A lithium battery pole piece edge detection method, device and system

Abstract

The present invention discloses a method and device for detecting the edge of a lithium battery pole piece. The method comprises: receiving an original lithium battery pole piece image; performing an opening operation on the original lithium battery pole piece image in a set direction to obtain an opening operation image; calculating the second-order gradient of the setting direction of the opening operation image to obtain a second-order gradient image; performing pixel mean and standard deviation calculation and condition judgment based on the second-order gradient image and the original lithium battery pole piece image to obtain a lithium battery pole piece edge detection result; the method can accurately identify the boundaries of different parts in the lithium battery pole piece image.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a method, device and system for detecting an edge of a lithium battery pole piece. Background Art

[0002] Lithium battery pole pieces are formed by coating a metal substrate with electrolyte and active materials, followed by drying, cutting, coating, compacting and other processes. The size of the pole piece needs to be strictly controlled to ensure the overall structure and performance of the battery. Especially during the cutting and winding process, if there are errors in the width and position of each part of the pole piece, it may cause difficulties in battery assembly and affect the battery's charging and discharging efficiency, life and safety. Therefore, before the lithium battery pole piece is rolled into a lithium battery, the width and position of each part of the pole piece need to be accurately measured to determine whether the pole piece is qualified.

[0003] The characteristics of lithium battery pole pieces are:

[0004] A complete lithium battery usually consists of three parts (ear tab, dressing, ear tab) or four parts (ceramic, dressing, ceramic, ear tab), and each lithium battery is made of rolled lithium battery electrodes.

[0005] In order to improve production efficiency, for a lithium battery composed of three parts, the entire lithium battery electrode may be coated into n strips (n=1,2,3...). When in use, the entire lithium battery electrode needs to be cut into n parts according to the set width, and each strip is rolled into a lithium battery. A lithium battery electrode with 1 strip consists of a tab, a dressing, and a tab; a lithium battery electrode with 2 strips consists of a tab, a dressing, a tab, a dressing, and a tab; a lithium battery electrode with 3 strips consists of a tab, a dressing, a tab, a dressing, and a tab, and so on.

[0006] For a lithium battery composed of four parts, the entire lithium battery electrode is often coated into one strip, that is, the entire lithium battery electrode is composed of ceramic, dressing, ceramic, and ear.

[0007] If there are 3 or more widths, the width of the lithium battery electrode is usually so large that the field of view of a camera cannot fully cover it. In this case, two cameras will be used to shoot at the same time and then stitched together into a complete lithium battery electrode image, such as Figure 1 As shown, Figure 1 In the figure, 1 is a camera, 2 is a roller, and 3 is a lithium battery electrode.

[0008] Since the edges of different parts of the lithium battery pole piece on the lithium battery pole piece image are a set of parallel vertical straight lines, the lithium battery pole piece image may show uneven brightness due to the influence of lighting or camera characteristics. For the lithium battery pole piece image spliced ​​by two cameras, if there is a slight difference between the exposure times of the two cameras, it is very likely that a gray value fault will appear at the splicing point, as shown in Figure 2 (b). The normal splicing is shown in Figure 2 (a).

[0009] The manufacturing process of lithium batteries includes batching, coating, rolling, slitting, winding / stacking, assembly, and injection. Among them, the image of the lithium battery pole piece is collected during rolling, so the left and right sides of the lithium battery pole piece image usually include the roller surface. Due to the high precision requirements for the production of lithium battery pole pieces, the grayscale values ​​of each part of the lithium battery pole piece that is retained and enters the pole piece edge detection program are usually relatively uniform, with only a few defects caused by camera shake, as shown in Figure 2 (c). On the contrary, the grayscale value of the roller surface may be larger or smaller, and there may be defects on the roller surface, especially stripe defects, which will interfere with the edge detection of the lithium battery pole piece, as shown in Figure 2 (d).

[0010] Patent document CN116416268 A discloses a method and device for detecting the edge position of a lithium battery pole piece based on a recursive binary division method, and performs detection through four steps: obtaining an intermittent coating image of a lithium battery pole piece; graying the intermittent coating image to obtain a grayed image; obtaining a target image based on the recursive binary division method and combining the analysis of the edge information in the grayed image; performing image processing on the target image to determine the edge position, and completing the detection of the edge position of the lithium battery pole piece. However, this method relies on the prior conditions that the approximate width of the pole ear and the dressing is known, the approximate width of each pole ear is the same, and the width of each dressing is roughly the same. When these three conditions are unknown, this method cannot be used to detect the edge of the lithium battery pole piece; in addition, the method disclosed in this patent is not applicable to the case where the lithium battery pole piece is composed of four parts: ceramic, dressing, ceramic, and pole ear; finally, this method uses a relatively fixed grayscale value to determine whether the upper and lower halves after binary division contain edges, and cannot adapt well to the situation where the grayscale value on the lithium battery pole piece is uneven. Summary of the invention

[0011] The present invention provides a lithium battery pole piece edge detection method, device and system, which can accurately identify the boundaries of different parts in a lithium battery pole piece image.

