Hole edge microcrack defect detection method and system, electronic equipment and storage medium

Abstract

The invention relates to the technical field of panel detection, in particular to a hole edge microcrack defect detection method and system, electronic equipment and a storage medium, and the method comprises the steps: obtaining an original bright field image and an original dark field image of a to-be-detected liquid crystal screen hole edge under telecentric imaging; inputting the original bright field image and the original dark field image into an improved microcrack defect detection model to obtain a detection result of the hole edge of the liquid crystal screen to be detected; the method is combined with an improved microcrack defect detection model, efficient and accurate detection of the liquid crystal screen hole edge microcracks is achieved, the detection speed and accuracy are improved, and the false detection rate and the omission ratio are reduced.

Description

Background Art

[0002] In the LCD manufacturing industry, with the continuous advancement of display technology and the increasing quality requirements of consumers, the detection of LCD defects has become particularly important. Among them, micro cracks on the edge of the hole, as a common defect type, not only affect the appearance of the product, but also have a potential impact on the performance of the display.

[0003] Traditional methods for detecting microcracks on the edge of holes mostly rely on manual visual inspection or the use of simple image processing technology. Although manual visual inspection is intuitive, it has problems such as strong subjectivity, low efficiency, and easy fatigue, and it is difficult to meet the needs of large-scale production lines. Although simple image processing technology can identify cracks to a certain extent, due to the subtlety and complexity of microcracks, there are often problems of missed detection and high false detection rates. In response to the above technical difficulties, automatic detection solutions based on computer vision and deep learning have begun to emerge. However, directly applying deep learning models to LCD screen defect detection faces the disadvantages of complex image features in the edge area of ​​the LCD screen hole, limited feature extraction capabilities, small size of microcrack defects, low accuracy of defect identification and positioning capabilities, and the inability to guarantee both detection efficiency and detection accuracy.

[0004] Therefore, it is urgent to provide a technical solution to solve the above problems. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a method, system, electronic device and storage medium for detecting micro-crack defects at the edge of a hole.

[0006] In a first aspect, the present invention provides a method for detecting microcrack defects at the edge of a hole, and the technical solution of the method is as follows:

[0007] Obtaining the original bright field image and the original dark field image of the edge of the hole of the LCD screen to be tested under telecentric imaging;

[0008] The original bright field image and the original dark field image are input into the improved microcrack defect detection model to obtain the detection result of the edge of the liquid crystal screen hole to be tested; wherein the improved microcrack defect detection model includes: an ROI area extraction module, a semantic segmentation module and a connected domain merging module which are connected in sequence; the ROI area extraction module is used to: perform ROI area extraction on the target image and perform image segmentation to obtain a block ROI image; the semantic segmentation module is used to: perform defect prediction on the block ROI image to obtain a first defect prediction image; the connected domain merging module is used to: merge the connected domains of each first defect prediction image that meets the preset conditions to obtain a merged defect prediction image.

[0009] The beneficial effects of the hole edge microcrack defect detection method of the present invention are as follows:

[0010] The method of the present invention is combined with an improved microcrack defect detection model to achieve efficient and accurate detection of microcracks at the edge of a liquid crystal screen hole, which not only improves the detection speed and accuracy, but also reduces the false detection and missed detection rates.

[0011] Based on the above scheme, the hole edge microcrack defect detection method of the present invention can also be improved as follows.

[0012] In an optional manner, the step of obtaining an original bright field image and an original dark field image of the edge of the hole of the liquid crystal panel to be tested under telecentric imaging includes:

[0013] Using a camera, a telecentric lens and bright field illumination, the original bright field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging is obtained;

[0014] The original dark field image of the edge of the hole of the liquid crystal panel to be tested under telecentric imaging is obtained by using the camera, the telecentric lens and dark field illumination.

[0015] Wherein, the bright field lighting highlights the color and shape characteristics of the rainbow pattern; the dark field lighting highlights the microcracks and enhances the shape characteristics of the microcracks.

[0016] In the above optional method, by using a camera, a telecentric lens and different lighting conditions to obtain the original images respectively, the different features of the edge of the LCD screen hole, especially the shape features of the microcracks, can be more effectively highlighted, thereby improving the subsequent detection model's ability to recognize microcracks.

[0017] In an optional manner, the improved microcrack defect detection model further includes: a preprocessing module;

[0018] The preprocessing module is used to perform size adjustment processing on the original image to obtain the target image.

