A method, device, equipment and medium for intelligent detection of lithium battery pole piece defects

Abstract

The present application relates to an intelligent detection method, device, equipment and medium for lithium battery pole piece defects. An intelligent detection method for lithium battery pole piece defects includes calling defect history data, generating a defect distribution heat map for each calibration sub-image based on the defect history data; determining the detection mode to be switched based on the thermal threshold comparison results of each calibration sub-image in the defect distribution heat map; if the detection mode is a focus mode, the second current detection image is divided into detection sub-images corresponding to each calibration sub-image, and each detection sub-image is subjected to defect detection according to the detection rules corresponding to the thermal threshold comparison results. Targeted detection rules are adopted to enhance the pertinence and accuracy of detection. The accuracy of pole piece defect detection is improved, invalid calculations are reduced, and efficient and high-accuracy lithium battery pole piece defect detection is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of lithium battery pole piece quality detection, and in particular to an intelligent detection method, device, equipment and medium for lithium battery pole piece defects. Background Art

[0002] At present, lithium battery pole pieces are widely used in new energy vehicles, energy storage equipment and various electronic products. Their manufacturing quality directly affects the safety, service life and performance stability of the battery. Therefore, in the production process, efficient and accurate defect detection of lithium battery pole pieces has become a key link in improving product quality.

[0003] Currently, lithium battery pole defect detection mainly relies on traditional image processing or manual visual inspection. Manual detection relies on experience, has subjective errors, and is difficult to ensure consistency. Traditional image processing-based methods usually use fixed thresholds or edge detection algorithms to identify defects. However, in cases where the pole edge is blurred, the lighting is uneven, or the defect types are complex, the detection accuracy decreases, and it is easy to cause false detection or missed detection. In addition, although some machine learning-based detection methods have improved the detection accuracy, they still cannot dynamically adjust the detection area, resulting in a waste of computing resources and low detection efficiency. Summary of the invention

[0004] In order to solve the problems that existing lithium battery pole piece defect detection methods have limited detection accuracy in complex environments and lack an intelligent detection mode optimization mechanism based on historical data, the present application provides an intelligent detection method, device, equipment and medium for lithium battery pole piece defects.

[0005] The above-mentioned invention objective of the present application is achieved through the following technical solutions: An intelligent detection method for lithium battery pole piece defects, the intelligent detection method for lithium battery pole piece defects comprising: Acquire a calibration detection image, and divide the calibration monitoring image into a plurality of calibration sub-images according to a preset division rule; Calling defect history data, and generating a defect distribution heat map for each calibrated sub-image according to the defect history data; Determining the detection mode to be switched based on the comparison results of the thermal thresholds of each calibrated sub-image in the defect distribution thermal map; If the detection mode is the full-image mode, a first current detection image of the lithium battery electrode to be detected is obtained, and defect detection is performed on the first current detection image based on image processing technology; If the detection mode is a focusing mode, a second current detection image of the lithium battery pole piece to be detected is obtained, and the second current detection image is divided into detection sub-images corresponding to each of the calibration sub-images based on the preset division rule, and defect detection is performed on each of the detection sub-images according to the detection rule corresponding to the thermal threshold comparison result.

[0006] By adopting the above technical solution, by introducing a dynamic detection mode optimization mechanism based on defect history data, combined with image segmentation, heat map analysis and adaptive detection rules, the problems of limited detection accuracy, waste of computing resources and lack of intelligent optimization in the prior art are solved. Specifically, the method first obtains a calibration detection image and divides it into multiple calibration sub-images according to preset rules, thereby establishing a fine-grained detection unit to ensure the flexibility and local accuracy of the detection. Subsequently, the historical defect data is called to generate a defect distribution heat map for each calibration sub-image, and the historical statistical information is used to improve the pertinence and reliability of the detection. Based on the comparison results of the thermal thresholds in the heat map, the detection mode is adaptively determined-if the defect area is evenly distributed, the full image mode is adopted to improve the detection coverage; if the defect has a local concentration trend, it is switched to the focus mode, and key detection is performed in the high defect risk area, thereby reducing the waste of computing resources and improving the detection efficiency. In the full image mode, the system obtains the current detection image and performs defect detection based on image processing technology to ensure overall quality control. In the focus mode, the detection sub-image is further divided, and targeted detection rules are adopted according to the distribution of thermal thresholds, thereby enhancing the pertinence and accuracy of the detection. This linkage optimization mechanism based on historical data and real-time detection not only improves the accuracy of pole piece defect detection, but also reduces invalid calculations through intelligent mode switching, thereby achieving efficient and high-accuracy lithium battery pole piece defect detection.

[0007] In a preferred example, the present application may be further configured as follows: the step of acquiring a calibration detection image and dividing the calibration monitoring image into a plurality of calibration sub-images according to a preset division rule includes: Based on the image acquisition device, a preliminary detection image is obtained; Preprocessing the preliminary detection image to generate a corresponding preprocessed calibration image, calculating a clarity index of the preprocessed calibration image, and determining whether the clarity index meets a preset index. If not, reacquiring a new preliminary detection image. If yes, determining that the preliminary detection image is a calibration detection image; According to a preset division rule, a corresponding division ratio is determined, and a corresponding ratio code is generated based on the division ratio; Determine the size data of the calibration detection image, divide the calibration monitoring image according to the size data and the division ratio, generate corresponding calibration sub-images, and generate corresponding quantity codes according to the generated number of the calibration sub-images; Each calibration sub-image is sequentially encoded, and a corresponding number code is added to each encoded calibration sub-image according to the ratio code and the quantity code.

[0008] By adopting the above technical solution, the clarity index is calculated after the image is preprocessed and it is determined whether it meets the preset requirements, which can effectively ensure that the image quality meets the detection requirements. If the clarity of the initial detection image does not meet the requirements, the system will automatically re-acquire a new image, thereby avoiding subsequent detection errors caused by blurred or poor image quality. Further, by accurately calculating the size data and division ratio of the calibration detection image and generating the corresponding ratio code and quantity code, the standardization and normalization of image division and numbering can be ensured, human errors can be reduced, and more systematic and consistent data support can be provided for subsequent detection.

[0009] In a preferred example, the present application may be further configured as follows: the step of calling defect history data and generating a defect distribution heat map for each calibrated sub-image according to the defect history data includes: According to the ratio code and the quantity code, calling corresponding defect history data; Based on the coding code, performing defect classification statistics on the defect history data to generate regional statistical data corresponding to each coding code; A heat map template type is selected, and based on the visualization rule of the heat map template type, a corresponding defect distribution heat map is drawn according to the encoding code and the regional statistical data.

[0010] By adopting the above technical solution, calling defect history data to generate defect distribution heat map and performing defect classification statistics, the defect detection system can make intelligent decisions based on past data experience, thereby avoiding blind detection that relies solely on the current detection image. The generated heat map can intuitively display the defect density in each area, which helps to accurately judge the defect risk and importance of different areas. Selecting a suitable heat map template and drawing it according to the visualization rules of the heat map not only optimizes the display effect of the heat map, but also improves the accuracy and efficiency of subsequent defect detection. It can effectively combine historical defect information with real-time image information to improve the overall intelligence of detection.

[0011] In a preferred example, the present application may be further configured as follows: in the step of determining the detection mode to be switched based on the thermal threshold comparison results of each calibration sub-image in the defect distribution thermogram, the thermal threshold comparison results include the regional difference corresponding to each calibration sub-image, including: Determine thermal preset thresholds; Calculate the regional difference value of the thermal preset threshold minus the regional statistical data; If all the regional differences are positive numbers, the detection mode is determined to be the full-image mode; If any of the area difference values ​​is a negative number, it is determined that the detection mode is the focus mode.