[0012] A lithium battery pole piece edge detection method, comprising:

[0013] Receive the original lithium battery electrode image;

[0014] Performing an opening operation on the original lithium battery pole piece image according to a set direction to obtain an opening operation image;

[0015] Calculating the second-order gradient of the set direction of the open operation image to obtain a second-order gradient image;

[0016] Based on the second-order gradient image and the original lithium battery pole piece image, the mean and standard deviation of pixels are calculated and conditional judgment is performed to obtain the lithium battery pole piece edge detection result.

[0017] Furthermore, the direction perpendicular to the extension direction of the dressing in the original lithium battery pole piece image is used as the set direction.

[0018] Further, calculating the second-order gradient of the set direction of the opening operation image to obtain the second-order gradient image includes:

[0019] The absolute value of the difference between the pixel value of the i-th row and j-th column in the opening operation image and the pixel value of the i-th row and j+2-th column in the second-order gradient image is calculated as the pixel value of the i-th row and j-th column in the second-order gradient image.

[0020] Furthermore, the mean and standard deviation of pixels are calculated and conditional judgment is performed based on the second-order gradient image and the original lithium battery pole piece image to obtain the lithium battery pole piece edge detection result, including:

[0021] Initialize a first edge set, calculate the mean and standard deviation of pixels based on the second-order gradient image, and add 1 to the column pixel index that meets the first preset condition according to the calculation result and store it in the first edge set;

[0022] Initialize a second edge set, perform mean calculation on pixels in the second-order gradient image based on the column pixel index in the first edge set minus 1, and store the column pixel index that meets the second preset condition into the second edge set according to the calculation result;

[0023] Initialize the third edge set, and judge the left and right neighborhood mean standard deviation based on the column pixel index in the second edge set corresponding to the pixels in the original lithium battery pole piece image, and store the column pixel index that meets the third preset condition into the third edge set according to the judgment result. The obtained third edge set is the column index of all real edges in the original lithium battery pole piece image.

[0024] Further, calculating the mean and standard deviation based on the second-order gradient image, adding 1 to the pixel index that meets the first preset condition according to the calculation result and storing it in the first edge set, including:

[0025] Traversing and calculating a first mean and a standard deviation of each column of pixels in the second-order gradient image;

[0026] The column pixel indexes whose first mean is greater than the preset mean and whose standard deviation is less than the preset standard deviation are incremented by 1 and then stored in the first edge set.

[0027] Further, based on the column pixel index in the first edge set minus 1, a mean value calculation is performed on the pixels in the second-order gradient image, and according to the calculation result, the column pixel index that meets the second preset condition is stored in the second edge set, including:

[0028] Calculate a second mean value of column pixels in the second-order gradient image corresponding to consecutive column pixel indices in the first edge set minus 1;

[0029] The column pixel index corresponding to the maximum value of the second mean obtained in the same boundary is stored in the second edge set.

[0030] Further, based on the column pixel index in the second edge set corresponding to the pixel in the original lithium battery pole piece image, the left and right neighborhood mean standard deviation is judged, and the column pixel index that meets the third preset condition is stored in the third edge set according to the judgment result, including:

[0031] Set the left and right mean difference threshold, left and right deviation threshold, and left and right standard deviation threshold;

[0032] Traversing the second edge set, calculating the average of the left neighboring column pixel means, the average of the left neighboring column pixel standard deviations, the average of the right neighboring column pixel means, and the average of the right neighboring column pixel standard deviations corresponding to the column pixel index in the second edge set in the original lithium battery pole piece image;

[0033] If the column pixel index in the second edge set corresponds to the column pixel in the original lithium battery electrode image and satisfies one of the following conditions, the corresponding column pixel index is stored in the third edge set:

[0034] The absolute value of the difference between the average value of the left neighbor column pixel means and the average value of the right neighbor column pixel means is greater than or equal to the left and right mean difference threshold, and the minimum value of the average value of the left neighbor column pixel standard deviation and the average value of the right neighbor column pixel standard deviation is less than or equal to the left and right standard deviation threshold; or,

[0035] The absolute value of the difference between the corresponding column pixel index and half of the number of column pixel indices in the second edge set is greater than the left and right deviation thresholds, and the minimum value of the average value of the left neighboring column pixel standard deviation and the average value of the right neighboring column pixel standard deviation is less than or equal to the left and right standard deviation thresholds.

[0036] Further, if the absolute value of the difference between the average value of the left neighboring column pixel means and the average value of the right neighboring column pixel means of the corresponding column pixel index is less than the left and right mean difference threshold, and the absolute value of the difference between the corresponding column pixel index and half of the number of column pixel indices in the second edge set is less than or equal to the left and right deviation threshold, it is determined that the column pixel corresponding to the column pixel index is a false edge caused by image splicing;

[0037] If the minimum value of the average value of the standard deviation of the left neighboring column pixels and the average value of the standard deviation of the right neighboring column pixels of the corresponding column pixel index is greater than the left and right standard deviation thresholds, it is determined that the column pixel corresponding to the column pixel index is a false edge caused by stripes introduced by the roller surface.

