Methods, apparatus, computer equipment, media and systems for detecting defects in manifolds

Abstract

This application relates to a method, apparatus, computer equipment, medium, and system for detecting defects in current collectors. The method includes: acquiring first illumination image information obtained by image acquisition of the current collector under a first illumination environment; performing preliminary defect detection on the current collector based on the first illumination image information and determining that the current collector has a defect of undetermined type; acquiring second illumination image information obtained by image acquisition of the current collector under a second illumination environment, wherein a first illumination angle of the first illumination environment is higher than a second illumination angle of the second illumination environment; performing three-dimensional contamination defect detection on the undetermined type defect based on the first and second illumination image information to determine the defect category of the undetermined type defect; and determining the defect detection result of the current collector based on the defect category. This method can accurately identify the defect category of the current collector, reduce the probability of missed or over-detection, and improve battery production efficiency.

Description

Technical Field

[0001] This application relates to the field of battery testing technology, and in particular to a method, apparatus, computer equipment, storage medium, system, and computer program product for detecting defects in manifolds. Background Technology

[0002] In the production process of cylindrical batteries, current collector welding is a crucial step in achieving a complete connection between the positive and negative electrodes and the current collector. The welding quality of the current collector directly affects the stability and safety of the battery. If metal particles generated during welding, such as weld slag larger than 100μm, are not effectively removed, the battery will exhibit severe self-discharge behavior. Therefore, to improve battery operational safety, defect detection of the welded current collector is necessary to reduce the probability of defective batteries being produced.

[0003] Currently, the common method for defect detection in current collectors involves acquiring images of the welded current collectors and then using image recognition models to identify defects in the images. However, because weld slag and dirt defects in current collectors often have similar morphologies, and are even more difficult to distinguish when they are small, current defect detection methods cannot accurately differentiate between defect categories. This can easily lead to misjudgments, resulting in either missed or over-detection, which negatively impacts the production efficiency of cylindrical batteries. Summary of the Invention

[0004] Therefore, it is necessary to provide a current collector defect detection method, apparatus, computer equipment, computer-readable storage medium, computer program product, and system that can accurately identify the defect type of the current collector, reduce the probability of missed or over-detection, and improve battery production efficiency, in order to address the above-mentioned technical problems.

[0005] Firstly, this application provides a method for detecting defects in a manifold, the method comprising:

[0006] Acquire the first illumination image information obtained by image acquisition of the collector disk under the first illumination environment;

[0007] If a preliminary defect detection is performed on the collector disk based on the first lighting image information, and it is determined that the collector disk has a defect of undetermined category, then second lighting image information is obtained by acquiring images of the collector disk under a second lighting environment; the first illumination angle of the first lighting environment is higher than the second illumination angle of the second lighting environment; the defect of undetermined category refers to a defect whose category cannot be distinguished by using the first lighting image information alone;

[0008] Based on the first lighting image information and the second lighting image information, a three-dimensional dirt defect detection is performed on the undetermined category of the defect to determine the defect category; the imaging of the three-dimensional dirt defect differs between the first lighting environment and the second lighting environment;

[0009] The defect detection result of the manifold is determined based on the defect category.

[0010] The method in the above embodiments only requires providing the current collector with a first illumination environment and a second illumination environment with different illumination angles, and adding one image acquisition. It can utilize the imaging differences of the current collector under the first illumination environment and the second illumination environment to highlight the morphological features of defects, thereby achieving the effect of accurately identifying three-dimensional and non-three-dimensional dirt defects. While effectively improving the detection accuracy of current collector defects and thus improving battery production efficiency, it can also effectively reduce the detection cost of accurate current collector defect detection.

[0011] In some embodiments, determining the defect detection result of the manifold based on the defect category includes:

[0012] In the case where the defect category is the three-dimensional dirt defect, the second defect size of the three-dimensional dirt defect is determined based on the second defect contour of the defect to be determined in the second illumination image information;

[0013] If the second defect size is less than the first acceptable size threshold set for three-dimensional contamination defects, the three-dimensional contamination defect detection of the manifold is determined to be qualified.

[0014] In the above embodiments, determining the defect size of a three-dimensional contamination defect using the second defect profile allows the determined defect size to be closer to the actual size of the three-dimensional contamination defect, effectively improving the accuracy of three-dimensional contamination defect conformity testing. Simultaneously, performing size compliance re-judgment for three-dimensional contamination defects can effectively reduce the risk of over-testing, improving battery production efficiency and reducing battery production costs.

[0015] In some embodiments, the method further includes:

[0016] If, based on the first illumination image information, it is determined that there is a defect to be detected in the collector disk, the first defect contour of the defect to be detected is extracted;

[0017] The first defect size of the defect to be detected is determined based on the first defect profile.

[0018] If the first defect size is less than a second acceptable size threshold set for planar contamination defects, and is greater than or equal to the first acceptable size threshold, the outline position of the first defect outline is determined; the second acceptable size threshold is greater than the first acceptable size threshold.

[0019] If, based on the contour position, it is determined that the defect to be detected requires three-dimensional dirt defect detection, the defect to be detected is classified as a defect of undetermined category.

[0020] In the above embodiments, by comparing the first defect size of the defect to be detected with the first qualified size threshold and the second qualified size threshold respectively, it is possible to quickly determine whether the defect to be detected is an obviously unqualified defect or an obviously qualified defect. Only when it is determined that the defect to be detected is not an obviously unqualified defect or an obviously qualified defect will the contour position of the defect to be detected be followed to determine whether the defect to be detected needs to be subjected to three-dimensional dirt defect detection, which effectively improves the detection efficiency of manifold defect detection.

[0021] In some embodiments, the method further includes:

[0022] For the weld bead area of ​​the manifold, the first lighting image information is used for region recognition to obtain the region outline position of the weld bead area;

[0023] The contour position of the defect to be detected is matched with the contour position of the region to obtain the defect position matching result;

[0024] If the defect location matching result indicates that the defect to be detected is not located in the weld area, it is determined that the defect to be detected needs to undergo three-dimensional contamination defect detection.

[0025] In the above embodiments, by matching the contour position of the defect to be detected with the regional contour position of the weld area, it is possible to quickly determine whether the defect to be detected is a weld defect. If it is determined that it is not a weld defect, subsequent three-dimensional contamination defect detection can be performed, which effectively improves the detection efficiency and detection accuracy of manifold defect detection.

[0026] In some embodiments, the method further includes:

[0027] If the defect location matching result indicates that the defect to be detected is located in the weld area, the defect to be detected is determined to be a pinhole burst defect;

[0028] If the first defect size is less than the third acceptable size threshold set for the pinhole burst defect, the pinhole burst defect detection of the manifold is determined to be qualified; the third acceptable size threshold is greater than the first acceptable size threshold and less than the second acceptable size threshold.

[0029] In the above embodiments, for pinhole burst defects located in the weld area, the corresponding third qualified size threshold is used to perform qualification testing, which can make the defect qualification testing process match the actual defect category more accurately and effectively improve the detection accuracy of manifold defect detection.

[0030] In some embodiments, the step of performing region recognition on the first illumination image information to obtain the region contour position of the weld bead area for the manifold includes:

[0031] Use a preset box of interest to extract the initial region information containing the weld area from the first lighting image information;

[0032] The initial region information is input into a pre-set weld width detection model. The initial region information is then truncated and corrected based on the weld width of the weld region to obtain the region outline position of the weld region.

[0033] In the above embodiments, a large area of ​​the weld bead region is captured from the first illumination image by pre-setting a bounding box of interest, and then the weld bead width obtained by identification is used to cut and correct the large area of ​​the weld bead region to obtain the region outline position of the weld bead region. This can effectively improve the recognition accuracy of the region outline position of the weld bead region and provide an accurate data basis for subsequent judgment on whether the defect to be detected is within the weld bead region.

[0034] In some embodiments, the outline position of the defect to be detected includes the defect outline position coordinates of the defect to be detected, and the region outline position includes the region outline position coordinates of the weld bead region.

[0035] The step of matching the contour position of the defect to be detected with the contour position of the region to obtain the defect position matching result includes:

[0036] The defect contour position coordinates of the defect to be detected are mapped to the region contour position coordinates of the weld area to determine the area of ​​the overlapping region between the defect contour of the defect to be detected and the region contour of the weld area.

[0037] Determine the area ratio between the area of ​​the overlapping region and the area of ​​the defect contour;

[0038] If the area ratio is greater than a preset area ratio threshold, the defect location matching result indicates that the defect to be detected is located in the weld area.

[0039] In the above embodiments, by using a preset area ratio threshold, the area ratio of the overlapping area obtained by mapping the defect contour position coordinates and the region contour position coordinates to the contour area is compared with the preset area ratio threshold. This allows for a quick and accurate determination of whether the defect to be detected is located in the weld area, thus improving the accuracy and efficiency of manifold defect detection.

[0040] In some embodiments, determining the conformity test result of the undetermined defect category based on the defect category includes:

[0041] If the defect category is the planar dirt defect, the planar dirt defect is determined to be a qualified defect.

[0042] In the above embodiments, when the undetermined defect is a planar contamination defect, directly identifying the planar contamination defect as a qualified defect can reduce the possibility of qualified defects being over-tested, leading to an increase in defective battery products. This improves the production efficiency of battery products and reduces their production costs.

[0043] In some embodiments, the defect categories include three-dimensional dirt defects and planar dirt defects; the first illumination image information includes bright field image information, and the second illumination image information includes dark field image information;

[0044] The step of performing three-dimensional dirt defect detection on the undetermined defects based on the first lighting image information and the second lighting image information, and determining the defect category of the undetermined defects, includes:

[0045] Gray-scale features are extracted from the bright-field image information and the dark-field image information respectively to obtain the bright-field gray-scale features and dark-field gray-scale features of the undetermined category of defects;

[0046] If the feature difference between the bright field grayscale feature and the dark field grayscale feature is greater than the three-dimensional dirt difference threshold, the defect category of the undetermined defect is determined to be a three-dimensional dirt defect.

[0047] If the feature difference is less than or equal to the three-dimensional dirt difference threshold, the defect category of the undetermined defect is determined to be a planar dirt defect.

[0048] In the above embodiments, by extracting the bright field grayscale features and dark field grayscale features of the undetermined category defects, the imaging differences of the undetermined category defects under the two illumination modes can be accurately and intuitively reflected. Then, the defect type of the undetermined category defects can be quickly determined based on the imaging differences, which effectively improves the efficiency and accuracy of defect category determination and identification of undetermined category defects.

[0049] In some embodiments, the step of extracting grayscale features from the first illumination image information and the second illumination image information respectively to obtain the bright-field grayscale features and dark-field grayscale features of the undetermined category of defects includes:

[0050] Based on the first defect contour of the undetermined defect of the category, grayscale features are extracted from the first illumination image information to obtain the bright field grayscale features of the undetermined defect of the category.

[0051] Using the first defect contour as the region of interest for detection, region mapping is performed on the second illumination image information to obtain the second defect contour of the defect of the undetermined category.

[0052] Based on the second defect contour, grayscale features are extracted from the second illumination image information to obtain the dark field grayscale features of the undetermined category of defects.

