An optical film surface defect detection method, system, product and medium

Abstract

An optical film surface defect detection method, system, product, and medium. The method includes: obtaining an edge detection image; performing connected component segmentation on the edge detection image; calculating the shape features of the connected components; comparing the shape features of every two connected components, and merging the connected components with similar shape features into a merged connected component; calculating the gray value distribution feature and gradient change feature of the merged connected component and inputting them into a single defect feature model to obtain a merged single defect error value; splitting the merged connected component with a merged single defect error greater than a preset single defect error threshold into the original similar connected components as defect connected components, and the defect connected components also include the merged connected component with a merged single defect error not greater than the preset single defect error threshold and non-similar connected components; constructing a defect type feature library and performing defect type matching on all defect connected components. Implementing the technical solution provided by this application improves the accuracy of defect recognition.

Description

Technical Field

[0001] This application relates to the field of new materials industry, and particularly to a method, system, product and medium for detecting surface defects of optical films. Background Art

[0002] Optical films play a key role in many industries such as electronic displays and optical instruments. Surface defects on them will significantly reduce the optical performance and reliability of products. Detecting surface defects of optical films can quickly and accurately locate the defects, avoiding defective products from entering the market. In recent years, the detection technology has developed from manual visual inspection to intelligent and automated detection, greatly improving the detection accuracy and efficiency, effectively guaranteeing the quality of optical film products, promoting the refined production and technological progress of related industries, and having extremely important industrial value.

[0003] Currently, machine vision detection technology is mainly used for detecting surface defects of optical films. The light source irradiates from different angles to highlight the defect features; the image acquisition system transmits these images to computer software, and the software will perform preprocessing operations on the images, such as grayscale conversion, filtering, etc., to improve the image quality. Then, algorithms such as edge detection and threshold segmentation are used to extract the defective parts, and finally, the defect types are determined according to the features such as the size, shape, and grayscale of the defects.

[0004] However, when there are dense areas with multiple defects on the surface of the optical film, when dividing the defect areas and determining the defect features according to the features such as the size, shape, and grayscale of the defects, due to the uneven distribution of internal features within the defects, it may lead to incorrect judgments of features such as the number and type of defects, thus affecting the accuracy of defect recognition of optical films. Summary of the Invention

[0005] This application provides a method, system, product and medium for detecting surface defects of optical films, aiming to improve the accuracy of defect recognition.

[0006] In a first aspect, this application provides a method for detecting surface defects of optical films, including:

[0007] Obtaining an edge detection image; performing connected component segmentation on the edge detection image to obtain connected components;

[0008] Calculate the shape features of the connected components; the shape features include circularity, rectangularity, and aspect ratio; compare the shape features of every two connected components to determine whether the shape features are similar; in the case where two connected components are similar connected components, merge the similar connected components into a merged connected component; calculate the merged gray value distribution feature and the merged gradient change feature of the merged connected component, input the merged gray value distribution and the merged gradient change feature into a single defect feature model to obtain a merged single defect error value; split the merged connected component with the merged single defect error greater than a preset single defect error threshold into the original similar connected components as defect connected components; the defect connected components also include the merged connected components and non-similar connected components with the merged single defect error not greater than the preset single defect error threshold; construct a defect type feature library and perform defect type matching on all defect connected components in the edge detection image.

[0009] In the above embodiment, the edge detection image is segmented into multiple connected components, the shape features are calculated and the similarity is compared, the connected components with similar shapes are merged, then the gray and gradient features of the merged connected component are calculated and input into the model to obtain the error value, and accordingly, it is selected to split into the original connected components or retain the merged connected component. After constructing the feature library, the types are matched for all defect connected components. Through multi-feature analysis and model evaluation, the defect connected components are identified and classified, the misjudgment rate is reduced, and the accuracy of optical film defect detection is improved.

[0010] Combined with some embodiments of the first aspect, in some embodiments, performing connected component segmentation in the edge detection image to obtain connected components specifically includes:

[0011] Calculate the gray average value and the gray standard deviation of the local area of each pixel point in the edge detection image; the local area of the pixel point is the area covered by sliding a window with a preset size centered on the pixel point; determine a defect judgment threshold through the gray average value and the gray standard deviation; connect the pixel points that are adjacent and have the gray average value greater than the defect judgment threshold among all pixel points in the edge detection image to obtain connected components.

[0012] In the above embodiment, the threshold is determined by calculating the local area features of the pixel points, and then the relevant pixel points are connected according to the threshold, so that the connected components can be accurately divided, providing a basis for subsequent accurate detection and analysis of optical film defects, and improving the accuracy and efficiency of defect recognition.

[0013] Combined with some embodiments of the first aspect, in some embodiments, the calculation of the gray average value and the gray standard deviation of the local area of each pixel point in the edge detection image specifically includes:

[0014] Starting from the upper left corner of the edge detection image, each pixel point is sequentially selected as the central pixel point in row-major order; for the current central pixel point, an empty data group is created, and the pixel gray values within the area covered by sliding a window of a preset size centered on the current central pixel point are stored in this data group; the gray average value of the local area of the current central pixel point is calculated using the first formula; where the first formula is: The gray standard deviation of the local area of the current central pixel point is calculated using the second formula; where the second formula is: Where is the gray average value of the local area of the current central pixel point, σ is the gray standard deviation of the local area of the current central pixel point, the size of the preset-size window is m×n, in this edge detection image, the coordinate axes are established with the upper left corner fixed point as the origin, the direction from the upper left corner fixed point to the lower left corner fixed point is the x-axis, the direction from the upper left corner fixed point to the upper right corner fixed point is the y-axis, the coordinate axis plane is divided into squares with a size of one pixel unit, and the coordinate position of the current central pixel point in these coordinate axes is (i,j).

