobile phone glass cover plate, calculate an initial global threshold, then divide the image into multiple non-overlapping small regions, calculate the local threshold for each small region, and use the calculated local threshold to perform a preliminary binaryzation process on each small region to generate a series of partially binaryzated images; and mark the position and bounding box of each crack in the binaryzated image.

3. The method for detecting the quality of a mobile phone glass cover plate based on an image sensor according to claim 1, wherein, The identification of the crack concentration areas specifically includes: Divide the image into multiple non-overlapping small regions, calculate the crack distribution feature outliers in the corresponding regions according to the crack distribution in each region, and judge whether the crack distribution feature outliers in each region are greater than or equal to the preset threshold. If so, mark it as a crack concentration area.

4. The method for detecting the quality of a mobile phone glass cover plate based on an image sensor according to claim 3, wherein, The process of obtaining the crack distribution feature outliers is as follows: Divide the mobile phone glass cover plate into several non-overlapping regions of the same size, and extract the crack distribution feature values for each region; Perform K-means clustering analysis on the crack distribution feature values, and initialize K clustering centers. For each region, find the nearest clustering center according to the Euclidean distance formula and assign it to the corresponding cluster; And calculate and update the position of the new center point of each cluster, repeat the assignment and update process until the clustering center reaches the maximum number of iterations; After obtaining the final clustering result, for each cluster, calculate its average crack distribution feature value, for each region, calculate the sum of the squares of the differences between the crack distribution feature value and the average feature value of the cluster to which it belongs, and take the square root to obtain the crack distribution feature outliers.

5. The method for detecting the quality of a mobile phone glass cover plate based on an image sensor according to claim 1, wherein Based on the crack concentration areas, for each crack concentration area, extract the curvature fluctuation and crack propagation rate characteristics of the cracks, and analyze the curvature fluctuation and crack propagation rate characteristics of the cracks. Specifically include: For each crack concentration area, extract the curvature fluctuation characteristics and crack propagation rate characteristics of the cracks. According to the degree of curvature fluctuation of the cracks, calculate the curvature fluctuation eigenvalue, and according to the crack propagation rate, calculate the crack propagation anomaly eigenvalue. Construct the curvature fluctuation eigenvalue and the crack propagation anomaly eigenvalue into a comprehensive feature vector, and use the comprehensive feature vectors in all areas as the input of the machine learning model. The output of the model is the quality score of each mobile phone glass cover plate, and judge the quality of each mobile phone glass cover plate according to the quality score.

6. The method for detecting the quality of a mobile phone glass cover plate based on an image sensor according to claim 5, wherein The process of obtaining the curvature fluctuation eigenvalue is as follows: For each point on all crack centerlines, approximately calculate the first derivative and second derivative of the arc length by the finite difference method, and calculate the curvature of each crack through the curvature calculation formula; Apply the Haar wavelet transform to the calculated curvature sequence, decompose the curvature sequence into different frequency components, and obtain a series of wavelet coefficients; Based on the wavelet coefficients after wavelet transform, calculate the sum of the standard deviations of all wavelet coefficients to obtain the curvature fluctuation eigenvalue.

7. The method for detecting the quality of a mobile phone glass cover plate based on an image sensor according to claim 5, characterized in that, The process of obtaining the crack propagation anomaly eigenvalue is as follows: Obtain and standardize the crack propagation rate data set, determine the B-distance and B-neighborhood for each standardized data point, and calculate the reachability density of each data point relative to other points; Calculate the local outlier factor of each data point according to the reachability density, and calculate the mean value of all local outlier factors to obtain the crack propagation anomaly eigenvalue.

8. The method for detecting the quality of a mobile phone glass cover plate based on an image sensor according to claim 5, wherein, The judgment of the quality of each mobile phone glass cover plate specifically includes: Judge whether the quality score of each mobile phone glass cover plate is greater than or equal to the preset threshold. If so, mark it as a high-quality mobile phone glass cover plate. If not, mark it as a low-quality mobile phone glass cover plate.

9. The method for detecting the quality of a mobile phone glass cover plate based on an image sensor according to claim 5, wherein The construction process of the machine learning model is as follows: Construct the curvature fluctuation eigenvalue and the crack propagation anomaly eigenvalue of each crack concentration area into a comprehensive feature vector; Use the comprehensive feature vectors in all areas as the input of the machine learning model, where the machine learning model is a support vector machine model; use the support vector machine model to output the quality score of each mobile phone glass cover plate, and judge the quality of each mobile phone glass cover plate according to the quality score; Collect and label a series of mobile phone glass cover plate samples with known quality scores to form a training set, and use the training set to train the support vector machine model; during the training process, use the cross-validation technique to adjust the model parameters. After multiple iterations, obtain the final support vector machine model; input the comprehensive feature vector of each newly input mobile phone glass cover plate into the trained support vector machine model, and output the quality score of the mobile phone glass cover plate.

Images