Detection method, device and equipment for obtaining plate edge optical point defect and storage medium

Abstract

The invention discloses a plate edge optical point defect detection method, device and equipment and a storage medium, and relates to the field of image detection. Scanning a panorama of the PNL material plate through a line scanning camera, and intercepting an optical point Mark image around the plate edge from the panorama; inputting the Mark image into a defect detection model, and performing defect detection according to the Mark point type; and determining a defect type of the Mark point, matching a defect label from the label library according to the defect type, and outputting a defect identification result. The Mark image is intercepted and input to the defect detection model, parallel detection of multiple input small images can be supported, and the problems that panoramic image recognition consumes too much computing power and the recognition progress is not high are solved. And for each cut Mark image, higher-precision identification can be carried out according to a model of a training number, defect types of Mark points are marked one by one, and then output is carried out, so that directional detection output is realized, and the yield of detection is improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of netlist files, and in particular to a method, device, equipment and storage medium for obtaining optical point defect detection at the edge of a board. Background Art

[0002] In chip and PCB design, optical positioning points are commonly known as "Mark points". They are the points used by the machine to locate when the PCB board is taken to the machine for soldering. Generally, the PCB board with surface mount components needs to set Mark points. In mass production, the placement machine can be calibrated by the operator manually or automatically by the machine to find the Mark points, providing a common and locatable circuit pattern for all steps in the assembly process. Therefore, Mark points are crucial to SMT production.

[0003] In the related technology, when the accuracy of the patch process or production process does not meet the requirements, the mark points may be damaged or offset. Failure to perform preliminary defect detection may result in the inability to calibrate and align the PCB board later, or the alignment accuracy may not meet the requirements or be offset, resulting in a significant reduction in the yield rate. Summary of the invention

[0004] The embodiments of the present application provide a method, device, equipment and storage medium for obtaining optical point defect detection on the edge of a board, so as to solve the problem that the defect of the mark point in the related technology affects the yield rate of the material board.

[0005] On the one hand, the present application provides a method for obtaining optical point defect detection of a board edge, the method comprising:

[0006] Scan the panoramic image of the PNL material board with a line scan camera, and capture the optical point Mark image around the board edge from the panoramic image;

[0007] Input the Mark image into a defect detection model, and perform defect detection according to the Mark point type;

[0008] Determine the defect type of the Mark point, match the defect label from the label library according to the defect type, and output the defect recognition result.

[0009] Specifically, before defect detection, it also includes:

[0010] Collect optical point defect images, label the defect images according to the mark point type, and build a training set and defect label library;

[0011] The labeled defect images are sent to the defect detection model for learning and training, and the loss function weight is adjusted according to the output results until the loss is less than the set value.

[0012] Specifically, the step of intercepting the optical point Mark image around the board edge from the panoramic image includes:

[0013] Obtain the size structure of the material board corresponding to the panoramic image, and determine the board type; different board types correspond to different sizes;

[0014] Determine the board edge area, number of Mark points and corresponding coordinate positions of the panoramic image according to the board structure;

[0015] Capture the corresponding Mark image according to the Mark point coordinate position.

[0016] Specifically, during model training, a database is constructed according to different board structures, and the board data includes material board size, segmentation area, number and type of Mark points, and Mark point coordinates in the panoramic view;

[0017] After the board type is determined, the board type data is matched from the database to determine the number of mark points, the type of mark points, and the coordinates of the mark points.

[0018] Specifically, the step of intercepting the corresponding Mark image according to the Mark point coordinate position includes:

[0019] Determine the image center and expansion size according to the Mark point type, locate the Mark point coordinates from the panoramic image, and determine the size of the captured Mark image according to the image center and expansion size.

[0020] Specifically, the defect detection process includes:

[0021] Identify the bottom pad and the center mark point outline in the mark image and match it with the determined mark point type; different mark point types correspond to bottom pads and mark points of different shapes;

[0022] When the bottom pads identified from the image match each other, the identified Mark point contour is matched with the theoretical Mark point contour;

[0023] When the similarity difference between the identified Mark point contour and the theoretical Mark point contour matches exceeds the set threshold, it is determined to be a defective Mark point and the corresponding defect type.