[0015] In the above embodiment, by calculating the local features of pixel points in the edge detection image one by one using the mean and standard deviation calculation formulas combined with the Gaussian formula, the situation of each local area of the image can be accurately quantified, laying a foundation for accurately identifying defects and improving the accuracy of defect detection in the subsequent process.

[0016] Combined with some embodiments of the first aspect, in some embodiments, for constructing the defect type feature library and performing defect type matching on all defect connected regions in the edge detection image, it specifically includes:

[0017] Collect a variety of known type optical film defect samples. For the sample connected region of each optical film defect sample, extract the sample feature data of the sample connected region; the sample feature data is the sample gray value distribution feature, sample gradient change feature, sample circularity, and sample rectangularity of the sample connected region; associate and store the sample feature data with the corresponding sample defect type of the sample connected region to construct a defect type feature library; calculate the defect feature data of the defect connected region; the defect feature data includes defect gray value distribution feature, defect gradient change feature, defect circularity, and defect rectangularity; calculate the similarity between the defect feature data and each sample feature data in the defect type feature library to obtain a similarity value group; take the maximum defect type as the defect type matched with the defect connected region; the maximum defect type is the defect type of the sample connected region corresponding to the maximum similarity data in the similarity value group.

[0018] In the above embodiments, by constructing a feature library and comparing similarities, it is possible to accurately determine the defect type of the defect connected region based on existing samples, improving the accuracy and efficiency of the defect classification of the optical film.

[0019] Combined with some embodiments of the first aspect, in some embodiments, after calculating the similarity between the defect feature data and each sample feature data in the defect type feature library to obtain a similarity value group, it further includes:

[0020] In the case where all the similarity data in the similarity data group are lower than a preset minimum similarity threshold, the defect connected region is marked as an unknown type connected region.

[0021] In the above embodiments, by marking the connected regions with relatively low similarities to all defect types in the feature library as unknown type connected regions, misclassification is avoided, which helps to improve the defect classification system.

[0022] Combined with some embodiments of the first aspect, in some embodiments, after the defect type of the defect connected region is matched to be the maximum defect type, it further includes:

[0023] Calculate the area of the defect connected region; based on the area of the defect connected region, the defect gray mean value in the defect gray value distribution feature, and the aspect ratio of the defect, judge the defect level of the defect connected region with a preset level division threshold.

[0024] In the above embodiments, by calculating the area of the defect connected region, and at the same time combining features such as the gray mean value and the aspect ratio in its gray value distribution feature, and referring to the preset level division threshold to comprehensively judge the defect level to which the defect connected region belongs, it is possible to perform a more detailed quantitative evaluation of the defects on the optical film.

[0025] Combined with some embodiments of the first aspect, in some embodiments, after splitting the merged connected region with a merged single defect greater than a preset single defect error threshold into the original similar connected regions as defect connected regions, it further includes: determining the central position of the defect connected region as the center of the connected region; calculating the center distance between every two centers of the connected regions and the center density of the connected regions in the edge detection image; in the case where the short distance ratio value in the center distances of the connected regions in the edge detection image is greater than a preset short distance ratio threshold or the center density of the connected regions is greater than a preset maximum density threshold, marking the optical film corresponding to the edge detection image as a high-defect optical film; the short distance ratio value is the ratio of the center distances of the connected regions less than a preset distance threshold to all the center distances of the connected regions in the edge detection image.

[0026] In the above embodiments, by analyzing the relevant parameters of the center of the connected component, it is possible to judge the density of optical film defects according to specific criteria, and then identify high-defect optical films with relatively dense defects or a large number of defects, which is convenient for screening out products with poor quality and ensuring the overall quality level of the products.

[0027] In a second aspect, an embodiment of the present application provides an optical film surface defect detection system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, and the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the optical film surface defect detection system to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0028] In a third aspect, an embodiment of the present application provides a computer program product containing instructions. When the computer program product runs on an optical film surface defect detection system, the optical film surface defect detection system is enabled to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0029] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, including instructions. When the instructions run on an optical film surface defect detection system, the optical film surface defect detection system is enabled to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0030] It can be understood that the optical film surface defect detection system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the optical film surface defect detection method provided in the embodiments of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be elaborated here.

[0031] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0032] 1. In the present application, the edge detection image is segmented into multiple connected components, the shape features are calculated and compared for similarity, the connected components with similar shapes are merged, and then the gray scale and gradient features of the merged connected components are calculated and input into the model to obtain an error value. According to this, the connected components are selected to be split into the original ones or the merged connected components are retained. After constructing the feature library, the types of all defect connected components are matched. Through multi-feature analysis and model evaluation, the defect connected components are identified and the defects are classified, reducing the misjudgment rate and improving the accuracy of optical film defect detection.

[0033] 2. By constructing a feature library and comparing similarities, this application enables accurate determination of the defect type of a defect connected region based on existing samples, improving the accuracy and efficiency of optical film defect classification.

[0034] 3. By calculating the area of the defect connected region and combining features such as the average gray value and aspect ratio in its gray value distribution characteristics, and referring to a preset grading threshold for comprehensive judgment of the defect grade to which the defect connected region belongs, this application enables more detailed quantitative evaluation of defects on the optical film. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a schematic structural diagram of an applicable system architecture for the optical film surface defect detection method in an embodiment of this application;

[0036] Figure 2 is a schematic flowchart of an optical film surface defect detection method in an embodiment of this application;

[0037] Figure 3 is another schematic flowchart of an optical film surface defect detection method in an embodiment of this application;

[0038] Figure 4 is a schematic diagram of an exemplary hardware structure of an optical film surface defect detection system in an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] The terms used in the following embodiments of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. As used in the specification and appended claims of this application, the singular forms "a", "an", "the", "above", "said", "this" are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in this application refers to and includes any or all possible combinations of one or more of the listed items.