[0019] In the above optional manner, the preprocessing module adjusts the size of the original image, thereby ensuring the uniformity of the input image, reducing the detection error caused by different image sizes, and improving the stability and accuracy of the detection model.

[0020] In an optional manner, the step of extracting the ROI region of the target image and dividing the image into blocks to obtain a block ROI image includes:

[0021] The ROI region of the target image is acquired, and the ROI region is equally divided into rectangles of a preset size to obtain the divided ROI image containing a plurality of rectangular frame regions.

[0022] In the above optional method, the ROI area of ​​the target image is extracted and processed in blocks, which effectively reduces the amount of calculation and improves the detection efficiency. At the same time, it enables the microcrack defect detection model to focus more on key areas, further improving the accuracy of detection.

[0023] In an optional manner, the step of performing defect prediction on the block ROI image to obtain a first defect prediction image includes:

[0024] The block ROI images are grouped, and the Bisenet model is used to perform pixel-level classification prediction on the grouped block ROI images to obtain the first defect prediction image.

[0025] In the above optional method, by grouping and using the Bisenet model to perform pixel-level classification prediction on the block ROI image, the efficiency and accuracy of the Bisenet model can be used to quickly and accurately identify microcrack defects, thereby improving the detection speed and accuracy.

[0026] In an optional manner, the preset condition is: the distance between the bounding rectangles of any two connected domains in the first defect prediction image is less than a preset threshold, and if the types of defects corresponding to the any two connected domains are the same; wherein the bounding box function in the image processing library is used to calculate the bounding rectangle of the connected domain.

[0027] In the above optional manner, connected domains are merged by setting preset conditions, which effectively solves the problems of false detection and missed detection caused by image noise or slight errors, further reduces the false detection and missed detection rates, and improves the reliability of the detection results.

[0028] In a second aspect, the present invention provides a hole edge microcrack defect detection system, the technical solution of the system is as follows:

[0029] The hole edge microcrack defect detection system comprises: an acquisition unit and a detection unit;

[0030] The acquisition unit is used to: acquire an original bright field image and an original dark field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging;

[0031] The detection unit is used to: input the original bright field image and the original dark field image into the improved microcrack defect detection model to obtain the detection result of the edge of the liquid crystal screen hole to be tested; wherein the improved microcrack defect detection model includes: an ROI area extraction module, a semantic segmentation module and a connected domain merging module which are connected in sequence; the ROI area extraction module is used to: perform ROI area extraction on the target image and perform image segmentation to obtain a block ROI image; the semantic segmentation module is used to: perform defect prediction on the block ROI image to obtain a first defect prediction image; the connected domain merging module is used to: merge the connected domains of each first defect prediction image that meets the preset conditions to obtain a merged defect prediction image.

[0032] The beneficial effects of the hole edge microcrack defect detection system of the present invention are as follows:

[0033] The system of the present invention is combined with an improved microcrack defect detection model to achieve efficient and accurate detection of microcracks at the edge of liquid crystal screen holes, which not only improves the detection speed and accuracy, but also reduces the false detection and missed detection rates.

[0034] Based on the above scheme, the hole edge microcrack defect detection system of the present invention can also be improved as follows.

[0035] In an optional manner, the acquisition unit is specifically used to:

[0036] Using a camera, a telecentric lens and bright field illumination, the original bright field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging is obtained;

[0037] Using the camera, the telecentric lens and dark field illumination, the original dark field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging is obtained;

[0038] Wherein, the bright field lighting highlights the color and shape characteristics of the rainbow pattern; the dark field lighting highlights the microcracks and enhances the shape characteristics of the microcracks.

[0039] In the above optional method, by using a camera, a telecentric lens and different lighting conditions to obtain the original images respectively, the different features of the edge of the LCD screen hole, especially the shape features of the microcracks, can be more effectively highlighted, thereby improving the subsequent detection model's ability to recognize microcracks.

[0040] In a third aspect, a technical solution of an electronic device of the present invention is as follows:

[0041] The invention comprises 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 the hole edge microcrack defect detection method of the present invention are realized.

[0042] In a fourth aspect, the present invention provides a computer-readable storage medium having the following technical solution:

[0043] Instructions are stored in the computer-readable storage medium. When the computer-readable storage medium reads the instructions, the computer-readable storage medium executes the steps of the hole edge microcrack defect detection method of the present invention.