[0053] In the above embodiments, by using the first defect contour as the detection region of interest, the second illumination image information is mapped to obtain the second defect contour. Then, grayscale features are extracted from the image information based on the first and second defect contours respectively. This effectively reduces the amount of data processing during grayscale feature extraction and improves the efficiency of grayscale feature extraction. Simultaneously, the region mapping method can also improve the feature matching degree between bright-field and dark-field grayscale features, thereby improving the accuracy of defect category determination.

[0054] Secondly, this application also provides a manifold defect detection device, the device comprising:

[0055] The first lighting information acquisition module is used to acquire the first lighting image information obtained by image acquisition of the collector disk under the first lighting environment;

[0056] The second lighting information acquisition module is used to acquire second lighting image information obtained by image acquisition of the collector under a second lighting environment when the collector is determined to have a category of undetermined defect after preliminary defect detection of the collector based on the first lighting image information; the first illumination angle of the first lighting environment is higher than the second illumination angle of the second lighting environment; the category of undetermined defect refers to a defect whose category cannot be distinguished by using the first lighting image information alone;

[0057] The defect category determination module is used to perform three-dimensional dirt defect detection on the undetermined defects based on the first lighting image information and the second lighting image information, and determine the defect category of the undetermined defects; the imaging of three-dimensional dirt defects differs between the first lighting environment and the second lighting environment;

[0058] The defect detection module is used to determine the defect detection result of the collector plate based on the defect category.

[0059] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method.

[0060] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0061] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method.

[0062] Sixthly, this application also provides a manifold defect detection system, the system including a first lighting environment component, a second lighting environment component, an image acquisition component, and a visual defect detection component communicatively connected to the image acquisition component;

[0063] The first lighting environment component is used to provide a first lighting environment for the collector disk;

[0064] The second lighting environment component is used to provide a second lighting environment for the collector disk; the first illumination angle of the first lighting environment is higher than the second illumination angle of the second lighting environment;

[0065] The image acquisition component is used to acquire first lighting image information of the collector disk under a first lighting environment and second lighting image information under a second lighting environment;

[0066] The visual defect detection component is used to implement the above-described manifold defect detection method.

[0067] In some embodiments, the system further includes an acquisition and control component that is communicatively connected to the visual defect detection component;

[0068] The acquisition control component is electrically connected to the first lighting environment component, the second lighting environment component, and the image acquisition component, respectively. It controls the opening and closing of the first lighting environment component or the second lighting environment component by sending electrical signals to the first lighting environment component or the second lighting environment component, and controls the image acquisition component to acquire images of the collector disk by sending electrical signals to the image acquisition component.

[0069] In some embodiments, the first lighting environment component includes a bowl-shaped light source with a first illumination angle and a coaxial light source;

[0070] The second lighting environment component includes a ring light source with a second illumination angle.

[0071] The aforementioned method, apparatus, computer equipment, storage medium, and computer program product for detecting defects in the manifold perform preliminary defect detection on the manifold using first illumination image information acquired under a first illumination environment at a first illumination angle. If it is determined that the manifold has defects of undetermined category, it indicates that there are defects in the manifold that cannot be distinguished by using the first illumination image alone. It is necessary to use the second illumination image information of the manifold for auxiliary identification to determine the defect category of the undetermined defect. The manifold is then provided with a second illumination environment at a second illumination angle, and second illumination image information is acquired by acquiring images of the manifold under the second illumination environment. Since the first illumination angle is higher than the second illumination angle of the second illumination environment, the imaging of three-dimensional dirt defects will differ between the first and second illumination environments. Based on the first and second illumination image information, three-dimensional dirt defect detection is performed on the undetermined defect to determine the defect category of the undetermined defect, and the defect detection result of the manifold is determined based on the defect category. The above method only requires providing the current collector with a first illumination environment and a second illumination environment with different illumination angles, and adding one image acquisition. It can utilize the imaging differences of the current collector under the first illumination environment and the second illumination environment to highlight the morphological characteristics of defects, and achieve the effect of accurately identifying three-dimensional and non-three-dimensional dirt defects. While effectively improving the detection accuracy of current collector defects and thus improving battery production efficiency, it can also effectively reduce the detection cost of accurate current collector defect detection. Attached Figure Description

[0072] Figure 1 This is a schematic diagram of the structure of the manifold defect detection system in some embodiments;

[0073] Figure 2 This is a schematic diagram of the structure of the manifold defect detection system in some other embodiments;

[0074] Figure 3 This is a schematic diagram of a bright field lighting environment in some embodiments;

[0075] Figure 4 This is a schematic diagram of a dark field lighting environment in some embodiments;

[0076] Figure 5 This is a flowchart illustrating the method for detecting defects in the manifold in some embodiments;

[0077] Figure 6 This is a flowchart illustrating the process of performing three-dimensional dirt defect detection on undetermined defects based on first illumination image information and second illumination image information in some embodiments, and determining the defect category of undetermined defects.

[0078] Figure 7 This is a flowchart illustrating the process of extracting grayscale features from bright-field image information and dark-field image information in some embodiments to obtain the bright-field grayscale features and dark-field grayscale features of defects of undetermined categories.

[0079] Figure 8 This is a flowchart illustrating the method for detecting defects in the manifold in other embodiments;

[0080] Figure 9 This is a flowchart illustrating the method for detecting defects in the manifold in other embodiments;

[0081] Figure 10 This is a flowchart illustrating the process of matching the contour position of the defect to be detected with the contour position of the region in some embodiments to obtain the defect position matching result.

[0082] Figure 11 This is a flowchart illustrating the preliminary defect detection portion of the manifold defect detection method in some embodiments;

[0083] Figure 12 This is a flowchart illustrating the defect category re-inspection part of the manifold defect detection method in some embodiments;

[0084] Figure 13 This is a structural block diagram of the manifold defect detection device in some embodiments;

[0085] Figure 14 1 is a diagram of the internal structure of a computer device in some embodiments. Detailed Implementation

[0086] The embodiments of the technical solution of this application will now be described in detail with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solution of this application and are therefore merely examples, and should not be used to limit the scope of protection of this application.

[0087] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0088] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least some of the embodiments of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0089] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0090] In the description of the embodiments of this application, the technical terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the embodiments of this application and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of this application.

[0091] In the description of the embodiments of this application, unless otherwise expressly specified and limited, technical terms such as "installation," "connection," "joining," and "fixing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0092] Current collector welding is a production step that achieves a complete connection between the positive and negative electrodes and the current collector. It involves precisely connecting the cell end face of the battery electrode to the current collector, thus ensuring a smooth flow of current within the battery. The welding quality of the current collector directly affects the stability and safety of the battery. If metal particles generated during welding, such as weld slag larger than 100μm, are not effectively removed, the battery will exhibit severe self-discharge behavior. Therefore, to improve battery operational safety, defect detection of the welded current collector is necessary to reduce the probability of defective batteries being discharged.

[0093] Currently, the common method for defect detection in current collectors involves acquiring images of the welded current collectors and then using image recognition models to identify defects in the images. However, because weld slag and dirt defects in current collectors often have similar morphologies, and are even more difficult to distinguish when they are small, current defect detection methods cannot accurately differentiate between defect categories. This can easily lead to misjudgments, resulting in either missed or over-detection, which negatively impacts the production efficiency of cylindrical batteries.

[0094] To improve the accuracy of manifold defect detection, preliminary defect detection can be performed on the manifold using the first illumination image information acquired under a first illumination environment at a first illumination angle. If it is determined that the manifold has a defect of undetermined category, it indicates that there are defects in the manifold that cannot be distinguished by the first illumination image alone. It is necessary to use the second illumination image information of the manifold for auxiliary identification to determine the defect category of the undetermined defect. Then, a second illumination environment with a second illumination angle is provided to the manifold, and the second illumination image information of the manifold is acquired under the second illumination environment. Since the first illumination angle is higher than the second illumination angle of the second illumination environment, the imaging of three-dimensional dirt defects will differ between the first and second illumination environments. Based on the first and second illumination image information, three-dimensional dirt defect detection is performed on the undetermined defect to determine the defect category of the undetermined defect. Based on the defect category, the defect detection result of the manifold is determined. The above method only requires providing the current collector with a first illumination environment and a second illumination environment with different illumination angles, and adding one image acquisition. It can utilize the imaging differences of the current collector under the first illumination environment and the second illumination environment to highlight the morphological characteristics of defects, and achieve the effect of accurately identifying three-dimensional and non-three-dimensional dirt defects. While effectively improving the detection accuracy of current collector defects and thus improving battery production efficiency, it can also effectively reduce the detection cost of accurate current collector defect detection.

[0095] The manifold defect detection method provided in this application embodiment can be applied to, for example, Figure 1The illustrated manifold defect detection system 100 includes a first lighting environment component 101, a second lighting environment component 102, an image acquisition component 103, and a visual defect detection component 104 that is communicatively connected to the first lighting environment component 101, the second lighting environment component 102, and the image acquisition component 103.

[0096] The first lighting environment component 101 is a device component used to provide a first lighting environment for the collector plate, and the second lighting environment component 102 is a device component used to provide a second lighting environment for the collector plate. The first illumination angle of the first lighting environment is different from the second illumination angle of the second lighting environment. The first illumination angle is higher than the second illumination angle. By acquiring images of the collector plate at different illumination angles, the imaging of three-dimensional dirt defects in the first lighting environment will be different from that in the second lighting environment. The difference in imaging of the collector plate in the first lighting environment and the second lighting environment can be used to highlight the morphological characteristics of the defects and achieve the effect of accurately identifying three-dimensional dirt defects and non-three-dimensional dirt defects.

[0097] It is understandable that the first lighting environment component 101 and the second lighting environment component 102 can be any lighting component that can provide lighting environments with different lighting angles, as long as the first image of the three-dimensional dirt defect in the undetermined category is different from the second image in the second lighting environment, and the first image of the non-three-dimensional dirt defect in the undetermined category is not different from the second image.

[0098] In some embodiments, the first lighting environment component 101 may include a device component for providing a bright field lighting environment, and the second lighting environment component 101 may include a device component for providing a dark field lighting environment.

[0099] Bright field illumination works by directly illuminating the surface of the object being measured with a light source. The strong light reflected from the object's surface forms a high-contrast image. By providing an environment for the current collector, the smooth surface of the current collector will appear as a bright area due to specular reflection, while the concave, foreign object, or textured areas will form dark areas due to scattering or occlusion, thus highlighting the differences in surface features.

[0100] When the first lighting environment component 101 is a bright field lighting environment component, the light emitted by the first lighting environment component 101 is also called bright field light or high-angle light. The first lighting environment component 101 can be any device component capable of providing bright field light.

[0101] Dark field illumination is an optical imaging technique that highlights the three-dimensional protrusions on the surface of an object by using low-angle scattered light. The principle is that the light source illuminates the object under test at a large incident angle, such as >75°, so that the light reflected from the smooth surface can avoid the image acquisition component 103, while the light signal generated by the scattering or refraction of the protrusion edge enters the image acquisition component 103. Therefore, in dark field imaging, flat areas appear as dark backgrounds, while protrusion defects are displayed as bright spots, forming a high contrast between light and dark.

[0102] When the second lighting environment component 102 is a dark field lighting environment component, the light emitted by the second lighting environment component 102 is also called dark field light or low-angle light. The second lighting environment component 102 can be any device component capable of providing dark field light.

