An automatic repair method, system, device and storage medium for panel defects

Abstract

This invention provides a method, system, device, and storage medium for automatic panel defect repair, relating to the field of automatic panel defect repair technology. The method comprises: using a pre-trained defect detection model to detect defects in a panel image to obtain defect target boxes; performing periodic matching and periodic migration of the panel image based on a preset matching template to locate repair anchor boxes in the panel image; performing intersection calculation and threshold verification on the defect target boxes and repair anchor boxes in the panel image to obtain repair boxes; automatically repairing panel defects based on the repair boxes and acquiring a panel defect repair image; and using a pre-trained defect repair effect detection model to detect defect repair in the panel defect repair image to obtain defect repair detection results. This invention utilizes deep learning technology and computer vision technology for defect detection, defect repair, and defect repair effect detection, meeting the needs of panel manufacturers for automatic defect detection and repair.

Description

Technical Field

[0001] This invention relates to the field of automatic panel defect repair technology, and more specifically, to an automatic panel defect repair method, system, device, and storage medium. Background Technology

[0002] In the manufacturing process of display panels, the complex and tedious production process easily introduces various defects. Traditional defect detection relies on manual visual inspection, which is not only susceptible to differences in personnel experience and mental state, leading to a high misjudgment rate, but also has relatively high labor costs. Therefore, many panel manufacturers have begun to introduce automated defect detection and classification systems (ADC systems) to replace manual defect detection, and these systems have achieved good results in actual production activities.

[0003] Existing ADC systems can detect panel defects, but they cannot automatically repair panel defects or detect the effect of defect repair based on the defect detection results, thus failing to meet the needs of panel manufacturers for automatic defect detection and repair. Summary of the Invention

[0004] This invention provides a method, system, device, and storage medium for automatic panel defect repair, which solves the problem that existing technologies cannot achieve automatic repair of panel defects based on defect detection results and detect the effect of defect repair.

[0005] In a first aspect, embodiments of the present invention provide an automatic repair method for panel defects, the method flow of which is as follows:

[0006] A target detection algorithm is used to train a model on a defect sample set and a defect repair sample set to obtain a defect detection model and a defect repair effect detection model.

[0007] A periodic pixel unit is obtained from the defect sample image as a matching template, and a repair anchor box is preset based on the matching template;

[0008] A defect detection model is used to detect defects in the panel image to obtain the defect target box;

[0009] Periodic matching and periodic migration of the panel image are performed based on the matching template to locate the repair anchor box corresponding to all periodic pixel units in the panel image.

[0010] Intersection calculations are performed on the defect target box and the repair anchor box corresponding to all periodic pixel units in the panel image to obtain all intersection boxes, and threshold verification is performed on all intersection boxes to obtain the final repair box.

[0011] Automatic panel defect repair is performed based on the final repair box, and panel defect repair images are obtained;

[0012] A defect repair effect detection model is used to detect defects in panel defect repair images to obtain defect repair detection results.

[0013] In the above embodiments, in order to quickly complete defect detection, automatic defect repair, and defect repair effect detection, the present invention first trains a defect detection model and a defect repair effect detection model based on deep learning technology, and generates matching templates and repair anchor boxes based on computer vision technology; then, it performs defect detection on the panel image based on deep learning technology, and performs periodic matching and periodic migration on the panel image based on computer vision technology; next, it performs threshold verification based on computer vision post-processing technology to obtain the final repair box; finally, it performs automatic panel defect repair based on the repair box, and performs defect repair detection based on deep learning technology; the defect detection, defect repair, and defect repair effect detection based on deep learning technology and computer vision technology can meet the needs of panel manufacturers for automatic defect detection and repair, has good practicality in panel production activities, and can significantly improve panel production efficiency.

[0014] As some optional implementations of this application, the process of using a target detection algorithm to train a model on a defect sample set and a defect repair sample set to obtain a defect detection model and a defect repair effect detection model is as follows:

[0015] Defect sample images at different magnifications are collected, and defect categories are labeled on the defect sample images to obtain a defect sample set. The defect sample images include a first defect sample image and a second defect sample image, and the magnification of the first defect sample image is less than that of the second defect sample image.

[0016] A defect detection model is obtained by training a model on defect sample images in a defect sample set using an object detection algorithm.

[0017] Collect defect repair sample images of different defect types and annotate the repair results of the defect repair sample images to obtain a defect repair sample set. The defect repair sample images include sample images of successful defect repair and sample images of failed defect repair.

