A detection method and system for in-cell and edge defects

Abstract

The present invention relates to a method and system for detecting defects in the cell surface and edges, including the following steps: obtaining a first image of a liquid crystal display screen; determining an edge region to be detected on the first image according to the first image; detecting the edge of the liquid crystal display screen according to the edge region to be detected to obtain a first detection result; obtaining a second image of the liquid crystal display screen; determining a panel display region on the second image according to the second image; determining feature points meeting set conditions according to the panel display region; determining a display region to be detected according to the feature points meeting the set conditions; detecting the cell surface of the liquid crystal display screen according to the display region to be detected to obtain a second detection result. The problem of poor generality of traditional liquid crystal display screen detection methods is solved.

Description

Background Art

[0002] Liquid crystal displays have been developed for more than two decades. From the laboratory to large-scale production, they have been applied in almost all fields. With the rapid development of the liquid crystal display industry, the quality requirements for liquid crystal panels are getting higher and higher, and visual inspection equipment has become an indispensable part of the liquid crystal panel production process.

[0003] Among them, in the cell manufacturing process, mechanical cutting and grinding operations need to be performed on the produced liquid crystal panels, which are likely to cause defects to the liquid crystal panels, especially the edge parts of the liquid crystal panels. As the middle section of the entire liquid crystal production, the cell section needs to detect as many defective screens as possible to avoid defective products flowing to the downstream. Therefore, there are extremely high requirements for the reliability and efficiency of the detection equipment.

[0004] At present, the appearance defect method based on machine vision has widely replaced manual visual inspection and is applied in various industrial fields. Traditional defect detection methods based on machine vision usually use traditional image processing algorithms or manually designed defect features plus classifiers to detect and identify appearance defects. Machine vision algorithms have achieved good results in some specific applications, but there are still many deficiencies, such as: the image preprocessing steps are numerous and highly targeted, the algorithm iteration speed is slow, and the generality is poor. Summary of the Invention

[0005] In order to overcome the problem of poor generality of traditional liquid crystal panel detection methods, the present invention provides a detection method and system for in-plane and edge defects of the cell.

[0006] In the first aspect, to solve the above technical problems, the present invention provides a detection method for in-plane and edge defects of the cell, including the following steps:

[0007] Obtain a first image of the liquid crystal panel, where the first image is an image obtained by scanning the liquid crystal panel with a line scan camera having a resolution lower than a first set value;

[0008] According to the first image, determine an edge region to be detected on the first image, where the edge region to be detected includes at least one of a grinding region, a corner region, a marking point region, and a defect region. The grinding region is a region on the liquid crystal panel with grinding marks, the corner region is a region at the corners of the liquid crystal panel, the marking point region is a region where artificial marking points are drawn on the liquid crystal panel, and the defect region is a region on the liquid crystal panel with breakage, spots, or scratches;

[0009] According to the edge region to be detected, detect the edge of the liquid crystal panel to obtain a first detection result;

[0010] Obtain a second image for the liquid crystal screen, where the second image is an image obtained by scanning the liquid crystal screen with a line scan camera having a resolution lower than a second set value, and a first set value is less than the second set value;

[0011] According to the second image, determine the panel display area on the second image, where the panel display area is the area on the liquid crystal screen excluding the edge part and used for displaying the picture;

[0012] According to the panel display area, determine the feature points that meet the set conditions, where the set conditions are the pixel points in the panel display area with a pixel value greater than the area standard deviation or a gray value greater than the gray threshold;

[0013] According to the feature points that meet the set conditions, determine the display area to be detected, where the display area to be detected is the area with defects on the panel display area;

[0014] According to the display area to be detected, detect the cell surface of the liquid crystal screen to obtain a second detection result.

[0015] The beneficial effect of a detection method for in-cell and edge defects provided by the present invention is as follows: Through the first image, determine the edge area to be detected, so as to detect the edge of the liquid crystal screen through the edge area to be detected. Through the second image, determine the panel display area, and then extract the area with defects on the panel display area to obtain the display area to be detected, so as to detect the cell surface of the liquid crystal screen through the display area to be detected. For different liquid crystal screens, this method only needs to replace the line scan camera for obtaining the first image and the second image, and has strong versatility, solving the problem of poor versatility of traditional liquid crystal screen detection methods.

[0016] On the basis of the above technical solution, a detection method for in-cell and edge defects of the present invention can be further improved as follows.

[0017] Further, the above-mentioned determining the edge area to be detected on the first image according to the first image includes:

[0018] According to the first image, through a deep learning segmentation algorithm, determine the edge area to be detected on the first image, where the deep learning segmentation algorithm is the Bisenet-v2 segmentation network.