[0103] In some embodiments, the first lighting environment component may include a bowl-shaped light source and a coaxial light source with a first illumination angle. The second lighting environment component may include a ring light source with a second illumination angle, wherein the first illumination angle is greater than the second illumination angle.

[0104] Bowl-shaped light sources typically consist of a bowl-shaped reflector and a built-in light source, primarily used to control the direction, focus, or scattering of light. Coaxial light sources are illumination devices where the direction of the light source aligns with the optical axis of the camera, eliminating shadows and specular reflections to achieve high-contrast imaging of subtle surface features. Ring light sources usually consist of multiple light sources arranged along a ring structure, with the light rays symmetrically distributed around the camera lens axis to form a uniform ring of light illuminating the object's surface, achieving effects such as surface texture enhancement and shadow elimination.

[0105] The first illumination angle is the illumination angle of the bowl-shaped light source, and the second illumination angle is the illumination angle of the ring-shaped light source. The first illumination angle is higher than the second illumination angle. The bowl-shaped light source uses a high-angle first illumination angle floodlight or spotlight design to directly and evenly illuminate the surface of the collector plate, reducing shadows. The ring-shaped light source uses a low-angle second illumination angle to obliquely illuminate the light, which can make the light entering the camera diffused, achieving the effect of a dark background and bright defect areas.

[0106] In some embodiments, the second lighting environment component 102 is a 0° ring light source.

[0107] In some embodiments, a coaxial light source, a bowl-shaped light source, or a 0° ring light source can all be strobe light sources.

[0108] In the above embodiments, the manifold defect detection system adopts a light source design of high-angle bowl light plus coaxial light plus low-angle ring light, which provides light field illumination of different angles for the manifold. Through dual light field imaging, the characteristics of manifold defects can be effectively highlighted, thereby improving the accuracy of the first illumination image information and the second illumination image information acquired subsequently.

[0109] In other embodiments, the first lighting environment component 101 may include a device component that provides a bright field lighting environment, and the second lighting environment component 102 may include a device component that provides a 2.5D imaging lighting environment.

[0110] Among them, 2.5D imaging is a technology between two-dimensional and three-dimensional imaging. By acquiring limited depth information, it adds the ability to perceive changes in the height or depth of an object's surface to a two-dimensional image, thus presenting a visual effect similar to three dimensions. Its principle is to illuminate the target object from different angles, acquire multiple images with different lighting conditions, and use the changes in the intensity of reflected light from the object's surface in these images, combined with appropriate algorithms, to estimate the height information of various defects on the target object's surface.

[0111] When the second lighting environment component 102 is a 2.5D imaging lighting component, the second lighting environment component 102 can have multiple second illumination angles, each different and lower than the first illumination angle. By acquiring second illumination image information collected under the second illumination environment corresponding to the multiple second illumination angles, and together with the first illumination image information collected under the first illumination angle, the three-dimensional features of the undetermined category of the defect can be determined. For example, if the undetermined category of the defect is a three-dimensional dirt defect, the three-dimensional structural features of the three-dimensional dirt defect can be obtained by fitting the first illumination image information and the second illumination image information. If the undetermined category of the defect is a non-three-dimensional dirt defect, such as a planar defect, the defect features obtained by fitting the first illumination image information and the second illumination image information are only planar defect features.

[0112] In some embodiments, the second lighting environment component 102 may include multiple bowl-shaped light sources with different illumination angles. The image of the collector disk is acquired under each bowl-shaped light source to obtain the corresponding image information, and finally the second lighting image information of the collector disk can be obtained.

[0113] The image acquisition component 103 is a device component used to acquire first illumination image information of the collector disk under a first illumination environment and second illumination image information under a second illumination environment. It is understood that the image acquisition component 103 can be any device component capable of image acquisition, such as a camera or camcorder.

[0114] In some embodiments, in order to improve detection accuracy, the image acquisition component 103 can be a high-resolution color global area array camera. By setting a high-resolution color global area array camera to the image acquisition component 103, the detection requirements at the μm level can be met.

[0115] The visual defect detection component 104 is a system component used to perform defect detection on the current collector based on the image information acquired by the image acquisition component 103, and to determine the passability of the current collector by detecting the passability of the defects. The visual defect detection component 104 is equipped with a visual defect detection system based on preset defect detection processing logic and corresponding defect detection algorithms, which can accurately detect defects in the current collector, thereby improving battery production efficiency and reducing the risk of missed or over-detected defects.

[0116] In some embodiments, when the visual defect detection component 104 determines that the collector disk has arrived at the detection station, it can control the first lighting environment component 101 to turn on, providing a first lighting environment for the collector disk. Then, it controls the image acquisition component 103 to acquire images of the collector disk under the first lighting environment, obtaining first lighting image information of the collector disk. It receives the first lighting image information sent by the image acquisition component 103 and controls the first lighting environment component 101 to turn off. Based on the first lighting image information, it performs preliminary defect detection on the collector disk. If it determines that the collector disk has a defect of undetermined category, it indicates that there is a defect in the collector disk whose category cannot be distinguished using only the first lighting image information. It needs to use the second lighting image information of the collector disk for auxiliary identification to determine the defect category of the undetermined defect. Alternatively, the visual defect detection component 104 can control the second lighting environment component 102 to turn on, providing a second lighting environment for the collector disk. Then, it controls the image acquisition component 103 to acquire images of the collector disk under the second lighting environment, obtaining second lighting image information of the collector disk. It receives the second lighting image information sent by the image acquisition component 103 and controls the second lighting environment component 102 to turn off. Subsequently, based on the first and second illumination image information, a three-dimensional dirt defect detection is performed on the undetermined category defects to determine the defect category, and the pass / fail test result of the undetermined category defects is determined based on the defect category.

[0117] The current collector plate defect detection system in the above embodiments, by setting a first lighting environment component and a second lighting environment component, can provide the current collector plate with first lighting environment and second lighting environment with different illumination angles. Compared with the traditional defect detection method, only one image acquisition is required. It can utilize the imaging difference of the current collector plate under the first lighting environment and the second lighting environment to highlight the morphological features of the defect, and achieve the effect of accurately identifying three-dimensional dirt defects and non-three-dimensional dirt defects. While effectively improving the current collector plate defect detection accuracy and thus improving battery production efficiency, it can also effectively reduce the detection cost of accurate current collector plate defect detection.

[0118] In some embodiments, the manifold defect detection system further includes an acquisition control component connected to the visual defect detection component. The acquisition control component is electrically connected to a first lighting environment component, a second lighting environment component, and an image acquisition component, respectively. It controls the activation of the first or second lighting environment component by sending electrical signals to it. It also controls the image acquisition component to acquire images of the manifold by sending electrical signals to it.

[0119] In some embodiments, the acquisition and control component can be a programmable logic controller (PLC).

[0120] In the above embodiments, the acquisition control component is electrically connected to the first lighting environment component, the second lighting environment component, and the image acquisition component by means of electrical signals, such as I / O signals, thereby controlling the opening and closing of the first lighting environment component or the second lighting environment component, and triggering image acquisition of the collector disk, which can effectively save image acquisition time and improve the detection efficiency of collector disk defect detection.

[0121] In some embodiments, as Figure 2 As shown, a manifold defect detection system is provided. The system includes a camera 201, a camera lens 202, a coaxial light source 203, a bowl-shaped light source 204, a 0° ring light 205, and an acquisition controller (not shown in the figure) that is electrically connected to the camera 201, the coaxial light source 203, the bowl-shaped light source 204, and the 0° ring light 205, respectively, and a visual defect detection component (not shown in the figure) that is communicatively connected to the acquisition controller.

[0122] In some embodiments, camera 201 may be a high-resolution color global area array camera.

[0123] In some embodiments, the coaxial light source 203, the bowl-shaped light source 204, and the 0° ring light 205 can be strobe light sources. After the acquisition controller sends a hard trigger signal to light up the light source, the light source can automatically turn off after a preset lighting time without the need for the acquisition controller, which can effectively reduce the resource consumption during the image acquisition process.

[0124] In some embodiments, when the visual defect detection component determines that the battery cell has arrived at the detection station, it sends a bright field light source activation command to the acquisition controller. In response to the bright field light source activation command, the acquisition controller sends a position signal and a battery cell code to the coaxial light source 203 and the bowl-shaped light source 204, hard-triggering the light sources to illuminate and creating a bright field lighting environment for the current collector. A schematic diagram of the bright field lighting environment is shown below. Figure 3 As shown.

[0125] The visual defect detection component controls the camera 201 to take an initial picture of the collector plate surface, obtaining the first illumination image information of the collector plate, and sends the first illumination image information to the visual defect detection component. The coaxial light source 203 and the bowl-shaped light source 204 automatically turn off after being lit and running for a preset bright field light source illumination time.

[0126] The visual defect detection component performs preliminary defect detection on the current collector based on bright-field image information. If it determines that the current collector has a defect of undetermined category, it indicates that there are defects in the current collector that cannot be distinguished by bright-field images alone. Secondary illumination image information from the current collector is needed for auxiliary identification to determine the defect category of the undetermined defect. The visual defect detection component sends a dark-field light source activation command to the acquisition controller. In response to the dark-field light source activation command, the acquisition controller sends a cell code to the 0° ring light 205, hard-triggeredly illuminating the light source and creating a dark-field illumination environment for the current collector. A schematic diagram of the dark-field illumination environment is shown below. Figure 4 As shown.

[0127] The visual defect detection component controls camera 201 to take a second picture of the collector disk surface, obtaining the second illumination image information of the collector disk, and sends the second illumination image information to the visual defect detection component. The 0° ring light 205 automatically turns off after being lit and running for the preset dark field light source illumination time. Understandably, the illumination time of the bright field light source and the dark field light source can be determined by the designer according to actual usage requirements. The illumination time of the bright field light source and the dark field light source can be the same or different, as long as it can maintain the normal acquisition of collector disk image information under the corresponding light source illumination by the image acquisition equipment.

[0128] The visual defect detection component performs three-dimensional dirt defect detection on the undetermined category defect based on the first illumination image information and the second illumination image information, determines the defect category of the undetermined category defect, and determines the pass / fail test result of the undetermined category defect based on the defect category.

[0129] In some embodiments, as Figure 5 As shown, a method for detecting defects in a manifold is provided, which is then applied to... Figure 1 Taking the visual defect detection component 104 in the image as an example, the following steps are included:

[0130] S502, acquire the first lighting image information obtained by image acquisition of the collector disk under the first lighting environment.

[0131] The first illumination image information is information data used to characterize the imaging features of the collector disk surface under the first illumination environment. For example, the first illumination image information may include each pixel that makes up the first illumination image of the collector disk and the pixel value corresponding to each pixel.

[0132] In some embodiments, the visual defect detection component can acquire first illumination image information of the collector disk through the image acquisition component when the collector disk is in a first illumination environment.

[0133] S504, after performing preliminary defect detection on the collector plate based on the first lighting image information and determining that the collector plate has a defect of undetermined type, the second lighting image information obtained by image acquisition of the collector plate under the second lighting environment is acquired.