[0018] A target detection algorithm is used to train a model on the defect repair samples in the defect repair sample set to obtain a defect repair effect detection model.

[0019] In the above embodiments, in order to perform defect detection and defect repair detection quickly and accurately, the present invention trains a defect detection model and a defect repair effect detection model based on deep learning technology, so that the defect detection model can quickly and accurately detect defects in the panel, and the defect repair effect detection model can quickly and accurately detect defect repairs in the panel.

[0020] As some optional implementations of this application, the process of obtaining a periodic pixel unit from the defect sample image as a matching template and presetting a repair anchor box based on the matching template is as follows:

[0021] A repeating periodic pixel unit is extracted from the defect sample image as a matching template, where the defect sample image is the second defect sample image;

[0022] Extract all line regions from the matching template and preset repair anchor frames for all line regions in the matching template.

[0023] In the above embodiments, in order to perform repair positioning quickly and accurately, the present invention generates matching templates and repair anchor frames based on computer vision technology, so that subsequent repair positioning can be performed based on the matching templates and preset repair anchor frames.

[0024] As some optional implementations of this application, the process of using a defect detection model to detect defects in a panel image to obtain defect target boxes is as follows:

[0025] Panel images at different magnifications are acquired, wherein the panel images include a first panel image and a second panel image, and the magnification of the first panel image is the same as that of the first defect sample image, and the magnification of the second panel image is the same as that of the second defect sample image;

[0026] The first panel image to be detected is input into the defect detection model for defect detection to determine whether the panel has defects.

[0027] If a defect exists, the second panel image to be detected is input into the defect detection model for defect detection to obtain the defect target box.

[0028] In the above embodiments, in order to perform defect detection quickly and accurately, the present invention uses a pre-trained defect detection model to perform defect detection on the panel image, which can quickly and accurately obtain the defect target box.

[0029] As some optional implementations of this application, the process of performing periodic matching and periodic migration on the panel image based on the matching template to locate the repair anchor box corresponding to all periodic pixel units in the panel image is as follows:

[0030] A template matching algorithm is used to perform periodic matching between the matching template and the second panel image to be detected, and a periodic pixel unit is obtained from the second panel image based on the periodic matching result;

[0031] Periodic pixel units are periodically migrated based on their size, and all periodic pixel units are obtained from the second panel image based on the periodic migration results.

[0032] Based on the repair anchor frame preset by the matching template, the line region is located for all periodic pixel units in the second panel image to obtain the repair anchor frame corresponding to all periodic pixel units in the second panel image.

[0033] In the above embodiments, in order to perform repair positioning quickly and accurately, the present invention uses a pre-set matching template to perform periodic matching and periodic migration on the panel image, which can quickly and accurately obtain the repair anchor frame corresponding to all periodic pixel units in the panel image.

[0034] As some optional embodiments of this application, the process of performing intersection calculations on the defect target box and the repair anchor box corresponding to all periodic pixel units in the panel image to obtain all intersection boxes is as follows:

[0035] The coordinates of the defect target box are intersected with the coordinates of the repair anchor boxes corresponding to all periodic pixel units in the panel image to obtain the repair anchor boxes corresponding to all line areas in the defect target box.

[0036] Coordinate positioning is performed on the repair anchor frames corresponding to all line areas within the defect target frame to obtain all intersection frames.

[0037] In the above embodiments, in order to quickly and accurately locate the repair coordinates, the present invention adopts an intersection calculation method, which can quickly and accurately obtain the intersection coordinate information of the defect target box and the repair anchor box corresponding to all line areas.

[0038] As some optional embodiments of this application, the threshold verification includes repair anchor frame size threshold verification, minimum repair size threshold verification, and intersection frame size threshold verification.

[0039] In the above embodiments, the present invention can perform redundancy filtering and size optimization on the intersection boxes through computer vision-based post-processing to obtain the final repair box.

[0040] In a second aspect, the present invention provides an automatic panel defect repair system, the system comprising:

[0041] The model training unit is used to train the model on the defect sample set and the defect repair sample set using the target detection algorithm, so as to obtain the defect detection model and the defect repair effect detection model.

[0042] A matching template unit is used to obtain a periodic pixel unit from the defect sample image as a matching template, and to preset a repair anchor box based on the matching template;

[0043] A defect detection unit is used to perform defect detection on the panel image using a defect detection model to obtain a defect target box;

[0044] A periodic verification unit performs periodic matching and periodic migration on the panel image based on a matching template to locate the repair anchor box corresponding to all periodic pixel units in the panel image.

