Method and apparatus for defect detection of panel images

Abstract

This application discloses a method and apparatus for defect detection of panel images. The method includes: acquiring a first panel image according to a target detection cycle, wherein the target detection cycle is a dynamic detection cycle determined based on the load state and defect density of a macroscopic detection device; performing four-neighborhood calculation on the first panel image to obtain candidate defect regions, performing eight-neighborhood calculation on the candidate defect regions to obtain target defect regions; determining defects in the target defect regions based on a preset area threshold, and outputting defect detection results.

Description

Technical Field

[0001] This application relates to the field of image defect detection technology, and more specifically, to a method and apparatus for detecting defects in panel images. Background Technology

[0002] In high-end manufacturing fields such as semiconductors and LCD panels, macroscopic inspection equipment is widely used in the processing of ultra-large substrates, including common substrates such as silicon wafers and glass substrates. The substrate surface is prone to micron-level spot defects, such as bright spots, dark spots, and stains, due to process fluctuations, environmental interference, or material impurities. Failure to detect these defects in a timely manner will directly lead to a decrease in product yield and an increase in production costs. Therefore, efficient and accurate defect detection is a core requirement for ensuring production quality.

[0003] In existing technologies, spot defect detection mainly adopts fixed-cycle scanning and threshold segmentation related methods: one is to perform fixed-cycle inspections with a preset time or step size, and identify defects through grayscale comparison and template matching; the other is to use the traditional neighborhood comparison method, and achieve defect judgment through simple difference or binarization processing.

[0004] However, these solutions have significant drawbacks: fixed cycles cannot adapt to the varying operating speeds of macroscopic inspection equipment and panel sizes; cycles that are too long are prone to missed detections, while cycles that are too short result in wasted resources; traditional neighborhood comparisons are sensitive to noise and do not incorporate dynamic cycle adjustments, leading to high false detection rates against complex texture backgrounds; at the same time, single-neighbor comparisons, such as four-neighbor comparisons, are insufficient for capturing irregular defects, while eight-neighbor comparisons, although comprehensive, require a large amount of computation, resulting in time-consuming processing of ultra-large images, all of which fail to meet the real-time inspection needs of high-speed production lines. Summary of the Invention

[0005] The main purpose of this application is to provide a defect detection method and apparatus for panel images to solve the problems of rigid defect detection cycle, high false negative rate and long processing time of large images in macroscopic inspection equipment. It can realize adaptive adjustment of inspection cycle, improve detection efficiency and accuracy, and adapt to the real-time detection needs of high-speed production lines.

[0006] To achieve the above objectives, a first aspect of this application proposes a defect detection method for panel images, comprising: acquiring a first panel image according to a target detection cycle, wherein the target detection cycle is a dynamic detection cycle determined based on the load state and defect density of a macroscopic detection device; performing four-neighborhood calculation on the first panel image to obtain candidate defect regions; performing eight-neighborhood calculation on the candidate defect regions to obtain target defect regions; determining defects in the target defect regions based on a preset area threshold, and outputting defect detection results.

[0007] According to the defect detection method for panel images provided in this application, before acquiring the first panel image according to the target detection cycle, the method further includes: acquiring the operating load data and historical defect detection data of the macroscopic inspection equipment, wherein the operating load data includes the central processing unit (CPU) utilization rate and memory utilization rate; performing a weighted calculation on the CPU utilization rate and the memory utilization rate to obtain the load state; calculating the ratio of the total number of defects detected within a preset statistical window to the total area of ​​the panel that has been inspected based on the historical defect detection data to obtain the defect density; and dynamically adjusting the image inspection cycle according to the load state and the defect density to obtain the target detection cycle.

[0008] According to the defect detection method for panel images provided in this application, the step of dynamically adjusting the image inspection cycle based on the load state and the defect density to obtain the target detection cycle includes: obtaining an initial detection cycle. The target detection period is calculated according to the linear weighting formula. The linear weighting formula is as follows: ; in, Indicates the load adjustment coefficient. This represents the defect density adjustment coefficient. Indicates the load status. Indicates the load status baseline value. Indicates defect density, This represents the baseline value for defect density.

[0009] According to the defect detection method for a panel image provided in this application, obtaining the initial detection cycle includes: determining the initial detection cycle based on the maximum throughput of the macroscopic detection device and the resolution of the first panel image.

[0010] According to the defect detection method for panel images provided in this application, the step of performing four-neighborhood calculation on the first panel image to obtain candidate defect regions, and performing eight-neighborhood calculation on the candidate defect regions to obtain target defect regions, includes: when a first preset condition is met, first performing four-neighborhood calculation on the first panel image to obtain candidate defect regions, and then performing eight-neighborhood calculation on the candidate defect regions to obtain target defect regions; wherein, the first preset condition includes: the load state of the macroscopic detection device is not higher than a first load threshold, and the image background uniformity of the first panel image is lower than a first uniformity threshold; the load state of the macroscopic detection device is not higher than the first load threshold, and the image texture complexity of the first panel image is higher than a first texture complexity threshold.