[0134] In this process, the first illumination angle of the first lighting environment is higher than the second illumination angle of the second lighting environment. Preliminary defect detection is a step to determine whether there are defects on the manifold surface whose category cannot be determined. By performing preliminary defect detection on the manifold using the first illumination image information, it can be determined whether there are defects in the manifold that cannot be distinguished by the first illumination image alone, requiring the use of the second illumination image information for auxiliary identification. Simultaneously, by performing preliminary defect detection on the manifold, obvious non-conforming defects on the manifold surface can be quickly detected, further improving the efficiency of manifold conformity testing.

[0135] In some embodiments, preliminary defect detection may be a defect detection operation that uses a pre-trained defect detection model to identify defects on the surface of the manifold. The visual defect detection component can input the first illumination image information of the manifold into the defect detection model, perform defect category detection on the surface of the manifold based on the defect detection model, obtain the category detection results of each defect on the surface of the manifold, and determine whether there are undetermined defects in the manifold based on the category detection results.

[0136] In some embodiments, if there are defects on the surface of the manifold that cannot be classified, the manifold is identified as having a pending defect category.

[0137] In some embodiments, the current collector is only considered to have a category of undetermined defect if all identified defects in the preliminary defect detection are acceptable, and if there are defects whose category cannot be determined. If unacceptable defects, such as large planar dirt defects or three-dimensional dirt defects, are detected on the surface of the current collector during the preliminary defect detection process, the current collector can be directly determined to be unacceptable, and the cells associated with the current collector are transferred to the defective product area for processing. If all identified defects in the current collector surface that can be detected by the preliminary defect detection are acceptable, such as planar dirt defects or pinhole defects of compliant size, the preliminary defect detection results can be followed up to determine if there are any defects whose category cannot be identified. If so, the current collector is considered to have a category of undetermined defect, and further testing of the undetermined defect is required. The acceptance test result of the undetermined defect is used to determine the acceptance test result of the current collector, and then the next step of the cell product flow is determined.

[0138] The second illumination image information is information data used to characterize the imaging features of the collector disk surface under the second illumination environment. For example, the second illumination image information may include each pixel that makes up the second illumination image of the collector disk and the pixel value corresponding to each pixel.

[0139] In some embodiments, the visual defect detection component can perform preliminary defect detection on the collector plate based on the first illumination image information. If it is determined that the collector plate has a defect of undetermined type based on the preliminary defect detection result, the image acquisition component can acquire the second illumination image information obtained by image acquisition of the collector plate under the second illumination environment.

[0140] S506, Based on the first illumination image information and the second illumination image information, perform three-dimensional dirt defect detection on the undetermined category defect to determine the defect category of the undetermined category defect.

[0141] Among them, 3D contamination defect detection is a defect detection operation used to determine whether a defect of unknown category is a 3D contamination defect. By performing 3D contamination defect detection on a defect of unknown category, it is possible to accurately determine whether the defect of unknown category is a 3D contamination defect, such as welding slag or 3D dust, or a non-3D contamination defect, such as a planar contamination defect. The imaging of 3D contamination defects differs between the first lighting environment and the second lighting environment.

[0142] Among them, the defect category is a category parameter obtained by classifying the defects of the manifold according to the defect morphology. For example, the defect category can include three-dimensional dirt defects, planar dirt defects, pinhole burst defects, etc.

[0143] In some embodiments, the visual defect detection component can perform three-dimensional dirt defect detection on defects of undetermined category based on the first illumination image information and the second illumination image information. That is, based on the feature difference between the imaging features of the defects of undetermined category under the first illumination environment and the imaging features under the second illumination environment, the component can perform three-dimensional dirt defect detection on defects of undetermined category to determine the defect category of the defects of undetermined category.

[0144] In some embodiments, the visual defect detection component can extract a first defect feature of a defect of undetermined category from the first illumination image information, extract a second defect feature of a defect of undetermined category from the second illumination image information, call a pre-trained stereo dirt defect detection model, input the first defect feature and the second defect feature into the stereo dirt defect detection model, perform stereo dirt defect detection on the defect of undetermined category, and output the defect category of the defect of undetermined category.

[0145] In some embodiments, when the first lighting environment includes bright-field lighting and the second lighting environment includes a 2.5D imaging lighting environment, the first lighting image information may include bright-field image information, and the second lighting image information may include 2.5D image information. The visual defect detection component can extract stereo features of the undetermined defect based on the bright-field image information and the 2.5D image information to obtain the stereo features of the undetermined defect. Based on the stereo features, the component can fit the stereo morphology of the undetermined defect to obtain its stereo morphology. If the stereo morphology of the undetermined defect meets the morphological requirements of a stereo dirt defect, the defect category of the undetermined defect can be determined as a stereo dirt defect.

[0146] In some embodiments, the visual defect detection component can determine the height information of a defect of unknown category based on its three-dimensional morphology. If the height of the defect, which indicates that the defect of unknown category is higher than a preset height threshold, the component determines that the three-dimensional morphology of the defect of unknown category meets the morphological requirements of a three-dimensional contamination defect. It is understood that the preset height threshold is a pre-set threshold parameter used to determine whether a defect is a three-dimensional contamination defect.

[0147] S508, Determine the defect detection results of the manifold based on the defect category.

[0148] Among them, the defect detection results are used to characterize whether a defect is acceptable or unacceptable. By determining the defect detection results of the manifold, a data basis can be provided for subsequently determining the overall acceptance of the manifold. Acceptance detection results can include whether the defect detection is acceptable or unacceptable.

[0149] In some embodiments, if a defect exists on the surface of the current collector, the current collector can be deemed unqualified in the conformity test, and the battery cell product to which the current collector belongs can be transported to the defective product area for processing.

[0150] In some embodiments, after obtaining the defect category of a defect to be determined, the visual defect detection component can determine the defect detection result of the manifold based on the defect category.

[0151] In the above-mentioned manifold defect detection method, the manifold is initially detected using the first illumination image information acquired under a first illumination environment at a first illumination angle. If it is determined that the manifold has a defect of undetermined category, it indicates that there is a defect in the manifold that cannot be distinguished by the first illumination image alone. It is necessary to use the second illumination image information of the manifold for auxiliary identification to determine the defect category of the undetermined defect. The manifold is then provided with a second illumination environment at a second illumination angle, and the second illumination image information of the manifold is acquired under the second illumination environment. Since the first illumination angle is higher than the second illumination angle of the second illumination environment, the imaging of three-dimensional dirt defects will differ between the first and second illumination environments. Based on the first and second illumination image information, three-dimensional dirt defect detection is performed on the undetermined defect to determine the defect category of the undetermined defect. Based on the defect category, the defect detection result of the manifold is determined. The above method only requires providing the current collector with a first illumination environment and a second illumination environment with different illumination angles, and adding one image acquisition. It can utilize the imaging differences of the current collector under the first illumination environment and the second illumination environment to highlight the morphological characteristics of defects, and achieve the effect of accurately identifying three-dimensional and non-three-dimensional dirt defects. While effectively improving the detection accuracy of current collector defects and thus improving battery production efficiency, it can also effectively reduce the detection cost of accurate current collector defect detection.

[0152] Welding slag and dirt have similar morphologies, and they are even more difficult to distinguish when they are small. However, by performing three-dimensional dirt defect detection, three-dimensional dirt defects and planar dirt defects in the manifold can be accurately identified. The following will explain how to determine the defect category of the undetermined defect through several examples.

[0153] In some embodiments, the defect categories include three-dimensional dirt defects and planar dirt defects, the first illumination image information includes bright field image information, and the second illumination image information includes dark field image information. For example... Figure 6 As shown, in step S506, based on the first illumination image information and the second illumination image information, a three-dimensional dirt defect detection is performed on the defects of undetermined category to determine the defect category of the undetermined defect, including:

[0154] S602, extract grayscale features from bright-field image information and dark-field image information respectively to obtain bright-field grayscale features and dark-field grayscale features of defects with undetermined categories.

[0155] Among them, grayscale feature extraction refers to the operation of extracting grayscale features corresponding to undetermined defects from image information. Brightfield grayscale features are feature information used to characterize the imaging of undetermined defects under brightfield illumination. They can reflect the reflection of brightfield illumination light by undetermined defects under brightfield illumination, and thus reflect the planar shape of undetermined defects.

[0156] Dark field grayscale features are characteristic information used to characterize the imaging of undetermined defects in a dark field illumination environment. They can reflect the scattering of dark field illumination light by undetermined defects in a dark field illumination environment, and thus reflect the three-dimensional morphology of undetermined defects.

[0157] In some embodiments, the visual defect detection component can extract grayscale features from bright-field image information and dark-field image information respectively to obtain bright-field grayscale features and dark-field grayscale features of defects of undetermined category.

[0158] In some embodiments, the visual defect detection component is equipped with a grayscale extraction model. The visual defect detection component first extracts the bright-field defect information and dark-field defect information of the undetermined category defect from the bright-field image information and dark-field image information, respectively. Then, it calls the grayscale extraction model and inputs the bright-field defect information and dark-field defect information into the grayscale extraction model, thereby obtaining the bright-field grayscale features and dark-field grayscale features of the undetermined category defect output by the grayscale extraction model.

[0159] In some embodiments, the visual defect detection component can extract the bright-field grayscale values ​​of each pixel corresponding to the undetermined defect from the bright-field image information, and then take the average of the bright-field grayscale values ​​to obtain the bright-field grayscale features of the undetermined defect. Similarly, it can extract the dark-field grayscale values ​​of each pixel corresponding to the undetermined defect from the dark-field image information, and then take the average of the dark-field grayscale values ​​to obtain the dark-field grayscale features of the undetermined defect.

[0160] In some embodiments, the visual defect detection component may preprocess the bright field image information and dark field image information before performing grayscale feature extraction, such as noise reduction processing and image size adjustment processing, in order to improve the accuracy of grayscale feature extraction.

[0161] S604, when the feature difference between the bright field grayscale feature and the dark field grayscale feature is greater than the three-dimensional dirt difference threshold, the defect category of the undetermined defect is determined as a three-dimensional dirt defect.

[0162] Among them, the feature difference degree can reflect the imaging difference of undetermined defects under two lighting conditions. The three-dimensional contamination difference degree threshold is a preset judgment threshold used to determine whether an undetermined defect is a three-dimensional contamination defect. The three-dimensional contamination difference degree threshold can be determined by the designer based on experimental data or empirical data.

[0163] Among them, three-dimensional contamination defects refer to the defects of the manifold that have a three-dimensional shape, such as welding slag or dust.

[0164] Understandably, both three-dimensional and planar dirt defects reflect bright field light under bright field lighting. Therefore, both three-dimensional and planar dirt defects appear black under bright field lighting. However, due to their three-dimensional nature, three-dimensional dirt defects scatter low-angle light. Therefore, under dark field lighting, three-dimensional dirt defects will appear white, while planar dirt defects will still appear black. Thus, the defect category of a defect can be determined by the difference in imaging between the two lighting conditions.

[0165] In some embodiments, the visual defect detection component can determine the feature difference between bright field grayscale features and dark field grayscale features, compare the feature difference with a preset three-dimensional dirt difference threshold, and if the feature difference is greater than the three-dimensional dirt difference threshold, it indicates that the imaging difference of the undetermined defect is large under the two lighting conditions, which is consistent with the imaging habits of three-dimensional dirt defects, and the defect category of the undetermined defect can be determined as a three-dimensional dirt defect.