[0040] Hereinafter, the terms "first" and "second" are only for descriptive purposes and cannot be construed as implying or indicating relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the embodiments of this application, unless otherwise stated, the meaning of "a plurality" is two or more.

[0041] Figure 1 is a schematic structural diagram of an applicable system architecture for the optical film surface defect detection method in an embodiment of this application.

[0042] Please refer to Figure 1, the system architecture includes an imaging device 110, a data transmission device 120, and a server 130.

[0043] The imaging device 110 includes an industrial camera 111, an optical lens 112, and an illumination system 113, which is used to obtain the original image of the optical film surface and provide basic data for subsequent defect detection. It is the data source of the entire detection system. Among them, a high-resolution and high-precision industrial camera 111 is used to capture clear images of the optical film surface at an appropriate frame rate to meet the requirements of the detection for image details and provide a high-quality source of original image data for subsequent defect analysis. According to different requirements of the detection scenario, optical lenses 112 with different focal lengths and apertures and adapted to the industrial camera 111 can be selected to ensure that the field of view and clarity of the imaging meet the detection standards. The illumination system 113 includes various types of light sources, such as uniform surface light sources, ring light sources, etc. By reasonably arranging and adjusting parameters such as the brightness, angle, and color of the light source, sufficient and uniform illumination conditions are provided for the optical film, reducing the impact of shadows and reflections on the image quality, so that the captured images can clearly present the true situation of the optical film surface.

[0044] The data transmission device 120 is a data cable 121, which stably and quickly transmits the image data collected by the imaging device to the data processing server, ensuring the efficiency and accuracy of data transmission. It is a bridge connecting the data acquisition and processing links. The data cable 121 uses high-speed and stable data cables, such as optical fibers, high-speed network cables, etc., to transmit the image data collected by the industrial camera to the subsequent processing hardware devices in real time, ensuring the coherence and timeliness of the detection process.

[0045] In some embodiments of the present application, the data transmission device uses data cable devices such as optical fibers and network cables for data transmission; in some other embodiments of the present application, the data transmission method further includes a network switch, which uses a network transmission method to transmit data, and no limitation is made here.

[0046] The server 130 includes a processor 131 and a memory and storage device 132, which performs complex computational processing on the transmitted image data and is the core computing unit of the entire system. Among them, the processor 131 is a high-performance central processing unit or a graphics processing unit, which can improve the detection efficiency and reduce the detection time for complex image data calculation tasks; the storage device 132 is used to temporarily store the image data being processed, intermediate calculation results, etc., to ensure that the data read and write speed can keep up with the processor operation speed. At the same time, a large amount of historical detection data, training model data, and detection result data are stored long-term for subsequent data analysis.

[0047] The following describes the optical film surface defect detection method in an embodiment of the present application:

[0048] Please refer toFigure 2 , which is a schematic flowchart of an optical film surface defect detection method provided by an embodiment of the present application. This method can be applied to Figure 1 the system architecture shown in the figure, and includes the following steps:

[0049] S201. Obtain an edge detection image;

[0050] First, use an imaging device, such as an industrial camera combined with a suitable optical lens and illumination system, to capture an optical film to obtain its original image. Then, apply the Canny edge detection algorithm to process the original image. By identifying features such as pixel gray value changes in the image, the object edges are highlighted, thereby obtaining an edge detection image.

[0051] In some embodiments of the present application, the Canny edge detection algorithm is used to process the original image; in some other embodiments of the present application, image processing algorithms such as the Laplacian edge detection algorithm can also be used to process the original image, which is not limited here.

[0052] The above technical steps use the edge detection algorithm for the original image of the optical film, converting the originally complex image that is not easy to directly analyze defects into an image that highlights edge information, providing an image basis for subsequent operations and improving the detection accuracy.

[0053] S202. Perform connected component segmentation on the edge detection image to obtain connected components;

[0054] In the edge detection image, first scan the image pixel points, and judge which pixels belong to defective pixel points according to the gray value of the pixel points. Group the defective pixel points and gradually mark and distinguish different connected components. By continuously traversing all the pixels of the image, the entire image is finally divided into independent connected components.

[0055] In some embodiments of the present application, it is judged whether a pixel point belongs to a defective pixel point according to the gray value of the pixel point; in some other embodiments of the present application, it can also be judged whether a pixel point belongs to a defective pixel point through the gradient feature or other features of the pixel point, which is not limited here.

[0056] The above technical steps divide the connected components according to the gray value characteristics of the pixel points, further refining the originally edge-presenting image into different connected regions, providing a more targeted object for subsequent accurate analysis of the characteristics of each region and precise judgment of defects, which is conducive to improving the detection accuracy.

[0057] S203. Calculate the shape features of the connected components;

[0058] For each connected component, corresponding mathematical algorithms and geometric calculation methods are used to obtain its shape features. For example, the circularity of the connected component is calculated through a specific formula to measure its similarity to a circle; the rectangularity is calculated using relevant algorithms to see how close it is to a rectangle; and the aspect ratio is calculated to know the ratio of its length to width, thereby obtaining these key feature data of each connected component in terms of shape.

[0059] In some embodiments of the present application, the shape features of the connected component include circularity, rectangularity, and aspect ratio; in some other embodiments of the present application, the shape features of the connected component may further include fractal dimension, etc., which are not limited herein.

[0060] The above technical steps calculate the shape features of the connected component, enabling the description and differentiation of different connected components from the geometric morphology perspective, facilitating the subsequent comparison of the similarity between connected components, providing an important basis for accurately merging similar connected components and judging the defect categories, etc., and helping to improve the accuracy and efficiency of defect detection.