[0011] According to the defect detection method for panel images provided in this application, after acquiring a first panel image according to a target detection cycle, the method further includes: when a second preset condition is met, performing four-neighborhood calculation only on the first panel image to obtain the target defect region; when a third preset condition is met, performing eight-neighborhood calculation only on the first panel image to obtain the target defect region; wherein, the second preset condition includes: the load state of the macroscopic detection device is higher than a first load threshold; the load state of the macroscopic detection device is not higher than the first load threshold, and the image background uniformity of the first panel image is higher than the first uniformity threshold, and the image texture complexity of the first panel image is not higher than the first texture complexity threshold, and the defect density is not higher than the first defect density threshold; the third preset condition includes: the defect density is higher than the first defect density threshold.

[0012] This application also provides a defect detection device for panel images, comprising the following modules: an acquisition module, a processing module, and an output module; the acquisition module is used to acquire a first panel image according to a target detection cycle, wherein the target detection cycle is a dynamic detection cycle determined based on the load state and defect density of a macroscopic detection device; the processing module is used to perform four-neighborhood calculation on the first panel image to obtain candidate defect regions, perform eight-neighborhood calculation on the candidate defect regions to obtain target defect regions, and determine defects in the target defect regions based on a preset area threshold; the output module is used to output the defect detection results.

[0013] According to the defect detection device for panel images provided in this application, the acquisition module is further configured to acquire the operating load data and historical defect detection data of the macroscopic inspection equipment, wherein the operating load data includes CPU utilization and memory utilization; the processing module is further configured to perform a weighted calculation on the CPU utilization and the memory utilization to obtain the load status; calculate the ratio of the total number of defects detected within a preset statistical window to the total area of ​​the panel that has been inspected based on the historical defect detection data to obtain the defect density; and dynamically adjust the image inspection cycle according to the load status and the defect density to obtain the target detection cycle.

[0014] According to the defect detection device for panel images provided in this application, the processing module is used to obtain an initial detection cycle. The target detection period is calculated according to the linear weighting formula. The linear weighting formula is as follows: ; in, Indicates the load adjustment coefficient. This represents the defect density adjustment coefficient. Indicates the load status. Indicates the load status baseline value. Indicates defect density, This represents the baseline value for defect density.

[0015] According to the present application, a defect detection device for a panel image is provided, wherein the processing module is configured to determine the initial detection cycle based on the maximum throughput of the macroscopic detection device and the resolution of the first panel image.

[0016] According to the defect detection device for panel images provided in this application, the processing module is used to, when a first preset condition is met, first perform four-neighborhood calculation on the first panel image to obtain candidate defect regions, and then perform eight-neighborhood calculation on the candidate defect regions to obtain target defect regions; wherein, the first preset condition includes: the load state of the macroscopic detection device is not higher than a first load threshold, and the image background uniformity of the first panel image is lower than a first uniformity threshold; the load state of the macroscopic detection device is not higher than the first load threshold, and the image texture complexity of the first panel image is higher than a first texture complexity threshold.

[0017] According to the defect detection device for panel images provided in this application, the processing module is configured to perform four-neighborhood calculation on the first panel image only when a second preset condition is met to obtain the target defect region; and to perform eight-neighborhood calculation on the first panel image only when a third preset condition is met to obtain the target defect region; wherein, the second preset condition includes: the load state of the macroscopic detection device is higher than a first load threshold; the load state of the macroscopic detection device is not higher than the first load threshold, and the image background uniformity of the first panel image is higher than the first uniformity threshold, and the image texture complexity of the first panel image is not higher than the first texture complexity threshold, and the defect density is not higher than the first defect density threshold; the third preset condition includes: the defect density is higher than the first defect density threshold.

[0018] This application also provides an electronic 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 a defect detection method for a panel image as described above.

[0019] This application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a defect detection method for a panel image as described above.

[0020] This application also provides a computer program product, including a computer program that, when executed by a processor, implements a defect detection method for a panel image as described above.