[0166] In some embodiments, the bright-field grayscale features and the dark-field grayscale features can be bright-field grayscale values ​​and dark-field grayscale values, and the feature difference between the two is the absolute value of the difference between the bright-field grayscale value and the dark-field grayscale value.

[0167] In some embodiments, the bright-field grayscale features and the dark-field grayscale features can be bright-field grayscale feature vectors and dark-field grayscale feature vectors, and the feature difference between the two can be determined by the vector distance between the feature vectors.

[0168] S606, when the feature difference degree is less than or equal to the three-dimensional dirt difference degree threshold, the defect category of the undetermined defect is determined to be a planar dirt defect.

[0169] Among them, planar contamination defects refer to manifold defects that do not have a three-dimensional shape and only have a two-dimensional planar shape, such as oil stains on the surface of the manifold.

[0170] In some embodiments, if the visual defect detection component determines that the feature difference between the bright field grayscale feature and the dark field grayscale feature is less than or equal to the three-dimensional dirt difference threshold, it indicates that the imaging difference of the undetermined defect is small under the two lighting conditions, which is consistent with the imaging habits of planar dirt defects, and the defect category of the undetermined defect can be determined as a planar dirt defect.

[0171] In the above embodiments, by extracting the bright field grayscale features and dark field grayscale features of the undetermined category defects, the imaging differences of the undetermined category defects under the two illumination modes can be accurately and intuitively reflected. Then, the defect type of the undetermined category defects can be quickly determined based on the imaging differences, which effectively improves the efficiency and accuracy of defect category determination and identification of undetermined category defects.

[0172] In some embodiments, as Figure 7 As shown in step S602, grayscale features are extracted from the bright-field image information and the dark-field image information respectively to obtain the bright-field grayscale features and dark-field grayscale features of the defects of undetermined category, including:

[0173] S702, based on the first defect contour of the defect of undetermined category, perform grayscale feature extraction on the brightfield image information to obtain the brightfield grayscale features of the defect of undetermined category.

[0174] The first defect contour of a defect of undetermined category refers to the boundary of the defect region in the bright-field image, which is a polygon or a closed curve. Understandably, the first defect contour of a defect of undetermined category can be identified and determined during the initial defect detection of the manifold. That is, the visual defect detection component can obtain the first defect contour of the defect of undetermined category from the initial defect detection results, for example, by obtaining the coordinates of each boundary point of the defect of undetermined category from the initial defect detection results to determine the first defect contour of the defect of undetermined category.

[0175] In some embodiments, the visual defect detection component can divide the bright field defect image information of the undetermined category from the bright field image information according to the first defect outline of the undetermined category defect, and then extract the grayscale features of the bright field defect image information to obtain the bright field grayscale features of the undetermined category defect.

[0176] S704: Using the first defect contour as the region of interest for detection, region mapping is performed on the dark field image information to obtain the second defect contour of the defect with undetermined category.

[0177] In this context, detecting the region of interest (ROI) refers to the local area in the image that needs to be detected. Region mapping, on the other hand, refers to the process of determining the corresponding contour position of the first defect contour in the dark field image based on the first defect contour in the bright field image. Understandably, region mapping can be achieved through coordinate projection.

[0178] The second defect contour refers to the boundary of the defect region in the dark field image where the category of the defect is to be determined. Similarly, the second defect contour is also a polygon or a closed curve.

[0179] In some embodiments, the visual defect detection component can define a region of interest (ROI) based on a first defect contour, and map the boundary coordinates of the ROI to a dark field image coordinate system using a preset light and dark image transformation matrix to obtain a second defect contour of a defect of unknown category.

[0180] S706, Based on the second defect contour, grayscale features are extracted from the dark field image information to obtain the dark field grayscale features of the defect of undetermined category.

[0181] In some embodiments, the visual defect detection component can classify dark field defect image information of undetermined categories from dark field image information based on the second defect contour, and then extract grayscale features from the dark field defect image information to obtain the dark field grayscale features of undetermined categories.

[0182] In the above embodiments, by using the first defect contour as the detection region of interest, the dark-field image information is mapped to obtain the second defect contour. Then, grayscale features are extracted from the image information based on the first and second defect contours respectively. This effectively reduces the amount of data processing during grayscale feature extraction and improves the efficiency of grayscale feature extraction. Simultaneously, the region mapping method also improves the feature matching degree between bright-field and dark-field grayscale features, thereby improving the accuracy of defect category determination.

[0183] When determining the conformity of three-dimensional contamination defects, in order to reduce the risk of over-inspection, it is necessary to consider the dimensional compliance of the three-dimensional contamination defects. Therefore, in some embodiments, S508, the defect detection result of the manifold is determined based on the defect category, including: if the defect category is a three-dimensional contamination defect, determining the second defect size of the three-dimensional contamination defect according to the second defect profile. If the second defect size is less than the first acceptable size threshold set for the three-dimensional contamination defect, the three-dimensional contamination defect detection of the manifold is determined to be qualified.

[0184] The defect size is a parameter used to quantify the geometric characteristics of a defect, reflecting its projected area on a two-dimensional plane. The second defect size is determined based on the second defect contour. Understandably, the second illumination image, such as a dark-field image, possesses three-dimensional morphological sensitivity. Therefore, the contrast of three-dimensional contamination defects in the second illumination image will be higher than that in the first illumination image, and the defect size in the second illumination image is closer to the actual size of the defect under metallographic inspection.

[0185] The first acceptable size threshold set for three-dimensional contamination defects is a preset threshold parameter used to determine whether a three-dimensional contamination defect is an acceptable defect. If the defect size is smaller than the first acceptable size threshold, it indicates that although a three-dimensional contamination defect exists on the current collector surface, it is an unavoidable production defect and will not affect the battery's operational reliability; therefore, the three-dimensional contamination defect can be determined as an acceptable defect. Conversely, if the defect size is greater than or equal to the first acceptable size threshold, it indicates that the three-dimensional contamination defect on the current collector surface will affect the battery's operational reliability; therefore, the three-dimensional contamination defect can be determined as an unacceptable defect. Understandably, the first acceptable size threshold can be determined by the designer based on actual production requirements. For example, if the designer determines that a three-dimensional contamination defect with a size less than 0.1 mm is an acceptable defect based on production documents such as the production manual, then 0.1 mm can be set as the first acceptable size threshold for three-dimensional contamination defects.

[0186] In some embodiments, when the visual defect detection component determines that the undetermined defect is a three-dimensional dirt defect, it can determine the second defect size of the three-dimensional dirt defect based on the second defect profile, and compare the second defect size with a first qualified size threshold set in advance for the three-dimensional dirt defect. If the second defect size is less than the first qualified size threshold, the three-dimensional dirt defect is determined to be a qualified defect, and the three-dimensional dirt defect detection of the manifold is qualified.

[0187] In the above embodiments, determining the defect size of a three-dimensional contamination defect using the second defect profile allows the determined defect size to be closer to the actual size of the three-dimensional contamination defect, effectively improving the accuracy of three-dimensional contamination defect conformity testing. Simultaneously, performing size compliance re-judgment for three-dimensional contamination defects can effectively reduce the risk of over-testing, improving battery production efficiency and reducing battery production costs.

[0188] In the process of defect detection of manifolds, preliminary defect detection is an effective means of quickly locating defects and detecting obviously unqualified defects. The following will explain how to perform preliminary defect detection of manifolds through several examples.

[0189] In some embodiments, as Figure 8 As shown, the method for detecting defects in the manifold also includes the following steps:

[0190] S802, if it is determined that there is a defect to be detected in the collector plate based on the first illumination image information, the first defect contour of the defect to be detected is extracted.

[0191] In some embodiments, the visual defect detection component invokes a pre-trained first defect recognition model, inputs the first illumination image information into the first defect recognition model, performs defect recognition on the first illumination image, obtains the initial defect detection result of the collector disk, and extracts the first defect contour of the defect to be detected from the first illumination image information when the initial defect detection result indicates that there is a defect to be detected in the collector disk.

[0192] In some embodiments, the visual defect detection component can enhance the defect boundary contrast of the defect to be detected based on the first illumination image information, and extract the first defect contour of the defect to be detected from the first illumination image through a segmentation algorithm.

[0193] S804, determine the first defect size of the defect to be detected based on the first defect profile.

[0194] The first defect size is the defect size determined based on the first defect profile.

[0195] In some embodiments, the visual defect detection component can determine the first defect size of the defect to be detected based on the first defect profile.

[0196] In some embodiments, the visual defect detection component can determine the minimum bounding rectangle of the defect profile based on the defect profile, and determine the rectangular area of ​​the minimum bounding rectangle as the defect size.

[0197] In some embodiments, the visual defect detection component can use a pixel counting method to count the total number of pixels in the defect outline and determine the defect size by multiplying the total number of pixels by the actual size of the pixels.

[0198] In some embodiments, the visual defect detection component can calculate the defect area and obtain the defect size by using Green's formula and the coordinates of the defect contour vertices.

[0199] S806, if the first defect size is less than the second acceptable size threshold set for planar contamination defects, and is greater than or equal to the first acceptable size threshold, determine the contour position of the first defect contour.

[0200] The second acceptable size threshold set for planar contamination defects is a preset threshold parameter used to determine whether the defect to be inspected is an obviously unacceptable defect. Understandably, the second acceptable size threshold can be determined by the designer based on actual production requirements. For example, if the designer determines, based on production manuals or other production documents, that a planar contamination defect with a size less than 0.5mm is an acceptable defect, then 0.5mm can be set as the second acceptable size threshold for planar contamination defects. Since planar contamination defects are generally large, the second acceptable size threshold set for planar contamination defects will be greater than the first acceptable size threshold set for three-dimensional contamination defects.

[0201] If the size of the first defect to be detected is greater than or equal to the second acceptable size threshold, it means that regardless of whether the defect to be detected is a planar defect or not, or in other words, regardless of what kind of defect it is, its size is too large and will affect the stable operation of the battery. Therefore, the defect to be detected can be directly identified as an unacceptable defect.

[0202] When the size of the first defect to be detected is less than the first acceptable size threshold, it means that regardless of whether the defect to be detected is a three-dimensional defect, or in other words, regardless of the type of defect, because its size is small, it will not affect the stable operation of the battery, and the defect to be detected can be directly determined as an acceptable defect.

[0203] If neither of the above two conditions is met, that is, the first defect size of the defect to be tested is less than the second acceptable size threshold, and is greater than or equal to the first acceptable size threshold, it means that the defect to be tested is neither obviously unacceptable nor obviously acceptable. Defect testing needs to be carried out on the defect to be tested in order to determine its acceptance test result.

[0204] Therefore, in some embodiments, the visual defect detection component compares the first defect size of the defect to be detected with a first acceptable size threshold and a second acceptable size threshold, respectively. If the first defect size is less than the second acceptable size threshold set for planar dirt defects and is greater than or equal to the first acceptable size threshold, the component determines the contour position of the first defect contour.

[0205] S808, when it is determined that the defect to be detected needs to be inspected in three dimensions based on its contour position, the defect to be detected is identified as a defect of undetermined category.

[0206] In some embodiments, the visual defect detection component can determine whether the defect to be detected needs to be detected in three-dimensional dirt defect detection based on the contour position. Only when the defect to be detected needs to be detected in three-dimensional dirt defect detection will the defect to be detected be identified as a defect of undetermined category.