[0019] The beneficial effect of adopting the above further solution is as follows: Through the deep learning segmentation algorithm, it is possible to automatically extract features for a variety of defects, and has a good detection rate for defect features. Therefore, using deep learning, extract the defect features on the first image to obtain the edge area to be detected.

[0020] Further, when the edge area to be detected includes a grinding area, it further includes:

[0021] Determine the width of the grinding marks in the grinding area according to the grinding area;

[0022] Detect the edge of the liquid crystal display screen according to the edge area to be detected, and obtain a first detection result, including:

[0023] If the width of the grinding marks is greater than a preset width, it is determined that the first detection result is that there is a defect in the edge of the liquid crystal display screen.

[0024] The beneficial effect of adopting the above further solution is that when the edge area to be detected is the grinding area, the edge of the liquid crystal display screen is detected according to the width of the grinding marks.

[0025] Further, when the edge area to be detected includes a corner area, it further includes:

[0026] Construct a preset coordinate system according to the edge area to be detected;

[0027] Determine the abscissa and ordinate of the corner point in the corner area on the preset coordinate system according to the corner area;

[0028] Detect the edge of the liquid crystal display screen according to the edge area to be detected, and obtain a first detection result, including:

[0029] If the abscissa is greater than a first threshold value or the ordinate is greater than a second threshold value, it is determined that the first detection result is that there is a defect in the edge of the liquid crystal display screen.

[0030] The beneficial effect of adopting the above further solution is that when the edge area to be detected is the corner area, the edge of the liquid crystal display screen is detected according to the positions of the abscissa and ordinate of the corner point in the corner area on the preset coordinate system.

[0031] Further, when the edge area to be detected includes a marked point area, it further includes:

[0032] Determine the edge line according to the marked point area by using a line fitting algorithm;

[0033] Determine the distance between the marked point and the edge line according to the edge line and the position of the marked point in the marked point area;

[0034] Detect the edge of the liquid crystal display screen according to the edge area to be detected, and obtain a first detection result, including:

[0035] If the distance is greater than a third threshold value, it is determined that the first detection result is that there is a defect in the edge of the liquid crystal display screen.

[0036] The beneficial effect of adopting the above further solution is that when the edge area to be detected is the marked point area, the edge of the liquid crystal display screen is detected by the distance between the edge line and the marked point.

[0037] Further, the above-mentioned detection of the edge of the liquid crystal display screen based on the edge area to be detected to obtain the first detection result includes:

[0038] If the edge area to be detected includes a defective area, it is determined that the edge of the liquid crystal display screen has a defect in the first detection result.

[0039] The beneficial effect of adopting the above further solution is that when the edge detection area is a defective area, it is directly determined that the edge of the liquid crystal display screen has a defect.

[0040] Further, the above-mentioned determination of the feature points that meet the set conditions according to the panel display area includes:

[0041] According to the panel display area, the regional standard deviation is determined through the first formula, where the first formula is:

[0042]

[0043] where δ represents the regional standard deviation, x i represents the i-th pixel point in the panel display area, N represents the total number of pixel points in the panel display area, and l represents the panel display area;

[0044] The pixel points in the panel display area whose pixel values are greater than the regional standard deviation are determined as the feature points that meet the set conditions.

[0045] The beneficial effect of adopting the above further solution is that the regional standard deviation is determined through the first formula. When the pixel value of a pixel point in the panel display area is greater than the regional standard deviation, it indicates that the pixel point is abnormal, and this pixel point is used as the feature point that meets the conditions.

[0046] Further, the above-mentioned detection of the cell surface of the liquid crystal display screen based on the display area to be detected to obtain the second detection result includes:

[0047] According to the display area to be detected, the classification label is determined through the deep learning classification algorithm. The classification label is background class, damage class, spot class, scratch class or water stain class;

[0048] When the classification label is damage class, spot class, scratch class or water stain class, it is determined that the cell surface of the liquid crystal display screen has a defect in the second detection result.

[0049] The beneficial effect of adopting the above further solution is that the deep learning classification algorithm does not need to manually extract features and can directly determine the classification of the features of the display area to be detected. Therefore, the features of the display area to be detected are directly classified by the deep learning classification algorithm to obtain the classification label, and the cell surface is detected through the classification label.