[0038] A lithium battery pole piece edge detection device, comprising:

[0039] A receiving module, used for receiving an original lithium battery electrode image;

[0040] An opening operation calculation module, used to perform an opening operation on the original lithium battery electrode image according to a set direction to obtain an opening operation image;

[0041] A second-order calculation module, used to calculate the second-order gradient of the set direction of the open operation image to obtain a second-order gradient image;

[0042] The judgment module is used to calculate the mean and standard deviation of pixels and make conditional judgments based on the second-order gradient image and the original lithium battery pole piece image to obtain the lithium battery pole piece edge detection result.

[0043] Furthermore, the opening operation calculation module and the second-order calculation module use the direction perpendicular to the extension direction of the dressing in the original lithium battery pole piece image as the set direction.

[0044] Furthermore, the second-order calculation module calculates the second-order gradient of the set direction of the open operation image to obtain the second-order gradient image, including:

[0045] The absolute value of the difference between the pixel value of the i-th row and j-th column in the opening operation image and the pixel value of the i-th row and j+2-th column in the second-order gradient image is calculated as the pixel value of the i-th row and j-th column in the second-order gradient image.

[0046] Furthermore, the judgment module calculates the mean and standard deviation of pixels and makes conditional judgment based on the second-order gradient image and the original lithium battery pole piece image to obtain the lithium battery pole piece edge detection result, including:

[0047] Initialize a first edge set, calculate the mean and standard deviation of pixels based on the second-order gradient image, and add 1 to the column pixel index that meets the first preset condition according to the calculation result and store it in the first edge set;

[0048] Initialize a second edge set, perform mean calculation on pixels in the second-order gradient image based on the column pixel index in the first edge set minus 1, and store the column pixel index that meets the second preset condition into the second edge set according to the calculation result;

[0049] Initialize the third edge set, and judge the left and right neighborhood mean standard deviation based on the column pixel index in the second edge set corresponding to the pixels in the original lithium battery pole piece image, and store the column pixel index that meets the third preset condition into the third edge set according to the judgment result. The obtained third edge set is the column index of all real edges in the original lithium battery pole piece image.

[0050] Furthermore, the judgment module calculates the mean and standard deviation based on the second-order gradient image, and adds 1 to the pixel index that meets the first preset condition according to the calculation result and stores it in the first edge set, including:

[0051] Traversing and calculating a first mean and a standard deviation of each column of pixels in the second-order gradient image;

[0052] The column pixel indexes whose first mean is greater than the preset mean and whose standard deviation is less than the preset standard deviation are incremented by 1 and then stored in the first edge set.

[0053] Further, the judgment module performs mean calculation on the pixels in the second-order gradient image corresponding to the column pixel index in the first edge set minus 1, and stores the column pixel index that meets the second preset condition into the second edge set according to the calculation result, including:

[0054] Calculate a second mean value of column pixels in the second-order gradient image corresponding to consecutive column pixel indices in the first edge set minus 1;

[0055] The column pixel index corresponding to the maximum value of the second mean obtained in the same boundary is stored in the second edge set.

[0056] Furthermore, the judgment module judges the standard deviation of the left and right neighborhood means based on the column pixel indexes in the second edge set corresponding to the pixels in the original lithium battery pole piece image, and stores the column pixel indexes that meet the third preset condition into the third edge set according to the judgment result, including:

[0057] Set the left and right mean difference threshold, left and right deviation threshold, and left and right standard deviation threshold;

[0058] Traversing the second edge set, calculating the average of the left neighboring column pixel means, the average of the left neighboring column pixel standard deviations, the average of the right neighboring column pixel means, and the average of the right neighboring column pixel standard deviations corresponding to the column pixel index in the second edge set in the original lithium battery pole piece image;

[0059] If the column pixel index in the second edge set corresponds to the column pixel in the original lithium battery electrode image and satisfies one of the following conditions, the corresponding column pixel index is stored in the third edge set:

[0060] The absolute value of the difference between the average value of the left neighbor column pixel means and the average value of the right neighbor column pixel means is greater than or equal to the left and right mean difference threshold, and the minimum value of the average value of the left neighbor column pixel standard deviation and the average value of the right neighbor column pixel standard deviation is less than or equal to the left and right standard deviation threshold; or,

[0061] The absolute value of the difference between the corresponding column pixel index and half of the number of column pixel indices in the second edge set is greater than the left and right deviation thresholds, and the minimum value of the average value of the left neighboring column pixel standard deviation and the average value of the right neighboring column pixel standard deviation is less than or equal to the left and right standard deviation thresholds.

[0062] Further, if the absolute value of the difference between the average value of the left neighboring column pixel means and the average value of the right neighboring column pixel means of the corresponding column pixel index is less than the left and right mean difference threshold, and the absolute value of the difference between the corresponding column pixel index and half of the number of column pixel indices in the second edge set is less than or equal to the left and right deviation threshold, it is determined that the column pixel corresponding to the column pixel index is a false edge caused by image splicing;

[0063] If the minimum value of the average value of the standard deviation of the left neighboring column pixels and the average value of the standard deviation of the right neighboring column pixels of the corresponding column pixel index is greater than the left and right standard deviation thresholds, it is determined that the column pixel corresponding to the column pixel index is a false edge caused by stripes introduced by the roller surface.