[0207] In the above embodiments, by comparing the first defect size of the defect to be detected with the first qualified size threshold and the second qualified size threshold respectively, it is possible to quickly determine whether the defect to be detected is an obviously unqualified defect or an obviously qualified defect. Only when it is determined that the defect to be detected is not an obviously unqualified defect or an obviously qualified defect will the contour position of the defect to be detected be followed to determine whether the defect to be detected needs to be subjected to three-dimensional dirt defect detection, which effectively improves the detection efficiency of manifold defect detection.

[0208] In some embodiments, as Figure 9 As shown, the method for detecting defects in the manifold also includes the following steps:

[0209] S902, for the weld bead area of ​​the manifold, perform region recognition on the first illumination image information to obtain the region contour position of the weld bead area.

[0210] The weld bead region of the manifold is the continuous molten pool trajectory formed on the surface of the manifold by laser welding. The region contour position of the weld bead region is information data used to characterize the region position of the weld bead region in the first illumination image. For example, the region contour position can include the position coordinates of the boundary points of each region contour that make up the weld bead region.

[0211] In some embodiments, the visual defect detection component can perform region recognition on the first illumination image information for the weld area of ​​the manifold to obtain the region contour position of the weld area.

[0212] In some embodiments, the visual defect detection component is pre-configured with a weld bead region recognition model. By inputting the first illumination image information into the weld bead region recognition model, the region contour position of the weld bead region output by the weld bead region recognition model can be obtained.

[0213] S904, perform position matching between the contour position of the defect to be detected and the contour position of the region to obtain the defect position matching result.

[0214] In some embodiments, the visual defect detection component can perform position matching between the contour position of the defect to be detected and the contour position of the region to obtain the defect position matching result.

[0215] In some embodiments, the defect location matching result may include a match or a mismatch, wherein a match means that the defect to be detected is located within the weld bead area, and a mismatch means that the defect to be detected is not located within the weld bead area.

[0216] In some embodiments, the visual defect detection component is equipped with a pre-trained position matching model. By inputting the contour position of the defect to be detected and the contour position of the region into the position matching model, the defect position matching result can be obtained.

[0217] S906, if the defect location matching result indicates that the defect to be detected is not located in the weld area, it is determined that the defect to be detected needs to be subjected to three-dimensional contamination defect detection.

[0218] In some embodiments, if the defect location matching result indicates that the defect to be detected is not located in the weld bead area, it means that the defect to be detected is located in the planar area of ​​the manifold. In this case, it cannot be detected according to the defect detection rules in the weld bead area, but needs to be detected in three dimensions. Therefore, the visual defect detection component determines that the defect to be detected needs to be detected in three dimensions.

[0219] In the above embodiments, by matching the contour position of the defect to be detected with the regional contour position of the weld area, it is possible to quickly determine whether the defect to be detected is a weld defect. If it is determined that it is not a weld defect, subsequent three-dimensional contamination defect detection can be performed, which effectively improves the detection efficiency and detection accuracy of manifold defect detection.

[0220] In other embodiments, the manifold defect detection method further includes: determining the defect to be detected as a pinhole burst defect if the defect location matching result indicates that the defect to be detected is located in the weld area; and determining that the pinhole burst defect detection of the manifold is qualified if the first defect size is less than a third qualified size threshold set for pinhole burst defects; the third qualified size threshold is greater than the first qualified size threshold and less than the second qualified size threshold.

[0221] Among them, pinhole burst defects are defects in the manifold caused by tiny voids or local bulges and cracks in the weld area. Due to the complexity of the welding conditions in the weld area, such as excessively deep molten pools or solder accumulation causing wavy weld surfaces, when performing defect detection in the weld area, there is no need to subdivide the defect categories. All defects in the weld area are directly identified as pinhole burst defects, and the third acceptable size threshold set for pinhole burst defects is used for acceptance testing.

[0222] The third acceptable size threshold set for pinhole defects is a preset threshold parameter used to determine whether a pinhole defect is an unacceptable defect. Understandably, this third acceptable size threshold can be determined by the designer based on actual production requirements. For example, if the designer determines that a pinhole defect with a size less than 0.2mm is acceptable based on production manuals or other production documents, then 0.2mm can be set as the third acceptable size threshold for pinhole defects. Since pinhole defects are generally smaller than large planar dirt defects but larger than small three-dimensional dirt defects, the third acceptable size threshold set for pinhole defects will be greater than the first acceptable size threshold set for three-dimensional dirt defects but less than the second acceptable size threshold set for planar dirt defects.

[0223] When the first defect size of the pinhole explosion defect is greater than or equal to the third qualified size threshold, it indicates that the size of the pinhole explosion defect is too large and will affect the stable operation of the battery. The pinhole explosion defect can be determined as a non-qualified defect. On the contrary, when the first defect size of the pinhole explosion defect is less than the third qualified size threshold, it indicates that the size of the pinhole explosion defect will not affect the stable operation of the battery, and the pinhole explosion defect can be determined as a non-qualified defect.

[0224] In some embodiments, when the visual defect detection component determines that the defect position matching result indicates that the defect to be detected is located in the weld area, it can determine that the defect to be detected is a pinhole explosion defect. Compare the first defect size of the pinhole explosion defect with the third qualified size threshold set for the pinhole explosion defect. When the first defect size of the pinhole explosion defect is less than the third qualified size threshold, determine that the pinhole explosion defect is a qualified defect, and the pinhole explosion defect detection of the current collector plate is qualified.

[0225] In the above embodiments, for the pinhole explosion defect located in the weld area, using the corresponding third qualified size threshold to perform the qualification detection on it can make the defect qualification detection process have a higher matching degree with the actual defect category, effectively improving the detection accuracy of the current collector plate defect detection.

[0226] In order to further improve the detection accuracy of the current collector plate defect detection, in some embodiments, S902, for the weld area of the current collector plate, perform region recognition on the first illumination image information to obtain the region contour position of the weld area, including: using a preset region of interest to frame the initial region information containing the weld area from the first illumination image information. Input the initial region information into a preset weld width detection model, and based on the weld width of the weld area, intercept and correct the initial region information to obtain the region contour position of the weld area.

[0227] Among them, the preset region of interest is a region extraction frame set in advance. The size of the preset region of interest can be set by the designer in advance according to the actual situation of the weld area. It can be understood that the size of the preset region of interest should be larger than the region size of the weld area. Through the preset region of interest, the initial region information containing the weld area can be quickly framed from the first illumination image information. The initial region information is the image sub-region information framed by the preset ROI, including the region information of the weld area and part of the current collector plate plane area.

[0228] Among them, the weld width detection model is a model tool used to quantitatively locate the weld width of the current collector plate, which can be trained by the designer in advance using historical image information. The weld width is the vertical distance between the edges on both sides of the weld center line.

[0229] In some embodiments, the visual defect detection component can use a preset bounding box of interest to extract initial region information containing the weld area from the first illumination image bounding box represented by the first illumination image information. Then, it calls a pre-trained weld width detection model, inputs the initial region information into the weld width detection model, and the weld width detection model will identify the weld width of the weld area and truncate and correct the initial region information based on the weld width of the weld area to obtain the region contour position of the weld area.

[0230] In some embodiments, the visual defect detection component can shrink the boundary of a preset box of interest based on the weld width of the weld area, extract and correct the weld area image information that matches the actual weld area from the initial area information, and then obtain the area contour position of the weld area based on the weld area image information.

[0231] In the above embodiments, a large area of ​​the weld bead region is captured from the first illumination image by pre-setting a bounding box of interest, and then the weld bead width obtained by identification is used to cut and correct the large area of ​​the weld bead region to obtain the region outline position of the weld bead region. This can effectively improve the recognition accuracy of the region outline position of the weld bead region and provide an accurate data basis for subsequent judgment on whether the defect to be detected is within the weld bead region.

[0232] In some embodiments, the contour location of the defect to be detected includes the coordinates of the defect contour location, and the region contour location includes the coordinates of the region contour location of the weld bead region. For example... Figure 10 As shown in S904, the contour position of the defect to be detected is matched with the contour position of the region to obtain the defect position matching result, including:

[0233] S1002, perform region mapping between the defect contour position coordinates of the defect to be detected and the region contour position coordinates of the weld area to determine the area of ​​overlap between the defect contour of the defect to be detected and the region contour of the weld area.

[0234] Among them, the defect contour position coordinates are a set of defect boundary point coordinates extracted by image processing algorithms, which can characterize the shape, area, and position of the defect contour to be detected. The region contour position coordinates are a set of weld bead region boundary coordinates extracted by image processing algorithms, which can characterize the shape, area, and position of the actual weld bead contour.

[0235] Region mapping refers to the mapping operation that transforms the position coordinates of the defect outline of the defect to be detected and the region outline of the weld area to the same coordinate system, establishing a spatial correspondence. By performing region mapping between the position coordinates of the defect outline of the defect to be detected and the position coordinates of the region outline of the weld area, the overlapping area of ​​the defect outline of the defect to be detected and the region outline of the weld area can be determined intuitively and accurately, and then the area of ​​the overlapping area can be calculated.

[0236] The overlapping area is the geometric area of ​​the part where the defect profile and the weld bead region profile overlap in space.

[0237] In some embodiments, the vision inspection component can perform a region mapping between the defect contour position coordinates of the defect to be inspected and the region contour position coordinates of the weld area, thereby determining the overlapping area between the defect contour of the defect to be inspected and the region contour of the weld area, and then determine the area of ​​the overlapping area through a region area calculation function or model.

[0238] S1004, determine the area ratio between the overlapping area and the contour area of ​​the defect.

[0239] The area of ​​the defect contour is the complete geometric area representing the coverage of the defect to be detected. The ratio of the overlapping area to the contour area is the percentage of the overlap area between the defect contour and the weld area to the total defect area. It can characterize the spatial intersection state of the defect to be detected in the weld area, such as inclusion, intersection, or contact.

[0240] In some embodiments, the visual inspection component can calculate the outline area of ​​the defect to be detected based on the defect outline location coordinates, and then calculate the area ratio between the overlapping area and the outline area of ​​the defect.

[0241] S1006, when the area ratio is greater than the preset area ratio threshold, the defect location matching result is determined to indicate that the defect to be detected is located in the weld area.

[0242] The preset area percentage threshold is a parameter used to determine whether a defect to be detected is located within the weld bead area. This threshold can be determined by the designer based on the actual detection accuracy. For example, with high detection accuracy, 95% can be set as the preset area percentage threshold; that is, the defect is considered to be located within the weld bead area only if 95% of the defect is within it. Conversely, with lower detection accuracy, 70% can be set as the preset area percentage threshold; that is, the defect is considered to be located within the weld bead area only if 70% of the defect is within it.

[0243] In some embodiments, the visual inspection component can compare the area ratio between the overlapping region area and the contour area of ​​the defect with a preset area ratio threshold. If the area ratio is greater than the preset area ratio threshold, the defect location matching result indicates that the defect to be detected is located in the weld bead region. If the area ratio is less than or equal to the preset area ratio threshold, the defect location matching result indicates that the defect to be detected is not located in the weld bead region.