[0050] In a second aspect, the present invention provides a detection system for in-cell and edge defects of a cell, including:

[0051] A first image acquisition module, configured to acquire a first image of a liquid crystal display screen, where the first image is an image obtained by scanning the liquid crystal display screen with a line-scan camera having a resolution lower than a first set value;

[0052] An edge region to be detected acquisition module, configured to determine an edge region to be detected on the first image according to the first image, where the edge region to be detected includes at least one of a grinding region, a corner region, a marking point region, and a defect region. The grinding region is a region on the liquid crystal display screen where there are grinding marks, the corner region is a region at the corners of the liquid crystal display screen, the marking point region is a region where artificial marking points are drawn on the liquid crystal display screen, and the defect region is a region on the liquid crystal display screen where there are breakages, spots, or scratches;

[0053] A first detection result acquisition module, configured to detect the edge of the liquid crystal display screen according to the edge region to be detected, and obtain a first detection result;

[0054] A second image acquisition module, configured to acquire a second image of the liquid crystal display screen, where the second image is an image obtained by scanning the liquid crystal display screen with a line-scan camera having a resolution lower than a second set value, and where the first set value is less than the second set value;

[0055] A panel display region acquisition module, configured to determine a panel display region on the second image according to the second image, where the panel display region is a region on the liquid crystal display screen excluding the edge part and used for displaying images;

[0056] A feature point acquisition module, configured to determine feature points meeting set conditions according to the panel display region, where the set conditions are pixel points in the panel display region whose pixel values are greater than the region standard deviation or whose gray values are greater than the gray threshold;

[0057] A display region to be detected acquisition module, configured to determine a display region to be detected according to the feature points meeting the set conditions, where the display region to be detected is a region on the panel display region where there are defects;

[0058] A second detection result acquisition module, configured to detect the cell surface of the liquid crystal display screen according to the display region to be detected, and obtain a second detection result.

[0059] In a third aspect, the present invention further provides an electronic device, including a memory, a processor, and a program stored in the memory and running on the processor. When the processor executes the program, the steps of a detection method for in-cell and edge defects of a cell as described above are implemented.

[0060] Fourthly, the present invention further provides a computer-readable storage medium, in which instructions are stored. When the instructions run on a terminal device, the terminal device is caused to execute the steps of a method for detecting defects in the cell surface and at the edge. Description of the Drawings

[0061] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be further described below with reference to the drawings and embodiments.

[0062] Figure 1 It is a flowchart of a method for detecting defects in the cell surface and at the edge according to an embodiment of the present invention;

[0063] Figure 2 It is a flowchart for liquid crystal screen detection;

[0064] Figure 3 It is the first image corresponding to the grinding area;

[0065] Figure 4 It is the area to be detected at the edge corresponding to the grinding area output by the Bisenet-v2 segmentation network;

[0066] Figure 5 It is the first image corresponding to the defect area;

[0067] Figure 6 It is the area to be detected at the edge corresponding to the defect area output by the Bisenet-v2 segmentation network;

[0068] Figure 7 It is a schematic structural diagram of a system for detecting defects in the cell surface and at the edge according to an embodiment of the present invention. Detailed Embodiments

[0069] The following embodiments are further explanations and supplements to the present invention and do not constitute any limitation to the present invention.

[0070] The following describes a method and a system for detecting defects in the cell surface and at the edge according to an embodiment of the present invention with reference to the drawings.

[0071] A method for detecting defects in the cell surface and at the edge according to an embodiment of the present invention is applied to a terminal device. In the solution of the present application, the terminal device is used as the execution subject to illustrate the solution of the present application. The terminal device can be a computer, a server, etc., and is used to execute the steps of a method for detecting defects in the cell surface and at the edge. The terminal device is also connected to each line scan camera, and each line scan camera is used to obtain the first image and the second image of the liquid crystal screen.

[0072] Among them, the detection of LCD screen mainly includes two aspects, one is the edge detection of LCD screen, the other is the detection of cell surface, that is, the display area of ​​the panel.

[0073] Optionally, since the edge detection of the LCD screen requires the measurement of the parameters of each edge (the parameters include the width of the grinding marks in the grinding area, the horizontal and vertical coordinates of the corners in the preset coordinate system, the position of the edge line and the marking point) and defect detection, the detection time is relatively long. Based on this, two channels can be set to perform edge detection at the same time to ensure that each edge of the LCD screen can be detected while shortening the detection time, such as Figure 1 As shown, the LCD screen 1 enters channel A and channel B for edge detection respectively (in order to shorten the detection time, when one channel A performs edge detection on the LCD screen, the other channel B can perform edge detection on another LCD screen, and then the LCD screens of channel A and channel B are exchanged). In channel A, the first image of the LCD screen 1 is first acquired by line scan cameras 2-1 and 2-2, and the first edge of the LCD screen 1 is detected by a method for detecting in-cell surface and edge defects provided by the present application. Then, by rotating, the first image of the LCD screen 1 is acquired by line scan cameras 3-1 and 3-2, and the second edge of the LCD screen 1 is detected by a method for detecting in-cell surface and edge defects provided by the present application. By analogy, the LCD screen 1 enters channel A and channel B for edge detection. Enter channel B, use line scan cameras 4-1 and 4-2 to obtain the first image of the LCD screen 1, and use a method for detecting in-cell surface and edge defects provided in the present application to detect the third edge of the LCD screen 1, then rotate, use line scan cameras 5-1 and 5-2 to obtain the first image of the LCD screen 1, and use a method for detecting in-cell surface and edge defects provided in the present application to detect the fourth edge of the LCD screen 1. At this point, all four edges of the LCD screen 1 are detected, and finally, use line scan cameras 6-1 and 6-2 to obtain the second image of the LCD screen 1, and use a method for detecting in-cell surface and edge defects provided in the present application to detect the cell surface of the LCD screen 1, thereby completing all inspections of the LCD screen.