[0044] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present invention. In addition, the same reference symbols are used to represent the same components throughout the accompanying drawings. In the accompanying drawings:

[0046] Figure 1 It is a schematic flow chart of an embodiment of a method for detecting micro-crack defects at the edge of a hole of the present invention;

[0047] Figure 2 This is a schematic diagram of the equipment composition of the telecentric imaging solution;

[0048] Figure 3 Schematic diagram of the training process of micro-crack defect detection model;

[0049] Figure 4 Schematic diagram of the ROI region extraction model training process;

[0050] Figure 5 It is a schematic diagram of a divided ROI image after the ROI area is divided into blocks;

[0051] Figure 6 is a schematic diagram of a first defect prediction image;

[0052] Figure 7 It is a schematic diagram of the defect detection method test flow;

[0053] Figure 8 Schematic diagram of the process flow for optimizing missed detection of micro cracks;

[0054] Fig. 9 It is a structural schematic diagram of an embodiment of a hole edge micro-crack defect detection system of the present invention;

[0055] Fig.10 The figure is a schematic structural diagram of an embodiment of an electronic device of the present invention. DETAILED DESCRIPTION

[0056] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0057] Figure 1 FIG. 1 is a flow chart of an embodiment of a method for detecting microcrack defects at the edge of a hole provided by the present invention, and the method is executed by a controller. Figure 1 As shown, the following steps are included:

[0058] S1. Obtain the original bright field image and the original dark field image of the edge of the hole of the LCD screen to be tested under telecentric imaging. In S1:

[0059] 1) Original bright field image refers to the image of the edge of the LCD screen hole to be tested, which is obtained by using a camera, a telecentric lens and bright field lighting under a telecentric imaging system. The original bright field image can highlight the color and shape characteristics of the rainbow pattern at the edge of the LCD screen hole, providing basic information for subsequent defect detection.

[0060] 2) Original dark field image refers to the image of the edge of the LCD screen hole to be tested, which is obtained by using a camera, a telecentric lens and dark field lighting under a telecentric imaging system. The original dark field image can highlight and enhance the shape characteristics of microcracks, helping the detection model to more accurately identify microcrack defects.

[0061] Specifically, the controller obtains an original bright field image and an original dark field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging.

[0062] S2, inputting the original bright field image and the original dark field image into the improved microcrack defect detection model to obtain the detection result of the edge of the hole of the liquid crystal screen to be tested; wherein the improved microcrack defect detection model includes: an ROI region extraction module, a semantic segmentation module and a connected domain merging module which are sequentially connected and arranged; the ROI region extraction module is used to: perform ROI region extraction on the target image and perform image block division to obtain a block ROI image; the semantic segmentation module is used to: perform defect prediction on the block ROI image to obtain a first defect prediction image; the connected domain merging module is used to: merge the connected domains of each first defect prediction image that meets the preset conditions to obtain a merged defect prediction image, in S2:

[0063] 1) The target image is any one of the original bright field image and the original dark field image after the size adjustment processing of the preprocessing module. The target image is used as the input of the improved microcrack defect detection model to ensure the consistency of the input image and improve the stability and accuracy of the detection model.

[0064] 2) The block ROI image is an image block obtained by extracting the ROI area (Region of Interest) from the target image and dividing it into equal parts according to a preset size rectangle. By extracting the ROI area, the amount of calculation can be reduced, the detection efficiency can be improved, and the model can be more focused on the key areas.

[0065] 3) The first defect prediction image is an image obtained after the semantic segmentation module predicts defects on the block ROI image, which is used to show the location and type of defects. The first defect prediction image can provide preliminary defect prediction results and provide a basis for subsequent connected domain merging.

[0066] 4) A connected domain refers to a set of interconnected pixels in an image that have the same pixel value. In defect detection, a connected domain represents a possible defect area. The shape, size, and position of the connected domain are important for determining the defect type.

[0067] 5) The merged defect prediction image is an image obtained by merging the connected domains in the first defect prediction image according to preset conditions by the connected domain merging module. By merging adjacent and qualified connected domains, the detection result of the edge of the LCD screen hole to be tested can be determined, false detection and missed detection can be reduced, and the accuracy and reliability of the final defect detection result can be improved.