[0012] By adopting the above technical solution, the detection mode is switched based on the comparison results of the thermal threshold in the defect distribution thermal map, and the detection strategy can be flexibly adjusted according to the defect distribution in different areas. By calculating the regional difference of each area, it can be determined whether it is necessary to switch to the focus mode based on the comparison results of the thermal preset threshold. When the difference of all areas is positive, the system can use the full-image mode for comprehensive detection, avoiding unnecessary local focusing processing; and when the difference of any area is negative, the system can quickly switch to the focus mode, focusing the detection accuracy on the defect-intensive areas, avoiding missed detection caused by the failure of traditional methods to intelligently identify key areas, and improving detection efficiency and accuracy.

[0013] In a preferred example, the present application may be further configured as follows: the step of performing defect detection on each of the detection sub-images according to the detection rule corresponding to the thermal threshold comparison result includes: Determine the regional difference of each of the detection sub-images according to the corresponding relationship between the detection sub-image and the calibration sub-image, and calculate the detection weight of each of the detection sub-images according to the regional difference; A corresponding detection order is generated according to the detection weight of each detection sub-image, and defect detection is performed on each detection sub-image according to the detection order.

[0014] By adopting the above technical solution, the regional difference calculation is performed based on the correspondence between the detection sub-image and the calibration sub-image, and the detection weight is generated, which can intelligently evaluate the importance and priority of each detection sub-image. This weight distribution mechanism ensures that areas with more serious defects are inspected first, and improves the detection accuracy of key areas. The detection order generated according to the detection weight helps to optimize the detection process and avoid the blind and disordered detection process in traditional methods. Through this intelligent weight and priority method, the detection accuracy can be effectively improved and the detection time can be reduced, ensuring that the detection process is efficient and accurate.

[0015] In a preferred example, the present application may be further configured as follows: if the detection mode is the full image mode, then obtaining a first current detection image of the lithium battery electrode to be detected, and performing defect detection on the first current detection image based on image processing technology, including: If the detection mode is the full-image mode, obtaining a first current detection image of the lithium battery electrode to be detected; Performing a preprocessing operation on the first current detection image, wherein the preprocessing operation includes a grayscale operation, a denoising operation, and a brightness equalization operation; Extracting defective areas in the first current detection image after preprocessing based on the image processing technology; Based on the preset division rule of the calibration sub-image, the pixel coordinates of the defect area are determined, and according to the mapping relationship between the pixel coordinates and the coordinate range corresponding to each calibration sub-image, the coverage area of ​​the defect area in each calibration sub-image is determined, and according to each coverage area, the calibration sub-image to which the defect area belongs is determined as the target sub-image; If there is one target sub-image, adding an abnormal mark to the target sub-image; If there are multiple target sub-images, the area ratio of the coverage area is determined, and a new preset division rule is re-determined according to the determination result of the area ratio determination. The new preset division rule is used to divide the new target sub-image, and then an abnormal mark is added to the new target sub-image; When performing secondary inspection of the lithium battery electrode to be inspected, it is detected whether the abnormal mark exists. If it exists, it is directly triggered to switch the inspection mode to the focusing mode to perform secondary inspection on the target sub-image.

[0016] By adopting the above technical solution, mapping calculations are performed based on the pixel coordinates of the defect area and the coordinate range of the calibration sub-image, so that the ownership of the defect area can be determined more accurately. By calculating the coverage area of ​​the defect area on the calibration sub-image, the system can accurately assign a target sub-image to each defect area, thereby avoiding misjudgment caused by uneven distribution of defect areas in traditional methods. When multiple target sub-images are detected, the system readjusts the division rules based on area ratio judgment and adds abnormal marks to the new target sub-image. This mechanism can effectively identify and process potential defect areas during secondary detection, further improving the intelligence and accuracy of detection. Through this dynamic adjustment and marking mechanism, not only the accuracy of the detection process is guaranteed, but also abnormal situations can be responded to in a timely manner to reduce the occurrence of false detection and missed detection.

[0017] The second object of the invention is achieved by the following technical solutions: An intelligent detection device for lithium battery pole piece defects, the intelligent detection device for lithium battery pole piece defects comprising: An acquisition module is used to acquire a calibration detection image and divide the calibration monitoring image into a plurality of calibration sub-images according to a preset division rule; A generation module, used to call defect history data and generate a defect distribution heat map for each calibration sub-image according to the defect history data; A determination module, configured to determine a detection mode to be switched based on a comparison result of thermal thresholds of each calibrated sub-image in the defect distribution thermal map; A first detection module, configured to obtain a first current detection image of the lithium battery electrode to be detected if the detection mode is a full-image mode, and perform defect detection on the first current detection image based on image processing technology; The second detection module is used to obtain a second current detection image of the lithium battery pole piece to be detected if the detection mode is a focusing mode, divide the second current detection image into detection sub-images corresponding to each of the calibration sub-images based on the preset division rule, and perform defect detection on each of the detection sub-images according to the detection rule corresponding to the thermal threshold comparison result.

[0018] The third objective of the present application is achieved through the following technical solutions: A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned intelligent detection method for lithium battery pole piece defects when executing the computer program.

[0019] The fourth objective of the present application is achieved through the following technical solutions: A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned intelligent detection method for lithium battery pole piece defects.

[0020] In summary, the present application includes at least one of the following beneficial technical effects: This application solves the problems of limited detection accuracy, waste of computing resources and lack of intelligent optimization in the prior art by introducing a dynamic detection mode optimization mechanism based on defect history data, combined with image segmentation, heat map analysis and adaptive detection rules. Specifically, the method first obtains a calibration detection image and divides it into multiple calibration sub-images according to preset rules, thereby establishing a fine-grained detection unit to ensure the flexibility and local accuracy of the detection. Subsequently, the historical defect data is called to generate a defect distribution heat map for each calibration sub-image, and the historical statistical information is used to improve the pertinence and reliability of the detection. Based on the comparison results of the thermal thresholds in the heat map, the detection mode is adaptively determined-if the defect area is evenly distributed, the full-image mode is adopted to improve the detection coverage; if the defect has a local concentration trend, it is switched to the focus mode, and key detection is performed in the high defect risk area, thereby reducing the waste of computing resources and improving the detection efficiency. In the full-image mode, the system obtains the current detection image and performs defect detection based on image processing technology to ensure overall quality control, while in the focus mode, the detection sub-image is further divided, and targeted detection rules are adopted according to the distribution of thermal thresholds, thereby enhancing the pertinence and accuracy of the detection. This linkage optimization mechanism based on historical data and real-time detection not only improves the accuracy of pole piece defect detection, but also reduces invalid calculations through intelligent mode switching, thereby achieving efficient and high-accuracy lithium battery pole piece defect detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flow chart of an intelligent detection method for lithium battery electrode defects in one embodiment of the present application.

[0022] Figure 2 It is a flowchart for implementing step S10 in a method for intelligently detecting defects in a lithium battery electrode in one embodiment of the present application; Figure 3 It is a flowchart for implementing step S20 in a method for intelligently detecting defects in a lithium battery electrode in one embodiment of the present application; Figure 4 It is a flowchart for implementing step S30 in a method for intelligently detecting defects in a lithium battery electrode in one embodiment of the present application; Figure 5 It is a flowchart for implementing step S50 in a method for intelligently detecting defects in a lithium battery electrode in an embodiment of the present application; Figure 6 It is a flowchart for implementing step S40 in a smart detection method for lithium battery pole piece defects in one embodiment of the present application; Figure 7 It is a principle block diagram of an intelligent detection device for lithium battery pole piece defects in one embodiment of the present application; Figure 8It is a schematic diagram of a device in an embodiment of the present application. DETAILED DESCRIPTION

[0023] The present application is further described in detail below in conjunction with the accompanying drawings.