[0061] S204. Determine whether every two connected components are similar connected components;

[0062] If two connected components are similar connected components, then execute the following step S205;

[0063] If one connected component and all other connected components are non-similar connected components, then execute the following step S207;

[0064] For each connected component, select every two connected components one by one, extract their shape features, and then compare whether the shapes are similar according to a preset difference range. For example, whether the circularity difference is within the preset circularity difference threshold range, and the same applies to the rectangularity and aspect ratio differences; in the case where the shapes of these two connected components are similar, determine whether the distance between these two connected components is less than the preset distance threshold. If the distance is less than the preset distance threshold, it is determined that these two connected components are similar connected components, and execute the following step S205 to merge the similar connected components into one connected component; if one connected component does not meet the requirements of the above-mentioned similar connected components with all other connected components, then it is determined that this one connected component is a non-similar connected component, and execute the following step S207 to add the non-similar connected component to the defective connected component.

[0065] In some embodiments of the present application, the distance between two connected components can be determined by the centroid method to determine the center of the connected component and then calculate the distance between the centers of the two connected components; in some other embodiments of the present application, the distance between two connected components can also be determined by methods such as calculating the distance between the two closest pixel points in the two connected components, etc., which are not limited herein.

[0066] The above technical steps distinguish similarity or dissimilarity by comparing the shape features and distances of connected components, enabling subsequent targeted merging or other processing of connected components, avoiding misjudgment of features such as the number and type of defects due to uneven internal feature distribution within the defects, thereby wrongly dividing the connected components, and helping to improve the accuracy of processing connected components in defect detection.

[0067] S205. Merge the similar connected components into a merged connected component;

[0068] After determining that two connected components are similar connected components, an image region merging algorithm is used for operation. For example, a pixel in one of the connected components is selected as the starting point, and then the pixels in the other connected component are added to the region where the starting point is located one by one, thereby achieving merging and completing the merging operation.

[0069] In some embodiments of the present application, merging similar connected components into one connected component uses the reverse effect of the region growing method; in other embodiments of the present application, merging similar connected components into one connected component can also use image merging algorithms such as the combination of dilation and complex screening operations, which are not limited herein.

[0070] The above technical steps avoid the situation of dividing a single defect into multiple connected components due to uneven internal feature distribution of the defect by merging similar connected components, and help to improve the accuracy of operations such as matching defect types.

[0071] S206. Calculate the gray value distribution feature and gradient change feature of the merged connected component, and input the merged gray value distribution and merged gradient change feature into the single defect feature model to obtain the merged single defect error value;

[0072] Specifically, the single defect feature model is obtained by training a pre-set neural network with multiple groups of historical gray value distribution features, historical gradient change features, and corresponding marked single defect error values, and is used to output a single defect error value according to a set of input gray value distribution features and historical gradient change features.

[0073] First, corresponding algorithms are used to statistically calculate the gray value distribution feature and gradient change feature of the pixels in the merged connected component. For example, numerical values such as the gray mean and gray median are calculated as the gray distribution feature, and the Sobel operator is used to calculate the gradient change feature at the same time. Then, these feature data are input into the single defect feature model that has been trained in advance using historical data, and after internal operations of the model, the merged single defect error value is finally output.

[0074] In some embodiments of the present application, the gray distribution feature is the gray mean and gray median; in other embodiments of the present application, the gray distribution feature may also include numerical values such as the gray mode and gray range, which are not limited herein.

[0075] In some embodiments of the present application, the Sobel operator is used to calculate the gradient change feature; in other embodiments of the present application, the Prewitt operator or other methods can also be used to calculate the gradient change feature, which is not limited herein.

[0076] The above technical steps extract features and perform operations with the aid of a model to evaluate the difference degree between the merged connected regions and single defects, avoid mis-merging the connected regions of different defects, and if the single defect error value is large, it is determined that the similar connected regions are not the same defect, thereby improving the accuracy of defect detection.

[0077] S207. Split the merged connected region with a merged single defect error greater than the preset single defect error threshold into the original similar connected regions as the defect connected regions, and the defect connected regions also include the merged connected regions with a merged single defect error not greater than the preset single defect error threshold and non-similar connected regions;

[0078] When the merged single defect error value obtained from the merged connected region is greater than the preset single defect error threshold, it means that this merged connected region does not meet the requirements. According to the previously recorded division situation of the similar connected regions, it is restored and split back into the original similar connected regions, and together with the merged connected regions with an error value not exceeding the threshold and the non-similar connected regions that cannot form similar connected regions with all other connected regions, they are used as the defect connected regions.

[0079] The above technical steps are reasonably split through error comparison, so that the true defect connected regions can be accurately screened out, avoiding misjudgment of defects caused by mis-merging, and improving the accuracy of defining the surface defects of the optical film.

[0080] S208. Construct a defect type feature library and perform defect type matching for all defect connected regions in the edge detection image.

[0081] First, collect various known optical film defect samples, extract the characteristic data of the sample connected regions, and associate the corresponding defect types to construct the defect type feature library. Then, for the defect connected regions in the edge detection image, extract the above relevant characteristic data in the same way and compare and match them with the data in the feature library to determine their defect types.

[0082] The above technical steps construct a feature library and perform matching, so that the type to which the defect connected region belongs can be accurately judged with reference to the existing sample data, improving the accuracy and efficiency of optical film defect detection and classification, and helping to better control the quality of optical film products.

[0083] In the above embodiments, multiple connected components are obtained by performing connected component segmentation on the edge detection image, their shape features are calculated and the similarities are compared, the connected components with similar shapes are merged, then the gray scale and gradient features of the merged connected components are calculated and input into the model to obtain an error value, and accordingly, they are selected to be split into the original connected components or the merged connected components are retained, and after constructing a feature library, the types of all defective connected components are matched. Through multi-feature analysis and model evaluation, defective connected components are identified and defect classification is performed, reducing the misjudgment rate and improving the accuracy of optical film defect detection.