[0068] Specifically, the controller further inputs the original bright field image and the original dark field image into the improved micro-crack defect detection model to obtain the detection result of the edge of the liquid crystal screen hole to be tested.

[0069] The technical solution of this embodiment is combined with an improved microcrack defect detection model to achieve efficient and accurate detection of microcracks on the edge of the LCD screen hole, which not only improves the detection speed and accuracy, but also reduces the false detection and missed detection rates.

[0070] In an optional manner, the step of obtaining an original bright field image and an original dark field image of the edge of the hole of the liquid crystal panel to be tested under telecentric imaging includes:

[0071] Using a camera, a telecentric lens and bright field lighting, an original bright field image of the edge of the hole of the LCD screen to be tested under telecentric imaging is obtained;

[0072] Using a camera, a telecentric lens and dark field illumination, an original dark field image of the edge of the hole of the LCD screen to be tested under telecentric imaging is obtained;

[0073] Among them, bright field lighting highlights the color and shape characteristics of the rainbow pattern; dark field lighting highlights the microcracks and enhances the shape characteristics of the microcracks.

[0074] In this embodiment, if Figure 2As shown, bright field illumination uses point light sources in the form of coaxial light. Dark field illumination uses 4 evenly distributed low-angle point light sources and a ring light source. The telecentric lens is a 3x telecentric lens, the camera is a Hikvision 25M area array camera, the field of view size is 4.2mm*4.2mm, and the actual physical length corresponding to one pixel is 0.83 microns.

[0075] Specifically, the controller further uses the camera, telecentric lens and bright field lighting to obtain the original bright field image of the edge of the hole of the LCD screen to be tested under telecentric imaging; and uses the camera, telecentric lens and dark field lighting to obtain the original dark field image of the edge of the hole of the LCD screen to be tested under telecentric imaging.

[0076] In an optional manner, the improved microcrack defect detection model further includes: a preprocessing module;

[0077] The preprocessing module is used to resize the original image to obtain the target image.

[0078] In this embodiment, the original image size is adjusted to 1024*1024 pixels using an image processing library.

[0079] It should be noted that if Figure 3 As shown in the figure, the training process of the microcrack defect detection model is as follows: for the bright field image in the target image, use the labelme software to mark the "rainbow pattern" defect polygon area; for the dark field image in the target image, use the labelme software to mark the "crack" defect; after obtaining the annotation file, organize the format into the COCO format required by the defect detection area model, and put the training data into the specified folder to start training the defect area detection model.

[0080] Specifically, the controller further performs a size adjustment process on the original image to obtain a target image.

[0081] In an optional manner, the step of extracting the ROI region of the target image and dividing the image into blocks to obtain a block ROI image includes:

[0082] The ROI region of the target image is obtained, and the ROI region is equally divided into rectangles of a preset size to obtain a block ROI image containing multiple rectangular frame regions.

[0083] Among them, the BiseNet model is used to extract the ROI area of ​​the target image.

[0084] It should be noted that if Figure 4As shown in the figure, the training process of the ROI region extraction model is as follows: for the required hole edge ROI area, use the labelme software to mark the hole edge polygon area; after obtaining the annotation file, organize the format and adapt it to the COCO format required by the ROI region extraction model; put the training data in the specified folder and start training the ROI region extraction model.

[0085] The target image is sent to the trained ROI region extraction model to obtain the annular ROI region; the ROI region is divided into blocks, the ROI region is a circular ring region by default, and the circular ring is divided into equal parts according to a preset rectangle of a preset size, each of which is a circular ring region contained in a rectangular frame, and the following is obtained: Figure 5 The block ROI image shown contains multiple rectangular box areas to adapt to the image size requirements of the defect semantic segmentation model Bisenet.

[0086] Specifically, the ROI region of the target image is acquired, and the ROI region is equally divided into rectangles of a preset size to obtain a block ROI image containing a plurality of rectangular frame regions.

[0087] In an optional manner, the step of performing defect prediction on the block ROI image to obtain a first defect prediction image includes:

[0088] The block ROI images are grouped, and the Bisenet model is used to perform pixel-level classification prediction on the grouped block ROI images to obtain a first defect prediction image.

[0089] The block ROI images are grouped according to a preset number of groups.

[0090] In this embodiment, the block ROI images are grouped into a group of images (eg, 32) according to the batchsize parameter requirement and are input into the Bisenet model as parameters.

[0091] It should be noted that both the semantic segmentation model and the ROI region extraction model use the BiseNet model for parameter optimization.