[0064] A lithium battery pole piece edge detection system comprises an image acquisition device, a processor and a storage device, wherein the image acquisition device is used to acquire an original lithium battery pole piece image, the storage device stores a plurality of instructions, and the processor is used to read the instructions and execute the above method.

[0065] The lithium battery pole piece edge detection method, device and system provided by the present invention have at least the following beneficial effects:

[0066] (1) Through opening operation, calculation of lateral second-order gradient, determination of unique edge index in the neighborhood, and judgment of standard deviation of left and right neighborhood means, the edge detection of lithium battery can be adaptively performed when the components of lithium battery pole pieces are unknown, the widths of each part are unknown, there is defect interference, and the grayscale is uneven. There is no need to perform targeted optimization on each lithium battery pole piece image; parameters such as the width, width and position error of each part of the lithium battery pole piece can also be further obtained;

[0067] (2) Compared with edge detection methods such as Canny, Sobel, and Prewitt, it has stronger anti-interference ability and more accurate detection results;

[0068] (3) The proposed algorithm only involves basic operations and simple logical judgments. The principle is simple and easy to burn into the hardware circuit of the lithium battery electrode edge detection device;

[0069] (4) It is easy to reproduce and is applicable to different lithium battery manufacturers and different image acquisition methods, with strong versatility. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is a structural schematic diagram of a lithium battery pole piece image acquisition system in the prior art.

[0071] Figure 2 (a) is a schematic diagram of image stitching in a lithium battery electrode image.

[0072] Figure 2 (b) is a schematic diagram of the grayscale fault that appears at the joint in the lithium battery electrode image.

[0073] Figure 2 (c) is a schematic diagram of the edges of a lithium battery electrode image under normal circumstances.

[0074] Figure 2 (d) is a schematic diagram of the stripe-shaped defects on the roller surface in the lithium battery pole piece image.

[0075] Figure 3 The present invention provides a flow chart of an embodiment of a lithium battery pole piece edge detection method.

[0076] Figure 4 A flowchart of an embodiment of calculating mean and standard deviation and judging conditions in the lithium battery pole piece edge detection method provided by the present invention.

[0077] Figure 5 A flowchart of an embodiment of obtaining a first edge set in the lithium battery pole piece edge detection method provided by the present invention.

[0078] Figure 6 A flowchart of an embodiment of obtaining a second edge set in the lithium battery pole piece edge detection method provided by the present invention.

[0079] Figure 7 A flowchart of an embodiment of obtaining a third edge set in the lithium battery pole piece edge detection method provided by the present invention.

[0080] FIG8 (a) is an original lithium battery pole piece image in an application scenario of the lithium battery pole piece edge detection method provided by the present invention.

[0081] FIG8( b ) is an open operation image of the lithium battery pole piece edge detection method provided by the present invention in an application scenario.

[0082] FIG8 (c) is a second-order gradient image of the lithium battery pole piece edge detection method provided by the present invention in an application scenario.

[0083] FIG8 (d) is a schematic diagram of determining a first edge set in an application scenario of the lithium battery pole piece edge detection method provided by the present invention.

[0084] FIG8( e ) is a schematic diagram of determining a second edge set in an application scenario of the lithium battery pole piece edge detection method provided by the present invention.

[0085] FIG8( f ) is a schematic diagram of determining the third edge set in an application scenario of the lithium battery pole piece edge detection method provided by the present invention.

[0086] Fig. 9 A schematic structural diagram of an embodiment of a lithium battery pole piece edge detection device provided by the present invention. DETAILED DESCRIPTION

[0087] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0088] refer to Figure 3 In some embodiments, a method for detecting an edge of a lithium battery pole piece is provided, comprising:

[0089] S1, receiving the original lithium battery electrode image;

[0090] S2, performing an opening operation on the original lithium battery electrode image according to a set direction to obtain an opening operation image;

[0091] S3, calculating the second-order gradient of the set direction of the open operation image to obtain a second-order gradient image;

[0092] S4. Calculate the mean and standard deviation of pixels and make conditional judgments based on the second-order gradient image and the original lithium battery pole piece image to obtain a lithium battery pole piece edge detection result.

[0093] Specifically, in step S1, the original lithium battery pole piece image is acquired by an image acquisition device, wherein the number of image acquisition devices can be determined according to the width of the lithium battery pole piece, and can be one image acquisition device or two image acquisition devices for simultaneous acquisition, and the original lithium battery pole piece image can be an image acquired by a single image acquisition device or a spliced ​​image after images acquired by two image acquisition devices are spliced.