[0084] In some embodiments, some special situations may be encountered during the detection of surface defects of an optical film. For example, in the case where there are too many surface defects on some optical films or some defects are too dense, these optical films need to be specially processed. In such a case, the optical film surface defect detection method can, according to the distance and density between the surface defects of the optical film, specially mark the optical films with too many defects or too many closely spaced defects. As Figure 3 shown, it is another flow schematic diagram of the optical film surface defect detection method provided by the embodiment of the present application, and this method can be used for Figure 1 the system architecture shown, and includes the following steps:

[0085] S301. Obtain an edge detection image;

[0086] S302. Starting from the upper left corner of the edge detection image, sequentially select each pixel point as the central pixel point in row-major order;

[0087] For the edge detection image, the image is regarded as a matrix composed of numerous pixel points. Starting from the first pixel point in the upper left corner of the image, in the order of column by column from left to right and row by row from top to bottom, each pixel point is sequentially regarded as the central pixel point, preparing for subsequent operations such as calculating local region features for each pixel point.

[0088] The above technical steps, by selecting the central pixel points in a specific order, enable the processing of each pixel point in the image regularly and without omission, facilitating the unified calculation of relevant features.

[0089] S303. For the current central pixel point, create an empty data group, and store the pixel gray scale values in the area covered by the sliding of a window with a preset size centered on the current central pixel point into this data group;

[0090] After selecting the current central pixel point, first initialize an empty data group. According to the preset window size (such as 3×3, 5×5, etc.), let the window slide on the image centered on the current central pixel point, sequentially extract the gray scale values corresponding to each pixel point in the area covered by the window, and then store them in the empty data group created before in order.

[0091] The above technical steps collect the pixel gray values in the window coverage area and store them in a data group, enabling subsequent accurate calculation of statistics such as the mean and standard deviation based on these data, providing an effective numerical basis for operations such as defect judgment and connected component segmentation, and improving the accuracy of defect detection.

[0092] S304. Calculate the gray average value of the local area of the current central pixel point using the first formula;

[0093] Specifically, in this edge detection image, an axis is established with the upper left fixed point as the origin. The direction from the upper left fixed point to the lower left fixed point is the x-axis, and the direction from the upper left fixed point to the upper right fixed point is the y-axis. The coordinate plane of this axis is divided into grids with a size of one pixel unit. The coordinate position of the current central pixel point in the coordinate axis is (i, j). The first formula is: Among them, is the gray average value of the local area of the current central pixel point, and the size of the window with a preset size is m×n.

[0094] In the calculation of local area features, it is necessary to determine a window range centered on the current pixel point. The integral domain calculation form in the formula involved here, such as and is to accurately define the window area centered on the pixel point (i, j).

[0095] In the image coordinate system, when we want to construct a local window with a size of m×n centered on a pixel point (i, j), it is necessary to accurately determine the start and end positions of the window. Using to (similarly for the y direction) this form can ensure that the window is centered on (i, j). When m is odd (this is a common window size selection in image processing because odd numbers can ensure there is a clear central pixel), this calculation method can accurately determine the range of the window, making the window center exactly at (i, j).

[0096] When performing local area calculations for each pixel point of the entire image, this calculation form can ensure a relatively consistent processing method at the image edge. Although the window may partially exceed the image boundary at the image edge, this calculation form can ensure the coherence and accuracy of the calculation through reasonable boundary processing (such as ignoring the part exceeding the boundary or adopting a specific filling method).

[0097] By limiting the calculation of the gray average value within this local area, local gray features can be better captured. In image processing, many features (such as edges, textures, etc.) often appear in local areas, and this local calculation helps to accurately discover and describe these features.

[0098] By selecting the integration domain based on the central pixel, boundary effects can be avoided or reduced to a certain extent. If a global or inappropriate integration domain is selected, inaccurate calculation results may occur at the image edge or at the junction of different feature regions. This selection of the integration domain for the local region makes the calculation of each local region relatively independent and reduces the mutual interference between different regions.

[0099] S305. Calculate the standard deviation of the gray level of the local region of the current central pixel using the second formula;

[0100] Specifically, the second formula is: where σ is the standard deviation of the gray level of the local region of the current central pixel.

[0101] When calculating the standard deviation using the second formula, it is based on the average gray level calculated for the above-mentioned local region The calculation of the standard deviation is closely related to the gray level distribution of the local region and can better reflect the degree of dispersion of the pixel gray level values within the local region.

[0102] The second formula is similar to the first formula. Its calculation based on the local region helps to describe the change of the pixel gray level within the local region. In image processing, the standard deviation of the gray level can reflect features such as the texture complexity of the image. Through this local calculation method, the local texture features of the image can be more accurately characterized.

[0103] Calculating the standard deviation within the local region can, to a certain extent, suppress the influence of noise. If the standard deviation is calculated for the entire image, the noise points may have a greater interference on the result. By calculating through the local region, only the noise within the local region will affect the result, and the normal pixels within the local region can balance the influence of the noise to a certain extent, thus obtaining a more reliable result of the standard deviation of the gray level.

[0104] In the above two steps, by calculating the average gray level and the standard deviation of the gray level using the first formula and the second formula, the interference of noise pixels within the local region on the result can be reduced, which helps to improve the accuracy of determining the threshold and connected component segmentation based on the average gray level and the standard deviation of the gray level in the subsequent process.

[0105] S306. Determine the defect judgment threshold based on the average gray level and the standard deviation of the gray level;

[0106] After obtaining the average gray level and the standard deviation of the gray level of the local region of each pixel, determine the defect judgment threshold according to the average gray level and the standard deviation of the gray level. For example, add several times the standard deviation to the average value as the defect judgment threshold.

[0107] In some embodiments of the present application, the defect judgment threshold is obtained by calculating the weighted average of the grayscale average value and the grayscale standard deviation; in some other embodiments of the present application, the defect judgment threshold can also be obtained by machine learning methods, training a model using known defective and defect-free samples, and enabling the model to output a threshold based on the input average value and grayscale standard deviation, etc. The specific method is not limited herein.