[0074] like Figure 2 As shown, the present invention provides a method for detecting defects in a cell surface and at an edge, comprising the following steps:

[0075] S1, acquiring a first image for a liquid crystal screen, where the first image is an image obtained by scanning the liquid crystal screen with a line scan camera having a resolution lower than a first set value;

[0076] S2. Based on the first image, determine the edge region to be detected on the first image. The edge region to be detected includes at least one of a grinding region, a corner region, a marking point region, and a defect region. The grinding region is the region on the liquid crystal screen with grinding marks. The corner region is the region at the corners of the liquid crystal screen. The marking point region is the region where artificial marking points are drawn on the liquid crystal screen. The defect region is the region on the liquid crystal screen with breakage, spots, or scratches.

[0077] S3. Detect the edge of the liquid crystal screen according to the edge region to be detected, and obtain the first detection result.

[0078] S4. Obtain a second image of the liquid crystal screen. The second image is an image obtained by scanning the liquid crystal screen with a line scan camera having a resolution lower than a second set value, where the first set value is less than the second set value.

[0079] S5. Based on the second image, determine the panel display region on the second image. The panel display region is the region on the liquid crystal screen excluding the edge part and used for displaying the picture.

[0080] S6. Determine the feature points meeting the set conditions according to the panel display region. The set condition is that the pixel value in the panel display region is greater than the region standard deviation or the gray value is greater than the gray threshold.

[0081] S7. Determine the display region to be detected according to the feature points meeting the set conditions. The display region to be detected is the region with defects on the panel display region.

[0082] S8. Detect the cell surface of the liquid crystal screen according to the display region to be detected, and obtain the second detection result.

[0083] Optionally, the above step of determining the edge region to be detected on the first image based on the first image includes:

[0084] Based on the first image, determine the edge region to be detected on the first image through a deep learning segmentation algorithm, where the deep learning segmentation algorithm is the Bisenet-v2 segmentation network.

[0085] In this embodiment, the Bisenet-v2 segmentation network includes a spatial path with a small stride for retaining spatial position information to generate a high-resolution feature map. After the spatial path, there is a semantic path with a fast downsampling rate for obtaining the feature map of the objective receptive field of the original image. It also includes a feature fusion module that fuses the outputs of the spatial path and the semantic path and outputs the final result to achieve a balance between speed and accuracy.

[0086] Based on the above, the first image is input into the Bisenet-v2 segmentation network, and an edge region to be detected is output. The edge region to be detected includes at least one of a grinding region, a corner region, a marking point region, and a defect region. For example, as Figure 3 shown, the first image is input into the Bisenet-v2 segmentation network, and an edge region to be detected corresponding to the grinding region as shown in Figure 4 is obtained. At this time, the line width in Figure 4 is the width of the grinding trace. According to the width of the grinding trace, the edge of the liquid crystal display screen can be detected.

[0087] Optionally, when the edge region to be detected includes a grinding region, it further includes:

[0088] Determine the width of the grinding trace in the grinding region according to the grinding region;

[0089] Detect the edge of the liquid crystal display screen according to the edge region to be detected, and obtain a first detection result, including:

[0090] If the width of the grinding trace is greater than a preset width, it is determined that there is a defect in the edge of the liquid crystal display screen in the first detection result.

[0091] In this embodiment, during the production process of the liquid crystal display screen, a grinding process is involved to polish the edge part of the liquid crystal display screen to make the edge smoother. Due to its own defects or defects in the grinding equipment, some liquid crystal display screens will leave grinding traces during the grinding process, making the liquid crystal display screen fail to meet the qualified standard. Therefore, it is necessary to detect the grinding traces. As Figure 4 shown, it is the grinding trace obtained through the Bisenet-v2 segmentation network ( Figure 4 a line in the middle part). If the width of the grinding trace is greater than the preset width, it is determined that there is a defect in the edge of the liquid crystal display screen in the first detection result. If the width of the grinding trace is less than the preset width, it is determined that the edge detection of the liquid crystal display screen is qualified.

[0092] Optionally, the preset width is the specification requirement customized by the customer.