[0045] The repair positioning unit is used to perform intersection calculations on the defect target box and the repair anchor box corresponding to all periodic pixel units in the panel image to obtain all intersection boxes, and to perform threshold verification on all intersection boxes to obtain the final repair box.

[0046] An automatic repair unit performs automatic panel defect repair based on the final repair frame and acquires a panel defect repair image;

[0047] The repair detection unit is used to perform defect repair detection on the panel defect repair image using a defect repair effect detection model to obtain defect repair detection results.

[0048] In a third aspect, the present invention provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned automatic panel defect repair method.

[0049] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned automatic panel defect repair method.

[0050] The beneficial effects of this invention are as follows: This invention is based on deep learning technology and computer vision technology for defect detection, defect repair and defect repair effect detection, which can meet the needs of panel manufacturers for automatic defect detection and repair, has good practicality in panel production activities and can significantly improve panel production efficiency. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a schematic diagram of the computer device structure of the hardware operating environment described in the embodiments of the present invention;

[0053] Figure 2 This is a flowchart of the automatic panel defect repair method described in the embodiments of the present invention;

[0054] Figure 3 This is a schematic diagram illustrating the acquisition of all pixel units of a panel image according to an embodiment of the present invention;

[0055] Figure 4 This is a comparative diagram of the panel defect repair before and after the present invention embodiment;

[0056] Figure 5 This is a structural block diagram of the automatic panel defect repair system described in an embodiment of the present invention. Detailed Implementation

[0057] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0058] To address the problem that existing technologies cannot achieve automatic repair of panel defects and detect the effect of defect repair based on defect detection results, this invention provides a method, system, device, and storage medium for automatic panel defect repair. Before introducing the specific technical solution of this invention, the hardware operating environment involved in the embodiments of this invention will be described first.

[0059] Please see Figure 1 , Figure 1 This is a schematic diagram of the computer device structure of the hardware operating environment involved in the embodiments of the present invention.

[0060] like Figure 1 As shown, the computer device may include: a processor, such as a central processing unit (CPU), a communication bus, a user interface, a network interface, and memory. The communication bus is used to enable communication between these components. The user interface may include a display screen and an input unit such as a keyboard; optionally, the user interface may also include a standard wired interface or a wireless interface. The network interface may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk storage device; optionally, the memory may also be a storage device independent of the aforementioned processor.

[0061] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the computer device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0062] like Figure 1 As shown, a memory, as a storage medium, may include an operating system, a data storage module, a network communication module, a user interface module, and an electronic program module.

[0063] exist Figure 1 In the computer device shown, the network interface is mainly used for data communication with the network server; the user interface is mainly used for data interaction with the user; the processor and memory in the computer device of this application can be set in the computer device, and the computer device calls the panel defect automatic repair system stored in the electronic program module through the processor and executes the panel defect automatic repair method provided in the embodiment of this application.

[0064] Based on the hardware environment of the foregoing embodiments, embodiments of this application provide an automatic panel defect repair method. Please refer to [link to relevant documentation]. Figure 2 , Figure 2 The flowchart shows a method for automatic repair of panel defects. The method flow is as follows:

[0065] (1) The defect sample set and the defect repair sample set are trained by the target detection algorithm to obtain the defect detection model and the defect repair effect detection model.

[0066] In this embodiment of the invention, the training process of the defect detection model is as follows:

[0067] (1.1) Collect defect sample images at different magnifications and label the defect sample images with defect categories to obtain a defect sample set; wherein, the defect sample images include a first defect sample image and a second defect sample image, and the magnification of the first defect sample image is less than that of the second defect sample image; specifically, the magnification of the first defect sample image includes, but is not limited to, 20 times, and the magnification of the second defect sample image includes, but is not limited to, 50 times.

[0068] (1.2) A defect detection model is obtained by training the defect sample images in the defect sample set using a target detection algorithm; wherein the target detection algorithm includes, but is not limited to, Faster RCNN.

[0069] In this embodiment of the invention, the training process of the defect repair effect detection model is as follows:

[0070] (1.3) Collect defect repair sample images of different defect types under positive light and backlight conditions, and label the defect repair sample images with repair results to obtain a defect repair sample set; wherein, the defect repair sample images include defect repair successful sample images and defect repair failed sample images, and the magnification of the defect repair sample images is the same as that of the second defect sample images.