[0024] In one embodiment, if Figure 1 As shown, the present application discloses an intelligent detection method for lithium battery pole piece defects, and the intelligent detection method for lithium battery pole piece defects includes: S10, obtain a calibration detection image, and divide the calibration monitoring image into multiple calibration sub-images according to a preset division rule; in this embodiment, the calibration detection image refers to an image used to establish a standardized detection benchmark in the detection system, which is usually obtained by an industrial camera or a high-precision scanning device to ensure that the image has clear details and stable lighting conditions, thereby improving the accuracy of the detection. The process of obtaining the calibration detection image usually includes adjusting the aperture, shutter speed and exposure parameters of the camera to adapt it to different batches of lithium battery pole pieces to avoid unstable image quality due to factors such as lighting changes and equipment jitter. The preset division rules for dividing the calibration sub-images are usually set based on the physical size of the image, the distribution characteristics of the defect area, and the detection accuracy requirements. For example, a grid division method can be used to divide the entire calibration detection image into rectangular areas of equal size, each rectangular area corresponding to a calibration sub-image to ensure that the detection area is evenly covered. If the defect frequency in some areas is high, the division density can be adaptively increased to make the sub-image size of these areas smaller to improve the accuracy of defect detection. In addition, during the segmentation process, the boundary relationship between the edge of the pole piece and the background needs to be considered to ensure that the sub-image after segmentation is not interfered by the background and to avoid false detection or missed detection.

[0025] S20, calling defect history data, and generating defect distribution heat maps for each calibration sub-image according to the defect history data; in this embodiment, defect history data refers to defect information collected from the detection process of multiple batches of lithium battery pole pieces in the past, which usually includes the type of defect (such as scratches, bubbles, and coating leakage), the frequency of occurrence of defects, and the location of defect occurrence. The purpose of calling defect history data is to use existing statistical data to predict the area where defects may occur in the current pole piece to be detected, so as to improve the pertinence and efficiency of detection. The generation process of defect distribution heat map usually adopts data visualization technology to map the frequency of defect occurrence at different locations to the color gradient, for example, the area with higher defect frequency is displayed in red, and the area with lower defect frequency is displayed in blue. In the process of generating the heat map, the defect data can be aggregated according to the divided area based on the encoding information of the calibration sub-image, and the data can be smoothed by the interpolation algorithm to ensure that the heat map can truly reflect the distribution of defects. For example, in a batch of lithium battery pole pieces, the probability of occurrence of pole piece edge defects in a certain area is high, then the heat map of the area will show a higher heat value, while the area with fewer defects will show a lower heat value. In this way, through the defect distribution heat map, we can focus on the high defect risk areas in the subsequent inspection process, thereby optimizing the inspection strategy and improving the inspection efficiency and accuracy.

[0026] S30, based on the comparison results of the thermal thresholds of each calibrated sub-image in the defect distribution thermodynamic map, determine the detection mode that needs to be switched; in this embodiment, the thermal threshold refers to a preset value on the defect distribution thermodynamic map used to determine whether a certain area needs to be focused on, and the value is usually obtained by historical data analysis and experience setting. The thermal threshold can be determined by calculating the defect density of each calibrated sub-image, that is, the probability of defects in the area in the past detection. If the probability exceeds a certain threshold, the area is considered to be a high-risk area. The switching of the detection mode is based on the comparison results of the thermal threshold. When the defect density of all calibrated sub-images is lower than the thermal threshold, the system determines that the overall defect distribution of the pole piece is uniform. At this time, the full image mode is used for detection, that is, the entire pole piece is scanned uniformly to ensure the overall detection coverage; and when the defect density of some calibrated sub-images is higher than the thermal threshold, the system will determine that the area has a high defect risk, and automatically switch to the focus mode at this time, and only focus on the high-risk area for detection, thereby improving the detection efficiency and avoiding unnecessary calculations and resource consumption for defect-free areas. For example, in some production batches, due to slight errors in the coating equipment, the defect rate on the left side of the pole piece is higher, while that on the right side is lower. The system will automatically recognize this feature and focus the inspection on the left area, optimizing resource allocation and improving the accuracy of inspection.

[0027] S40, if the detection mode is the full-image mode, then obtain the first current detection image of the lithium battery pole piece to be detected, and perform defect detection on the first current detection image based on the image processing technology; in this embodiment, the first current detection image refers to the detection image of the complete lithium battery pole piece directly obtained by the image acquisition device in the full-image mode. In the process of obtaining the first current detection image, it is necessary to ensure that the camera parameters are consistent with the parameters of the calibration detection image to ensure the consistency of the detection results. After obtaining the detection image, the system will perform image processing technology analysis on it, including grayscale, denoising, edge enhancement and other processing steps to improve the identifiability of the defective area. Defect detection mainly adopts a method combining edge detection, morphological analysis and deep learning algorithm. First, edge detection technology (such as Sobel operator or Canny operator) is used to extract the edge information of the pole piece to determine whether there are obvious scratches or defects. Then, morphological operations (such as expansion and corrosion) are used to remove noise to ensure that the detected defects are real, rather than false detections caused by factors such as dust and light changes. In addition, if the system integrates a deep learning model, it can further analyze the image through a convolutional neural network (CNN) and automatically classify the defect type to improve the accuracy of detection. For example, during the detection process, if the system finds that the color change in a certain area is abnormal and the edge is discontinuous, it can be determined that the area may have a coating defect and mark it for subsequent processing.

[0028] S50, if the detection mode is the focus mode, then obtain the second current detection image of the lithium battery pole piece to be detected, divide the second current detection image into detection sub-images corresponding to each calibration sub-image based on the preset division rule, and perform defect detection on each detection sub-image according to the detection rule corresponding to the thermal threshold comparison result. In this embodiment, the second current detection image refers to the detection image of the local area of ​​the pole piece obtained by the image acquisition device in the focus mode. The acquisition of this image usually depends on the guidance of the thermal map, so that the system only performs detailed detection on the high defect risk area. After obtaining the second current detection image, the system will divide it into multiple detection sub-images based on the preset division rule to ensure the refinement of the detection. The division method of the detection sub-image can be consistent with the calibration sub-image, or it can be dynamically adjusted according to the real-time situation. For example, if the defect frequency of a certain area is high, the detection sub-image of the area can be divided into smaller ones to improve the detection resolution. For each detection sub-image, the system will apply the corresponding detection rules according to the comparison result of the thermal threshold, such as using stricter detection standards for high threshold areas and looser standards for low threshold areas to reduce the false alarm rate. For example, in some production batches, some areas of the pole piece may have serious uneven coating. The system can increase the weight of edge detection for these areas based on the detection rules, and combine machine learning algorithms to identify whether they are coating defects, thereby improving the accuracy and reliability of detection.

[0029] In summary, by introducing a dynamic detection mode optimization mechanism based on defect history data, combined with image segmentation, heat map analysis and adaptive detection rules, the problems of limited detection accuracy, waste of computing resources and lack of intelligent optimization in the prior art are solved. Specifically, the method first obtains a calibration detection image and divides it into multiple calibration sub-images according to preset rules, thereby establishing a fine-grained detection unit to ensure the flexibility and local accuracy of the detection. Subsequently, the historical defect data is called to generate a defect distribution heat map for each calibration sub-image, and the historical statistical information is used to improve the pertinence and reliability of the detection. Based on the comparison results of the thermal thresholds in the heat map, the detection mode is adaptively determined - if the defect area is evenly distributed, the full image mode is used to improve the detection coverage; if the defect has a local concentration trend, it is switched to the focus mode, and key detection is performed in the high defect risk area, thereby reducing the waste of computing resources and improving the detection efficiency. In the full image mode, the system obtains the current detection image and performs defect detection based on image processing technology to ensure overall quality control. In the focus mode, the detection sub-image is further divided, and targeted detection rules are adopted according to the distribution of thermal thresholds, thereby enhancing the pertinence and accuracy of the detection. This linkage optimization mechanism based on historical data and real-time detection not only improves the accuracy of pole piece defect detection, but also reduces invalid calculations through intelligent mode switching, thereby achieving efficient and high-accuracy lithium battery pole piece defect detection.