[0093] Optionally, when the edge region to be detected includes a corner region, it further includes:

[0094] Construct a preset coordinate system according to the edge region to be detected;

[0095] Determine the abscissa and ordinate of the corner of the corner region on the preset coordinate system according to the corner region;

[0096] Detect the edge of the liquid crystal display screen according to the edge region to be detected, and obtain a first detection result, including:

[0097] If the abscissa is greater than the first threshold or the ordinate is greater than the second threshold, it is determined that there is a defect at the edge of the liquid crystal screen in the first detection result.

[0098] Optionally, since the image specifications of the first image are the same, the image specifications of the obtained corner regions are the same, and the placement positions of the liquid crystal screens in the channel are the same, therefore, the corners of different liquid crystal screens should be in the same position in the corner region. Based on this, a preset coordinate system is constructed. If the abscissa is greater than the first threshold or the ordinate is greater than the second threshold, it is determined that there is a defect at the edge of the liquid crystal screen in the first detection result. If the abscissa is less than the first threshold and the ordinate is less than the second threshold, it is determined that the edge detection of the liquid crystal screen is qualified in the first detection result.

[0099] Optionally, both the first threshold and the second threshold are customer-defined specifications.

[0100] Optionally, when the edge region to be detected includes a marked point region, it further includes:

[0101] According to the marked point region, determine the edge line through a line fitting algorithm;

[0102] According to the position of the edge line and the marked points in the marked point region, determine the distance between the marked points and the edge line;

[0103] According to the edge region to be detected, detect the edge of the liquid crystal screen to obtain the first detection result, including:

[0104] If the distance is greater than the third threshold, it is determined that there is a defect at the edge of the liquid crystal screen in the first detection result.

[0105] In this embodiment, the line fitting algorithm can adopt the RANSAC method. The edge line is mainly determined by the following method, and the following method includes:

[0106] S11, according to the edge line region, randomly select two points, and determine a line l from these two points;

[0107] S12, according to the threshold t, determine the geometric distance between the remaining points in the edge line region and the line l, and take the points with the geometric distance less than t as the data point set S(l), and call it the consensus set of the line l. Among them, each point in the data point set S(l) is called an inlier;

[0108] S13, repeat the random selection several times, that is, S11 - S12, to obtain the consensus sets S(l1), S(l2),..., S(ln) corresponding to the lines l1, l2,..., ln;

[0109] S14, take the data point set with the largest number of inliers in each consensus set as the target data point set, and use the inliers in the target data point set to fit a line by the least squares method as the edge line.

[0110] Optionally, if the distance between the marked point and the edge line is greater than the third threshold, it is determined that there is a defect in the edge of the liquid crystal screen in the first detection result; if the distance between the marked point and the edge line is less than the third threshold, it is determined that the first detection result is that the edge detection of the liquid crystal screen is qualified.

[0111] Optionally, the third threshold is a custom specification of the customer.

[0112] Optionally, the above-mentioned detection of the edge of the liquid crystal screen according to the edge area to be detected to obtain the first detection result includes:

[0113] If the edge area to be detected includes a defective area, it is determined that there is a defect in the edge of the liquid crystal screen in the first detection result.

[0114] Optionally, if the edge detection area is a defective area, it is directly determined that there is a defect in the edge of the liquid crystal screen in the first detection result. As Figure 5 shown, the first image corresponding to the defective area is obtained. The first image is input into the Bisenet-v2 segmentation network to obtain the edge area to be detected corresponding to the defective area as Figure 6 shown. At this time, a defect can be clearly seen from the Figure 6 solid color background, that is, it is determined that there is a defect in the edge of the liquid crystal screen in the first detection result.

[0115] Optionally, when there are interferences such as water stains on the liquid crystal screen, the water stains will affect the recognition of the first image by the Bisenet-v2 segmentation network, resulting in misjudgment. Therefore, a deep learning classification algorithm can be used to screen the output result of the Bisenet-v2 segmentation network again to remove the interference images and improve the accuracy of the edge detection of the liquid crystal screen.

[0116] Optionally, the above-mentioned determination of the feature points that meet the set conditions according to the panel display area includes:

[0117] According to the panel display area, the regional standard deviation is determined through the first formula, where the first formula is:

[0118]

[0119] where δ represents the regional standard deviation, x i represents the i-th pixel point in the panel display area, N represents the total number of pixel points in the panel display area, and l represents the panel display area;

[0120] The pixel points with pixel values greater than the regional standard deviation among the pixel points in the panel display area are determined as the feature points that meet the set conditions.

[0121] Optionally, the grayscale threshold is a custom specification of the customer. The grayscale value of the panel display area is a value from 0 to 255. The display area to be detected can also be determined through the second formula, where the second formula is:

[0122]

[0123] where g(x, y) is the display area to be detected, f(x, y) is the grayscale value of the pixel points in the panel display area, and T represents the grayscale threshold.

