Battery pole piece burr detection method based on machine vision

Abstract

The invention discloses a pole piece burr detection method based on machine vision, and belongs to the technical field of battery pole piece burr detection, and the method comprises the steps: carrying out the imaging of a to-be-detected slice through an area-array camera with a telecentric lens, obtaining a to-be-detected picture, and storing the to-be-detected picture in a buffer queue; the to-be-detected picture is preprocessed; performing monitoring area positioning on the preprocessed to-be-detected picture; performing burr detection on each positioning area to obtain a burr detection result; according to the invention, automatic focus tracking of the camera can be realized; the requirement of double-sided detection can be met in an extremely small space; high-speed optical imaging of transverse and longitudinal burrs of the laser edge / slitting edge of the pole piece is realized; the detection equipment can automatically avoid the material belt to adapt to the production process of belt threading and the like; full inspection at the speed of 120 meters per second is supported; the appearance size can be adjusted according to the installation position, the dustproof function is achieved, and shading measures are taken.

Description

Technical Field

[0001] The present invention belongs to the technical field of burr detection for battery electrodes, and particularly relates to a method for detecting burrs on battery electrodes based on machine vision. Background Art

[0002] The global attention to clean energy and sustainable development is increasing day by day, and fields such as new energy vehicles, energy storage devices, and 3C devices are developing rapidly. As the core energy component in these fields, the market demand for lithium batteries has increased sharply. In 2024, the sales volume of new energy vehicles has continued to rise, and the safety of batteries has always been a key item for market promotion. Quality defects of batteries are one of the important factors leading to safety hazards.

[0003] In the manufacturing process of battery cells, the detection and control of electrode burrs are crucial parts for manufacturing high-quality batteries. Electrode burrs refer to the tiny and sharp metal protrusions formed on the edges of electrodes during the processes of slitting and die-cutting after the electrodes have gone through pulping, coating, and rolling in the manufacturing process of battery cells. These protrusions have various shapes and uneven sizes, and they are caused by factors such as tool wear during cutting, material tearing, or unreasonable setting of processing parameters. Their sizes range from several micrometers to hundreds of micrometers. Burrs or beads on the electrodes not only reduce the energy density of the battery, but also, due to the relatively thin battery separator (from ten to dozens of micrometers), when the burrs pierce the separator, it will cause the positive and negative electrodes to be connected, leading to self-discharge or internal short-circuit faults. In severe cases, there will be safety hazards such as battery swelling and explosion, causing huge losses.

[0004] To improve the performance and safety of batteries, Section 5.3.6 "Burr Control" of the IEEE 1725-2021 standard recommends that the height of electrode burrs should be lower than 50% of the lower limit of the separator thickness tolerance. Therefore, in the manufacturing of battery cells, there is an urgent need for more efficient and advanced detection equipment to timely detect electrode burrs, optimize the processing technology in a timely manner, and effectively control the generation of burrs by selecting the tool material and geometry, and precisely regulating the slitting and die-cutting process parameters (such as cutting speed, pressure, etc.).

[0005] In the manufacturing of battery cells, the control of electrode burrs is extremely crucial. Most manufacturers currently use manual detection with microscopes, but long-term "hand-shaking and eye-watching" is prone to missing burrs, beads, and other defects on the electrodes due to visual fatigue.

[0006] The size of burrs is usually in the micrometer range, and their tiny scale makes the detection difficulty significantly increase. Moreover, factors such as the complex production process and large production demand of battery electrodes further exacerbate the complexity of detection. In the actual work process, it is also necessary to consider the compatible detection of various complex defects. Therefore, there are many challenges in the manufacturing and detection of battery electrodes:

[0007] It is difficult to balance efficiency and precision. Traditional methods require equipment such as microscopes and rely on manual operation and visual inspection, which have high requirements for personnel skills, low detection efficiency, high misjudgment rates, and cannot achieve full inspection.

[0008] Full automation detection cannot be achieved. In the workshop, full automation detection of burrs on the electrode sheet is required, replacing manual focus adjustment, including calibration, detection, and data analysis.

[0009] There is a lack of a customized data management system. For traditional manual microscope burr detection, on-site personnel need to manually collect and statistically analyze the results again, with low data analysis and management efficiency.

[0010] It is difficult to balance size detection and burr detection. For traditional battery electrode sheet detection, an image measuring instrument is needed to detect the size, and then the machine is replaced to use a microscope to observe burrs, resulting in low efficiency and inaccurate detection results.