[0071] (1.4) A target detection algorithm is used to train a model on the defect repair samples in the defect repair sample set to obtain a defect repair effect detection model; wherein, the target detection algorithm includes, but is not limited to, Faster RCNN.

[0072] (2) Obtain a periodic pixel unit from the defect sample image as a matching template, and pre-set the repair anchor box based on the matching template, wherein a periodic pixel unit includes a complete and overlapping line area in the panel image.

[0073] In this embodiment of the invention, the process for obtaining the matching template is as follows:

[0074] (2.1) Extract a repeating periodic pixel unit from the defect sample image as a matching template; wherein the defect sample image is a second defect sample image.

[0075] (2.2) Extract all line regions in the matching template and preset repair anchor frames for all line regions in the matching template that may have short circuits.

[0076] (3) Use a defect detection model to detect defects in the panel image to obtain the defect target box.

[0077] In this embodiment of the invention, the defect detection process is as follows:

[0078] (3.1) Receive the first panel image transmitted from the repair machine, and perform a Laplace transform on the first panel image. Compare the Laplace transform result with the blur threshold to determine whether the first panel image is blurry. If the first panel image transmitted from the repair machine is blurry, call the repair machine to automatically focus and obtain a clear first panel image again. Wherein, the magnification of the first panel image is the same as that of the first defect sample image.

[0079] (3.2) Input the first panel image to be detected into the defect detection model to perform defect detection in order to determine whether there is a defect in the panel.

[0080] (2.3) If a defect exists, the second panel image transmitted from the repair machine is received, and the second panel image to be detected is input into the defect detection model for defect detection to obtain the defect target box; wherein the defect target box includes defect category information and defect coordinate information, and the magnification of the second panel image is the same as that of the second defect sample image.

[0081] (4) Perform periodic matching and periodic migration on the panel image based on the matching template to locate the repair anchor box corresponding to all periodic pixel units in the panel image.

[0082] In this embodiment of the invention, the process for repairing anchor frames corresponding to all periodic pixel units in the positioning panel image is as follows:

[0083] (4.1) A template matching algorithm is used to perform periodic matching between the matching template and the second panel image to be detected, and a periodic pixel unit is obtained from the second panel image based on the periodic matching result.

[0084] (4.2) Periodic pixel units are periodically migrated based on their size, and all periodic pixel units are obtained from the second panel image based on the periodic migration results. Please refer to [link to relevant documentation]. Figure 3 , Figure 3 A schematic diagram of obtaining all pixel units for a panel image.

[0085] (4.3) Based on the repair anchor frame preset by the matching template, the line region is located for all periodic pixel units in the second panel image to obtain the repair anchor frame corresponding to all periodic pixel units in the second panel image.

[0086] In this embodiment of the invention, the process for obtaining all intersection boxes is as follows:

[0087] (4.4) Perform intersection calculations between the coordinates of the defect target box and the coordinates of the repair anchor boxes corresponding to all periodic pixel units in the panel image to obtain the repair anchor boxes corresponding to all line areas in the defect target box.

[0088] (4.5) Perform coordinate positioning on all repair anchor frames corresponding to all line areas in the defect target frame to obtain all intersection frames, wherein the intersection frames include defect repair coordinate information.

[0089] (5) Perform intersection calculations on the defect target box and the repair anchor box corresponding to all periodic pixel units in the panel image to obtain all intersection boxes, and perform threshold verification on all intersection boxes to obtain the final repair box.

[0090] In this embodiment of the invention, the threshold verification includes the repair anchor frame size threshold verification, the minimum repair size threshold verification, and the intersection frame size threshold verification. That is, through computer vision-based post-processing, the intersection frame can be redundantly filtered and its size optimized to obtain the final repair frame.

[0091] (6) Automatically repair panel defects based on the final repair frame and obtain panel defect repair images; that is, input the final repair frame into the repair machine and call the repair machine to perform laser repair or grinding repair on the panel.

[0092] Please see Figure 4 , Figure 4 This is a schematic diagram comparing the panel defect before and after repair. After the panel defect is automatically repaired, the repaired defect area is image-captured to obtain panel defect repair images under positive light and backlight conditions. The magnification of the panel defect repair images is the same as that of the defect repair sample images.

[0093] (7) The defect repair effect detection model is used to detect the defect repair of the panel defect repair image to obtain the defect repair detection result.

[0094] In this embodiment of the invention, the defect repair detection process is as follows:

[0095] (7.1) Input the panel defect repair image to be detected into the defect repair effect detection model to detect the defect repair and determine whether the defect repair is successful.