[0094] Further, in step S2, the original lithium battery pole piece image img_0 is opened in a set direction to obtain an opened image img_1. The direction perpendicular to the extension direction of the dressing in the original lithium battery pole piece image is the set direction. For example, the extension direction of the dressing in the original lithium battery pole piece image is taken as the y-axis, and the original lithium battery pole piece image is opened in the x-axis direction using a convolution kernel, which can remove the brighter or darker fine stripes that may appear at the image joints and on the roller surface.

[0095] Further, in step S3, the second-order gradient of the set direction of the open operation image is calculated to obtain a second-order gradient image, including:

[0096] The absolute value of the difference between the pixel value of the i-th row and j-th column in the opening operation image and the pixel value of the i-th row and j+2-th column in the second-order gradient image is calculated as the pixel value of the i-th row and j-th column in the second-order gradient image.

[0097] Specifically, assuming that the original lithium battery electrode image img_0 includes m rows and n columns of pixels, and the open operation image img_1 also includes m rows and n columns of pixels, the second-order gradient of the open operation image img_1 in the x-axis direction is calculated to obtain the second-order gradient image img_2 with m rows (n-2) columns. Assuming that the pixel value of the i-th row and j-th column (i=0, 1, ...m-1; j=0, 1, ..., n-1) of the open operation image img_1 is value1[i][j], then the pixel value value2[i][j] of the i-th row and j-th column (i=0, 1, ...m-1; j=0, 1, ..., n-3) of the second-order gradient image img_2 is:

[0098] ; (1)

[0099] i is the row pixel index and j is the column pixel index.

[0100] By calculating the second-order gradient, we can preliminarily lock the approximate column number where the edge is located.

[0101] Further, refer to Figure 4In step S4, the mean and standard deviation of pixels are calculated and conditional judgment is performed based on the second-order gradient image and the original lithium battery pole piece image to obtain the lithium battery pole piece edge detection result, including:

[0102] S41, initializing a first edge set, calculating the mean and standard deviation of pixels based on the second-order gradient image, and adding 1 to the column pixel index that meets the first preset condition according to the calculation result and storing it in the first edge set;

[0103] S42, initializing a second edge set, performing mean calculation on pixels in the second-order gradient image corresponding to consecutive column pixel indices in the first edge set minus 1, and storing column pixel indices that meet a second preset condition in the second edge set according to the calculation result;

[0104] S43, initialize the third edge set, and judge the left and right neighborhood mean standard deviation based on the column pixel index in the second edge set corresponding to the pixels in the original lithium battery pole piece image, and store the column pixel index that meets the third preset condition into the third edge set according to the judgment result, and the obtained third edge set is the column index of all real edges in the original lithium battery pole piece image.

[0105] Further, in step S41, the mean and standard deviation are calculated based on the second-order gradient image, and the pixel index that meets the first preset condition is increased by 1 according to the calculation result and then stored in the first edge set, including:

[0106] S41a, traversing and calculating the first mean and standard deviation of each column of pixels in the second-order gradient image;

[0107] S41b, adding 1 to the column pixel index whose first mean is greater than the preset mean and whose standard deviation is less than the preset standard deviation, and then storing it in the first edge set.

[0108] Specifically, first set the preset mean value mean_t and the preset standard deviation var_t, and initialize the first edge set set_1, traverse and calculate the first mean value and the first standard deviation of each column of pixels in the second-order gradient image img_2, and add 1 to the column pixel index whose first mean value is greater than the preset mean value mean_t and the first standard deviation is less than the preset standard deviation var_t, and store it in the first edge set set_1. The specific flow chart is as follows: Figure 5 shown.

[0109] Assume that the first mean of the pixels in the jth column (j=0, 1, ..., n-3) of the second-order gradient image img_2 is mean2[j] and the first standard deviation is var2[j], then:

[0110] ; (2)

[0111] ; (3)

[0112] Where m represents the number of rows of pixels in the second-order gradient image img_2, and vula2[k][j] represents the pixel value of the pixel in the k-th row and j-th column.

[0113] If mean2[j]>mean_t and var2[j]<var_t, the column pixel index colIndex=j+1 is stored in the first edge set set_1.

[0114] Further, in step S42, based on the column pixel index in the first edge set minus 1, a mean value calculation is performed on the pixels in the second-order gradient image, and the column pixel index that meets the second preset condition is stored in the second edge set according to the calculation result, including:

[0115] S42a, calculating a second mean value of column pixels in the second-order gradient image corresponding to the index of consecutive column pixels in the first edge set minus 1;

[0116] S42b, storing the column pixel index corresponding to the maximum value of the second mean obtained in the same boundary into the second edge set.

[0117] Specifically, the second edge set set_2 is initialized, and the column neighbor threshold neighborT is set. If the difference between the i-th column pixel index and the i-1-th column pixel index in the second edge set set_2 is less than the column neighbor threshold, it is determined that they are the same boundary. For the consecutive column pixel indexes in the first edge set set_1 minus 1, the second mean of the second-order gradient image img_2 at that location is calculated in sequence, and the column pixel index with the largest second mean is stored in the second edge set set_2. Since the pixel value at the boundary does not change suddenly, but has a gradual process, it is possible that a boundary will have multiple column pixel indexes placed in the first edge set set_1.