[0011] Large-size measurement cannot be satisfied. When using manual measurement with a microscope, the platform is small and only supports small-size measurement, unable to meet the detection requirements for the external dimensions of larger electrode sheets.

[0012] There are various defect detection requirements. In addition to traditional burr detection, automation detection of exposed foil, decarburization, etc. can also be carried out.

[0013] In summary, due to the technical defects in the existing technology, a method for detecting burrs on electrode sheets based on machine vision is needed to meet the stability requirements of detection. Summary of the Invention

[0014] The technical problem to be solved by the present invention is to provide a method for detecting burrs on electrode sheets based on machine vision in view of the deficiencies in the background technology. It is equipped with a special algorithm for electrode sheet burrs and a customized optical system, which can synchronously detect burrs on the front and side of battery electrode sheets, and can also detect defects such as decarburization and exposed foil. It integrates the basic external dimension function of an image measuring instrument and has characteristics such as high detection efficiency, diverse functions, and automated detection.

[0015] The present invention adopts the following technical solutions to solve the above technical problems:

[0016] A method for detecting burrs on battery electrode sheets based on machine vision specifically includes the following steps:

[0017] Step 1: Image the slice to be detected through an area array camera with a telecentric lens, obtain the image to be detected, and store it in the buffer queue.

[0018] Step 2: Preprocess the image to be detected to obtain the burr area to be detected.

[0019] Step 3: Perform image preprocessing on the burr area to reduce noise interference, and screen out possible burrs based on the position of the pole piece film area according to different thresholds. Determine the number of burrs and their coordinates based on the area and length regional features according to the actual production size requirements, and perform monitoring area positioning.

[0020] Step 4: Perform burr detection on each positioning area to obtain the burr detection result.

[0021] Compared with the prior art, the present invention adopts the above technical solutions and has the following technical effects:

[0022] A method for detecting burrs in pole piece slitting based on vision detection in the present invention; imaging the slice to be detected through an area array camera with a telecentric lens, obtaining the picture to be detected and storing it in a buffer queue; preprocessing the picture to be detected; performing monitoring area positioning on the preprocessed picture to be detected; performing burr detection on each positioning area to obtain the burr detection result; the present invention can realize automatic focusing of the camera; can meet the requirements of double-sided detection in a very small space; realize high-speed optical imaging of the transverse and longitudinal burrs on the laser edge / slitting edge of the pole piece; the detection device can automatically avoid the material belt and adapt to the production process such as threading; support full inspection at a speed of 120 meters per second; the external dimensions can be adjusted according to the installation position, with dust-proof function and light-shielding measures. Description of the Drawings

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0024] Figure 1 is a flowchart of the method for detecting burrs in pole pieces based on machine vision provided by the present invention;

[0025] Figure 2 is a schematic diagram of the camera imaging effect of the present invention;

[0026] Figure 3 is a schematic diagram of the effect of finding burr defects in the foil area in the pole piece of the present invention;

[0027] Figure 4 is a schematic diagram of the final display panel of the burr detection effect of the present invention. Detailed Embodiments

[0028] The following will further elaborate on the technical solutions of the present invention in conjunction with the drawings:

[0029] To enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention. It should be noted that the terms "first" and the like in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0030] A burr detection method for battery electrodes based on machine vision, as Figure 1 shown, specifically includes the following steps;

[0031] Step 1, image the slice to be detected through an area array camera with a telecentric lens, obtain the image to be detected and store it in the buffer queue;

[0032] Step 2, preprocess the image to be detected to obtain the burr area to be detected;

[0033] Step 3, preprocess the image of the burr area to reduce noise interference, and screen out possible burrs according to different thresholds based on the position of the electrode film area. Determine the number of burrs and their coordinates according to the area and length regional features of the actual production size requirements, and perform monitoring area positioning;

[0034] Step 4, perform burr detection on each positioning area to obtain the burr detection result.