[0024] Specifically, the contour matching process includes:

[0025] Determine the rectangular Mark border based on the identified Mark point edge contour and the enclosing area formed by the contour;

[0026] Taking the four sides of the rectangular Mark as the center, rectangular images of set widths are respectively intercepted from the binarized image as damage analysis images;

[0027] The erosion operation is performed on the intercepted damage analysis images respectively, and the overall variance of the pixel value fluctuation after the erosion operation is calculated. When the overall variance value is greater than the corresponding fluctuation threshold, it is determined that the Mark is defiled.

[0028] On the other hand, the present application provides a device for obtaining optical point defect detection on a plate edge, the device comprising:

[0029] An extraction module is used to scan a panoramic image of a PNL material plate through a line scan camera, and capture an optical point Mark image around the edge of the plate from the panoramic image;

[0030] A detection module, used for inputting the Mark image into a defect detection model and performing defect detection according to the Mark point type;

[0031] The output module is used to determine the defect type of the mark point, match the defect label from the label library according to the defect type, and output the defect recognition result.

[0032] On the other hand, the present application provides a computer device, comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set are loaded and executed by the processor to implement the method for obtaining optical point defect detection on the edge of a plate as described in the above aspects.

[0033] On the other hand, the present application provides a computer-readable storage medium, wherein the readable storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement the method for obtaining optical point defect detection on the edge of a plate as described in the above aspects.

[0034] The beneficial effects of the technical solution provided by the embodiment of the present application include at least: the present application directly extracts the optical point Mark image around the edge of the plate after extracting the panoramic image, and directly inputs the intercepted image into the defect detection model for recognition. This small image recognition can support multi-input parallel detection, avoiding the problem of excessive computing power consumption and low recognition progress in panoramic image recognition. For each cut Mark image, a higher-precision recognition can be performed based on the model of the training number, and the defect type of each molecular Mark point can be output after labeling to achieve directional detection output and improve the yield rate of detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a flow chart of a method for obtaining optical point defect detection on a plate edge provided in an embodiment of the present application;

[0036] Figure 2A possible form of Mark point location schematic diagram is shown;

[0037] Figure 3 The defect diagrams of some typical mark points are shown;

[0038] Figure 4 A schematic diagram of a material plate structure layout is disclosed;

[0039] Figure 5 A schematic diagram of extracting a Mark image is shown;

[0040] Figure 6 A schematic diagram of defacement detection analysis for images captured in four directions is listed;

[0041] Figure 7 A structural block diagram of a device for detecting optical point defects on a plate edge provided in an embodiment of the present application is shown;

[0042] Figure 8 A structural block diagram of a computer device provided by an exemplary embodiment of the present application is shown. DETAILED DESCRIPTION

[0043] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0044] The term "multiple" as used herein refers to two or more than two. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the related objects are in an "or" relationship.

[0045] Figure 1 : is a flow chart of a method for obtaining optical point defect detection on a plate edge provided in an embodiment of the present application, comprising the following steps:

[0046] S1, scanning a panoramic image of a PNL material board by a line scan camera, and capturing an optical point Mark image around the board edge from the panoramic image;

[0047] The panoramic image is a high-definition image of the entire PNL material board, on which there are several PCS particles arranged in arrays or non-arrays. PCS particles are SOC or combined circuit boards used for subsequent production tape-out and cutting, while optical points are markers located outside the PCS particles and arranged around the PNL material board. Figure 2 A possible form of the Mark point position diagram is shown. The distance from the Mark point (edge) to the edge of the circuit board must be ≥5.0MM. This is to provide the minimum spacing requirement for the machine to clamp the circuit board.

[0048] S2, input the Mark image into the defect detection model, and perform defect detection according to the Mark point type;

[0049] Before executing this process, the defect detection model needs to be trained. The training process first requires collecting optical point defect images, labeling the defect images according to the Mark point type, and building a training set and defect label library. In some embodiments, the Mark point types include the center of a circle, square, rectangle, and other structures. The defect types may include offset, erosion, notch, and foreign matter. Figure 3 The defect images of some typical mark points are shown, from left to right, they are mark point erosion defect, mark point abnormal defect, mark point notch defect, and the mark point types are round pad round mark point, square pad square mark point, and square pad round mark point.