[0118] Assume that there are num_1 column pixel indexes in the first edge set set_1, namely index_0, index_1,...index_(num_1-1), traverse the first edge set set_1, set the initial value of the second mean maximum value tmpV of the column pixels to mean2[index_0], and the initial value of the second mean maximum column pixel index tmpInd to index_0.

[0119] For the i-th (i=1, 2, ..., num-1) column pixel index index_i, if index_i-index_(i-1) is less than the column adjacent threshold neighborT, it means that the column pixel index index_i and index_(i-1) represent the same boundary, compare the second mean value mean2[index_i-1] and tmpV of the second-order gradient image img_2 at the column index index_i, if mean2[index_i-1]>tmpV, then update tmpV to mean2[index_i-1], and update tmpInd to index_i; otherwise, it means that the column index representing the previous boundary in the first edge set set_1 has been traversed, store tmpInd in the second edge set set_2, and update tmpV to mean2[index_i], tmpInd to index_i, repeat the above steps until i=num_1, at this time it means that all num_1 column indexes have been traversed and screened, and store tmpInd in the second edge set set_2. The specific algorithm flow chart is as follows Figure 6 shown.

[0120] Further, in step S43, based on the column pixel index in the second edge set corresponding to the pixel in the original lithium battery pole piece image, the left and right neighborhood mean standard deviation is judged, and the column pixel index that meets the third preset condition is stored in the third edge set according to the judgment result, including:

[0121] S43a, setting left and right mean difference thresholds, left and right deviation thresholds, and left and right standard deviation thresholds;

[0122] S43b, traversing the second edge set, calculating the average value of the left neighboring column pixel mean, the average value of the left neighboring column pixel standard deviation, the average value of the right neighboring column pixel mean, and the average value of the right neighboring column pixel standard deviation corresponding to the column pixel index in the second edge set in the original lithium battery electrode image;

[0123] S43c, if the column pixel index in the second edge set corresponds to the column pixel in the original lithium battery electrode image and satisfies one of the following, the corresponding column pixel index is stored in the third edge set:

[0124] The absolute value of the difference between the average value of the left neighbor column pixel means and the average value of the right neighbor column pixel means is greater than or equal to the left and right mean difference threshold, and the minimum value of the average value of the left neighbor column pixel standard deviation and the average value of the right neighbor column pixel standard deviation is less than or equal to the left and right standard deviation threshold; or,

[0125] The absolute value of the difference between the corresponding column pixel index and half of the number of column pixel indices in the second edge set is greater than the left and right deviation thresholds, and the minimum value of the average value of the left neighboring column pixel standard deviation and the average value of the right neighboring column pixel standard deviation is less than or equal to the left and right standard deviation thresholds.

[0126] Wherein, if the absolute value of the difference between the average value of the left neighboring column pixel means and the average value of the right neighboring column pixel means of the corresponding column pixel index is less than the left and right mean difference threshold, and the absolute value of the difference between the corresponding column pixel index and half of the number of column pixel indices in the second edge set is less than or equal to the left and right deviation threshold, it is determined that the column pixel corresponding to the column pixel index is a false edge caused by image splicing;

[0127] If the minimum value of the average value of the standard deviation of the left neighboring column pixels and the average value of the standard deviation of the right neighboring column pixels of the corresponding column pixel index is greater than the left and right standard deviation thresholds, it is determined that the column pixel corresponding to the column pixel index is a false edge caused by stripes introduced by the roller surface.

[0128] Among them, the mean of the left neighboring column pixels of a column pixel is the average value of the preset number of column pixels adjacent to the left of the column pixel, the mean of the right neighboring column pixels of the column pixel is the average value of the preset number of column pixels adjacent to the right of the column pixel, the standard deviation of the left neighboring column pixels of the column pixel is the standard deviation of the preset number of column pixels adjacent to the left of the column pixel, and the standard deviation of the right neighboring column pixels of the column pixel is the standard deviation of the preset number of column pixels adjacent to the right of the column pixel.

[0129] Specifically, assuming that there are num_2 column pixel indexes in the second edge set set_2, and the original lithium battery electrode image has col columns in total, set the left and right neighborhood thresholds rangeT, the left and right mean difference thresholds subMeanT, the left and right deviation thresholds camT, and the left and right standard deviation thresholds VarT.

[0130] Traverse the column pixel indices set_2[i] in the second edge set set_2 (i = 0, 1, 2, ..., num_2 - 1), and calculate the mean and standard deviation of the columns from the (set_2[i] - 1)-th column, the (set_2[i] - 2)-th column... to the max(set_2[i] - rangeT, 0)-th column respectively, that is, the mean of the left neighborhood column pixels and the standard deviation of the left neighborhood column pixels. Then calculate the average value lmm of the means of these min(rangeT, set_2[i]) column pixels and the average value lvm of the standard deviations. Then calculate the mean and standard deviation of the columns from the (set_2[i] + 1)-th column, the (set_2[i] + 2)-th column... to the min(set_2[i] + rangeT, col - 1)-th column respectively, that is, the mean of the right neighborhood column pixels and the standard deviation of the right neighborhood column pixels. Then calculate the average value rmm of the means of these max(rangeT, col - 1 - set_2[i]) columns and the average value rvm of the standard deviations.