[0021] The technical solutions provided by the embodiments of this application may include the following beneficial effects: By adopting a dynamic target detection cycle image acquisition method based on the load status and defect density of macroscopic detection equipment, it is possible to accurately adapt to the variation characteristics of equipment operating speed and panel size, effectively avoid the missed detection problem and resource waste that are easily caused by fixed cycles, and significantly improve detection efficiency. By employing a step-by-step detection logic that first performs four-neighbor domain calculation to quickly screen candidate defect regions and then verifies the candidate regions using eight-neighbor domain calculation, the detection speed and comprehensive coverage are cleverly balanced. This not only solves the problem of insufficient ability of a single four-neighbor domain to capture irregular defects, but also avoids the drawbacks of large computational load and long time consumption caused by using eight-neighbor domain calculation throughout the process. By combining preset area thresholds to make a final judgment on the target defect area, noise interference can be effectively filtered out, and the false detection rate under complex texture backgrounds can be significantly reduced.

[0022] In summary, this application achieves efficient and accurate detection of speckle defects by working synergistically from three dimensions: adaptive detection cycle, neighborhood comparison optimization, and noise filtering. It can fully meet the real-time detection needs of high-speed production lines and provide strong support for improving product yield and reducing production costs. Attached Figure Description

[0023] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application. In the drawings: Figure 1 One of the flowcharts for the defect detection method of panel images provided in this application; Figure 2 This is a schematic diagram of the four-neighborhood pixel calculation provided in this application; Figure 3 This is a schematic diagram of the eight-neighborhood pixel calculation provided in this application; Figure 4 The second schematic flowchart of the defect detection method for panel images provided in this application; Figure 5 This is a schematic diagram of the structure of the defect detection device for panel images provided by the present invention; Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0026] In this application, the terms "upper," "lower," "left," "right," "front," "rear," "top," "bottom," "inner," "outer," "middle," "vertical," "horizontal," "lateral," and "longitudinal" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are primarily for the purpose of better describing this application and its embodiments, and are not intended to limit the indicated device, element, or component to having a specific orientation, or to be constructed and operated in a specific orientation.

[0027] Furthermore, in addition to indicating location or positional relationship, some of the aforementioned terms may also have other meanings. For example, the term "above" may also be used in certain circumstances to indicate a certain dependency or connection relationship. Those skilled in the art can understand the specific meaning of these terms in this application based on the specific circumstances.

[0028] Furthermore, the terms "installation," "setup," "equipped with," "connection," "linked," and "socketing" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral structure; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or an internal connection between two devices, components, or parts. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0029] This application describes some exemplary embodiments for illustrative purposes. It should be understood that this application may be implemented in other ways not specifically shown in the accompanying drawings.

[0030] like Figure 1 As shown, this application provides a method for detecting defects in a panel image, which can be applied to a panel image defect detection device. The method may include steps S101-S103: S101, The panel image defect detection device acquires the first panel image according to the target detection cycle.

[0031] The target detection cycle mentioned above is a dynamic detection cycle determined based on the load status and defect density of the macroscopic detection equipment. The load status reflects the real-time resource utilization of the equipment, and the defect density reflects the quality status of the production process. In other words, the target detection cycle is not a fixed value and can change dynamically according to the load status and defect density.

[0032] It should be noted that the aforementioned macroscopic inspection equipment refers to manufacturing equipment used to process ultra-large substrates such as semiconductors and LCD panels. It typically includes a high-resolution line scan camera and an industrial personal computer (PC) processing platform, which can realize the acquisition and processing of images on the substrate surface.

[0033] It should be noted that the aforementioned first panel image refers to an image containing information about the surface of a semiconductor or liquid crystal panel substrate, captured by a high-resolution camera of a macroscopic inspection device. This first panel image may contain micron-level spot defects, such as bright spots, dark spots, and stains.

[0034] S102. The panel image defect detection device performs four-neighborhood calculation on the first panel image to obtain candidate defect regions, and performs eight-neighborhood calculation on the candidate defect regions to obtain target defect regions.

[0035] Optionally, the step of performing four-neighborhood calculation on the first panel image to obtain candidate defect regions, and performing eight-neighborhood calculation on the candidate defect regions to obtain target defect regions, includes: when a first preset condition is met, first performing four-neighborhood calculation on the first panel image to obtain candidate defect regions, and then performing eight-neighborhood calculation on the candidate defect regions to obtain target defect regions; wherein, the first preset condition includes: the load state of the macroscopic detection device is not higher than a first load threshold, and the image background uniformity of the first panel image is lower than a first uniformity threshold; the load state of the macroscopic detection device is not higher than the first load threshold, and the image texture complexity of the first panel image is higher than a first texture complexity threshold.

[0036] Specifically, when the load of the macroscopic inspection equipment is not higher than the first load threshold, it means that the equipment has sufficient computing resources to support high-precision inspection. When the uniformity of the background of the first panel image is lower than the first uniformity threshold or the complexity of the image texture is higher than the first uniformity threshold, the pixel point with a sudden change in gray value can be quickly found by four-neighbor calculation alone, but it is easy to miss irregular defects or misjudge texture. Therefore, eight-neighbor calculation can be further combined to improve the detection accuracy, thereby achieving a balance between resource adaptation and detection accuracy.