[0030] In one embodiment, if Figure 2 As shown, in step S10, that is, obtaining a calibration detection image, and dividing the calibration monitoring image into a plurality of calibration sub-images according to a preset division rule, the steps include: S101. Based on the image acquisition device, a preliminary detection image is obtained; in this embodiment, the image acquisition device refers to an imaging device for acquiring images of lithium battery pole pieces, usually including an industrial camera, a CCD camera or a high-precision scanner to ensure that the acquired image has sufficient resolution and detail information. The preliminary detection image refers to the original image without any preprocessing, which comes directly from the image acquisition device, usually with possible noise, uneven lighting or slight distortion. In practical applications, fixed or mobile cameras can be used for acquisition. Fixed cameras are suitable for assembly line inspection and can ensure that the acquisition parameters of each image are consistent, while mobile cameras can be used for shooting at different angles to improve the comprehensiveness of the inspection. For example, on a production line, a high-speed industrial camera can be used to continuously shoot the pole piece, and the exposure time can be controlled by a trigger signal to adapt to different lighting conditions and ensure image clarity.

[0031] S102, preprocessing the preliminary detection image, generating a corresponding preprocessing calibration image, calculating the clarity index of the preprocessing calibration image, judging whether the clarity index meets the preset index, if not, reacquiring a new preliminary detection image, if so, determining the preliminary detection image as a calibration detection image; in this embodiment, the preprocessing calibration image refers to a standardized image that can be used for subsequent detection after a series of image enhancement and optimization. The preprocessing operation usually includes steps such as graying, denoising and brightness equalization to ensure that the contrast, clarity and texture details of the image meet the detection requirements. The clarity index is used to measure the sharpness and edge details of the image. Common calculation methods include Laplace transform, Fourier transform or gradient statistical analysis. For example, the variance of the Laplace operator can be used to calculate the edge strength of the image. If the variance is lower than the preset threshold, the image is considered to be blurred and needs to be re-collected. If the clarity requirements are met, the image is determined as a calibration detection image to ensure the accuracy and stability of subsequent defect detection. Through this process, low-quality images can be effectively filtered to avoid the impact of blur or noise on the detection results, and improve the reliability and consistency of the overall detection.

[0032] S103. According to the preset division rules, determine the corresponding division ratio, and generate the corresponding ratio code based on the division ratio; in this embodiment, the preset division rule refers to the image division method set according to the size of the lithium battery pole piece, the detection accuracy requirements and the defect distribution characteristics. The division ratio determines the size and number of the calibration sub-images, and its calculation method is usually based on the pixel resolution and physical size of the overall image. For example, if the width of the pole piece is 300mm and the height is 400mm, the system can set the division ratio to 50mm×50mm, so that the entire image is evenly divided into 6×8 grid-shaped calibration sub-images. The ratio code is used to record the coding information of the division ratio, so that the subsequent steps can correctly parse the division scheme to ensure that the image division method of different inspection batches is consistent. For example, the ratio code can be encoded in binary or hexadecimal to store the division size and ratio information in the system for subsequent calls.

[0033] S104, determine the size data of the calibration detection image, divide the calibration monitoring image according to the size data and the division ratio, generate corresponding calibration sub-images, and generate corresponding quantity codes according to the number of calibration sub-images generated; in this embodiment, the size data refers to the physical size or pixel resolution of the calibration detection image, which is used to guide the accurate division of the image to ensure that the size of each calibration sub-image is consistent. When dividing, the system first obtains the total width and height data of the image, and calculates the number of calibration sub-images to be generated in combination with the division ratio. For example, if the image resolution is 6000×8000 pixels and the division ratio is 1000×1000 pixels, the entire image can be divided into 6×8=48 calibration sub-images. After the calibration sub-image is generated, the system will assign it a unique quantity code, which is used to identify the total number of sub-images in the current detection batch for subsequent data management and storage. For example, if the current batch of detection images generates 48 calibration sub-images, the system will store the quantity code "48" to ensure that all related processing is based on the same number of images.

[0034] S105. Sequentially encode each calibrated sub-image, and add a corresponding number code to each coded calibrated sub-image according to the ratio code and the quantity code. In this embodiment, sequential coding refers to numbering the calibrated sub-images in a specific order so that the specific position of each sub-image can be accurately located during subsequent detection and data analysis. The sequential coding method can be performed in a row-first or column-first order. For example, if the row-first coding method is used, the sub-image in the upper left corner is numbered "01", and the numbers on its right are incremented in sequence until the end of the row, and the next row continues to be encoded from the left. This numbering method can ensure that the positional relationship of each sub-image in the entire image remains consistent. The numbering code refers to the unique identification information of each calibrated sub-image, and the numbering information is usually combined with the ratio code and the quantity code to ensure data consistency. For example, if the scale code of a calibrated sub-image is "05050" (indicating 50mm×50mm division) and the quantity code is "48" (indicating that 48 sub-images are generated in the current inspection image), its number code may be "05050-48-01", where "01" represents the order of the sub-image in the entire image. By adding number codes to the calibrated sub-images, it is possible to accurately track and manage each sub-image during subsequent defect detection, data storage, and result analysis, thereby improving the standardization and automation of the inspection process.

[0035] In one embodiment, if Figure 3 As shown, in step S20, that is, calling the defect history data, the step of generating the defect distribution heat map for each calibration sub-image according to the defect history data includes: S201. According to the ratio code and the quantity code, the corresponding defect history data is called; in this embodiment, the ratio code is the parameter information generated by the calibration test image during the division process, indicating the division ratio of the calibration sub-image, and the quantity code is used to record the total number of calibration sub-images of the current test image. Defect history data refers to defect information recorded in the past detection process, including information such as the frequency of occurrence, distribution location, defect type, and production batch of defects. The purpose of calling defect history data is to retrieve the data that matches the current detection image division method in history by comparing the division ratio and quantity of the current detection image to ensure the accuracy of statistical analysis. For example, if the current calibration test image is divided according to the ratio of 50mm×50mm and 48 calibration sub-images are generated, the system will retrieve all historical defect data that are also divided according to the ratio and have the same number of sub-images in the database for subsequent statistical analysis. This can ensure the consistency of historical data with current detection data, avoid data deviation caused by different division methods, and improve the accuracy of defect prediction.

[0036] S202, based on the coding code, the defect historical data is subjected to defect classification statistics, and regional statistical data corresponding to each coding code is generated; in this embodiment, the coding code is parameter information used to uniquely identify the calibration sub-image, which is usually composed of a ratio code, a quantity code, and the numbering information of the calibration sub-image, to ensure that each sub-image has a unique corresponding identification in the entire data system. Defect classification statistics refers to the classification and calculation of the defect types recorded in the historical data to analyze the frequency of occurrence of each defect in different calibration sub-images. The specific implementation method is to first group all historical defect data by sub-image according to the coding code, and then perform statistics according to the defect type (such as scratches, missing coating, foreign matter, wrinkles, etc.). For example, if a calibrated sub-image has 20 scratches, 10 missing coatings, and 5 foreign matter in the past 100 batches of inspections, the regional statistical data of the sub-image will contain the statistical information of these three defects and calculate the probability of their defect occurrence. In this way, it is possible to intuitively analyze whether each sub-image is a high defect occurrence area in the past inspection records, and provide data support for the subsequent drawing of the thermal map.