[0244] In the above embodiments, by using a pre-set area ratio threshold, the ratio of the overlapping area obtained by mapping the defect contour coordinates to the region contour coordinates to the contour area is compared with the pre-set area ratio threshold. This allows for quick and accurate determination of whether the defect to be detected is located in the weld bead region, improving the accuracy and efficiency of manifold defect detection. The above embodiments provide a detailed explanation of how to perform preliminary defect detection on the manifold. Since a threshold judgment has been made during the preliminary defect detection process to determine whether the defect size is greater than the second acceptable size threshold set for planar contamination defects, in some embodiments, S508, the acceptance test result of the defect to be detected based on the defect category is determined, including: if the defect category is a planar contamination defect, determining the planar contamination defect as an acceptable defect.

[0245] In some embodiments, if the defect to be detected is a defect of undetermined category, it can be said that the defect size of the defect to be detected is less than the second acceptable size threshold. Based on this, if the defect category of the defect of undetermined category is a planar dirt defect, it can be directly determined that the defect size of the planar dirt defect is less than the second acceptable size threshold, and the visual defect detection component can directly determine the planar dirt defect as an acceptable defect.

[0246] In the above embodiments, when the undetermined defect is a planar contamination defect, directly identifying the planar contamination defect as a qualified defect can reduce the possibility of qualified defects being over-tested, leading to an increase in defective battery products. This improves the production efficiency of battery products and reduces their production costs.

[0247] In some embodiments, a manifold defect detection method is provided, which can be applied to, for example, Figure 2 The following explanation uses a current collector defect detection system as an example. It is understood that such a system can simultaneously detect multiple current collectors, for example, four current collectors. This application describes the defect detection process for any one of these current collectors, and the steps are consistent across all current collectors. The current collector defect detection method specifically includes two parts: a preliminary defect detection part and a defect category re-inspection part.

[0248] like Figure 11 As shown, the preliminary defect detection section includes the following steps:

[0249] S1101, acquire bright-field image information obtained by image acquisition of the collector disk under bright-field illumination.

[0250] When the battery cell reaches the defect detection station of the current collector, the visual defect detection component sends a bright field light source activation command to the acquisition controller. In response to the bright field light source activation command, the acquisition controller sends a position signal and a battery cell code to the coaxial light source and the cup-shaped light source, which hard-trigger the light source to create a bright field lighting environment for the current collector. Then, the visual defect detection component controls the camera to take the first picture of the surface of the current collector and obtain the bright field image information of the current collector.

[0251] S1102, Based on the bright field image information, determine whether there is a defect to be detected in the collector disk. If not, proceed to S1103; if yes, proceed to S1104.

[0252] S1103, after confirming that the collector plate has passed inspection, the battery cell is transported to the next production station.

[0253] S1104, Extract the first defect contour of the defect to be detected, and determine the first defect size based on the first defect contour.

[0254] S1105, determine whether the first defect size is greater than the second acceptable size threshold. If yes, proceed to S1106; otherwise, proceed to 1007.

[0255] In some embodiments, the second acceptable size threshold set for planar contamination defects can be 0.5 mm. Even if the defect to be detected is a planar contamination defect, if its defect size is >0.5 mm, the defect to be detected can be directly determined as an unacceptable defect.

[0256] S1106, the defect to be tested is determined to be a non-conforming defect. The current collector fails the test, and the battery cell is discharged from the defective product transport tank.

[0257] Among them, the defective product transport trough is the NG trough in the production line. If there is even one defect in the manifold, it can be judged as a defective product (NG).

[0258] S1107, determine whether the first defect size is less than the first acceptable size threshold. If yes, proceed to S1108; otherwise, proceed to S1109.

[0259] In some embodiments, the first acceptable size threshold for three-dimensional contamination defects, such as weld slag, can be 0.1 mm. Even if the defect to be detected is a three-dimensional contamination defect, it can be directly determined as an acceptable defect if its defect size is <0.1 mm.

[0260] S1108, the defect to be tested is determined to be a qualified defect.

[0261] It should be noted that the collector plate can only be deemed to have passed inspection (OK) if all defects in the collector plate are acceptable.

[0262] S1109, Determine the contour position of the first defect contour.

[0263] S1110: For the weld area of ​​the manifold, perform region recognition on the bright field image information to obtain the region contour position of the weld area.

[0264] The visual defect detection component can first use a preset ROI to extract the initial region information containing the weld area from the bright field image information, and then input the initial region information into the preset weld width detection model. Based on the weld width of the weld area, the initial region information is cropped and corrected to obtain the region contour position of the weld area.

[0265] S1111, perform position matching between the contour position and the region contour position to obtain the defect position matching result.

[0266] S1112, determine whether the defect location matching result indicates that the defect to be detected is located in the weld area. If not, proceed to S1113; if yes, proceed to S1114.

[0267] S1113, the defect to be detected is determined to be a defect of undetermined category.

[0268] In cases where the defect location matching result indicates that the defect to be detected is not located in the weld bead area, it is determined that the defect to be detected needs to undergo three-dimensional contamination defect detection.

[0269] S1114, determine whether the first defect size is less than the third acceptable size threshold. If yes, execute S1108; otherwise, return to execute S1106.

[0270] In cases where the defect location matching result indicates that the defect to be detected is located in the weld area, the defect to be detected is determined to be a pinhole burst defect.

[0271] In some embodiments, the third acceptable size threshold set for pinhole burst defects can be 0.2 mm.

[0272] If the initial defect detection determines that the manifold has a defect of undetermined category, a defect category re-inspection can be performed on the defect. The method and steps for the defect category re-inspection are as follows: Figure 12 As shown, the specific steps include:

[0273] S1201, acquire dark field image information obtained by image acquisition of the collector disk under dark field illumination.

[0274] In cases where the current collector plate is found to have defects of undetermined type, it indicates that the current collector plate contains defects whose category cannot be distinguished using only bright-field images. Dark-field image information from the current collector plate is needed for auxiliary identification to determine the defect category of the undetermined defect. The visual defect detection component sends a dark-field light source activation command to the acquisition controller. In response to this command, the acquisition controller sends a cell code to the 0° ring light, hard-triggered to illuminate the light source, creating a dark-field illumination environment for the current collector plate. The visual defect detection component then controls the camera to take a second picture of the current collector plate surface, obtaining the dark-field image information of the current collector plate.

[0275] S1202, Based on the first defect contour of the defect of undetermined category, perform grayscale feature extraction on the brightfield image information to obtain the brightfield grayscale features of the defect of undetermined category.

[0276] S1203, using the first defect contour as the region of interest for detection, region mapping is performed on the dark field image information to obtain the second defect contour of the defect with undetermined category.

[0277] S1204, Based on the second defect contour, grayscale features are extracted from the dark field image information to obtain the dark field grayscale features of the defect with undetermined category.

[0278] S1205 calculates the feature difference between bright-field grayscale features and dark-field grayscale features.

[0279] S1206, determine whether the feature difference degree is greater than the three-dimensional dirt difference degree threshold. If not, proceed to S1207. If yes, proceed to S1208-S1210.

[0280] S1207, the defect category of the undetermined defect is a planar contamination defect, and it is a qualified defect.

[0281] S1208, the defect category of the undetermined defect is determined to be a three-dimensional dirt defect, and the second defect size of the three-dimensional dirt defect is determined according to the second defect profile.

[0282] S1209, determine whether the second defect size is less than the first acceptable size threshold. If yes, execute S1210; otherwise, execute S1211.

[0283] S1210, the three-dimensional dirt defect is determined to be a qualified defect.

[0284] S1211, the three-dimensional dirt defect is determined to be a non-conforming defect, the collector plate test is unqualified, and the battery cell is discharged from the defective product transport tank.

[0285] The aforementioned manifold defect detection method employs a dual-light-field composite detection approach. This enables precise identification of both three-dimensional and planar contaminants that are difficult to locate with a single light field, effectively reducing the risk of misjudgment. Compared to traditional methods, it only introduces a low-angle ring light, increasing the number of images to two, achieving low cost, simplicity, and high efficiency. Furthermore, this vision solution detects not only three-dimensional contaminants, such as weld slag, but also weld beads and other inspection items. The combination of bright and dark field methods provides better imaging results for each inspection item. On the other hand, it classifies particles of different sizes within the manifold plane and weld beads, optimizing the inspection sequence and reducing the amount of information processed. Using the bright field to identify the location of weld slag and the dark field to calculate the size of the weld slag makes the detection more accurate and faster.

[0286] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0287] Based on the same inventive concept, this application also provides a manifold defect detection device for implementing the aforementioned manifold defect detection method. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more manifold defect detection device embodiments provided below can be found in the limitations of the manifold defect detection method described above, and will not be repeated here.

[0288] In some embodiments, as Figure 13 As shown, a manifold defect detection device 1300 is provided, including: a first lighting information acquisition module 1301, a second lighting information acquisition module 1302, a defect category determination module 1303, and a defect detection module 1304, wherein:

[0289] The first lighting information acquisition module 1301 is used to acquire the first lighting image information obtained by image acquisition of the collector disk under the first lighting environment.

[0290] The second lighting information acquisition module 1302 is used to acquire second lighting image information obtained by image acquisition of the collector disk under a second lighting environment when preliminary defect detection of the collector disk is performed based on the first lighting image information and it is determined that the collector disk has a defect of undetermined category; the first illumination angle of the first lighting environment is higher than the second illumination angle of the second lighting environment. The defect category determination module 1303 is used to perform three-dimensional dirt defect detection on the defect of undetermined category based on the first lighting image information and the second lighting image information, and determine the defect category of the defect of undetermined category.

[0291] The defect detection module 1304 is used to determine the conformity test results of undetermined defects based on the defect category.

[0292] In some embodiments, the defect categories include three-dimensional dirt defects and planar dirt defects, the first illumination image information includes bright field image information, and the second illumination image information includes dark field image information; the defect category determination module 1303 is used to: extract grayscale features from the bright field image information and the dark field image information respectively to obtain the bright field grayscale features and dark field grayscale features of the defect to be determined; if the feature difference between the bright field grayscale features and the dark field grayscale features is greater than the three-dimensional dirt difference threshold, determine the defect category of the defect to be determined as a three-dimensional dirt defect; if the feature difference is less than or equal to the three-dimensional dirt difference threshold, determine the defect category of the defect to be determined as a planar dirt defect.

[0293] In some embodiments, the defect category determination module 1303 is configured to: extract grayscale features from bright-field image information based on the first defect contour of the defect to be determined, to obtain the bright-field grayscale features of the defect to be determined; map the dark-field image information using the first defect contour as the region of interest, to obtain the second defect contour of the defect to be determined; and extract grayscale features from the dark-field image information based on the second defect contour, to obtain the dark-field grayscale features of the defect to be determined.

[0294] In some embodiments, the defect detection module 1304 is configured to: determine the second defect size of the three-dimensional dirt defect based on the second defect outline in the second illumination image information when the defect category is a three-dimensional dirt defect; and determine the three-dimensional dirt defect as a qualified defect when the second defect size is less than a first qualified size threshold set for the three-dimensional dirt defect.

[0295] In some embodiments, the manifold defect detection device 1300 further includes:

[0296] The defect contour extraction module is used to extract the first defect contour of the defect to be detected when it is determined that there is a defect to be detected in the manifold based on the first illumination image information.