[0037] The four-neighbor calculation for the first panel image includes: like Figure 2 As shown, for each pixel in the first panel image First, obtain the grayscale values ​​of the four adjacent pitch distance pixels (top, bottom, left, and right), which are as follows: , , , The range of values ​​for pitch is: , Indicates the width of the rectangular area to be inspected. Indicates the height of the rectangular area to be inspected; Then, the mean gray value of the four neighborhoods is calculated using these four gray values. ; Calculate the pixels again With the mean gray level Gray variance: Gray-scale variance It can be used to measure pixels. The degree of grayscale difference between the pixel and its four surrounding points. The larger the value, the more incompatible the pixel is with its surrounding environment, and the higher the probability that it is a defect.

[0038] Finally, the grayscale variance Exceeding the preset variance threshold The pixels are marked as candidate defect points, and the area composed of all candidate defect points is determined as the candidate defect region.

[0039] It should be noted that the candidate defect region obtained by four-neighbor calculation has the characteristics of wide coverage and fast screening speed, but the candidate defect region may contain noise points and pseudo-defects formed by texture interference, resulting in low screening accuracy.

[0040] The eight-neighbor calculation for the candidate defect region includes: like Figure 3 As shown, for each pixel in the candidate defect region, the gradient components can be obtained by first approximating the first-order partial derivative using the central difference method. ; ; Then, the gradient magnitude of the eight-neighborhood is calculated based on the gradient components. : Gradient magnitude It represents the rate and direction of grayscale change in the image at that point; true defect edges usually have large gradient magnitudes.

[0041] Finally, the gradient magnitude Exceeding the preset gradient threshold The target defect area is obtained by preserving the affected area.

[0042] It should be noted that the target defect region eliminates noise points and false defects in the candidate defect region through gradient magnitude analysis, retaining only the real defect region with obvious edge features, which significantly improves the accuracy of defect identification. At the same time, since only the candidate defect region is calculated in eight neighborhoods, the problem of large amount of computation and long time consumption caused by the calculation of eight neighborhoods in the whole region is avoided.

[0043] Optionally, after acquiring the first panel image according to the target detection cycle, the method further includes: when a second preset condition is met, performing four-neighborhood calculation only on the first panel image to obtain the target defect region; when a third preset condition is met, performing eight-neighborhood calculation only on the first panel image to obtain the target defect region; wherein, the second preset condition includes: the load state of the macroscopic detection device is higher than a first load threshold; the load state of the macroscopic detection device is not higher than the first load threshold, and the image background uniformity of the first panel image is higher than the first uniformity threshold, and the image texture complexity of the first panel image is not higher than the first texture complexity threshold, and the defect density is not higher than the first defect density threshold; the third preset condition includes: the defect density is higher than the first defect density threshold.

[0044] Specifically, when the load of the macroscopic inspection equipment exceeds the first load threshold, the equipment's operating resources are strained. Performing complex dual-neighborhood calculations can easily lead to system lag. In this case, a four-neighborhood calculation can quickly screen out suspected defect areas. Although these areas may contain a small amount of interference information, they can avoid resource overload while ensuring the continuity of inspection. When the load of the macroscopic inspection equipment is not higher than the first load threshold, but the uniformity of the background of the first panel image is higher than the first uniformity threshold and the image texture complexity is not higher than the first texture complexity threshold, and the defect density is not higher than the first defect density threshold, it means that the production process is stable and there is little image interference. The grayscale variance analysis of a single four-neighborhood calculation can meet the defect detection requirements without the need for additional eight-neighborhood verification, which ensures both detection accuracy and simplifies the calculation process.

[0045] When the defect density is higher than the first defect density threshold, it indicates that there may be an abnormality in the production process. It is necessary to minimize the risk of missed detection. At this time, regardless of the load status of the macroscopic detection equipment, the background uniformity and texture complexity of the first panel image, the four-neighbor preliminary screening step is skipped, and the eight-neighbor calculation is directly performed on the entire area of ​​the first panel image.

[0046] It should be noted that the above solution can dynamically match the optimal detection mode according to the equipment operating status, image features, and production quality. When equipment resources are scarce, it prioritizes detection efficiency and continuity. When production is stable and interference is minimal, it balances efficiency and basic accuracy. When production is abnormal, it fully guarantees detection comprehensiveness and accuracy. This effectively solves the shortcomings of traditional fixed detection modes that cannot adapt to complex scenarios. It avoids resource waste or system overload, and reduces the risk of missed detections and false detections. Ultimately, it achieves the optimal balance between detection efficiency and accuracy in different scenarios, and can fully meet the real-time detection needs of high-speed production lines.