[0035] The present invention can achieve automatic focusing of the camera; can meet the requirements of double-sided detection in a very small space; achieve high-speed optical imaging of the transverse and longitudinal burrs on the laser edge / slit edge of the electrode; the detection device can automatically avoid the material tape and adapt to the production process such as threading the tape; support full inspection at a speed of 120 meters per second; the external dimensions can be adjusted according to the installation position, with dust-proof function and light-shielding measures

[0036] The specific steps of Step 1 are as follows:

[0037] Step 1.1, adjust the autofocus module of the detection device according to actual needs;

[0038] Step 1.2, adjust the appropriate light source intensity according to the material characteristics;

[0039] Step 1.3, take a photo of the slice to be detected through an area array camera with a telecentric lens for imaging;

[0040] Step 1.4, the device sensor detects the position of the pole piece in real time;

[0041] Step 1.5, the motor drives the coordinate axis to move, adjusts the position of the camera, obtains the picture to be detected and stores it in the buffer queue.

[0042] The specific steps of step 2 are as follows;

[0043] Step 2.1, name the initial image of the slice to be detected collected as A;

[0044] Step 2.2, perform threshold processing on the initial image A. By setting one or more thresholds, the pixels of the image are divided into different categories for binary processing to separate the target area from the background; adopt the fixed threshold method, and select the concave point at the boundary between the background and the foreground in the histogram as the threshold through histogram analysis;

[0045] Step 2.3, after selecting the appropriate threshold according to the image boundary, separate the pole piece area B from the original image A;

[0046] Step 2.4, perform preprocessing operations of smoothing and denoising on the obtained pole piece area B by using the denoising method of image morphology operations;

[0047] The opening operation is denoted as: Where: I represents the input image, and B represents the structuring element;

[0048] Among them, erosion Under the action of the structuring element B, take the minimum gray value in the neighborhood with (x, y) as the center position, where (s, t) ∈ B, and s and t represent the offset coordinates in the structuring element B; used to remove small areas, shrink the target area, and eliminate isolated noise points;

[0049] Dilation Under the action of the structuring element B, take the maximum gray value in the neighborhood with (x, y) as the center position, where (s, t) ∈ B, and s and t represent the offset coordinates in the structuring element B; used to restore the main structure of the target area after erosion.

[0050] Step 2.5, after performing the opening operation on the pole piece area B, obtain the pole piece area C with small noise removed and the target boundary smoothed;

[0051] In step 2.5, in a digital image, the connectivity of a pixel is based on the relationship of its neighboring pixels; for a pixel point

[0052] p(x, y), its neighborhood is defined by different connectivity rules: 4-neighborhood and 8-neighborhood. In the 4-neighborhood mode, the four adjacent pixels above, below, left, and right of a pixel are the 4-neighborhood;

[0053] Neighborhood definition formula: N4(p) = {(x, y - 1), (x - 1, y), (x + 1, y), (x, y + 1)}.

[0054] By using the seed filling algorithm, a pixel seed is selected and all adjacent pixels are recursively expanded until all connected pixels are labeled, obtaining different regions in the connected component labeled image C. In 4-connectivity, the connected region C is a set of pixel points such that any two pixel points are connected within C, that is

[0055]

[0056] where: p and q are pixel points in the image; f(p) is the gray value or label of pixel point p; P represents a path from p to q, and all pixel points on the path satisfy: connectivity and consistent gray value.

[0057] Step 2.6, perform connectivity analysis on the electrode tab region C; obtain the actual region film region D of the electrode tab film region after removing various noises and other influences;

[0058] Step 2.7, cut out the film region obtained after circumscribing a rectangle from the original image A to obtain the film region image E; subsequently, perform threshold processing on the film region image E, aiming to obtain the foil region F;

[0059] In step 2.6, after obtaining the set of pixel points, the required set is screened out through morphological features, and feature screening is performed through area, roundness, aspect ratio, and circumscribed rectangle size, using area as the screening feature;

[0060] The area is defined as the total number of all pixels inside the region, A = ∑1, (x, y) ∈ R; where: A is the number of pixels in the region, and R represents the target region;

[0061] Connect the screened set of pixel points into a whole; given a set of regions R = {R1, R2,..., Rn}, its union is defined as: where: R i represents each individual region, and R union is the merged region. After merging, the outer boundaries of all regions become a whole; obtain the actual region film region D of the electrode tab film region after removing various noise influences;

[0062] Perform an external rectangle operation on the membrane region D using the minimum external rotated rectangle operation. For a region, its minimum external rotated rectangle refers to a rectangle with the smallest area that can completely enclose the region; the rectangle is tilted and rotated, not horizontally aligned.