[0092] Among them, the Bisenet model adopts the following quantitative measures to streamline and optimize:

[0093] ① Reducing the size of the convolution kernel: The size of the original convolution kernel is 5x5 or 7x7. Reducing the size of the convolution kernel to 3x3 can reduce the computational complexity of the convolution operation.

[0094] ② Reduce the number of channels: The number of channels of the original convolutional layer is 64 or 128. Reducing the number of channels to 32 or 64 can reduce the number of model parameters.

[0095] The above two methods can significantly reduce the number of model parameters, thereby improving the speed of training convergence. Specifically, there are N convolutional layers in the backbone network of the original Bisenet model, and the number of parameters of each convolutional layer is C_in*C_out*K*K. Among them, C_in is the number of input channels, C_out is the number of output channels, and K is the size of the convolution kernel. If the size of the convolution kernel is reduced to 3x3 and the number of channels is reduced to C / 2, the number of parameters of each convolutional layer will become C_in / 2*C_out / 2*3*3. Therefore, the number of parameters of the entire backbone network will be reduced to 1 / 4 of the original.

[0096] Specifically, the controller further groups the block ROI images, and uses the Bisenet model to perform pixel-level classification prediction on the grouped block ROI images to obtain a first defect prediction image.

[0097] In an optional manner, the preset condition is: the distance between the bounding rectangles of any two connected domains in the first defect prediction image is less than a preset threshold, and if the types of defects corresponding to any two connected domains are the same; wherein the bounding box function in the image processing library is used to calculate the bounding rectangle of the connected domain.

[0098] like Figure 6 As shown, the distance between the circumscribed rectangles of the two connected domains can be clearly identified from the first defect prediction image.

[0099] It should be noted that the first defect prediction image is binarized to obtain a binary image, and the defective areas in the binary image are marked using the connected component marking algorithm of the image processing library, thereby determining each connected domain of the first defect prediction image.

[0100] In this embodiment, the preset threshold is set to 5 pixels by default. If the closest distance between the vertices of the circumscribed rectangles of two connected domains is less than 5 pixels and they belong to the same crack or rainbow pattern, the two connected domains are merged. The merged connected domains are screened according to the shape features to determine the target defect position and target defect type. If the area of ​​the connected domain is less than 10 pixels or the long side of the minimum circumscribed rectangle of the connected domain is less than 2 pixels, it is not output.

[0101] Specifically, the boundingbox function in the image processing library is used to calculate the bounding rectangle of the connected domain. If the distance between the bounding rectangles of any two connected domains in the first defect prediction image is less than a preset threshold, and if the types of defects corresponding to any two connected domains are the same, the connected domains are merged, and the merged connected domains are screened according to shape features to obtain a merged defect prediction image. The merged defect prediction image is used as the detection result image to determine the target defect location and target defect type.

[0102] like Figure 7 As shown in the figure, the overall process of the defect detection method test is as follows: read in the original image data, resize the original image, and obtain the target image; send the target image to the ROI region extraction model, extract the ROI region, and divide the ROI region into blocks to obtain a block ROI image; send the block ROI image to the semantic segmentation model to obtain the first defect prediction map; perform connected domain analysis on the first defect prediction map, and calculate the circumscribed rectangle of the connected domain. If the distance between the circumscribed rectangles of any two connected domains in the first defect prediction image is less than the preset threshold, and if the types of defects corresponding to any two connected domains are the same, the connected domains are merged, and the merged connected domains are screened according to the shape features to obtain a merged defect prediction image. The merged defect prediction image is used as the detection result map to determine the target defect location and target defect type.

[0103] It should be noted that if Figure 8 As shown in the figure, the hole edge microcrack defect detection also includes the microcrack missed detection optimization process, which specifically includes the following contents:

[0104] 1) When missed detection occurs, first query the test result map by ID to check whether the ROI area of ​​the test result map is correctly identified and whether the corresponding defect is successfully segmented:

[0105] ① If the ROI region extraction is wrong, collect the target image and use Labelme software to annotate the ROI region. After annotating, train the ROI region extraction model. After the training is completed, update the ROI region extraction model.

[0106] ② If the ROI area is extracted correctly, but no defect is found on the corresponding defect segmentation map, the target image is also collected and the defect is annotated through Labelme software. After annotation, the ROI area extraction model is trained. After the training is completed, the ROI area extraction model is updated.