[0047] S103. The panel image defect detection device determines the defect in the target defect area based on a preset area threshold and outputs the defect detection result.

[0048] Specifically, morphological filtering can be applied to the target defect area first. Dilation operations are used to fill in the tiny voids within the defect area, and erosion operations are then used to eliminate small protrusions at the edges of the defect area, making its outline more regular and avoiding area calculation errors caused by residual noise or irregular defect edges. Subsequently, the total number of pixels in the morphologically filtered defect area is counted. Combined with the mapping relationship between the pixel resolution of the first panel image and the actual panel size, the total number of pixels is converted into the actual physical area of ​​the target defect area.

[0049] The calculated actual physical area is then compared with a preset area threshold. If the actual physical area of ​​the target defect region is greater than or equal to the preset area threshold, the region is determined to be a real and valid defect. If the actual physical area of ​​the target defect region is less than the preset area threshold, the region is determined to be invalid interference and is removed.

[0050] Finally, all area information determined to be genuine and valid defects is integrated, including the coordinates, actual area, and shape of the defects, to generate and output a standardized defect detection report, providing accurate data support for subsequent production process adjustments and screening of non-conforming products.

[0051] In this embodiment, by employing a dynamic target detection cycle image acquisition based on the load state and defect density of the macroscopic detection equipment, it can accurately adapt to the variations in equipment operating speed and panel size, effectively avoiding the missed detection problems and resource waste that are easily caused by fixed cycles, and significantly improving detection efficiency. The step-by-step detection logic, which first performs rapid initial screening of candidate defect regions through four-neighbor domain calculation and then verifies the candidate regions through eight-neighbor domain calculation, cleverly balances detection speed and comprehensive coverage. This solves the problem of insufficient capture capability of a single four-neighbor domain for irregular defects, while avoiding the drawbacks of large computational load and long processing time caused by using eight-neighbor domain calculation throughout the process. By combining a preset area threshold for final determination of the target defect region, noise interference can be effectively filtered, significantly reducing the false detection rate against complex texture backgrounds. In summary, this application achieves efficient and accurate detection of speckle defects through synergistic efforts in three dimensions: adaptive detection cycle, optimized neighborhood comparison, and noise filtering. This fully meets the real-time detection needs of high-speed production lines, providing strong support for improving product yield and reducing production costs.

[0052] like Figure 4 As shown, before acquiring the first panel image according to the target detection cycle, the defect detection method for panel images provided in this application embodiment may further include S104-S107: S104. The panel image defect detection device collects the operating load data and historical defect detection data of the macroscopic detection equipment.

[0053] The operational load data includes CPU utilization and memory utilization. Both CPU utilization and memory utilization range from 0% to 100%. Historical defect detection data refers to all relevant data generated and stored during all panel defect detection processes completed by the macroscopic inspection equipment before acquiring the first panel image. This data may include: basic defect information, such as the number, coordinates, actual area, and type of actual valid defects identified in each detection; detection scenario information, such as the panel size at the time of defect detection, the resolution of the first panel image, and the detection cycle; and associated auxiliary information, such as the load status of the macroscopic inspection equipment at the time of detection and the defect density calculation results at that time.

[0054] S105. The panel image defect detection device performs a weighted calculation of the CPU usage rate and the memory usage rate to obtain the load status.

[0055] Specifically, the weighting coefficient of CPU utilization can be preset. Weighting coefficients for memory usage ,in, and The values ​​of all values ​​are in the range [0, 1], and the constraints are satisfied. This constraint ensures that the load condition is a reasonable weighted average of CPU utilization and memory utilization.

[0056] After obtaining the CPU and memory usage of the macroscopic detection equipment, it can be based on weighting coefficients. and weighting coefficients The load status is calculated using the following formula. : ; in, This indicates the CPU utilization rate collected in real time. This indicates the memory usage rate collected in real time.

[0057] S106. The panel image defect detection device calculates the ratio of the total number of defects detected within a preset statistical window to the total area of ​​the panel that has been inspected based on the historical defect detection data, and obtains the defect density.

[0058] Specifically, the duration or detection range of the statistical window can be preset, such as the most recent hour or the 10 most recently inspected panels; then, the number of all judged valid defects within the preset statistical window can be extracted from historical defect detection data and recorded as the total number of defects. Simultaneously, the sum of the areas of all panels that have undergone comprehensive inspection by the macroscopic inspection equipment within the preset statistical window is extracted and recorded as the total area of ​​the inspected panels. Finally, through the formula Defect density was calculated .

[0059] S107. The panel image defect detection device dynamically adjusts the image inspection cycle according to the load state and the defect density to obtain the target detection cycle.