[0108] The above technical steps determine the threshold by using the grayscale average value and the standard deviation, enabling clear distinction between normal pixel points and pixel points that may have defects, providing a quantitative standard for accurately screening out defective pixel points in the subsequent process and reducing misjudgment situations.

[0109] S307. Connect the pixel points that are adjacent and have a grayscale average value greater than the defect judgment threshold among all the pixel points in the edge detection image to obtain a connected domain.

[0110] Specifically, traverse all the pixel points in the edge detection image. For each pixel point, check the pixel points adjacent to it (such as adjacent up, down, left, and right, etc.). Then compare the grayscale average value of them with the determined defect judgment threshold. If the grayscale average value of the adjacent pixel points exceeds this threshold, connect these qualified adjacent pixel points through corresponding algorithms to form a whole, that is, obtain the connected domain.

[0111] The above technical steps judge defective pixel points based on the threshold and connect adjacent defective pixel points in the detection image, enabling the segmentation of the area that may have defects in the image as a connected domain, which is convenient for further analyzing the characteristics of these connected domains in the subsequent process.

[0112] S308. Calculate the shape characteristics of the connected domain.

[0113] S309. Judge whether every two connected domains are similar connected domains.

[0114] If two connected domains are similar connected domains, then execute the following step S205.

[0115] If one connected domain and all other connected domains are non-similar connected domains, then execute the following step S207.

[0116] S310. Merge the similar connected domains into a merged connected domain.

[0117] S311. Calculate the merged grayscale value distribution characteristics and merged gradient change characteristics of the merged connected domain, input the merged grayscale value distribution and merged gradient change characteristics into the single defect feature model, and obtain the merged single defect error value.

[0118] S312. Split the merged connected regions with a merged single defect error greater than the preset single defect error threshold into the original similar connected regions as defect connected regions, where the defect connected regions also include the merged connected regions with a merged single defect error not greater than the preset single defect error threshold and non-similar connected regions;

[0119] S313. Collect various known types of optical film defect samples, and for the sample connected regions of each optical film defect sample, extract the sample feature data of the sample connected regions;

[0120] Specifically, the sample feature data includes the sample gray value distribution feature, the sample gradient change feature, the sample roundness, and the sample rectangularity.

[0121] Collect known optical film defect samples covering different types such as scratches, spots, holes, etc. For the connected regions in each sample, use corresponding algorithms to extract feature data, such as calculating the gray value distribution feature, the gradient change feature, the roundness, and the rectangularity of the connected regions, so as to obtain the sample feature data of each sample connected region comprehensively.

[0122] The above technical steps enable the accumulation of feature information corresponding to various defects by extracting the feature data of the sample connected regions, providing a data basis for constructing a defect type feature library and facilitating the accurate matching of unknown defect types.

[0123] S314. Construct a defect type feature library;

[0124] When constructing a defect type feature library, first create a suitable data storage structure, such as a database table or a file in a specific format, etc., and then associate the collected and extracted sample feature data with the sample defect types of their corresponding sample connected regions one by one, and store them in the created storage structure in the form of key-value pairs or records to complete the associated storage and construct the defect type feature library.

[0125] The above technical steps enable the subsequent detection to find and judge the type of unknown defects according to feature comparison by associating and storing the sample feature data with the defect types.

[0126] S315. Calculate the defect feature data of the defect connected regions;

[0127] Specifically, the defect feature data includes the defect gray value distribution feature, the defect gradient change feature, the defect roundness, and the defect rectangularity.

[0128] The above technical steps enable the subsequent comparison and matching with the data in the defect type feature library by calculating the feature data of the defect connected regions, facilitating the accurate judgment of the defect type to which it belongs.

[0129] S316. Calculate the similarity between the defect feature data and each sample feature data in the defect type feature library to obtain a similarity value group;

[0130] For the defect feature data of the defect connected region, calculate it with each sample feature data in the defect type feature library in turn according to the similarity algorithm. Each time a comparison is made, a corresponding similarity value will be obtained. Summing up these values will result in a similarity value group containing multiple similarity values.

[0131] In some embodiments of the present application, the similarity algorithm used for similarity calculation is the Euclidean distance algorithm; in some other embodiments of the present application, other similarity algorithms such as the cosine similarity algorithm can also be used for similarity calculation, which is not limited here.

[0132] The above technical steps calculate the similarity value group, enabling the similarity between the defect connected region and the sample to be measured based on these values, and then finding the most matching sample to determine the defect type corresponding to the defect connected region.

[0133] S317. Match the defect type of the defect connected region as the maximum defect type;

[0134] After obtaining the similarity value group, compare the similarity values within the group to obtain the maximum similarity value, and determine the defect type of the defect connected region as the defect type of the sample corresponding to the maximum similarity value, that is, the maximum defect type.

[0135] The above technical steps accurately classify the defect connected region by matching the maximum defect type, clarifying its belonging defect type.

[0136] S318. In the case where all similarity data in the similarity data group are lower than the preset minimum similarity threshold, mark the defect connected region as an unknown type connected region;

[0137] After obtaining the similarity data group, compare the similarity values within the group to obtain the maximum similarity value. If the maximum similarity value is lower than the preset minimum similarity threshold, it indicates that the defect connected region has too large a difference from the sample features in the feature library and cannot accurately match the existing types. At this time, it is marked as an unknown type connected region.

[0138] The above technical steps mark the unknown type connected region by reasonably setting the threshold, enabling the identification of special defect situations not covered in the feature library, facilitating further analysis and research of these new situations, continuously improving the defect type feature library, and enhancing the comprehensiveness of defect identification for various types.

[0139] S319. Calculate the area of the defect connected region of the defect connected region;

[0140] Determine all the pixel points included in the defective connected region, and count the number of pixel points. Since each pixel can be regarded as occupying a certain unit area in the image (for example, the area of a single pixel is set to 1), the counted value of the number of pixel points is equal to the area size of the defective connected region.