[0037] S203, select the heat map template type, based on the visualization rules of the heat map template type, according to the coding code and regional statistical data, draw the corresponding defect distribution heat map. In this embodiment, the heat map template type is a parameter used to control the display mode of the heat map. Different color gradients, data mapping methods and smoothness of the heat area can be selected according to different visualization requirements. Visualization rules refer to the rules for how to convert regional statistical data into visual color information when drawing a heat map, usually including setting the color gradient range, the mapping method of defect distribution density and the outlier processing method. The process of drawing the defect distribution heat map is first to match the statistical data of each calibration sub-image according to the coding code, and then perform color mapping according to the visualization rules, setting the area with high defect frequency to red, the area with medium frequency to yellow, and the area with low frequency to blue. For example, in the detection of a batch of lithium battery pole pieces, if the calibration sub-image in the upper left corner has more defects in the historical detection, the thermal value of this area will be higher than other areas, and will appear darker red on the heat map, while other areas with fewer defects will be lighter in color. In this way, the defect distribution of each calibrated sub-image can be intuitively reflected, helping the inspection system to quickly lock high-risk areas and improve the pertinence and efficiency of inspection.

[0038] In one embodiment, if Figure 4 As shown, in step S30, that is, in the step of determining the detection mode to be switched based on the thermal threshold comparison results of each calibrated sub-image in the defect distribution thermogram, the thermal threshold comparison results include the regional differences corresponding to each calibrated sub-image, including: S301, determine the thermal preset threshold; in this embodiment, the thermal preset threshold is a reference value for determining whether a certain calibrated sub-image needs to be inspected in detail, and the value is usually obtained by analyzing historical defect data or setting by experience. Its role is to provide a measurement standard so that the system can determine the detection mode based on the difference between the value and the actual detection data. The method for determining the thermal preset threshold can adopt a statistical method or a machine learning algorithm. For example, the average value of the defect occurrence frequency of a certain area can be calculated based on the detection data of multiple batches in the past, and a safety factor can be set to ensure that the detection strategy can cover common defect areas and avoid excessive waste of computing resources. For example, if the average defect occurrence frequency of a calibrated sub-image in the past 50 detection batches is 0.3 (i.e., there is a 30% probability of defects), the thermal preset threshold can be set to 0.4 to ensure that only when the defect probability exceeds 40%, the area is identified as a high defect risk area, and in the subsequent steps, it is decided whether to switch to a more refined detection mode.

[0039] S302, calculate the regional difference of the thermal preset threshold minus the regional statistical data; in this embodiment, the regional difference refers to the value obtained by comparing the thermal preset threshold with the actual defect statistical data, which reflects whether the defect frequency of a calibrated sub-image exceeds the set safety range. The method of calculating the regional difference is to subtract the defect incidence rate of each calibrated sub-image from the thermal preset threshold. If the value is positive, it means that the defect incidence rate of the area is lower than expected and can be classified as a low-risk area; if the value is negative, it means that the defect incidence rate of the area exceeds the preset threshold and needs to be focused on. For example, assuming that the thermal preset threshold of a certain area is 0.4, and the actual defect frequency of the area is 0.6, the calculated regional difference is -0.2, indicating that the defect frequency of the area is high, which needs to attract system attention and may adjust the detection mode. By calculating the regional difference, it is possible to intuitively determine which areas have a higher defect risk and provide a basis for subsequent detection mode selection.

[0040] S303, if all the regional differences are positive, the detection mode is determined to be the full-image mode; in this embodiment, the full-image mode refers to uniformly scanning the entire detection image without paying extra attention to any specific area to ensure that all areas can be covered and detected. When all regional differences are positive, it means that the defect occurrence frequency of all current calibration sub-images does not exceed the set thermal preset threshold, indicating that the quality of the pole pieces of the current detection batch is relatively stable, and there is no need to perform local key detection, so the full-image mode can be used for detection. In the full-image mode, the system will perform the same image processing and defect detection algorithms on the entire detection image according to the standard process, such as edge detection, morphological analysis or deep learning model recognition, etc., without using higher detection accuracy or stricter detection standards for certain specific areas. For example, in the detection process of a batch of pole pieces, if the regional differences of all calibration sub-images are greater than 0.1, the system will select the full-image mode to process the entire image with standard detection accuracy to ensure maximum detection efficiency while avoiding the increase in computational burden due to unnecessary mode switching.

[0041] S304, if any of the regional differences is a negative number, the detection mode is determined to be a focus mode. In this embodiment, the focus mode refers to focusing on the detection of areas with a high defect incidence rate to improve the detection accuracy and reduce the false detection rate. When at least one of the calculated regional differences is a negative number, it means that the defect frequency in the area has exceeded the thermal preset threshold, indicating that there is a large defect risk in the area and a more sophisticated detection strategy needs to be adopted. Therefore, the system will automatically switch to the focus mode and only conduct in-depth detection on high-risk areas to improve the accuracy of defect identification. The focus mode usually adopts a higher image resolution, a more sophisticated feature extraction method or a stricter threshold judgment standard. For example, in this mode, a smaller detection sub-image can be used for analysis, or a deep learning model can be introduced to further classify the defect type in the area to ensure that the detection result is more accurate. For example, in a certain production batch, if the difference of the calibrated sub-image area in the left area is -0.3, and the differences in other areas are all positive, the system will automatically set the left area to the focus mode and increase the sampling rate or detection accuracy of the area to ensure that the defects in the area can be accurately identified and marked. In this way, by switching to the focus mode, the detection accuracy of the key area can be maximized, while avoiding unnecessary high-precision calculations of the entire image, thereby improving the efficiency and reliability of detection.

[0042] In one embodiment, if Figure 5 As shown, in step S50, that is, the step of performing defect detection on each detection sub-image according to the detection rule corresponding to the thermal threshold comparison result, it includes: S501. According to the correspondence between the detection sub-image and the calibration sub-image, the regional difference of each detection sub-image is determined, and the detection weight of each detection sub-image is calculated according to the regional difference; in this embodiment, the detection sub-image refers to the sub-region image divided from the current image to be detected after the detection mode is switched to the focus mode, and the calibration sub-image refers to the standardized reference sub-image divided according to the preset rules in the calibration detection image. The correspondence between the detection sub-image and the calibration sub-image is used to ensure that the current area to be detected can be accurately matched and analyzed based on historical data and known defect information during the detection process. The regional difference refers to the difference between the defect distribution data of the detection sub-image and the historical defect distribution data of the calibration sub-image. The difference reflects whether the defect incidence rate of a certain area in the current detection batch is higher or lower than the historical average. The method for calculating the regional difference is usually to compare the defect statistics of the detection sub-image with the historical data of the corresponding calibration sub-image. If the defect frequency of the area is higher than the set thermal preset threshold, the regional difference is negative, indicating that the area has a large defect risk and needs to be paid special attention; if the regional difference is positive, it means that the defect incidence rate of the area is low and a lower priority can be used for detection. The detection weight is a calculated value used to allocate detection resources and priorities. It is usually calculated based on the regional difference. The higher the weight, the greater the detection importance of the area. In practical applications, the detection weight can be calculated based on the regional difference and the set defect risk coefficient. For example, if the regional difference of a detection sub-image is -0.4 and the defect risk coefficient is set to 1.5, the detection weight of the detection sub-image can be set to 0.6, indicating that the area requires more detailed detection. For example, in the detection of a batch of lithium battery pole pieces, if the historical defect incidence rate of a certain area is 10%, and the defect incidence rate of the current batch reaches 30%, the calculated regional difference is -0.2, and the system will automatically increase the detection weight of the area to ensure that the defects can be accurately identified and marked. In this way, the allocation of detection resources can be effectively optimized and the targeting and accuracy of detection can be improved.