[0050] The process is to send this intercepted Mark image to the model for detection, and output the detection result after adjusting the model parameters. During the training process, the parameters of the dynamic loss function can be gradually adjusted according to the output results. The labeled defect image is sent to the defect detection model for learning and training, and the loss function weight is adjusted according to the output results until the loss is less than the set value.

[0051] For the above training process, it can be divided into multiple sub-losses according to the number of training types required for the model, and the sum of the sub-losses is the total loss. Obtain the t-th round loss ratio of various sub-losses after the t-th round of iterative training, and the t-1-th round loss ratio of various sub-losses after the t-1-th round of iterative training; then determine the change range of the loss ratio of various sub-losses based on the t-th round loss ratio and the t-1-th round loss ratio; then determine the t+1-th round loss weights corresponding to various sub-losses based on the change range of the loss ratio of various sub-losses and the t-th round loss weights. Stop after the target number of iterative trainings and the accuracy requirements are met to obtain the detection model.

[0052] S3. Determine the defect type of the Mark point, match the defect label from the label library according to the defect type, and output the defect recognition result.

[0053] For the detection process of the input model, the defect type of the Mark point is first determined, and then a label is added according to the defect type, and the label is a label type recorded in the label library, and then the recognition result can be output.

[0054] For a complete panorama, after determining and obtaining the images of each Mark point, it is equivalent to completing the directional inspection. For defective inspection results, the defects at the target position of the panorama can be directly output for prompt alarm. Correspondingly, when defects are determined, manual intervention can be made to adjust or repair them to ensure the yield rate of subsequent processes.

[0055] In summary, this application directly extracts the optical point Mark image around the edge of the plate after extracting the panoramic image, and directly inputs the intercepted image into the defect detection model for recognition. This small image recognition can support multi-input parallel detection, avoiding the problem of excessive computing power consumption and low recognition progress in panoramic image recognition. For each cut Mark image, a higher-precision recognition can be performed based on the model of the training number, and the defect type of each molecular Mark point can be output after labeling to achieve directional detection output and improve the yield rate of detection.

[0056] In some embodiments, the process of capturing the optical point Mark image around the board edge from the panoramic image may include the following steps:

[0057] A. Obtain the size structure of the material board corresponding to the panoramic image and determine the board type; different board types correspond to different sizes;

[0058] Different board types naturally represent different size structures. The size of the panoramic image can be determined after the line scan camera is scanned, or entered during synchronous loading of the machine or system, so the board type information can be obtained at this stage.

[0059] B. Determine the board edge area, number of Mark points and corresponding coordinate positions of the panoramic image according to the board structure;

[0060] The board edge area is determined based on the size, e.g. Figure 4 A schematic diagram of the material board structure layout is disclosed. The board edge area is usually defined as a range of a certain width inward from the edge, and the mark points are distributed in this area, and the number and position are determined according to design requirements.

[0061] Although in theory it is possible to directly read the pixel data of the panoramic image and then perform segmentation and recognition, this method consumes more computing power. In this application, a simpler recognition method is chosen, that is, a database matching method is used for rapid recognition.

[0062] Specifically, a database can be built according to different board structures during model training, and the board data includes material board size, segmentation area, number and type of Mark points, and Mark point coordinates in the panorama. After reading these data, the panorama can be quickly cut (the center point can be identified first, and then different areas can be determined and divided outward), located, and identified without the need for pixel-level analysis.

[0063] C. Capture the corresponding Mark image according to the Mark point coordinate position.

[0064] This process specifically determines the image center and the outer expansion size according to the Mark point type, and then locates the Mark point coordinates from the panoramic image, and determines the size of the captured Mark image according to the image center and the outer expansion size.

[0065] Figure 5 The schematic diagram of extracting the Mark image is shown. In the figure, the outer bottom pad is theoretically the largest outer contour area. However, in order to prevent the Mark point offset or pad offset from causing incomplete capture, the expansion method is used. The center of the image is theoretically the center coordinate. After determining the coordinate, the expansion method can be used according to the specification type of the Mark point.

[0066] In some embodiments, the detailed defect detection process includes the following:

[0067] A. Identify the bottom pad and the center mark point outline in the mark image and match it with the determined mark point type; different mark point types correspond to bottom pads and mark points of different shapes;

[0068] B. When the bottom pads identified from the image match each other, the identified Mark point contour is matched with the theoretical Mark point contour;

[0069] C. When the similarity difference between the identified Mark point contour and the theoretical Mark point contour exceeds the set threshold, it is determined to be a defective Mark point and the corresponding defect type.