[0107] ③ If the ROI area is extracted correctly and the defect is segmented correctly, it is judged that the defect does not meet the card control parameter requirements. Therefore, it is necessary to optimize the detection by adjusting the parameters of the corresponding defect module under the real-time parameters.

[0108] 2) When an over-inspection occurs, the original image corresponding to the target image is searched by ID, the corresponding original image is collected and placed in the corresponding training data folder, and training is started. After the training is completed, the micro-crack defect detection model is updated.

[0109] Fig. 9 FIG. 2 is a schematic diagram showing a structure of an embodiment of a hole edge microcrack defect detection system 200 provided by the present invention. Fig. 9As shown, the system 200 includes: an acquisition unit 210 and a detection unit 220 .

[0110] The acquisition unit 210 is used to: acquire an original bright field image and an original dark field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging;

[0111] The detection unit 220 is used to: input the original bright field image and the original dark field image into the improved microcrack defect detection model to obtain the detection result of the edge of the LCD screen hole to be tested; wherein, the improved microcrack defect detection model includes: an ROI area extraction module, a semantic segmentation module and a connected domain merging module which are connected in sequence; the ROI area extraction module is used to: extract the ROI area of ​​the target image and divide the image into blocks to obtain a block ROI image; the semantic segmentation module is used to: perform defect prediction on the block ROI image to obtain a first defect prediction image; the connected domain merging module is used to: merge the connected domains of each first defect prediction image that meets the preset conditions to obtain a merged defect prediction image.

[0112] In an optional manner, the acquiring unit 210 is specifically configured to:

[0113] Using a camera, a telecentric lens and bright field lighting, an original bright field image of the edge of the hole of the LCD screen to be tested under telecentric imaging is obtained;

[0114] Using a camera, a telecentric lens and dark field illumination, an original dark field image of the edge of the hole of the LCD screen to be tested under telecentric imaging is obtained;

[0115] Among them, bright field lighting highlights the color and shape characteristics of the rainbow pattern; dark field lighting highlights the microcracks and enhances the shape characteristics of the microcracks.

[0116] The technical solution of this embodiment is combined with an improved microcrack defect detection model to achieve efficient and accurate detection of microcracks on the edge of the LCD screen hole, which not only improves the detection speed and accuracy, but also reduces the false detection and missed detection rates.

[0117] The above parameters and steps for each module to implement corresponding functions in the hole edge microcrack defect detection system 200 of this embodiment can refer to the parameters and steps in the embodiment of the hole edge microcrack defect detection method above, and will not be repeated here.

[0118] like Fig.10 As shown, an electronic device 300 according to an embodiment of the present invention includes a processor 320, the processor 320 is coupled to a memory 310, and the memory 310 stores at least one computer program 330, and the at least one computer program 330 is loaded and executed by the processor 320, so that the electronic device 300 implements any of the above-mentioned methods for detecting micro-crack defects at the edge of a hole, specifically:

[0119] The electronic device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors 320 (Central Processing Units, CPU) and one or more memories 310, wherein the one or more memories 310 store at least one computer program 330, and the at least one computer program 330 is loaded and executed by the one or more processors 320, so that the electronic device 300 implements any of the hole edge microcrack defect detection methods provided in the above embodiments. Of course, the electronic device 300 may also have components such as a wired or wireless network interface, a keyboard, and an input and output interface for input and output, and the electronic device 300 may also include other components for implementing device functions, which will not be described in detail here.

[0120] A computer-readable storage medium according to an embodiment of the present invention stores at least one computer program, and the at least one computer program is loaded and executed by a processor so that a computer implements any of the above-mentioned methods for detecting microcrack defects at the edge of a hole.

[0121] Optionally, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, and the like.

[0122] In an exemplary embodiment, a computer program product or a computer program is also provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the electronic device performs any of the above-mentioned methods for detecting microcrack defects at the edge of a hole.

[0123] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects and represent the definition of a specific order or sequence. The order of use of similar objects can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than the order shown or described.

[0124] Those skilled in the art know that the present invention can be implemented as a system, method or computer program product. Therefore, the present disclosure can be specifically implemented in the following forms, namely: it can be complete hardware, it can be complete software (including firmware, resident software, microcode, etc.), or it can be a combination of hardware and software, which is generally referred to as "circuit", "module" or "system" herein. 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, and the computer-readable medium contains computer-readable program code.