[0043] S502. Generate a corresponding detection order according to the detection weight of each detection sub-image, and perform defect detection on each detection sub-image according to the detection order. In this embodiment, the detection order refers to the detection execution order set for each detection sub-image according to the size of the detection weight. This order is used to optimize the detection process and improve the detection efficiency of key areas. The method of generating the detection order is usually to arrange in descending order according to the detection weight, and the detection sub-image with a higher weight is detected first to ensure that high-risk areas can be quickly identified and processed. For example, in the process of detecting a batch of pole pieces, if the defect weight of some areas is high, the detection priority of these areas will be increased, so that the system can scan and analyze these high-risk areas first to ensure that key defects are not missed. When performing defect detection, each detection sub-image will be processed according to the established detection rules. These rules may include edge detection, color analysis, texture matching or deep learning recognition algorithms to ensure that different types of defects can be accurately identified. For example, if the detection weight of a detection sub-image is 0.8, and the detection weight of another detection sub-image is 0.5, the system will first perform detection on the sub-image with a weight of 0.8, and output the detection results first, so that the subsequent processing system can make decisions and respond as soon as possible. This weight-sorting detection strategy can significantly improve the efficiency of detection, avoid unnecessary waste of computing resources, and ensure that defects in key areas can be discovered and processed in a timely manner.

[0044] In one embodiment, if Figure 6 As shown, in step S40, that is, if the detection mode is the full image mode, a first current detection image of the lithium battery electrode to be detected is obtained, and the step of performing defect detection on the first current detection image based on image processing technology includes: S401. If the detection mode is the full-image mode, the first current detection image of the lithium battery pole piece to be detected is obtained; in this embodiment, the first current detection image refers to the detection image of the complete lithium battery pole piece obtained by the image acquisition device in the full-image mode. The full-image mode is used to uniformly detect the entire pole piece, so the acquired image needs to have sufficient resolution and uniform lighting conditions to ensure that the defect features can be clearly presented. In practical applications, image acquisition devices usually use industrial cameras or high-precision scanners to ensure the stability and consistency of the acquired images. For example, on an automated production line, the camera can be fixed on the inspection station, and the shutter time can be controlled by a trigger signal so that each inspection image is acquired under the same exposure parameters, thereby eliminating the impact of changes in ambient light on the detection accuracy.

[0045] S402, perform preprocessing operations on the first current detection image, the preprocessing operations include grayscale operation, denoising operation and brightness equalization operation; in this embodiment, the preprocessing operation refers to optimizing the detection image before the formal defect detection to improve the recognizability of the defect area. The grayscale operation is used to convert the color image into a single-channel grayscale image to reduce the computational complexity and enhance the contrast of the image to make the defect features more prominent. The denoising operation is used to remove random noise or texture interference in the image, usually using methods such as Gaussian filtering or median filtering to ensure that the detection system can identify defects more accurately. The brightness equalization operation is used to adjust the brightness distribution of the image to avoid false detection or missed detection due to uneven local illumination. For example, in practical applications, if the left side of the image has low brightness due to insufficient illumination, and the right side has too high brightness due to strong reflected light, the histogram equalization method can be used to make the brightness distribution of the entire image more uniform, thereby improving the stability and accuracy of the overall detection.

[0046] S403, based on image processing technology, extract the defective area in the first current detection image after the preprocessing operation; in this embodiment, the defective area refers to the area in the detection image that shows obvious deviation from the normal electrode surface, and these areas may contain scratches, wrinkles, pollution, missing coating and other defects. The process of extracting defective areas usually adopts image segmentation, edge detection or deep learning methods. For example, the Canny operator can be used for edge detection to extract the area with drastic grayscale changes in the image, and then identify the part that may have defects. In addition, convolutional neural network (CNN) can also be used to extract and classify features of the image to improve the recognition ability of complex defects. For example, in the detection process of lithium battery pole pieces, if the grayscale value of a certain area is abnormal and the edge has a large gradient change, then the area is likely to contain scratches or missing coating defects, and the system will automatically mark the area and enter the next step of analysis.

[0047] S404, based on the preset division rule of the calibration sub-image, determine the pixel coordinates of the defective area, and determine the coverage area of ​​the defective area in each calibration sub-image according to the mapping relationship between the pixel coordinates and the coordinate range corresponding to each calibration sub-image, and determine the calibration sub-image to which the defective area belongs as the target sub-image according to each coverage area; in this embodiment, the pixel coordinates are used to accurately describe the position of the defective area in the detection image. Through the preset division rule of the calibration sub-image, the entire detection image can be divided into a grid, and a unique calibration sub-image number is assigned to each grid. When the defective area is detected, the system calculates the pixel coordinates of the area and maps it to the corresponding calibration sub-image. If the defective area falls completely into a calibration sub-image, the calibration sub-image is identified as the target sub-image; if the defective area spans multiple calibration sub-images, it is necessary to further calculate the coverage area of ​​each calibration sub-image. For example, if the pixel coordinates of a defective area are located at the boundary of the calibration sub-image grid and occupy two calibration sub-images at the same time, it is necessary to calculate the coverage ratio of the area in each calibration sub-image, so as to subsequently determine the attribution of the target sub-image.

[0048] S405. If there is only one target sub-image, an abnormal mark is added to the target sub-image; in this embodiment, the abnormal mark refers to the identification information added by the system to the calibration sub-image where the defect is detected, which is used for subsequent detection mode adjustment or data storage. When the target sub-image is unique, it means that the defect area falls completely within a certain calibration sub-image, so an abnormal mark can be directly added to the sub-image so that the area can be quickly identified during subsequent detection or quality control. For example, in the detection system, a database can be used to record abnormal marks, and the calibration sub-image can be focused on in subsequent batch detection, or the abnormal information of the sub-image can be highlighted in the quality inspection report for production personnel to review and repair.

[0049] S406, if there are multiple target sub-images, then the area ratio of the coverage area is determined, and according to the determination result of the area ratio determination, a new preset division rule is re-determined. The new preset division rule is used to divide the new target sub-image, and then add an abnormal mark to the new target sub-image; in this embodiment, the area ratio determination of the coverage area means that when a defect area spans multiple calibration sub-images, the system calculates the coverage ratio of the area on different sub-images, and determines the final target sub-image according to the preset rules. Generally speaking, the system will set a minimum coverage ratio threshold. For example, if the coverage area of ​​a sub-image is less than 10%, the sub-image will not be selected as the target sub-image, but will be classified into the adjacent sub-image with the largest proportion. In addition, when the defect area spans multiple sub-images and the area is close to being evenly divided, the system may adjust the preset division rule and re-divide the boundary area to more accurately classify the defect. For example, in a certain detection, if the defect area occupies two sub-images, covering 45% and 55% of the area respectively, the division boundary can be adjusted so that the defect is completely attributed to the sub-image with a larger proportion, so that the detection model can mark the abnormal area more efficiently.

[0050] S407, when performing secondary inspection of the lithium battery pole piece to be inspected, detect whether there is an abnormal mark. If there is, directly trigger the switching of the detection mode to the focus mode to perform secondary inspection on the target sub-image. In this embodiment, secondary inspection means that after the abnormal mark is found during the inspection process, the system automatically adjusts the detection strategy and performs more detailed inspection on the target sub-image. The presence of abnormal marks indicates that the area has been identified as a potential defect area in the previous inspection, so the system will skip the full image mode and switch directly to the focus mode to improve the detection accuracy and efficiency. In the focus mode, the system will use a higher resolution image acquisition method or apply a more complex feature extraction algorithm to ensure that the defects are accurately identified. For example, in the full image mode, a calibrated sub-image is marked due to the detection of an abnormality. When the secondary inspection is started, the system will re-acquire the high-definition inspection image of the sub-image and use the deep learning model to classify the defect type to determine whether repair or further analysis is required. In this way, the computational burden of irrelevant areas can be effectively reduced and the detection accuracy of key areas can be improved.