[0070] This step mainly uses the contour matching method to determine the defect type. For problems such as "bite corrosion", "missing plating" and "foreign matter", the contour matching method can be used to calculate the area and similarity difference.

[0071] In the case of pixel color differences caused by dirt, the enclosing contour of the maximum enclosing area that can be formed by the edge contour can be compared with the circumscribed rectangular frame identified by target detection, and when the area difference between the enclosing contour and the circumscribed rectangular frame does not exceed the set threshold, the enclosing contour of the maximum enclosing area is determined as the Mark point frame;

[0072] This step is to extract the contour first, and then compare the contour that can enclose the largest area with the target contour in the aforementioned model database. In order to detect dirt and foreign objects, the determined Mark point border can also be designed as a rectangular border selected by the circumscribed shape, and the subsequent comparison is also based on the area of ​​the circumscribed rectangle. If the area difference does not exceed the set threshold, the contour with the largest enclosed area is used as the final high-precision Mark point border, because the accuracy of the Mark point border is crucial to the detection of edge contamination.

[0073] Taking the four sides of the Mark point border as the center, respectively, rectangular images of set width are intercepted from the binarized image as edge damage analysis images;

[0074] The four sides correspond to four small rectangular blocks respectively. Erosion operations are performed on the images of the four small rectangular areas respectively, and different structural elements are used in the horizontal and vertical directions for appropriate smoothing.

[0075] The erosion operation is performed on the intercepted damage analysis image respectively, and the overall variance of the pixel value fluctuation after the erosion operation is calculated. When the overall variance value is greater than the corresponding fluctuation threshold, it is determined that there is doubt that it is damaged.

[0076] Figure 6 The schematic diagram of the defacement detection analysis of the images intercepted in four directions is listed. This step takes the small rectangular area containing the left border as an example. It is assumed that an array is maintained, and the array elements are in order: the number of consecutive black pixels starting from the first pixel in the first row of the rectangular area, the number of consecutive black pixels starting from the first pixel in the second row of the rectangular area, and so on, until the number of consecutive black pixels starting from the first pixel in the last row.

[0077] For example, suppose the image is a 4*4 rectangular image. The following table lists a possible pixel representation, and the array it maintains is [2,4,1,2]. Of course, the actual image data contains much more pixel content than this, but this is just for the sake of illustration.

[0078] Table 1 Pixel distribution table of rectangular area

[0079] black black White White black black black black black White White White black black White White

[0080] Taking the small rectangular area containing the left border as an example, we calculate the overall variance of the array based on the maintained array, which represents the degree of fluctuation of the edge pixels. Once the degree of fluctuation is greater than a predefined threshold, it means that there is edge contamination on the left edge of the Mark point.

[0081] In actual situations, since the rectangular area may be very large, there may be a situation where only a small part of the rectangular area has relatively drastic fluctuations in pixel value, and as a whole, the fluctuation may not be that large. For example, if the array we maintain is [2,10,2,2,2,2,2,2,2,2,2,2,2], we can see that a small part of the area has drastic fluctuations, while the overall variance is pulled down by other areas and does not reach the predefined threshold. Therefore, we can also use the method of calculating the sliding window variance separately or simultaneously to more sensitively capture areas with drastic fluctuations.

[0082] Specifically, assuming that in the above example of [2,10,2,2,2,2,2,2,2,2,2,2,2], we define the size of the sliding window as 4 and the sliding step as 1, we can get a series of arrays representing smaller parts of the rectangular area [2,10,2,2], [10,2,2,2], [2,2,2,2]... Calculate the variance of the smaller arrays as a series of values ​​of the sliding window variance. We can also predefine a threshold to determine whether the sliding window variance reaches this threshold, so as to determine whether there is edge contamination on the edge of the mark point.

[0083] Figure 7 The structural block diagram of the device for detecting optical point defects on the edge of a plate provided in an embodiment of the present application is shown, and the device comprises:

[0084] An extraction module 710 is used to scan a panoramic image of a PNL material plate through a line scan camera, and capture an optical point Mark image around the plate edge from the panoramic image;

[0085] A detection module 720 is used to input the Mark image into a defect detection model and perform defect detection according to the Mark point type;

[0086] The output module 730 is used to determine the defect type of the Mark point, match the defect label from the label library according to the defect type, and output the defect recognition result.