[0125] Any combination of one or more computer-readable media may be used. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device.

[0126] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for detecting microcrack defects at the edge of a hole, characterized in that: include: Obtaining the original bright field image and the original dark field image of the edge of the hole of the LCD screen to be tested under telecentric imaging; The original bright field image and the original dark field image are input into the improved microcrack defect detection model to obtain the detection result of the edge of the liquid crystal screen hole to be tested; wherein the improved microcrack defect detection model includes: an ROI area extraction module, a semantic segmentation module and a connected domain merging module which are connected in sequence; the ROI area extraction module is used to: perform ROI area extraction on the target image and perform image segmentation to obtain a block ROI image; the semantic segmentation module is used to: perform defect prediction on the block ROI image to obtain a first defect prediction image; the connected domain merging module is used to: merge the connected domains of each first defect prediction image that meets the preset conditions to obtain a merged defect prediction image.

2. The hole edge microcrack defect detection method according to claim 1, characterized in that: The steps of obtaining the original bright field image and the original dark field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging include: Using a camera, a telecentric lens and bright field illumination, the original bright field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging is obtained; Using the camera, the telecentric lens and dark field illumination, the original dark field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging is obtained; Wherein, the bright field lighting highlights the color and shape characteristics of the rainbow pattern; the dark field lighting highlights the microcracks and enhances the shape characteristics of the microcracks.

3. The hole edge microcrack defect detection method according to claim 1, characterized in that: The improved microcrack defect detection model also includes: a preprocessing module; The preprocessing module is used to perform size adjustment processing on the original image to obtain the target image.

4. The hole edge microcrack defect detection method according to claim 1, characterized in that: The step of extracting the ROI region of the target image and dividing the image into blocks to obtain a block ROI image includes: The ROI region of the target image is acquired, and the ROI region is equally divided into rectangles of a preset size to obtain the divided ROI image containing a plurality of rectangular frame regions.

5. The hole edge microcrack defect detection method according to claim 1, characterized in that: The step of performing defect prediction on the block ROI image to obtain a first defect prediction image comprises: The block ROI images are grouped, and the Bisenet model is used to perform pixel-level classification prediction on the grouped block ROI images to obtain the first defect prediction image.

6. The hole edge microcrack defect detection method according to claim 1, characterized in that: The preset condition is: the distance between the bounding rectangles of any two connected domains in the first defect prediction image is less than a preset threshold, and if the types of defects corresponding to the any two connected domains are the same; wherein the bounding box function in the image processing library is used to calculate the bounding rectangle of the connected domain.

7. A hole edge microcrack defect detection system, characterized in that: include: Acquisition unit and detection unit; The acquisition unit is used to: acquire an original bright field image and an original dark field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging; The detection unit is used to: input the original bright field image and the original dark field image into the improved microcrack defect detection model to obtain the detection result of the edge of the liquid crystal screen hole to be tested; wherein the improved microcrack defect detection model includes: an ROI area extraction module, a semantic segmentation module and a connected domain merging module which are connected in sequence; the ROI area extraction module is used to: perform ROI area extraction on the target image and perform image segmentation to obtain a block ROI image; the semantic segmentation module is used to: perform defect prediction on the block ROI image to obtain a first defect prediction image; the connected domain merging module is used to: merge the connected domains of each first defect prediction image that meets the preset conditions to obtain a merged defect prediction image.

8. The hole edge microcrack defect detection system according to claim 7, characterized in that: The acquisition unit is specifically used for: Using a camera, a telecentric lens and bright field illumination, the original bright field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging is obtained; Using the camera, the telecentric lens and dark field illumination, the original dark field image of the edge of the hole of the liquid crystal screen to be tested under telecentric imaging is obtained; Wherein, the bright field lighting highlights the color and shape characteristics of the rainbow pattern; The dark field illumination highlights the microcracks and enhances the shape characteristics of the microcracks.

9. An electronic device, characterized in that: The electronic device includes a processor, the processor is coupled to a memory, at least one computer program is stored in the memory, and the at least one computer program is loaded and executed by the processor so that the electronic device implements the hole edge microcrack defect detection method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one computer program, and the at least one computer program is loaded and executed by the processor so that the computer-readable storage medium implements the hole edge microcrack defect detection method according to any one of claims 1 to 6.

Images