[0051] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0052] In one embodiment, a smart detection device for lithium battery pole piece defects is provided, and the smart detection device for lithium battery pole piece defects corresponds one-to-one to the smart detection method for lithium battery pole piece defects in the above embodiment. Figure 7As shown, the intelligent detection device for lithium battery pole piece defects includes an acquisition module, a generation module, a determination module, a first detection module and a second detection module. The detailed description of each functional module is as follows: An acquisition module is used to acquire a calibration detection image and divide the calibration monitoring image into a plurality of calibration sub-images according to a preset division rule; A generation module, used to call defect history data and generate a defect distribution heat map for each calibration sub-image according to the defect history data; A determination module, configured to determine a detection mode to be switched based on a comparison result of thermal thresholds of each calibrated sub-image in the defect distribution thermal map; A first detection module, configured to obtain a first current detection image of the lithium battery electrode to be detected if the detection mode is a full-image mode, and perform defect detection on the first current detection image based on image processing technology; The second detection module is used to obtain a second current detection image of the lithium battery pole piece to be detected if the detection mode is a focusing mode, divide the second current detection image into detection sub-images corresponding to each of the calibration sub-images based on the preset division rule, and perform defect detection on each of the detection sub-images according to the detection rule corresponding to the thermal threshold comparison result.

[0053] Optionally, the acquisition module includes: A first acquisition unit, used for acquiring a preliminary detection image based on an image acquisition device; A first generating unit is used to preprocess the preliminary detection image, generate a corresponding preprocessed calibration image, calculate the clarity index of the preprocessed calibration image, and determine whether the clarity index meets a preset index. If not, reacquire a new preliminary detection image. If yes, determine that the preliminary detection image is a calibration detection image; A first determining unit, configured to determine a corresponding division ratio according to a preset division rule, and generate a corresponding ratio code based on the division ratio; a division unit, used to determine the size data of the calibration detection image, divide the calibration monitoring image according to the size data and the division ratio, generate corresponding calibration sub-images, and generate corresponding quantity codes according to the generated number of the calibration sub-images; A first adding unit, used for sequentially encoding each of the calibration sub-images, and adding a corresponding numbering code to each of the encoded calibration sub-images according to the ratio code and the quantity code; Optionally, the generating module includes: A calling unit, used for calling corresponding defect history data according to the ratio code and the quantity code; A second generating unit is used to perform defect classification statistics on the defect history data based on the encoding code, and generate regional statistical data corresponding to each encoding code; A drawing unit, used for selecting a heat map template type, and drawing a corresponding defect distribution heat map based on the visualization rule of the heat map template type, the encoding code and the regional statistical data; Optionally, the determining module includes: A second determining unit, used to determine a preset thermal threshold; A first calculation unit, configured to calculate a regional difference value between the preset thermal threshold value and the regional statistical data; A third determining unit, configured to determine that the detection mode is a full-image mode if all of the regional differences are positive numbers; a fourth determining unit, configured to determine that the detection mode is a focusing mode if any of the area difference values ​​is a negative number; Optionally, the first detection module includes: A second calculation unit is used to determine the regional difference of each of the detection sub-images according to the corresponding relationship between the detection sub-image and the calibration sub-image, and calculate the detection weight of each of the detection sub-images according to the regional difference; A third generating unit is used to generate a corresponding detection order according to the detection weight of each detection sub-image, and perform defect detection on each detection sub-image according to the detection order; Optionally, the second detection module includes: A second acquisition unit, configured to acquire a first current detection image of the lithium battery electrode to be detected if the detection mode is a full image mode; An execution unit, configured to perform a preprocessing operation on the first current detection image, wherein the preprocessing operation includes a grayscale operation, a denoising operation, and a brightness equalization operation; An extraction unit, configured to extract a defect area in the first current detection image after the preprocessing operation based on the image processing technology; a fifth determining unit, configured to determine the pixel coordinates of the defective area based on a preset division rule of the calibration sub-image, and determine the coverage area of ​​the defective area in each calibration sub-image according to a mapping relationship between the pixel coordinates and a coordinate range corresponding to each calibration sub-image, and determine the calibration sub-image to which the defective area belongs as a target sub-image according to each coverage area; A second adding unit, configured to add an abnormal mark to the target sub-image if there is only one target sub-image; A third adding unit is used for determining the area ratio of the coverage area if there are multiple target sub-images, and re-determining a new preset division rule according to the determination result of the area ratio determination, wherein the new preset division rule is used to divide the new target sub-image and then add an abnormal mark to the new target sub-image; The detection unit is used to detect whether the abnormal mark exists when performing secondary detection of the lithium battery electrode to be detected. If it exists, it directly triggers the switching of the detection mode to the focusing mode to perform secondary detection on the target sub-image.

[0054] For the specific definition of an intelligent detection device for lithium battery pole piece defects, please refer to the definition of an intelligent detection method for lithium battery pole piece defects above, which will not be repeated here. Each module in the above-mentioned intelligent detection device for lithium battery pole piece defects can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0055] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 8 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an intelligent detection method for lithium battery pole piece defects is implemented.

[0056] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program: S10, obtaining a calibration detection image, and dividing the calibration monitoring image into a plurality of calibration sub-images according to a preset division rule; S20, calling defect history data, and generating a defect distribution heat map for each calibrated sub-image according to the defect history data; S30, determining the detection mode to be switched based on the comparison results of the thermal thresholds of each calibrated sub-image in the defect distribution thermal map; S40, if the detection mode is the full image mode, obtaining a first current detection image of the lithium battery electrode to be detected, and performing defect detection on the first current detection image based on image processing technology; S50. If the detection mode is the focusing mode, obtain the second current detection image of the lithium battery electrode to be detected, divide the second current detection image into detection sub-images corresponding to each calibration sub-image based on a preset division rule, and perform defect detection on each detection sub-image according to the detection rule corresponding to the thermal threshold comparison result.

[0057] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: S10, obtaining a calibration detection image, and dividing the calibration monitoring image into a plurality of calibration sub-images according to a preset division rule; S20, calling defect history data, and generating a defect distribution heat map for each calibrated sub-image according to the defect history data; S30, determining the detection mode to be switched based on the comparison results of the thermal thresholds of each calibrated sub-image in the defect distribution thermal map; S40, if the detection mode is the full image mode, obtaining a first current detection image of the lithium battery electrode to be detected, and performing defect detection on the first current detection image based on image processing technology; S50. If the detection mode is the focusing mode, obtain the second current detection image of the lithium battery electrode to be detected, divide the second current detection image into detection sub-images corresponding to each calibration sub-image based on a preset division rule, and perform defect detection on each detection sub-image according to the detection rule corresponding to the thermal threshold comparison result.

[0058] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and 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 embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0059] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0060] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. An intelligent detection method for lithium battery pole piece defects, characterized in that: The intelligent detection method for lithium battery pole piece defects comprises: Acquire a calibration detection image, and divide the calibration monitoring image into a plurality of calibration sub-images according to a preset division rule; Calling defect history data, and generating a defect distribution heat map for each calibrated sub-image according to the defect history data; Determining the detection mode to be switched based on the comparison results of the thermal thresholds of each calibrated sub-image in the defect distribution thermal map; If the detection mode is the full-image mode, a first current detection image of the lithium battery electrode to be detected is obtained, and defect detection is performed on the first current detection image based on image processing technology; If the detection mode is a focusing mode, a second current detection image of the lithium battery pole piece to be detected is obtained, and the second current detection image is divided into detection sub-images corresponding to each of the calibration sub-images based on the preset division rule, and defect detection is performed on each of the detection sub-images according to the detection rule corresponding to the thermal threshold comparison result.

2. The intelligent detection method for lithium battery pole piece defects according to claim 1, characterized in that: The step of acquiring a calibration detection image and dividing the calibration monitoring image into a plurality of calibration sub-images according to a preset division rule includes: Based on the image acquisition device, a preliminary detection image is obtained; Preprocessing the preliminary detection image to generate a corresponding preprocessed calibration image, calculating a clarity index of the preprocessed calibration image, and determining whether the clarity index meets a preset index. If not, reacquiring a new preliminary detection image. If yes, determining that the preliminary detection image is a calibration detection image; According to a preset division rule, a corresponding division ratio is determined, and a corresponding ratio code is generated based on the division ratio; Determine the size data of the calibration detection image, divide the calibration monitoring image according to the size data and the division ratio, generate corresponding calibration sub-images, and generate corresponding quantity codes according to the generated number of the calibration sub-images; Each calibration sub-image is sequentially encoded, and a corresponding number code is added to each encoded calibration sub-image according to the ratio code and the quantity code.