[0087] The device for obtaining optical point defect detection on the edge of a plate provided in the embodiment of the present application can be applied to the method for obtaining optical point defect detection on the edge of a plate provided in the above embodiment. For relevant details, refer to the above method embodiment. The implementation principle and technical effect are similar and will not be repeated here.

[0088] It should be noted that the device for obtaining optical point defect detection at the edge of a plate provided in the embodiment of the present application is only illustrated by the division of the above-mentioned functional modules / functional units. In practical applications, the above-mentioned functions can be assigned to different functional modules / functional units as needed, that is, the internal structure of the device for obtaining optical point defect detection at the edge of a plate is divided into different functional modules / functional units to complete all or part of the functions described above. In addition, the implementation method of the method for obtaining optical point defect detection at the edge of a plate provided in the above-mentioned method embodiment and the implementation method of the device for obtaining optical point defect detection at the edge of a plate provided in this embodiment belong to the same concept. The specific implementation process of the device for obtaining optical point defect detection at the edge of a plate provided in this embodiment is detailed in the above-mentioned method embodiment and will not be repeated here.

[0089] Figure 8 The block diagram of a computer device provided by an exemplary embodiment of the present application is shown. It is a computer device such as a desktop computer, a laptop computer, a PDA, and a cloud server. The computer device may include, but is not limited to, a processor and a memory. Among them, the processor and the memory may be connected by a bus or in other ways. Among them, the processor may be a central processing unit (CPU). The processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, graphics processors (GPU), embedded neural network processors (NPU) or other dedicated deep learning coprocessors, discrete gates or transistor logic devices, discrete hardware components and other chips, or a combination of the above-mentioned various chips.

[0090] The processor may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content that needs to be displayed on the display screen. In some embodiments, the processor may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0091] As a non-transient computer-readable storage medium, the memory can be used to store non-transient software programs, non-transient computer executable programs and modules, such as program instructions / modules corresponding to the methods in the above-mentioned embodiments of the present application. The processor executes various functional applications and data processing of the processor by running the non-transient software programs, instructions and modules stored in the memory, that is, the method in the above-mentioned method implementation is realized. The memory may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the processor, etc. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network and a combination thereof.

[0092] In some embodiments, the computer device may further optionally include: a peripheral device interface and at least one peripheral device. The processor, the memory and the peripheral device interface may be connected via a bus or a signal line. Each peripheral device may be connected to the peripheral device interface via a bus, a signal line or a circuit board. Specifically, the peripheral device includes: at least one of a radio frequency circuit, a display screen and a keyboard.

[0093] The peripheral device interface can be used to connect at least one peripheral device related to I / O (Input / Output) to the processor and the memory. In some embodiments, the processor, the memory, and the peripheral device interface are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor, the memory, and the peripheral device interface can be implemented on a separate chip or circuit board, which is not limited in this embodiment.

[0094] The display screen is used to display the UI (User Interface). The UI may include graphics, text, icons, videos and any combination thereof. When the display screen is a touch screen, the display screen also has the ability to collect touch signals on the surface or above the surface of the display screen. The touch signal can be input to the processor as a control signal for processing. At this time, the display screen can also be used to provide virtual buttons and / or virtual keyboards, also known as soft buttons and / or soft keyboards. In some embodiments, the display screen can be one, arranged on the front panel of the computer device; in other embodiments, the display screen can be at least two, respectively arranged on different surfaces of the computer device or in a folding design; in other embodiments, the display screen can be a flexible display screen, arranged on a curved surface or a folding surface of the computer device. Even, the display screen can be set to a non-rectangular irregular shape, that is, a special-shaped screen. The display screen can be made of materials such as LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode) and the like.

[0095] The power supply is used to power various components in the computer device. The power supply can be AC, DC, a disposable battery, or a rechargeable battery. When the power supply includes a rechargeable battery, the rechargeable battery can be a wired rechargeable battery or a wireless rechargeable battery. A wired rechargeable battery is a battery that is charged through a wired line, and a wireless rechargeable battery is a battery that is charged through a wireless coil. The rechargeable battery can also be used to support fast charging technology.