[0131] The condition for a true edge is:

[0132] !{[abs(lmm - rmm) < subMeanT && abs(set_2[i] - num_2 / 2) <= camT] || [min(lvm, rvm) > VarT]}; that is, if the absolute value of the difference between the average value lmm of the means of the left neighborhood column pixels and the average value rmm of the means of the right neighborhood column pixels is greater than or equal to the left - right mean difference threshold subMeanT, and the minimum value min(lvm, rvm) of the average value lvm of the standard deviations of the left neighborhood column pixels and the average value rvm of the standard deviations of the right neighborhood column pixels is less than or equal to the left - right standard deviation threshold VarT, then it is considered a true edge; or, if the absolute value of the difference between the corresponding column pixel index set_2[i] and half of the number of column pixel indices in the second edge set (num_2 / 2) is greater than the left - right deviation threshold camT, and the minimum value min(lvm, rvm) of the average value lvm of the standard deviations of the left neighborhood column pixels and the average value rvm of the standard deviations of the right neighborhood column pixels is less than or equal to the left - right standard deviation threshold VarT, then it is considered a true edge.

[0133] If abs(lmm - rmm) < subMeanT and abs(set_2[i] - num_2 / 2) <= camT, it indicates that this edge column index is a false edge caused by the stitching of two cameras; if min(lvm, rvm) > VarT, it indicates that this edge column index is a false edge caused by rough stripes or impurities on the roller surface. Column pixel indices other than these are considered true edges and stored in the third edge set set_3. The specific algorithm flowchart is as Figure 7 shown.

[0134] The method provided in this embodiment is further described below through specific application scenarios.

[0135] A set of intermittent coating images of lithium battery electrodes img_0 was selected, and the original image is shown in Figure 8 (a).

[0136] The original image is opened in the x-axis direction with a convolution kernel size of 1*3 to obtain the opened image img_1, as shown in Figure 8(b). This shows that the x-axis opening operation can effectively remove the brighter or darker fine stripes that may appear at the camera joints and on the roller surface.

[0137] The second-order gradient value in the x-axis direction is calculated by splitting the operation image to obtain the second-order gradient image img_2, as shown in Figure 8 (c). This shows that the second-order gradient calculation can normalize the non-edge area with large and small grayscale values.

[0138] Initialize the set set_1 to be empty, calculate the mean and standard deviation of the second-order gradient values ​​column by column, and retain the column indexes whose mean is greater than 10 and whose standard deviation is less than the mean, and store them in set_1. Set the columns corresponding to these column indexes to white, and the remaining columns to black, and obtain a schematic image, as shown in Figure 8 (d), which shows that the edge area and non-edge area can be preliminarily distinguished.

[0139] Assume that there are num_1 column indexes in set_1, initialize set tmp and set_2 to be empty, store the first column index set_1[0] in set_1 in tmp, traverse the column index set_1[i] (i=1, 2, ..., num_1-1) in set_1, if set_1[i]-set_1[i-1]≤5, store set_1[i] in tmp; otherwise, calculate the mean of all columns in tmp, store the column index with the largest mean in set_2, clear tmp, and then store set_1[i] in tmp. Set the columns corresponding to these column indexes to white and the remaining columns to black, and obtain a schematic image, as shown in Figure 8 (e), which shows that the unique edge column index in each edge region can be effectively retained.

[0140] Assume that there are num_2 column indices in set_2, and that there are col columns in the image. Initialize sets tmp and set_3 to be empty, store the first column index set_1[0] in set_1 in tmp, traverse the column index set_2[i] (i=0,1, 2, ..., num_2-1) in set_2, and calculate the mean and standard deviation of the set_2[i]-1th column, the set_2[i]-2th column, ... the max(set_2[i]-20, 0)th column, and then calculate the mean lmm and the mean lvm of the mean of the min(20, set_2[i])th column, and then calculate the mean and standard deviation of the set_2[i]+1th column, the set_2[i]+2th column, ... the min(set_2[i]+20, col-1)th column, and then calculate the max(20, col-1-set_2[i]) column mean rmm and standard deviation rvm; if abs(lmm-col_rmm)<30 and abs(set_2[i]-num_2 / 2)<=10, it means that this edge column index is a false edge caused by the splicing of two cameras; if min(lvm, rvm)>30, it means that this edge column index is a false edge caused by rough stripes or impurities on the roller surface. Other column indices are considered to be true edges and stored in set_3. The columns corresponding to these column indices are set to white and the remaining columns are set to black to obtain a schematic image, as shown in Figure 8 (f), which shows that the true edges can be retained while the false edges are eliminated, and the edge detection is finally completed.