[0141] In some embodiments of the present application, the pixel calculation method is used to calculate the area of the defective connected region; in some other embodiments of the present application, methods such as the integral method can also be used to calculate the area of the next connected region, which is not limited herein.

[0142] The above technical steps provide a basis for evaluating the impact degree of the defect on the optical film by calculating the area of the defective connected region, and provide a data basis for subsequent operations such as classifying the defects according to factors such as the area size.

[0143] S320. Determine the defect level of the defective connected region according to a preset level division threshold;

[0144] Obtain the area, defect gray mean value, and defect length-width ratio data of the defective connected region, and compare them with the corresponding level division thresholds respectively. For example, when the area reaches a certain range, the gray mean value is in a specific interval, and the length-width ratio meets the corresponding conditions, etc. After comprehensive judgment, determine which defect level the defective connected region conforms to.

[0145] The above technical steps determine the defect level by comparing multiple index thresholds, enabling quantitative differentiation of the severity of the defects, and effectively improving the scientificity and rationality of the defect control of the optical film.

[0146] S321. Determine the center position of the defective connected region, which is the center of the connected region;

[0147] For the defective connected region, mark all the pixel points belonging to this connected region, and then separately count the coordinate values of these pixel points in the horizontal and vertical directions. Take the average value of the horizontal direction coordinate values as the abscissa, and the average value of the vertical direction coordinate values as the ordinate. The coordinate point position determined thereby is the center of the connected region.

[0148] In some embodiments of the present application, the centroid method is used to determine the center position of the defective connected region; in some other embodiments of the present application, methods such as the center of the bounding box method can also be used to determine the center position of the defective connected region, which is not limited herein.

[0149] S322. Calculate the distance between the centers of every two connected components and the density of the centers of connected components in the edge-detection image. For each determined center of a connected component, calculate the distance between every two centers of connected components. The value can be calculated by substituting the coordinates into the distance formula between two points. The density of the centers of connected components is the number of centers of connected components divided by the total area of the edge-detection image, which is used to measure the distribution of the centers of connected components per unit area.

[0150] Through calculating the distance and density, the above technical steps enable the understanding of the distribution density and mutual positional relationship of defects in the image, and obtain the distribution of defects on the surface of the optical film.

[0151] S323. In the case where the short-distance ratio value among the distances between the centers of connected components in the edge-detection image is greater than the preset short-distance ratio threshold or the density of the centers of connected components is greater than the preset maximum density threshold, mark the optical film corresponding to the edge-detection image as a high-defect optical film.

[0152] Specifically, the short-distance ratio value is the ratio of the distances between the centers of connected components less than the preset distance threshold to all the distances between the centers of connected components in the edge-detection image.

[0153] Statistically analyze the proportion of short distances among the distances between the centers of connected components in the edge-detection image to calculate the short-distance ratio value. At the same time, calculate the density of the centers of connected components, and then compare these two values with the preset short-distance ratio threshold and maximum density threshold respectively. In the case where the short-distance ratio value is greater than the preset short-distance ratio threshold or the center density is greater than the maximum density threshold, mark the corresponding optical film as a high-defect optical film.

[0154] Through the above technical steps of setting threshold comparison and judgment, it is possible to quickly screen out optical films with relatively dense defect distributions or a large number of defects, which is convenient for preferentially performing special processing on such high-defect optical films and improving the efficiency of overall quality control of optical film products.

[0155] Steps S301, S308 - S312 are similar to Figure 2 Steps S201, S203 - S207 in the illustrated embodiment. For details, refer to the descriptions in Steps S201, S203 - S207 and will not be elaborated here.

[0156] In the embodiments of the present application, first, an edge detection image is obtained and the connected regions are segmented. The local region features of each pixel point are calculated to determine a threshold for dividing the connected regions, so as to accurately locate the possible defect regions. Secondly, the shape features of the connected regions are compared and the similar connected regions are merged. The features of the merged connected regions are calculated and input into the model for evaluation. According to the error, they are split or retained. Then, a feature library is constructed and the similarity is compared to match the defect type, which helps to accurately identify the types of defects and improve the classification accuracy. The features such as the area of the defect connected region are calculated to determine the defect level, and the central position is determined. The central distance and density are calculated to distinguish the high-defect optical film, so as to quantitatively evaluate and analyze the distribution of the defects, effectively reduce the misjudgment rate, ensure the quality of the optical film product, and improve the accuracy of defect detection.

[0157] The following introduces the exemplary optical film surface defect detection system 400 provided by the embodiments of the present application. Figure 4 It is an exemplary hardware structure diagram of the optical film surface defect detection system 400 provided by the embodiments of the present application.

[0158] In some embodiments, the optical film surface defect detection system 400 includes a computer device. The computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data. The network interface of the computer device is used to communicate with other external terminals or servers through a network connection. In some embodiments, the network interface can be a wired network interface, and in some embodiments, the network interface can also be a wireless network interface. When the computer program is executed by the processor, it realizes the method in the embodiments of the present application.

[0159] Those skilled in the art can understand that Figure 4 the structure shown in

[0160] is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements. As described above, the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0161] As used in the foregoing embodiments, depending on the context, the term "when" can be interpreted to mean "if" or "after" or "in response to determining" or "in response to detecting". Similarly, depending on the context, the phrase "when determining" or "if detecting (the stated condition or event)" can be interpreted to mean "if determining" or "in response to determining" or "when detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)".

[0162] In the foregoing embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more integrated available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid-state drive), etc.

[0163] Those of ordinary skill in the art can understand all or part of the processes in the methods of the foregoing embodiments. These processes can be completed by relevant hardware instructed by a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the foregoing method embodiments. The foregoing storage medium includes: various media that can store program codes such as ROM or random access memory RAM, magnetic disks, or optical discs.