[0124] Optionally, since the background of the panel display area is relatively simple, if there are defects, the pixel values and grayscale values of its image will surely be abnormal. Therefore, by using the above first formula and second formula, the regional standard deviation and grayscale threshold can be determined, and the display area to be detected can be determined through the regional standard deviation and grayscale threshold.

[0125] Optionally, the above-mentioned detection of the cell surface of the liquid crystal screen according to the display area to be detected obtains a second detection result, including:

[0126] According to the display area to be detected, the classification label is determined through a deep learning classification algorithm. The classification labels are background class, breakage class, spot class, scratch class, or water stain class;

[0127] When the classification label is breakage class, spot class, scratch class, or water stain class, it is determined that the second detection result indicates that there are defects on the cell surface of the liquid crystal screen.

[0128] Optionally, the deep learning classification algorithm uses the repvgg model of the CNN convolutional neural network. Inputting the display area to be detected into the repvgg model can obtain the classification label. Among them, the background class means that the display area to be detected is a background image, that is, the second detection result indicates that there are defects on the cell surface of the liquid crystal screen. The breakage class means that there are breakages in the display area to be detected, that is, the second detection result indicates that there are defects on the cell surface of the liquid crystal screen. The spot class means that there are spots in the display area to be detected, that is, the second detection result indicates that there are defects on the cell surface of the liquid crystal screen. The scratch class means that there are scratches in the display area to be detected, that is, the second detection result indicates that there are defects on the cell surface of the liquid crystal screen. The water stain class means that there are water stains in the display area to be detected, that is, the second detection result indicates that there are defects on the cell surface of the liquid crystal screen.

[0129] As Figure 7 shown, the embodiment of the present invention also provides a detection system for defects inside and at the edge of the cell surface, including:

[0130] The first image acquisition module 201 is used to acquire a first image of the liquid crystal screen. The first image is an image obtained by scanning the liquid crystal screen with a line-scan camera whose resolution is lower than the first set value;

[0131] The edge area to be detected acquisition module 202 is configured to determine the edge area to be detected on the first image according to the first image. The edge area to be detected includes at least one of a grinding area, a corner area, a marking point area, and a defect area. The grinding area is the area on the liquid crystal screen with grinding marks. The corner area is the area at the corners of the liquid crystal screen. The marking point area is the area where artificial marking points are drawn on the liquid crystal screen. The defect area is the area on the liquid crystal screen with breakage, spots, or scratches.

[0132] The first detection result acquisition module 203 is configured to detect the edge of the liquid crystal screen according to the edge area to be detected and obtain a first detection result.

[0133] The second image acquisition module 204 is configured to acquire a second image of the liquid crystal screen. The second image is an image obtained by scanning the liquid crystal screen with a line scan camera having a resolution lower than a second set value, where a first set value is less than the second set value.

[0134] The panel display area acquisition module 205 is configured to determine the panel display area on the second image according to the second image. The panel display area is the area on the liquid crystal screen excluding the edge part and used for displaying images.

[0135] The feature point acquisition module 206 is configured to determine feature points meeting set conditions according to the panel display area. The set conditions are pixel points in the panel display area with a pixel value greater than the area standard deviation or a gray value greater than the gray threshold.

[0136] The display area to be detected acquisition module 207 is configured to determine the display area to be detected according to the feature points meeting the set conditions. The display area to be detected is the area with defects on the panel display area.

[0137] The second detection result acquisition module 208 is configured to detect the cell surface of the liquid crystal screen according to the display area to be detected and obtain a second detection result.

[0138] Optionally, the above-mentioned edge area to be detected acquisition module 202 is specifically configured to:

[0139] According to the first image, determine the edge area to be detected on the first image through a deep learning segmentation algorithm, where the deep learning segmentation algorithm is the Bisenet-v2 segmentation network.

[0140] Optionally, the above-mentioned first detection result acquisition module 203 further includes:

[0141] The grinding mark acquisition module is configured to determine the width of the grinding marks in the grinding area according to the grinding area.

[0142] Then the first detection result acquisition module 203 is specifically configured to:

[0143] If the width of the grinding mark is greater than the preset width, it is determined that the first detection result is that there is a defect at the edge of the liquid crystal screen.

[0144] Optionally, the first detection result acquisition module 203 further includes:

[0145] A preset coordinate system acquisition module, configured to construct a preset coordinate system according to the edge area to be detected;

[0146] A coordinate acquisition module, configured to determine the abscissa and ordinate of the corner point in the corner area on the preset coordinate system according to the corner area;

[0147] Then the first detection result acquisition module 203 is specifically configured to:

[0148] If the abscissa is greater than the first threshold or the ordinate is greater than the second threshold, it is determined that the first detection result is that there is a defect at the edge of the liquid crystal screen.