[0096] (7.2) If the defect repair is successful, the next panel defect repair image to be detected is input into the defect repair effect detection model for defect repair detection. If the defect repair fails, the repair box is re-acquired and subsequent defect repair processing is performed.

[0097] In summary, this invention first trains a defect detection model and a defect repair effect detection model based on deep learning technology, and generates matching templates and repair anchor boxes based on computer vision technology. Then, it performs defect detection on panel images based on deep learning technology, and performs periodic matching and periodic migration on the panel images based on computer vision technology. Next, it performs threshold verification based on computer vision post-processing technology to obtain the final repair boxes. Finally, it automatically repairs panel defects based on the repair boxes, and performs defect repair detection based on deep learning technology. This invention, which uses deep learning and computer vision technologies for defect detection, defect repair, and defect repair effect detection, can meet the needs of panel manufacturers for automatic defect detection and repair, has good practicality in panel production activities, and can significantly improve panel production efficiency.

[0098] Furthermore, in one embodiment, based on the same inventive concept as the foregoing embodiments, this embodiment of the invention provides an automatic panel defect repair system, which corresponds one-to-one with the method described in Embodiment 1. Please refer to [link / reference]. Figure 5 , Figure 5 This is a structural block diagram of an automatic panel defect repair system, the system comprising:

[0099] The model training unit is used to train the model on the defect sample set and the defect repair sample set using the target detection algorithm, so as to obtain the defect detection model and the defect repair effect detection model.

[0100] A matching template unit is used to obtain a periodic pixel unit from the defect sample image as a matching template, and to preset a repair anchor box based on the matching template;

[0101] A defect detection unit is used to perform defect detection on the panel image using a defect detection model to obtain a defect target box;

[0102] A periodic verification unit performs periodic matching and periodic migration on the panel image based on a matching template to locate the repair anchor box corresponding to all periodic pixel units in the panel image.

[0103] The repair positioning unit is used to perform intersection calculations on the defect target box and the repair anchor box corresponding to all periodic pixel units in the panel image to obtain all intersection boxes, and to perform threshold verification on all intersection boxes to obtain the final repair box.

[0104] An automatic repair unit performs automatic panel defect repair based on the final repair frame and acquires a panel defect repair image;

[0105] The repair detection unit is used to perform defect repair detection on the panel defect repair image using a defect repair effect detection model to obtain defect repair detection results.

[0106] It should be noted that each unit in the automatic panel defect repair system in this embodiment corresponds one-to-one with each step in the automatic panel defect repair method in the aforementioned embodiment. Therefore, the specific implementation method and the technical effects achieved in this embodiment can be referred to the implementation method of the aforementioned automatic panel defect repair method, and will not be repeated here.

[0107] Furthermore, in one embodiment, this application also provides a computer device, the computer device including a processor, a memory, and a computer program stored in the memory, the computer program being executed by the processor to implement the methods in the foregoing embodiments.

[0108] In addition, in one embodiment, this application also provides a computer storage medium storing a computer program that is executed by a processor to implement the methods described in the foregoing embodiments.

[0109] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a device including one or any combination of the above-mentioned memories. The computer may be a variety of computing devices, including smart terminals and servers.

[0110] In some embodiments, executable instructions may take the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a standalone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0111] As an example, executable instructions may, but do not necessarily, correspond to files in a file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple collaborating files (e.g., a file that stores one or more modules, subroutines, or code sections).

[0112] As an example, executable instructions can be deployed to execute on a single computing device, or on multiple computing devices located in one location, or on multiple computing devices distributed across multiple locations and interconnected via a communication network.

[0113] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0114] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0115] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a multimedia terminal device (which may be a mobile phone, computer, television receiver, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0116] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for automatic repair of panel defects, characterized in that, The method flow is as follows: A target detection algorithm is used to train a model on a defect sample set and a defect repair sample set to obtain a defect detection model and a defect repair effect detection model. A periodic pixel unit is obtained from the defect sample image as a matching template, and a repair anchor box is preset based on the matching template; A defect detection model is used to detect defects in the panel image to obtain the defect target box; Periodic matching and periodic migration of the panel image are performed based on the matching template to locate the repair anchor box corresponding to all periodic pixel units in the panel image. Intersection calculations are performed on the defect target box and the repair anchor box corresponding to all periodic pixel units in the panel image to obtain all intersection boxes, and threshold verification is performed on all intersection boxes to obtain the final repair box. Automatic panel defect repair is performed based on the final repair box, and panel defect repair images are obtained; A defect repair effect detection model is used to detect defects in panel defect repair images to obtain defect repair detection results.