[0060] Optionally, the step of dynamically adjusting the image inspection cycle based on the load state and the defect density to obtain the target detection cycle includes: obtaining an initial detection cycle. The target detection period is calculated according to the linear weighting formula. The linear weighting formula is as follows: ; in, Indicates the load adjustment coefficient. This represents the defect density adjustment coefficient. Indicates the load status. Indicates the load status baseline value. Indicates defect density, This represents the baseline value for defect density.

[0061] Specifically, load status reference value The ideal load condition value for the macroscopic inspection equipment during normal and stable operation; the defect density benchmark value. Both are the ideal defect density values ​​when the production process quality is stable, and are pre-calibrated based on the actual production scenario and equipment performance.

[0062] Load adjustment coefficient A positive number is used to control the sensitivity of the load state to the target detection cycle. Greater than the load condition baseline value hour, A positive value can reduce the target detection cycle. Greater than the initial detection period This allows for an extension of the detection cycle, freeing up computing resources for macroscopic detection equipment and preventing resource overload. Similarly, when the load state... Less than the load condition baseline value hour, A negative value can reduce the target detection cycle. Less than the initial detection period In this way, the inspection cycle can be shortened and idle resources can be fully utilized.

[0063] Defect density adjustment coefficient A negative number is used to control the sensitivity of defect density to the target detection cycle. When the defect density is negative... Greater than the defect density benchmark value hour, It is a positive value. A negative value can reduce the target detection cycle. Less than the initial detection period This allows for a shorter inspection cycle, thereby increasing the frequency of monitoring the production process and enabling timely detection of production anomalies. Similarly, when the defect density... Less than the defect density benchmark value hour, It is a negative value. A positive value can reduce the target detection cycle. Greater than the initial detection period This allows for an extension of the testing cycle, freeing up computing resources for macroscopic testing equipment and preventing resource overload.

[0064] Optionally, obtaining the initial detection period includes: determining the initial detection period based on the maximum throughput of the macroscopic detection device and the resolution of the first panel image.

[0065] Specifically, the maximum throughput of the macroscopic inspection equipment refers to the maximum amount of image data that the equipment can stably process per unit time, which is determined by the hardware performance of the equipment. The higher the resolution of the first panel image, the greater the amount of computation and data transmission required for a single image processing. The initial inspection cycle is not a fixed value, but a function jointly determined by the equipment hardware capabilities and task requirements. Its specific functional relationship can be calibrated in combination with the actual equipment performance to ensure that the initial inspection cycle is a reasonable benchmark cycle for the equipment to complete the acquisition and basic processing of a single first panel image under the premise of stable operation at full load.

[0066] In this embodiment, the load status and defect density can be obtained through standardized weighted calculation based on the operating load data and historical defect detection data of the macroscopic detection equipment. Then, the detection cycle can be dynamically adjusted using a linear weighted formula, so that the target detection cycle can adapt to the equipment operating status and production quality in real time. This avoids the drawbacks of fixed cycles being unable to cope with equipment load fluctuations and production anomalies. Furthermore, the scientific and reasonable nature of the cycle adjustment is ensured through precise parameter calibration and formula calculation, effectively improving detection efficiency and resource utilization. At the same time, it provides a highly adaptable cycle basis for subsequent accurate detection.

[0067] The foregoing mainly describes the solutions provided by the embodiments of this application from a methodological perspective. To achieve the above functions, it includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, the embodiments of this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0068] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0069] It should be noted that the apparatus in the embodiments of this application includes a virtual device and a physical device. The virtual device can be a defect detection device for panel images, and the physical device can include electronic devices, computer storage media, and computer program products.

[0070] The panel image defect detection method provided in this application can be executed by a panel image defect detection device or a control module for panel image defect detection within that device. This application uses a panel image defect detection device executing the panel image defect detection method as an example to illustrate the panel image defect detection device provided in this application.

[0071] It should be noted that the embodiments of this application can divide the panel image defect detection device into functional modules according to the above method examples. For example, each function can be divided into its own functional modules, or two or more functions can be integrated into one processing module. The integrated modules can be implemented in hardware or as software functional modules. Optionally, the module division in the embodiments of this application is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0072] like Figure 5 As shown in the figure, this application embodiment provides a defect detection device 500 for panel images. The defect detection device 500 for panel images includes: an acquisition module 501, a processing module 502, and an output module 503; the acquisition module 501 is used to acquire a first panel image according to a target detection cycle, the target detection cycle being a dynamic detection cycle determined based on the load state and defect density of a macroscopic detection device; the processing module 502 is used to perform four-neighborhood calculation on the first panel image to obtain candidate defect regions, perform eight-neighborhood calculation on the candidate defect regions to obtain target defect regions, and determine defects in the target defect regions based on a preset area threshold; the output module 503 is used to output the defect detection results.