[0063] Let (x, y) be the coordinates of the center point of the rectangle; \(\phi\) be the rotation angle of the rectangle; L1 be the length of the rectangle, i.e., the major axis direction; L2 be the width of the rectangle, i.e., the minor axis direction.

[0064] The calculated rectangle satisfies: Among them: the area of the rectangle L1×L2 takes the minimum value, and the rectangle completely encloses the region R.

[0065] Step 2.8: Connect the small divided regions after screening out abnormal points into a whole region, and then expand the whole region according to the process standard foil thickness to obtain the expanded foil region G.

[0066] In Step 2.8, perform a connectivity analysis on the foil region F, and divide the region after the connectivity analysis of the foil region F into multiple small rectangles in the row and column directions.

[0067] Let the size of the input rectangular region R be W×H, and the division parameters: nr: the number of divisions in the row direction; nc: the number of divisions in the column direction.

[0068] The size of each sub-rectangle is: After division, nr×nc small rectangles will be obtained.

[0069] Perform morphological feature screening on the divided small regions. At this time, the lower limit of the area will be set relatively high, aiming to retain the effective target region, eliminate irrelevant small regions or noise points, and ensure that subsequent processing is only carried out on the effective region.

[0070] Then sort the regions after morphological feature screening for easy subsequent processing in order, and sort them according to the region order; the mathematical definition of sorting is: Let the input region set R = {R1, R2,..., Rn} contain n connected regions; let each region Ri have the following attributes:

[0071] Starting point of the region: (x i , y i ); Center of gravity of the region: Area of the region: A i ; After sorting, a new order is obtained: R σ(1) , R σ(2) , …, R σ(n) ;

[0072] After sorting, calculate the area and the center point of each region, and use the center point of the region. The definition of the center point of the region is as follows: Let the region R consist of a set of pixel points (xi, yi); then:

[0073] Let the calculated area A of the region be the total number of pixels within the region,

[0074] Then the center of gravity (cx, cy) of the region is calculated by the following formula:

[0075] where: Cx is the horizontal center point of the region, i.e., the center of gravity in the X direction; Cy is the vertical center point of the region, i.e., the center of gravity in the Y direction;

[0076] The calculation method is similar to the centroid calculation, that is, the sum of the coordinates of all points is taken and the average value is obtained;

[0077] After obtaining the center points of the above-mentioned segmented regions, discrete points among them are selected and removed as outliers through the method of normal distribution to enhance the stability of the algorithm;

[0078] The mathematical definition of normal distribution: where: μ is the mean value, which determines the central position of the normal distribution; σ is the standard deviation, which determines the width of the normal distribution; σ2 is the variance, which represents the degree of dispersion of the data, that is, the magnitude of the fluctuation of the data around the mean value;

[0079] Connect the small segmented regions after removing outliers into a whole region, and then expand the whole region according to the process standard to obtain the expanded foil region G.

[0080] Step 2.9, perform a regional difference operation on the actual foil region F and the expanded foil region G; remove the actual foil region F from the expanded foil region G to obtain the remaining region, and the remaining region is the region where burrs exist, that is, the burr detection region H.

[0081] In step 2.9, let regions A and B be two set regions in the image, and their pixel point coordinates are respectively expressed as:

[0082] A = {(x, y) ∣ (x, y) belongs to A}; B = {(x, y) ∣ (x, y) belongs to B}

[0083] The regional difference operation Difference(A, B) is calculated as follows: A - B = {(x, y) ∣ (x, y) ∈ A, (x, y) ∉ B};

[0084] That is, all pixel points within B are removed from A to obtain the remaining region. The actual foil region F is removed from the expanded foil region G to obtain the remaining region, and the remaining region is the region where burrs may exist, that is, the burr detection region H.

[0085] In H, we perform threshold processing (using the same method), followed by connectivity analysis (using the same method). Then, the regions obtained after the connectivity analysis are screened according to their areas for morphological feature selection, and the burr defect regions considered in actual process production are selected. Subsequently, these burr defect regions are sorted according to their areas to facilitate subsequent collation, statistics, and output.

[0086] Those of ordinary skill in the art can understand that the above are only preferred examples of the invention and are not used to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, for those skilled in the art, they can still modify the technical solutions described in the foregoing examples, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, etc. made within the spirit and principle of the invention shall be included within the protection scope of the invention. All technical features in this embodiment can be freely combined according to actual needs.