[0096] Those skilled in the art will appreciate that the structure shown in this embodiment does not constitute a limitation on the computer device, and may include more or fewer components than shown in the figure, or combine certain components, or adopt a different component arrangement.

[0097] The embodiment of the present application also discloses a computer-readable storage medium. Specifically, the computer-readable storage medium is used to store a computer program, and when the computer program is executed by the processor, the method in the above-mentioned method implementation is implemented. Those skilled in the art can understand that the implementation of all or part of the process in the above-mentioned implementation method of the present application can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the process of the implementation of the above-mentioned methods. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (Flash Memory), a hard disk (Hard Disk Drive, abbreviated as: HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above-mentioned types of memory.

[0098] This specific embodiment is merely an explanation of the present invention and is not a limitation of the present invention. After reading this specification, those skilled in the art may make non-creative modifications to the present embodiment as needed. However, as long as they are within the scope of the claims of the present invention, they are protected by the patent law.

Claims

1. A method for obtaining optical point defect detection on a plate edge, characterized in that: The method comprises: Scanning a panoramic image of the PNL material board with a line scan camera, and capturing an optical point Mark image around the board edge from the panoramic image; Input the Mark image into a defect detection model, and perform defect detection according to the Mark point type; Determine the defect type of the Mark point, match the defect label from the label library according to the defect type, and output the defect recognition result.

2. The method according to claim 1, characterized in that Before defect detection, it also includes: Collect optical point defect images, label the defect images according to the mark point type, and build a training set and defect label library; The labeled defect images are sent to the defect detection model for learning and training, and the loss function weight is adjusted according to the output results until the loss is less than the set value.

3. The method according to claim 1 or 2, characterized in that: The method of intercepting the optical point Mark image around the board edge from the panoramic image comprises: Obtain the size structure of the material board corresponding to the panoramic image, and determine the board type; different board types correspond to different sizes; Determine the board edge area, number of Mark points and corresponding coordinate positions of the panoramic image according to the board structure; Capture the corresponding Mark image according to the Mark point coordinate position.

4. The method according to claim 3, characterized in that During model training, a database is built based on different board structures, and the board data includes material board size, segmentation area, number and type of Mark points, and the coordinates of the Mark points in the panoramic view; After the board type is determined, the board type data is matched from the database to determine the number of mark points, the type of mark points, and the coordinates of the mark points.

5. The method according to claim 3, characterized in that: The step of intercepting the corresponding Mark image according to the Mark point coordinate position includes: Determine the image center and expansion size according to the Mark point type, locate the Mark point coordinates from the panoramic image, and determine the size of the captured Mark image according to the image center and expansion size.

6. The method according to claim 5, characterized in that The defect detection process includes: Identify the bottom pad and the center mark point outline in the mark image and match it with the determined mark point type; different mark point types correspond to bottom pads and mark points of different shapes; When the bottom pads identified from the image match each other, the identified Mark point contour is matched with the theoretical Mark point contour; When the similarity difference between the identified Mark point contour and the theoretical Mark point contour matches exceeds the set threshold, it is determined to be a defective Mark point and the corresponding defect type.

7. The method according to claim 6, characterized in that The contour matching process includes: Determine the rectangular Mark border based on the identified Mark point edge contour and the enclosing area formed by the contour; Taking the four sides of the rectangular Mark as the center, rectangular images of set widths are respectively intercepted from the binarized image as damage analysis images; The erosion operation is performed on the intercepted damage analysis images respectively, and the overall variance of the pixel value fluctuation after the erosion operation is calculated. When the overall variance value is greater than the corresponding fluctuation threshold, it is determined that the Mark is defiled.

8. A device for detecting optical point defects on a plate edge, characterized in that: The device comprises: An extraction module is used to scan a panoramic image of a PNL material plate through a line scan camera, and capture an optical point Mark image around the edge of the plate from the panoramic image; A detection module, used for inputting the Mark image into a defect detection model and performing defect detection according to the Mark point type; The output module is used to determine the defect type of the mark point, match the defect label from the label library according to the defect type, and output the defect recognition result.

9. A computer device, characterized in that: The computer device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the method for obtaining optical point defect detection on the edge of a plate as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The readable storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set are loaded and executed by the processor to implement the method for obtaining optical point defect detection on the edge of a plate as described in any one of claims 1 to 7.

Images