[0073] Optionally, the acquisition module 501 is further configured to acquire the operating load data and historical defect detection data of the macroscopic inspection equipment, wherein the operating load data includes CPU utilization and memory utilization; the processing module 502 is further configured to perform a weighted calculation on the CPU utilization and the memory utilization to obtain the load status; calculate the ratio of the total number of defects detected within a preset statistical window to the total area of ​​the panel that has been inspected based on the historical defect detection data to obtain the defect density; and dynamically adjust the image inspection cycle according to the load status and the defect density to obtain the target detection cycle.

[0074] Optionally, the processing module 502 is used to obtain the initial detection period. The target detection period is calculated according to the linear weighting formula. The linear weighting formula is as follows: ; in, Indicates the load adjustment coefficient. This represents the defect density adjustment coefficient. Indicates the load status. Indicates the load status baseline value. Indicates defect density, This represents the baseline value for defect density.

[0075] Optionally, the processing module 502 is configured to determine the initial detection cycle based on the maximum throughput of the macroscopic detection device and the resolution of the first panel image.

[0076] Optionally, the processing module 502 is configured to, when a first preset condition is met, first perform four-neighborhood calculation on the first panel image to obtain a candidate defect region, and then perform eight-neighborhood calculation on the candidate defect region to obtain a target defect region; wherein, the first preset condition includes: the load state of the macroscopic detection device is not higher than a first load threshold, and the image background uniformity of the first panel image is lower than a first uniformity threshold; the load state of the macroscopic detection device is not higher than the first load threshold, and the image texture complexity of the first panel image is higher than a first texture complexity threshold.

[0077] Optionally, the processing module 502 is configured to perform four-neighborhood calculation on the first panel image only when a second preset condition is met, to obtain the target defect region; and to perform eight-neighborhood calculation on the first panel image only when a third preset condition is met, to obtain the target defect region; wherein the second preset condition includes: the load state of the macroscopic detection device is higher than a first load threshold; the load state of the macroscopic detection device is not higher than the first load threshold, and the image background uniformity of the first panel image is higher than the first uniformity threshold, and the image texture complexity of the first panel image is not higher than the first texture complexity threshold, and the defect density is not higher than the first defect density threshold; the third preset condition includes: the defect density is higher than the first defect density threshold.

[0078] In this embodiment, by employing a dynamic target detection cycle image acquisition based on the load state and defect density of the macroscopic detection equipment, it can accurately adapt to the variations in equipment operating speed and panel size, effectively avoiding the missed detection problems and resource waste that are easily caused by fixed cycles, and significantly improving detection efficiency. The step-by-step detection logic, which first performs rapid initial screening of candidate defect regions through four-neighbor domain calculation and then verifies the candidate regions through eight-neighbor domain calculation, cleverly balances detection speed and comprehensive coverage. This solves the problem of insufficient capture capability of a single four-neighbor domain for irregular defects, while avoiding the drawbacks of large computational load and long processing time caused by using eight-neighbor domain calculation throughout the process. By combining a preset area threshold for final determination of the target defect region, noise interference can be effectively filtered, significantly reducing the false detection rate against complex texture backgrounds. In summary, this application achieves efficient and accurate detection of speckle defects through synergistic efforts in three dimensions: adaptive detection cycle, optimized neighborhood comparison, and noise filtering. This fully meets the real-time detection needs of high-speed production lines, providing strong support for improving product yield and reducing production costs.

[0079] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include a processor 610, a communication interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communication interface 620, and the memory 630 communicate with each other through the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a defect detection method for a panel image. The method includes: acquiring a first panel image according to a target detection cycle, wherein the target detection cycle is a dynamic detection cycle determined based on the load state and defect density of the macroscopic detection device; performing four-neighborhood calculation on the first panel image to obtain candidate defect regions; performing eight-neighborhood calculation on the candidate defect regions to obtain target defect regions; determining defects in the target defect regions based on a preset area threshold, and outputting the defect detection result.

[0080] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0081] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the panel image defect detection method provided by the above methods. The method includes: acquiring a first panel image according to a target detection cycle, wherein the target detection cycle is a dynamic detection cycle determined based on the load state and defect density of a macroscopic detection device; performing four-neighborhood calculation on the first panel image to obtain candidate defect regions; performing eight-neighborhood calculation on the candidate defect regions to obtain target defect regions; determining defects in the target defect regions based on a preset area threshold; and outputting defect detection results.

[0082] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements a defect detection method for a panel image provided by the methods described above. The method includes: acquiring a first panel image according to a target detection cycle, wherein the target detection cycle is a dynamic detection cycle determined based on the load state and defect density of a macroscopic detection device; performing four-neighborhood calculation on the first panel image to obtain candidate defect regions; performing eight-neighborhood calculation on the candidate defect regions to obtain target defect regions; determining defects in the target defect regions based on a preset area threshold; and outputting defect detection results.