[0087] Finally, it should be noted that the above are only preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A burr detection method for battery electrodes based on machine vision, characterized in that: Specifically, it includes the following steps; Step 1, image the slice to be detected by an area array camera with a telecentric lens, obtain the image to be detected and store it in the buffer queue; Step 2, preprocess the image to be detected to obtain the burr area to be detected; Step 3, preprocess the burr area image to reduce noise interference, and based on the position of the pole piece film area, screen out possible burrs according to different thresholds. According to the actual production size requirements, judge the number and coordinates of burrs based on the area and length regional features of the burrs, and perform monitoring area positioning; Step 4, perform burr detection on each positioning area to obtain the burr detection result.

2. The method for detecting burrs on battery electrode sheets based on machine vision according to claim 1, wherein: The specific steps of Step 1 are as follows: Step 1.1, adjust the autofocus module of the detection device according to actual needs; Step 1.2, adjust the appropriate light source intensity according to the material characteristics; Step 1.3, image the slice to be detected by an area array camera with a telecentric lens; Step 1.4, the device sensor real-time detects the position of the pole piece; Step 1.5, the motor drives the coordinate axis to move, adjusts the camera position, obtains the image to be detected and stores it in the buffer queue.

3. The method for detecting burrs on battery electrode sheets based on machine vision according to claim 1, wherein: The specific steps of Step 2 are as follows; Step 2.1, name the initial image of the slice to be detected collected as A; Step 2.2, perform threshold processing on the initial image A. By setting one or more thresholds, the pixels of the image are divided into different categories for binary processing, so that the target area is separated from the background; adopt the fixed threshold method, and select the concave point at the boundary between the background and the foreground in the histogram as the threshold through histogram analysis; Step 2.3, after selecting the appropriate threshold according to the image boundary, separate the pole piece area B from the original image A; Step 2.4, perform a preprocessing operation of smoothing and denoising on the obtained pole piece area B by using the denoising method of image morphology operation; Step 2.5, perform an opening operation on the pole piece area B to obtain the pole piece area C after removing small noises and smoothing the target boundary; Step 2.6, perform connectivity analysis on the pole piece area C; obtain the actual area film area D of the pole piece film area after removing the influence of various noises, etc.; Step 2.7, cut out the film area obtained after circumscribing the rectangle from the original image A to obtain the film area image E; then perform threshold processing on the film area image E to obtain the foil area F; Step 2.8, connect the small segmented areas after screening out abnormal points into a whole area, and then expand the whole area according to the process standard foil thickness to obtain the expanded foil area G; Step 2.9, perform regional difference operation on the actual foil area F and the expanded foil area G; remove the actual foil area F from the expanded foil area G to obtain the remaining area, and the remaining area is the area where burrs exist, that is, the burr detection area H.

4. The method for detecting burrs on battery electrodes based on machine vision according to claim 3, characterized in that: In Step 2.4, the principle of the opening operation is: Let the input image I(x,y) be a grayscale image; the structuring element SE is B; The opening operation is denoted as: where: I represents the input image, and B represents the structuring element; Among them, corrosion Under the action of the structural element B, the minimum gray value in the neighborhood is taken with (x, y) as the central position, where (s, t) ∈ B, and s and t represent the offset coordinates in the structural element B; it is used to remove small areas, shrink the target area, and eliminate isolated noise points; Dilation I⊕B: Under the action of the structuring element B, the maximum gray value in the neighborhood is taken with (x, y) as the central position, where (s, t) ∈ B, and s and t represent the offset coordinates in the structuring element B; it is used to restore the main structure of the target area after erosion.

5. The method for detecting burrs on battery electrodes based on machine vision according to claim 3, characterized in that: In step 2.5, in a digital image, the connectivity of a pixel is based on the relationship of its neighboring pixels; for a pixel point p(x,y), its neighborhood is defined by different connectivity rules: 4-neighborhood and 8-neighborhood. Using the 4-neighborhood method, that is, the four adjacent pixels above, below, left, and right of a pixel are the 4-neighborhood; Select a pixel seed through the seed filling algorithm and recursively expand all adjacent pixels until all connected pixels are labeled, obtaining different regions in the connected component labeled image C. In 4-connectivity, the connected region C is a set of pixel points such that any two pixel points are connected within C. Let the image be a two-dimensional pixel grid I, and the connected region C is defined as: where: p and q are pixel points in the image; f(p) is the gray value or label of pixel point p; P represents a path from p to q, and all pixel points on the path satisfy: the connectivity and gray value are the same.