3. The intelligent detection method for lithium battery pole piece defects according to claim 2, characterized in that: The step of calling defect history data and generating a defect distribution heat map for each calibrated sub-image according to the defect history data includes: According to the ratio code and the quantity code, calling corresponding defect history data; Based on the coding code, performing defect classification statistics on the defect history data to generate regional statistical data corresponding to each coding code; A heat map template type is selected, and based on the visualization rule of the heat map template type, a corresponding defect distribution heat map is drawn according to the encoding code and the regional statistical data.

4. The intelligent detection method for lithium battery pole piece defects according to claim 3, characterized in that: In the step of determining the detection mode to be switched based on the thermal threshold comparison results of each calibration sub-image in the defect distribution thermogram, the thermal threshold comparison results include the regional difference corresponding to each calibration sub-image, including: Determine thermal preset thresholds; Calculate the regional difference value of the thermal preset threshold minus the regional statistical data; If all the regional differences are positive numbers, the detection mode is determined to be the full-image mode; If any of the area difference values ​​is a negative number, it is determined that the detection mode is the focus mode.

5. The intelligent detection method for lithium battery pole piece defects according to claim 4, characterized in that: The step of performing defect detection on each of the detection sub-images according to the detection rule corresponding to the thermal threshold comparison result includes: Determine the regional difference of each of the detection sub-images according to the corresponding relationship between the detection sub-image and the calibration sub-image, and calculate the detection weight of each of the detection sub-images according to the regional difference; A corresponding detection order is generated according to the detection weight of each detection sub-image, and defect detection is performed on each detection sub-image according to the detection order.

6. The intelligent detection method for lithium battery pole piece defects according to claim 1, characterized in that: If the detection mode is the full image mode, the step of obtaining a first current detection image of the lithium battery electrode to be detected, and performing defect detection on the first current detection image based on image processing technology includes: If the detection mode is the full-image mode, obtaining a first current detection image of the lithium battery electrode to be detected; Performing a preprocessing operation on the first current detection image, wherein the preprocessing operation includes a grayscale operation, a denoising operation, and a brightness equalization operation; Extracting defective areas in the first current detection image after preprocessing based on the image processing technology; Based on the preset division rule of the calibration sub-image, the pixel coordinates of the defect area are determined, and according to the mapping relationship between the pixel coordinates and the coordinate range corresponding to each calibration sub-image, the coverage area of ​​the defect area in each calibration sub-image is determined, and according to each coverage area, the calibration sub-image to which the defect area belongs is determined as the target sub-image; If there is one target sub-image, adding an abnormal mark to the target sub-image; If there are multiple target sub-images, the area ratio of the coverage area is determined, and a new preset division rule is re-determined according to the determination result of the area ratio determination. The new preset division rule is used to divide the new target sub-image, and then an abnormal mark is added to the new target sub-image; When performing secondary inspection of the lithium battery electrode to be inspected, it is detected whether the abnormal mark exists. If it exists, it is directly triggered to switch the inspection mode to the focusing mode to perform secondary inspection on the target sub-image.

7. An intelligent detection device for lithium battery pole piece defects, characterized in that: The intelligent detection device for lithium battery pole piece defects comprises: An acquisition module is used to acquire a calibration detection image and divide the calibration monitoring image into a plurality of calibration sub-images according to a preset division rule; A generation module, used to call defect history data and generate a defect distribution heat map for each calibration sub-image according to the defect history data; A determination module, configured to determine a detection mode to be switched based on a comparison result of thermal thresholds of each calibrated sub-image in the defect distribution thermal map; A first detection module, configured to obtain a first current detection image of the lithium battery electrode to be detected if the detection mode is a full-image mode, and perform defect detection on the first current detection image based on image processing technology; The second detection module is used to obtain a second current detection image of the lithium battery pole piece to be detected if the detection mode is a focusing mode, divide the second current detection image into detection sub-images corresponding to each of the calibration sub-images based on the preset division rule, and perform defect detection on each of the detection sub-images according to the detection rule corresponding to the thermal threshold comparison result.

8. The intelligent detection device for lithium battery pole piece defects according to claim 7, characterized in that: The acquisition module comprises: A first acquisition unit, used for acquiring a preliminary detection image based on an image acquisition device; A first generating unit is used to preprocess the preliminary detection image, generate a corresponding preprocessed calibration image, calculate the clarity index of the preprocessed calibration image, and determine whether the clarity index meets a preset index. If not, reacquire a new preliminary detection image. If yes, determine that the preliminary detection image is a calibration detection image; A first determining unit, configured to determine a corresponding division ratio according to a preset division rule, and generate a corresponding ratio code based on the division ratio; a division unit, used to determine the size data of the calibration detection image, divide the calibration monitoring image according to the size data and the division ratio, generate corresponding calibration sub-images, and generate corresponding quantity codes according to the generated number of the calibration sub-images; A first adding unit, used for sequentially encoding each of the calibration sub-images, and adding a corresponding numbering code to each of the encoded calibration sub-images according to the ratio code and the quantity code; The generation module comprises: A calling unit, used for calling corresponding defect history data according to the ratio code and the quantity code; A second generating unit is used to perform defect classification statistics on the defect history data based on the encoding code, and generate regional statistical data corresponding to each encoding code; A drawing unit, used for selecting a heat map template type, and drawing a corresponding defect distribution heat map based on the visualization rule of the heat map template type, the encoding code and the regional statistical data; The determination module comprises: A second determining unit, used to determine a preset thermal threshold; A first calculation unit, configured to calculate a regional difference value between the preset thermal threshold value and the regional statistical data; A third determining unit, configured to determine that the detection mode is a full-image mode if all of the regional differences are positive numbers; a fourth determining unit, configured to determine that the detection mode is a focusing mode if any of the area difference values ​​is a negative number; The first detection module comprises: A second calculation unit is used to determine the regional difference of each of the detection sub-images according to the corresponding relationship between the detection sub-image and the calibration sub-image, and calculate the detection weight of each of the detection sub-images according to the regional difference; A third generating unit is used to generate a corresponding detection order according to the detection weight of each detection sub-image, and perform defect detection on each detection sub-image according to the detection order; The second detection module comprises: A second acquisition unit, configured to acquire a first current detection image of the lithium battery electrode to be detected if the detection mode is a full image mode; An execution unit, configured to perform a preprocessing operation on the first current detection image, wherein the preprocessing operation includes a grayscale operation, a denoising operation, and a brightness equalization operation; An extraction unit, configured to extract a defect area in the first current detection image after the preprocessing operation based on the image processing technology; a fifth determining unit, configured to determine the pixel coordinates of the defective area based on a preset division rule of the calibration sub-image, and determine the coverage area of ​​the defective area in each calibration sub-image according to a mapping relationship between the pixel coordinates and a coordinate range corresponding to each calibration sub-image, and determine the calibration sub-image to which the defective area belongs as a target sub-image according to each coverage area; A second adding unit, configured to add an abnormal mark to the target sub-image if there is only one target sub-image; A third adding unit is used for determining the area ratio of the coverage area if there are multiple target sub-images, and re-determining a new preset division rule according to the determination result of the area ratio determination, wherein the new preset division rule is used to divide the new target sub-image and then add an abnormal mark to the new target sub-image; The detection unit is used to detect whether the abnormal mark exists when performing secondary detection of the lithium battery electrode to be detected. If it exists, it directly triggers the switching of the detection mode to the focusing mode to perform secondary detection on the target sub-image.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the intelligent detection method for lithium battery pole piece defects as described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of an intelligent detection method for lithium battery pole piece defects as described in any one of claims 1 to 6 are implemented.

Images