[0083] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0084] Obviously, those skilled in the art should understand that the various units or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps into a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0085] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for defect detection in panel images, characterized in that, include: The first panel image is acquired according to the target detection cycle, which is a dynamic detection cycle determined based on the load state and defect density of the macroscopic detection equipment; the smaller the load state, the smaller the target detection cycle, and the larger the load state, the larger the target detection cycle. When the first preset condition is met, the first panel image is first subjected to four-neighbor calculation to obtain a candidate defect region, and then the candidate defect region is subjected to eight-neighbor calculation to obtain a target defect region; when the second preset condition is met, only the first panel image is subjected to four-neighbor calculation to obtain the target defect region. When the third preset condition is met, only the first panel image is subjected to eight-neighbor calculation to obtain the target defect area; The target defect area is determined based on a preset area threshold, and the defect detection result is output. The first preset conditions include: the load state of the macroscopic detection device is not higher than the first load threshold and the image background uniformity of the first panel image is lower than the first uniformity threshold; the load state of the macroscopic detection device is not higher than the first load threshold and the image texture complexity of the first panel image is higher than the first texture complexity threshold. The second preset conditions include: the load state of the macroscopic detection device is higher than the first load threshold, the load state of the macroscopic detection device is not higher than the first load threshold, the image background uniformity of the first panel image is higher than the first uniformity threshold, the image texture complexity of the first panel image is not higher than the first texture complexity threshold, and the defect density is not higher than the first defect density threshold. The third preset condition includes: the defect density is higher than the first defect density threshold.

2. The defect detection method for panel images according to claim 1, characterized in that, Before acquiring the first panel image according to the target detection cycle, the method further includes: Collect the operating load data and historical defect detection data of the macroscopic inspection equipment. The operating load data includes the CPU utilization rate and memory utilization rate. The load status is obtained by weighting the CPU utilization and the memory utilization. The defect density is obtained by calculating the ratio of the total number of defects detected within the preset statistical window to the total area of ​​the panel that has been inspected based on the historical defect detection data. The target detection cycle is obtained by dynamically adjusting the image inspection cycle based on the load state and the defect density.

3. The defect detection method for panel images according to claim 2, characterized in that, The step of dynamically adjusting the image inspection cycle based on the load state and the defect density to obtain the target detection cycle includes: Obtain the initial detection period ; The target detection period is calculated using the linear weighting formula. The linear weighting formula is as follows: ； in, Indicates the load adjustment coefficient. This represents the defect density adjustment coefficient. Indicates the load status. Indicates the load status baseline value. Indicates defect density, This represents the baseline value for defect density.

4. The defect detection method for panel images according to claim 3, characterized in that, The process of obtaining the initial detection period includes: The initial detection cycle is determined based on the maximum throughput of the macroscopic detection device and the resolution of the first panel image.

5. A defect detection device for panel images, characterized in that, include: The module consists of a data acquisition module, a processing module, and an output module. The acquisition module is used to acquire the first panel image according to the target detection cycle, which is a dynamic detection cycle determined based on the load state and defect density of the macroscopic detection equipment; the smaller the load state, the smaller the target detection cycle, and the larger the load state, the larger the target detection cycle. The processing module is configured to, when a first preset condition is met, first perform four-neighbor calculation on the first panel image to obtain a candidate defect region, and then perform eight-neighbor calculation on the candidate defect region to obtain a target defect region; when a second preset condition is met, only perform four-neighbor calculation on the first panel image to obtain the target defect region; and when a third preset condition is met, only perform eight-neighbor calculation on the first panel image to obtain the target defect region. Defect determination is performed on the target defect area based on a preset area threshold. The output module is used to output the defect detection results; The first preset conditions include: the load state of the macroscopic detection device is not higher than the first load threshold and the image background uniformity of the first panel image is lower than the first uniformity threshold; the load state of the macroscopic detection device is not higher than the first load threshold and the image texture complexity of the first panel image is higher than the first texture complexity threshold. The second preset conditions include: the load state of the macroscopic detection device is higher than the first load threshold, the load state of the macroscopic detection device is not higher than the first load threshold, the image background uniformity of the first panel image is higher than the first uniformity threshold, the image texture complexity of the first panel image is not higher than the first texture complexity threshold, and the defect density is not higher than the first defect density threshold. The third preset condition includes: the defect density is higher than the first defect density threshold.

6. An electronic 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 the computer program, it implements the defect detection method for the panel image as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the defect detection method for the panel image as described in any one of claims 1 to 4.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the defect detection method for the panel image as described in any one of claims 1 to 4.

Images