6. The method for detecting burrs on battery electrode plates based on machine vision according to claim 3, characterized in that: In step 2.6, after obtaining the set of pixel points, the required set is screened out through morphological features. Feature screening is performed through area, roundness, aspect ratio, and the size of the circumscribed rectangle, and the area is used as the screening feature; The area is defined as the total number of all pixels inside the region, A = ∑1, (x,y) ∈ R; where: A is the number of pixels in the region, and R represents the target region; Connect the set of filtered pixel points into a whole; given a set of regions \(R = \{R_1, R_2, \ldots, R_n\}\), its union is defined as: where: \(R\) i represents each individual region, \(R\) union is the merged region. After merging, the outer boundaries of all regions become a whole; obtain the actual region \(D\) of the pole piece film region after removing the influence of various noises. Perform the circumscribed rectangle operation on the membrane region D, using the minimum circumscribed rotated rectangle operation. For a region, its minimum circumscribed rotated rectangle refers to a rectangle with the smallest area that can completely enclose the region; this rectangle is tilted and rotated, not horizontally aligned; Let (x,y) be the coordinates of the center point of the rectangle; \(\phi\) is the rotation angle of the rectangle; L1 is the length of the rectangle, i.e., the major axis direction; L2 is the width of the rectangle, i.e., the minor axis direction; The calculated rectangle satisfies: where: the area L1×L2 of the rectangle takes the minimum value, and the rectangle completely encloses the region R.

7. A method for detecting burrs on battery electrodes based on machine vision according to claim 6, characterized in that: In step 2.8, perform a connectivity analysis on the foil region F, and divide the region after the connectivity analysis of the foil region F into multiple small rectangles in the row and column directions; Let the size of the input rectangular region R be W×H, and the division parameters: nr: the number of divisions in the row direction; nc: the number of divisions in the column direction; The size of each sub-rectangle is: After division, nr×nc small rectangles will be obtained; Perform morphological feature screening on the divided small regions. At this time, the lower limit of the area will be set relatively high, aiming to retain the effective target region, eliminate irrelevant small regions or noise points, and ensure that subsequent processing is only carried out on the effective region; Then sort the regions after morphological feature screening for subsequent processing in order. Sort them in the order of the regions; the mathematical definition of sorting is: let the input region set R = {R1, R2,..., Rn} contain n connected regions; let each region Ri have the attribute: Region starting point: (x i , y i ); Region centroid: Region area: A i ; After sorting, the new order is obtained: R σ(1) , R σ(2 ), …, R σ(n) ; After sorting, calculate the area and the center point of each region, and use the center point of the region; the definition of the center point of the region is: let the region R be composed of a set of pixel points (xi, yi); then: Let the area A of the region be calculated as the total number of pixels within the region, Then, the center of gravity (cx, cy) of the region is calculated by the following formula: where: Cx is the horizontal center point of the region, i.e., the centroid in the X direction; Cy is the vertical center point of the region, i.e., the centroid in the Y direction; The calculation method is similar to the centroid calculation, that is, the sum of the coordinates of all points is taken and averaged; After obtaining the center points of the above-mentioned divided regions, screen out the discrete points among them as outliers through the method of normal distribution to enhance the stability of the algorithm; Mathematical definition of normal distribution: Where: μ is the mean, which determines the central position of the normal distribution; σ is the standard deviation, which determines the width of the normal distribution; σ2 is the variance, representing the degree of data dispersion, that is, the magnitude of data fluctuations around the mean; Connect the small segmented regions after removing abnormal points into a whole region, and then expand the whole region according to the process standard to obtain the expanded foil region G.

8. The method for detecting burrs on battery electrodes based on machine vision according to claim 6, characterized in that: In step 2.9, let regions A and B be two set regions in the image, and their pixel coordinates are respectively represented as: A = {(x, y) | (x, y) belongs to A}; B = {(x, y) | (x, y) belongs to B} The region difference operation Difference(A, B) is calculated as follows: A - B = {(x, y) | (x, y) ∈ A, (x, y) ∈ / B}; That is, remove all the pixel points in B from A to obtain the remaining region. Remove the actual foil region F from the expanded foil region G to obtain the remaining region. The remaining region is the region where burrs may exist, that is, the burr detection region H.

Images