2. The method of claim 1, wherein, The process of using target detection algorithms to train models on defect sample sets and defect repair sample sets to obtain defect detection models and defect repair effect detection models is as follows: Defect sample images at different magnifications are collected, and defect categories are labeled on the defect sample images to obtain a defect sample set. The defect sample images include a first defect sample image and a second defect sample image, and the magnification of the first defect sample image is less than that of the second defect sample image. A defect detection model is obtained by training a model on defect sample images in a defect sample set using an object detection algorithm. Collect defect repair sample images of different defect types and annotate the repair results of the defect repair sample images to obtain a defect repair sample set. The defect repair sample images include sample images of successful defect repair and sample images of failed defect repair. A target detection algorithm is used to train a model on the defect repair samples in the defect repair sample set to obtain a defect repair effect detection model.

3. The method of claim 2, wherein, The process of obtaining a periodic pixel unit from the defect sample image as a matching template and pre-setting the repair anchor box based on the matching template is as follows: A repeating periodic pixel unit is extracted from the defect sample image as a matching template, where the defect sample image is the second defect sample image; Extract all line regions from the matching template and preset repair anchor frames for all line regions in the matching template.

4. The method of claim 2, wherein, The process of using a defect detection model to detect defects in panel images and obtain defect target boxes is as follows: Panel images at different magnifications are acquired, wherein the panel images include a first panel image and a second panel image, and the magnification of the first panel image is the same as that of the first defect sample image, and the magnification of the second panel image is the same as that of the second defect sample image; The first panel image to be detected is input into the defect detection model for defect detection to determine whether the panel has defects. If a defect exists, the second panel image to be detected is input into the defect detection model for defect detection to obtain the defect target box.

5. The method of claim 1, wherein, The process of locating the repair anchor boxes corresponding to all periodic pixel units in the panel image by performing periodic matching and periodic migration based on matching templates is as follows: A template matching algorithm is used to perform periodic matching between the matching template and the second panel image to be detected, and a periodic pixel unit is obtained from the second panel image based on the periodic matching result; Periodic pixel units are periodically migrated based on their size, and all periodic pixel units are obtained from the second panel image based on the periodic migration results. Based on the repair anchor frame preset by the matching template, the line region is located for all periodic pixel units in the second panel image to obtain the repair anchor frame corresponding to all periodic pixel units in the second panel image.

6. The automatic panel defect repair method according to claim 5, characterized in that, The process of calculating the intersection of the defect target box and the repair anchor box corresponding to all periodic pixel units in the panel image to obtain all intersection boxes is as follows: The coordinates of the defect target box are intersected with the coordinates of the repair anchor boxes corresponding to all periodic pixel units in the panel image to obtain the repair anchor boxes corresponding to all line areas in the defect target box. Coordinate positioning is performed on the repair anchor frames corresponding to all line areas within the defect target frame to obtain all intersection frames.

7. The automatic panel defect repair method according to claim 1, characterized in that, The threshold verification includes the anchor frame size threshold verification, the minimum repair size threshold verification, and the intersection frame size threshold verification.

8. An automatic panel defect repair system, characterized in that, The system includes: The model training unit is used to train the model on the defect sample set and the defect repair sample set using the target detection algorithm, so as to obtain the defect detection model and the defect repair effect detection model. A matching template unit is used to obtain a periodic pixel unit from the defect sample image as a matching template, and to preset a repair anchor box based on the matching template; A defect detection unit is used to perform defect detection on the panel image using a defect detection model to obtain a defect target box; A periodic verification unit performs periodic matching and periodic migration on the panel image based on a matching template to locate the repair anchor box corresponding to all periodic pixel units in the panel image. The repair positioning unit is used to perform intersection calculations on the defect target box and the repair anchor box corresponding to all periodic pixel units in the panel image to obtain all intersection boxes, and to perform threshold verification on all intersection boxes to obtain the final repair box. An automatic repair unit performs automatic panel defect repair based on the final repair frame and acquires a panel defect repair image; The repair detection unit is used to perform defect repair detection on the panel defect repair image using a defect repair effect detection model to obtain defect repair detection results.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes a computer program, it implements the automatic panel defect repair method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the automatic panel defect repair method according to any one of claims 1-7.

Images