[0149] Optionally, the first detection result acquisition module 203 further includes:

[0150] An edge line acquisition module, configured to determine an edge line according to the marked point area through a line fitting algorithm;

[0151] A distance acquisition module, configured to determine the distance between the marked point and the edge line according to the position of the edge line and the marked point in the marked point area;

[0152] Then the first detection result acquisition module 203 is specifically configured to:

[0153] If the distance is greater than the third threshold, it is determined that the first detection result is that there is a defect at the edge of the liquid crystal screen.

[0154] Optionally, the first detection result acquisition module 203 is specifically configured to:

[0155] If the edge area to be detected includes a defect area, it is determined that the first detection result is that there is a defect at the edge of the liquid crystal screen.

[0156] Optionally, the feature point acquisition module 206 specifically includes:

[0157] A regional standard deviation acquisition module, configured to determine a regional standard deviation according to the panel display area through a first formula, where the first formula is:

[0158]

[0159] where δ represents the regional standard deviation, x i represents the i-th pixel point in the panel display area, N represents the total number of pixel points in the panel display area, and l represents the panel display area;

[0160] The first judgment module is configured to determine, as feature points meeting the set conditions, the pixel points in the panel display area whose pixel values are greater than the regional standard deviation.

[0161] Optionally, the above-mentioned second detection result acquisition module 208 specifically includes:

[0162] The classification label module is configured to determine classification labels, which are background class, breakage class, spot class, scratch class or water stain class, according to the area to be detected for display through a deep learning classification algorithm.

[0163] The second judgment module is configured to judge that the second detection result indicates that there are defects on the cell surface of the liquid crystal screen when the classification label is the breakage class, the spot class, the scratch class or the water stain class.

[0164] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a program stored on the memory and running on the processor. When the processor executes the program, it implements some or all of the steps of the above-mentioned method for detecting defects inside and at the edge of the cell surface.

[0165] Among them, the electronic device can be a computer. Correspondingly, its program is computer software. Moreover, for the various parameters and steps in the above-mentioned electronic device of the present invention, reference can be made to the various parameters and steps in the embodiment of the method for detecting defects inside and at the edge of the cell surface in the foregoing text, which will not be elaborated herein.

[0166] Those skilled in the art of the present technology know that the present invention can be implemented as a system, a method or a computer program product. Therefore, the present disclosure can be specifically implemented in the following forms, that is: it can be completely hardware, can also be completely software (including firmware, resident software, microcode, etc.), and can also be in the form of a combination of hardware and software, which is generally referred to as "circuit", "module" or "system" in this article. In addition, in some embodiments, the present invention can also be implemented in the form of a computer program product in one or more computer-readable media, which contains computer-readable program codes. The computer-readable storage medium can be, for example, but not limited to - an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above.

[0167] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "examples", "specific examples", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0168] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A detection method for in-cell and edge defects of a cell, characterized in that, it includes the following steps: Obtain a first image of the liquid crystal screen, where the first image is an image obtained by scanning the liquid crystal screen with a line-scan camera having a resolution lower than a first set value; According to the first image, determine the edge region to be detected on the first image, where the edge region to be detected includes at least one of a grinding region, a corner region, a marking point region, and a defect region. The grinding region is the region on the liquid crystal screen where grinding marks exist, the corner region is the region at the corners of the liquid crystal screen, the marking point region is the region where artificial marking points are drawn on the liquid crystal screen, and the defect region is the region on the liquid crystal screen where there are breaks, spots, or scratches; According to the edge region to be detected, detect the edge of the liquid crystal screen to obtain a first detection result; Obtain a second image of the liquid crystal screen, where the second image is an image obtained by scanning the liquid crystal screen with a line-scan camera having a resolution lower than a second set value, where the first set value is less than the second set value; According to the second image, determine the panel display region on the second image, where the panel display region is the region on the liquid crystal screen excluding the edge part and used for displaying the picture; According to the panel display region, determine the feature points meeting the set conditions, where the set conditions are the pixel points in the panel display region with pixel values greater than the regional standard deviation or gray values greater than the gray threshold; According to the feature points meeting the set conditions, determine the display region to be detected, where the display region to be detected is the region with defects on the panel display region; According to the display region to be detected, detect the cell surface of the liquid crystal screen to obtain a second detection result; The determining the feature points meeting the set conditions according to the panel display region includes: According to the panel display region, determine the regional standard deviation through a first formula, where the first formula is: Among them, δ represents the regional standard deviation, and x i represents the i-th pixel point in the panel display area, N represents the total number of pixel points in the panel display area, and l represents the panel display area; Determine the pixel points in the panel display region with pixel values greater than the regional standard deviation as the feature points meeting the set conditions.