[0297] The first defect size determination module is used to determine the first defect size of the defect to be detected based on the first defect profile.

[0298] The contour position determination module is used to determine the contour position of the first defect contour when the first defect size is less than a second acceptable size threshold set for planar contamination defects, and is greater than or equal to the first acceptable size threshold; the second acceptable size threshold is greater than the first acceptable size threshold.

[0299] The category-pending defect determination module is used to determine the defect to be detected as a category-pending defect when it is determined that the defect to be detected needs to be detected by three-dimensional dirt defect detection based on the contour position.

[0300] In some embodiments, the manifold defect detection device 1300 further includes:

[0301] The region contour position recognition module is used to perform region recognition on the first illumination image information for the weld area of ​​the manifold, and obtain the region contour position of the weld area.

[0302] The position matching module is used to match the contour position of the defect to be detected with the contour position of the region to obtain the defect position matching result.

[0303] The results analysis module is used to determine whether the defect to be detected needs to be detected in a three-dimensional contamination defect detection process when the defect location matching results indicate that the defect to be detected is not located in the weld area.

[0304] In some embodiments, the manifold defect detection device 1300 further includes:

[0305] The pinhole burst defect determination module is used to determine the defect to be detected as a pinhole burst defect when the defect location matching result indicates that the defect to be detected is located in the weld area.

[0306] The size comparison module is used to determine that a pinhole explosion defect is a qualified defect if the first defect size is less than the third qualified size threshold set for the pinhole explosion defect; the third qualified size threshold is greater than the first qualified size threshold and less than the second qualified size threshold.

[0307] In some embodiments, the region contour location recognition module is configured to: use a preset bounding box to extract initial region information containing the weld bead region from the first illumination image information; input the initial region information into a preset weld bead width detection model; and truncate and correct the initial region information based on the weld bead width of the weld bead region to obtain the region contour location of the weld bead region.

[0308] In some embodiments, the defect detection module 1304 is used to: determine that a planar dirt defect is a qualified defect when the defect category is a planar dirt defect.

[0309] Each module in the aforementioned manifold defect detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0310] In some embodiments, a computer device is provided, which may be a visual defect detection component, and its internal structure diagram may be as follows: Figure 14 As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for detecting defects in a power supply disk. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0311] Those skilled in the art will understand that Figure 14 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0312] In some embodiments, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the specific steps of the above-described embodiments of the manifold defect detection method.

[0313] In some embodiments, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the specific steps of the above-described embodiments of the collector disk defect detection method.

[0314] In some embodiments, a computer program product is provided, including a computer program that, when executed by a processor, implements the specific steps of the above-described embodiments of the collector disk defect detection method.

[0315] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the acquisition, storage, processing, and transmission of the data all comply with relevant laws and regulations.

[0316] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), data processing logic devices based on quantum computing, and the like.

[0317] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0318] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for detecting defects in a manifold, characterized in that, The method includes: Acquire the first illumination image information obtained by image acquisition of the collector disk under the first illumination environment; If a preliminary defect detection is performed on the collector disk based on the first lighting image information, and it is determined that the collector disk has a defect of undetermined category, then second lighting image information is obtained by acquiring images of the collector disk under a second lighting environment; the first illumination angle of the first lighting environment is higher than the second illumination angle of the second lighting environment; the defect of undetermined category refers to a defect whose category cannot be distinguished by using the first lighting image information alone; Based on the first lighting image information and the second lighting image information, a three-dimensional dirt defect detection is performed on the undetermined category of the defect to determine the defect category; the imaging of the three-dimensional dirt defect differs between the first lighting environment and the second lighting environment; The defect detection result of the manifold is determined based on the defect category; The step of determining the defect detection result of the manifold based on the defect category includes: In the case where the defect category is a three-dimensional dirt defect, the second defect size of the three-dimensional dirt defect is determined based on the second defect contour of the defect to be determined in the second illumination image information; If the second defect size is less than the first acceptable size threshold set for three-dimensional contamination defects, the three-dimensional contamination defect detection of the manifold is determined to be qualified. If the defect category is planar contamination defect, the planar contamination defect of the manifold is determined to be qualified. The method further includes: if it is determined, based on the first illumination image information, that there is a defect to be detected in the collector disk, extracting a first defect contour of the defect to be detected; The first defect size of the defect to be detected is determined based on the first defect profile. If the first defect size is less than a second acceptable size threshold set for planar contamination defects, and is greater than or equal to a first acceptable size threshold, the outline position of the first defect outline is determined; the second acceptable size threshold is greater than the first acceptable size threshold. If, based on the contour position, it is determined that the defect to be detected requires three-dimensional dirt defect detection, the defect to be detected is classified as a defect of undetermined category.

2. The method according to claim 1, characterized in that, The method further includes: For the weld bead area of ​​the manifold, the first lighting image information is used for region recognition to obtain the region outline position of the weld bead area; The contour position of the defect to be detected is matched with the contour position of the region to obtain the defect position matching result; If the defect location matching result indicates that the defect to be detected is not located in the weld area, it is determined that the defect to be detected needs to undergo three-dimensional contamination defect detection.

3. The method according to claim 2, characterized in that, The method further includes: If the defect location matching result indicates that the defect to be detected is located in the weld area, the defect to be detected is determined to be a pinhole burst defect; If the first defect size is less than the third acceptable size threshold set for the pinhole burst defect, the pinhole burst defect detection of the manifold is determined to be qualified; the third acceptable size threshold is greater than the first acceptable size threshold and less than the second acceptable size threshold.

4. The method according to claim 2, characterized in that, The step of performing region recognition on the first illumination image information to obtain the region contour position of the weld bead area for the manifold includes: Use a preset box of interest to extract the initial region information containing the weld area from the first lighting image information; The initial region information is input into a pre-set weld width detection model. The initial region information is then truncated and corrected based on the weld width of the weld region to obtain the region outline position of the weld region.

5. The method according to claim 2, characterized in that, The outline position of the defect to be detected includes the defect outline position coordinates of the defect to be detected, and the region outline position includes the region outline position coordinates of the weld area. The step of matching the contour position of the defect to be detected with the contour position of the region to obtain the defect position matching result includes: The defect contour position coordinates of the defect to be detected are mapped to the region contour position coordinates of the weld area to determine the area of ​​the overlapping region between the defect contour of the defect to be detected and the region contour of the weld area. Determine the area ratio between the area of ​​the overlapping region and the area of ​​the defect contour; If the area ratio is greater than a preset area ratio threshold, the defect location matching result indicates that the defect to be detected is located in the weld area.

6. The method according to claim 1, characterized in that, The defect categories include three-dimensional dirt defects and two-dimensional dirt defects; the first illumination image information includes bright field image information, and the second illumination image information includes dark field image information. The step of performing three-dimensional dirt defect detection on the undetermined defects based on the first lighting image information and the second lighting image information, and determining the defect category of the undetermined defects, includes: Gray-scale features are extracted from the bright-field image information and the dark-field image information respectively to obtain the bright-field gray-scale features and dark-field gray-scale features of the undetermined category of defects; If the feature difference between the bright field grayscale feature and the dark field grayscale feature is greater than the three-dimensional dirt difference threshold, the defect category of the undetermined defect is determined to be a three-dimensional dirt defect. If the feature difference is less than or equal to the three-dimensional dirt difference threshold, the defect category of the undetermined defect is determined to be a planar dirt defect.

7. The method according to claim 6, characterized in that, The step of extracting grayscale features from the bright-field image information and the dark-field image information respectively to obtain the bright-field grayscale features and dark-field grayscale features of the undetermined category of defects includes: Based on the first defect contour of the undetermined defect category, grayscale features are extracted from the brightfield image information to obtain the brightfield grayscale features of the undetermined defect category. Using the first defect contour as the detection region of interest, region mapping is performed on the dark field image information to obtain the second defect contour of the defect of the undetermined category. Based on the second defect contour, grayscale features are extracted from the dark field image information to obtain the dark field grayscale features of the undetermined category defect.

8. A collector plate defect detection device, characterized in that, The device includes: The first lighting information acquisition module is used to acquire the first lighting image information obtained by image acquisition of the collector disk under the first lighting environment; The second lighting information acquisition module is used to acquire second lighting image information obtained by image acquisition of the collector under a second lighting environment when the collector is determined to have a category of undetermined defect after preliminary defect detection based on the first lighting image information; the category of undetermined defect refers to a defect whose category cannot be distinguished by using the first lighting image information alone. The defect category determination module is used to perform three-dimensional dirt defect detection on the undetermined defects based on the first lighting image information and the second lighting image information, and determine the defect category of the undetermined defects; the first illumination angle of the first lighting environment is higher than the second illumination angle of the second lighting environment; there are differences in the imaging of three-dimensional dirt defects in the first lighting environment and the second lighting environment; A defect detection module is used to determine the defect detection result of the collector plate based on the defect category; Specifically, the defect detection module is used to determine the second defect size of the three-dimensional dirt defect based on the second defect outline of the undetermined defect in the second lighting image information when the defect category is a three-dimensional dirt defect; determine that the three-dimensional dirt defect detection of the manifold is qualified when the second defect size is less than the first qualified size threshold set for the three-dimensional dirt defect; and determine that the planar dirt defect detection of the manifold is qualified when the defect category is a planar dirt defect. The device further includes: The defect contour extraction module is used to extract the first defect contour of the defect to be detected when it is determined that there is a defect to be detected in the collector disk based on the first illumination image information. The first defect size determination module is used to determine the first defect size of the defect to be detected based on the first defect profile. The contour position determination module is used to determine the contour position of the first defect contour when the first defect size is less than a second acceptable size threshold set for planar contamination defects, and is greater than or equal to a first acceptable size threshold; the second acceptable size threshold is greater than the first acceptable size threshold. The category-pending defect determination module is used to determine the defect to be detected as a category-pending defect when it is determined, based on the contour position, that the defect to be detected needs to be detected by three-dimensional dirt defect detection.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

12. A manifold defect detection system, characterized in that, The system includes a first lighting environment component, a second lighting environment component, an image acquisition component, and a visual defect detection component that is communicatively connected to the image acquisition component; The first lighting environment component is used to provide a first lighting environment for the collector disk; The second lighting environment component is used to provide a second lighting environment for the collector disk; the first illumination angle of the first lighting environment is higher than the second illumination angle of the second lighting environment; the image acquisition component is used to acquire first lighting image information of the collector disk under the first lighting environment and second lighting image information under the second lighting environment; The visual defect detection component is used to implement the manifold defect detection method as described in any one of claims 1 to 7.

13. The system according to claim 12, characterized in that, The system also includes an acquisition and control component that is communicatively connected to the visual defect detection component; The acquisition control component is electrically connected to the first lighting environment component, the second lighting environment component, and the image acquisition component, respectively. It controls the opening and closing of the first lighting environment component or the second lighting environment component by sending electrical signals to the first lighting environment component or the second lighting environment component, and controls the image acquisition component to acquire images of the collector disk by sending electrical signals to the image acquisition component.

14. The system according to claim 12 or 13, characterized in that, The first lighting environment component includes a bowl-shaped light source with a first illumination angle and a coaxial light source; The second lighting environment component includes a ring light source with a second illumination angle.

Images