Claims

1. An optical film surface defect detection method, characterized in that Including: Obtain an edge detection image; Perform connected component segmentation on the edge detection image to obtain connected components; Calculate the shape features of the connected components; the shape features include circularity, rectangularity, and aspect ratio; Compare the shape features of every two connected components to determine whether the shape features are similar; the shape features are considered similar when the difference in circularity is within a preset circularity difference threshold range, the difference in rectangularity is within a preset rectangularity difference threshold range, and the difference in aspect ratio is within a preset aspect ratio difference threshold range; When two connected components are similar connected components, merge the similar connected components into a merged connected component; The similar connected components are two connected components with similar shape features and a distance less than a preset distance threshold; Calculate the merged gray value distribution feature and the merged gradient change feature of the merged connected component, and input the merged gray value distribution and the merged gradient change feature into a single defect feature model to obtain a merged single defect error value; the single defect feature model is obtained by training a preset neural network using multiple groups of historical gray value distribution features, historical gradient change features, and corresponding marked single defect error values; Split the merged connected component with the merged single defect error greater than a preset single defect error threshold into the original similar connected components as defect connected components; the defect connected components also include the merged connected components with the merged single defect error not greater than the preset single defect error threshold and non-similar connected components; the non-similar connected components are the connected components in the edge detection image that cannot form similar connected components with all other connected components; Construct a defect type feature library and perform defect type matching on all defect connected components in the edge detection image.

2. The method according to claim 1, wherein The performing connected component segmentation on the edge detection image to obtain connected components specifically includes: Calculate the gray average value and the gray standard deviation of the local area of each pixel point in the edge detection image; the local area of the pixel point is the area covered by sliding a window of a preset size centered on the pixel point; Determine a defect judgment threshold through the gray average value and the gray standard deviation; Connect the adjacent pixel points in all pixel points of the edge detection image whose gray average value is greater than the defect judgment threshold to obtain connected components.

3. The method according to claim 2, characterized in that, The calculating the gray average value and the gray standard deviation of the local area of each pixel point in the edge detection image specifically includes: Sequentially select each pixel point from the upper left corner of the edge detection image as the central pixel point in row-major order; For the current central pixel point, create an empty data group, and store the pixel gray values in the area covered by sliding a window of a preset size centered on the current central pixel point into the data group; Use the first formula to calculate the gray average value of the local area of the current central pixel point; Wherein, the first formula is: Use the second formula to calculate the gray standard deviation of the local area of the current central pixel point; Wherein, the second formula is: Among them, is the average gray value of the local area of the current central pixel point, σ is the standard deviation of the gray value of the local area of the current central pixel point, the size of the window with the preset size is m×n. In the edge detection image, a coordinate axis is established with the upper left fixed point as the origin. The direction from the upper left fixed point to the lower left fixed point is the x-axis, and the direction from the upper left fixed point to the upper right fixed point is the y-axis. The coordinate axis plane is divided into squares with a size of one pixel unit. The coordinate position of the current central pixel point in the coordinate axis is (i, j).

4. The method according to claim 1, wherein The constructing a defect type feature library and performing defect type matching on all defect connected components in the edge detection image specifically includes: Collect samples of various known types of optical film defects. For the sample connected regions of each optical film defect sample, extract the sample feature data of the sample connected regions; the sample feature data is the sample gray value distribution feature, sample gradient change feature, sample circularity, and sample rectangularity of the sample connected regions. Associate and store the sample feature data with the sample defect types of the corresponding sample connected regions to construct a defect type feature library. Calculate the defect feature data of the defect connected regions; the defect feature data includes the defect gray value distribution feature, defect gradient change feature, defect circularity, and defect rectangularity. Calculate the similarity between the defect feature data and each sample feature data in the defect type feature library to obtain a similarity value group. Take the maximum defect type as the defect type matching the defect connected region; the maximum defect type is the defect type of the sample connected region corresponding to the maximum similarity data in the similarity value group.

5. The method according to claim 4, wherein After calculating the similarity between the defect feature data and each sample feature data in the defect type feature library to obtain a similarity value group, it further includes: in the case where all the similarity data in the similarity data group are lower than a preset minimum similarity threshold, mark the defect connected region as an unknown type connected region.

6. The method according to claim 4, characterized in that, After the defect type of the defect connected region is matched as the maximum defect type, it further includes: Calculate the area of the defect connected region of the defect connected region. Based on the area of the defect connected region, the defect gray mean value in the defect gray value distribution feature, and the defect aspect ratio, determine the defect level of the defect connected region using a preset level division threshold.

7. The method according to claim 1, wherein After splitting the merged connected regions with a merged single defect error greater than a preset single defect error threshold into the original similar connected regions as defect connected regions, it further includes: Determine the center position of the defect connected region as the connected region center. Calculate the connected region center distance between every two connected region centers and the connected region center density of the edge detection image. In the case where the short distance ratio value in the connected region center distances in the edge detection image is greater than a preset short distance ratio threshold or the connected region center density is greater than a preset maximum density threshold, mark the optical film corresponding to the edge detection image as a high-defect optical film; the short distance ratio value is the ratio of the connected region center distances less than a preset distance threshold to all the connected region center distances in the edge detection image.

8. An optical film surface defect detection system, characterized in that, The optical film surface defect detection system includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the optical film surface defect detection system to execute the method according to any one of claims 1 - 7.

9. A computer program product comprising instructions, characterized in that, When the computer program product runs on the optical film surface defect detection system, the optical film surface defect detection system is caused to execute the method according to any one of claims 1-7.

10. A computer-readable storage medium, comprising instructions, characterized in that, When the instruction runs on the optical film surface defect detection system, the optical film surface defect detection system is caused to execute the method according to any one of claims 1-7.

Images