2. The method according to claim 1, characterized in that, The determining the edge region to be detected on the first image according to the first image includes: According to the first image, determine the edge region to be detected on the first image through a deep learning segmentation algorithm, where the deep learning segmentation algorithm is the Bisenet-v2 segmentation network.

3. The method according to claim 1, characterized in that, When the edge region to be detected includes a grinding region, it further includes: According to the grinding region, determine the width of the grinding marks in the grinding region; The detecting the edge of the liquid crystal screen according to the edge region to be detected to obtain a first detection result includes: If the width of the grinding marks is greater than a preset width, then determine that the first detection result is that there are defects on the edge of the liquid crystal screen.

4. The method according to claim 1, characterized in that, When the edge region to be detected includes a corner region, it further includes: According to the edge region to be detected, construct a preset coordinate system; Determine the abscissa and ordinate of the corner of the corner region on the preset coordinate system according to the corner region; The edge of the liquid crystal screen is detected according to the edge region to be detected, and a first detection result is obtained, including: If the abscissa is greater than the first threshold or the ordinate is greater than the second threshold, it is determined that the first detection result is that there is a defect in the edge of the liquid crystal screen.

5. The method according to claim 1, characterized in that, when the edge region to be detected includes a marked point region, it further includes: Determine an edge line according to the marked point region by using a linear fitting algorithm; Determine the distance between the marked point and the edge line according to the position of the edge line and the marked point in the marked point region; The edge of the liquid crystal screen is detected according to the edge region to be detected, and a first detection result is obtained, including: If the distance is greater than the third threshold, it is determined that the first detection result is that there is a defect in the edge of the liquid crystal screen.

6. The method according to claim 1, characterized in that, The edge of the liquid crystal screen is detected according to the edge region to be detected, and a first detection result is obtained, including: If the edge region to be detected includes the defect region, it is determined that the first detection result is that there is a defect in the edge of the liquid crystal screen.

7. The method according to claim 1, characterized in that, The cell surface of the liquid crystal screen is detected according to the display region to be detected, and a second detection result is obtained, including: Determine a classification label according to the display region to be detected by using a deep learning classification algorithm, and the classification label is a background class, a breakage class, a spot class, a scratch class or a water stain class; When the classification label is the breakage class, the spot class, the scratch class or the water stain class, it is determined that the second detection result is that there is a defect in the cell surface of the liquid crystal screen.

8. A detection system for defects in the cell surface and at the edge, characterized in that, it includes: A first image acquisition module, configured to acquire a first image of a liquid crystal screen, where the first image is an image obtained by scanning the liquid crystal screen with a line scan camera having a resolution lower than a first set value; An edge region to be detected acquisition module, configured to determine an edge region to be detected on the first image according to the first image, where the edge region to be detected includes at least one of a grinding region, a corner region, a marked point region and a defect region, the grinding region is a region on the liquid crystal screen where there are grinding marks, the corner region is a region at the corner of the liquid crystal screen, the marked point region is a region where a marked point is manually drawn on the liquid crystal screen, and the defect region is a region on the liquid crystal screen where there are breakages, spots or scratches; A first detection result acquisition module, configured to detect the edge of the liquid crystal screen according to the edge region to be detected, and obtain a first detection result; A second image acquisition module, configured to acquire a second image of the liquid crystal screen, where the second image is an image obtained by scanning the liquid crystal screen with a line scan camera having a resolution lower than a second set value, and wherein the first set value is less than the second set value; A panel display area acquisition module, configured to determine a panel display area on the second image according to the second image, where the panel display area is an area on the liquid crystal screen excluding the edge part and used for displaying a picture; A feature point acquisition module, configured to determine feature points meeting set conditions according to the panel display area, where the set conditions are pixel points in the panel display area with a pixel value greater than the area standard deviation or a gray value greater than the gray threshold; A display area to be detected acquisition module, configured to determine a display area to be detected according to the feature points meeting the set conditions, where the display area to be detected is an area with defects on the panel display area; A second detection result acquisition module, configured to detect the cell surface of the liquid crystal screen according to the display area to be detected, and obtain a second detection result; The determining feature points meeting set conditions according to the panel display area includes: Determining an area standard deviation according to the panel display area through a first formula, where the first formula is: Among them, δ represents the regional standard deviation, and x i represents the i-th pixel point in the panel display area, N represents the total number of pixel points in the panel display area, and l represents the panel display area; Determining pixel points with a pixel value greater than the area standard deviation among the pixel points in the panel display area as feature points meeting the set conditions.

9. An electronic device, including a memory, a processor, and a program stored on the memory and running on the processor, wherein, when the processor executes the program, the steps of a detection method for in-cell and edge defects as described in any one of claims 1 to 7 are implemented.

Images