[0141] refer to Fig. 9 In some embodiments, a lithium battery pole piece edge detection device is also provided, comprising:

[0142] Receiving module 201, used for receiving original lithium battery pole piece image;

[0143] An opening calculation module 202 is used to perform an opening operation on the original lithium battery electrode image according to a set direction to obtain an opening operation image;

[0144] A second-order calculation module 203, used to calculate the second-order gradient of the set direction of the open operation image to obtain a second-order gradient image;

[0145] The judgment module 204 is used to calculate the mean and standard deviation of pixels and make conditional judgments based on the second-order gradient image and the original lithium battery pole piece image to obtain the lithium battery pole piece edge detection result.

[0146] Furthermore, the opening operation calculation module and the second-order calculation module use the direction perpendicular to the extension direction of the dressing in the original lithium battery pole piece image as the set direction.

[0147] Furthermore, the second-order calculation module calculates the second-order gradient of the set direction of the open operation image to obtain the second-order gradient image, including:

[0148] The absolute value of the difference between the pixel value of the i-th row and j-th column in the opening operation image and the pixel value of the i-th row and j+2-th column in the second-order gradient image is calculated as the pixel value of the i-th row and j-th column in the second-order gradient image.

[0149] Furthermore, the judgment module calculates the mean and standard deviation of pixels and makes conditional judgment based on the second-order gradient image and the original lithium battery pole piece image to obtain the lithium battery pole piece edge detection result, including:

[0150] Initialize a first edge set, calculate the mean and standard deviation of pixels based on the second-order gradient image, and add 1 to the column pixel index that meets the first preset condition according to the calculation result and store it in the first edge set;

[0151] Initialize a second edge set, perform mean calculation on pixels in the second-order gradient image based on the column pixel index in the first edge set minus 1, and store the column pixel index that meets the second preset condition into the second edge set according to the calculation result;

[0152] Initialize the third edge set, and judge the left and right neighborhood mean standard deviation based on the column pixel index in the second edge set corresponding to the pixels in the original lithium battery pole piece image, and store the column pixel index that meets the third preset condition into the third edge set according to the judgment result. The obtained third edge set is the column index of all real edges in the original lithium battery pole piece image.

[0153] Furthermore, the judgment module calculates the mean and standard deviation based on the second-order gradient image, and adds 1 to the pixel index that meets the first preset condition according to the calculation result and stores it in the first edge set, including:

[0154] Traversing and calculating a first mean and a standard deviation of each column of pixels in the second-order gradient image;

[0155] The column pixel indexes whose first mean is greater than the preset mean and whose standard deviation is less than the preset standard deviation are incremented by 1 and then stored in the first edge set.

[0156] Further, the judgment module performs mean calculation on the pixels in the second-order gradient image corresponding to the column pixel index in the first edge set minus 1, and stores the column pixel index that meets the second preset condition into the second edge set according to the calculation result, including:

[0157] Calculate the second mean of the column pixels in the second-order gradient image corresponding to the column pixel index in the first edge set minus 1;

[0158] The column pixel index corresponding to the maximum value of the second mean obtained in the same boundary is stored in the second edge set.

[0159] Furthermore, the judgment module judges the standard deviation of the left and right neighborhood means based on the column pixel indexes in the second edge set corresponding to the pixels in the original lithium battery pole piece image, and stores the column pixel indexes that meet the third preset condition into the third edge set according to the judgment result, including:

[0160] Set the left and right mean difference threshold, left and right deviation threshold, and left and right standard deviation threshold;

[0161] Traversing the second edge set, calculating the average of the left neighboring column pixel means, the average of the left neighboring column pixel standard deviations, the average of the right neighboring column pixel means, and the average of the right neighboring column pixel standard deviations corresponding to the column pixel index in the second edge set in the original lithium battery pole piece image;

[0162] If the column pixel index in the second edge set corresponds to the column pixel in the original lithium battery electrode image and satisfies one of the following conditions, the corresponding column pixel index is stored in the third edge set:

[0163] The absolute value of the difference between the average value of the left neighbor column pixel means and the average value of the right neighbor column pixel means is greater than or equal to the left and right mean difference threshold, and the minimum value of the average value of the left neighbor column pixel standard deviation and the average value of the right neighbor column pixel standard deviation is less than or equal to the left and right standard deviation threshold; or,

[0164] The absolute value of the difference between the corresponding column pixel index and half of the number of column pixel indices in the second edge set is greater than the left and right deviation thresholds, and the minimum value of the average value of the left neighboring column pixel standard deviation and the average value of the right neighboring column pixel standard deviation is less than or equal to the left and right standard deviation thresholds.

[0165] Further, if the absolute value of the difference between the average value of the left neighboring column pixel means and the average value of the right neighboring column pixel means of the corresponding column pixel index is less than the left and right mean difference threshold, and the absolute value of the difference between the corresponding column pixel index and half of the number of column pixel indices in the second edge set is less than or equal to the left and right deviation threshold, it is determined that the column pixel corresponding to the column pixel index is a false edge caused by image splicing;

[0166] If the minimum value of the average value of the standard deviation of the left neighboring column pixels and the average value of the standard deviation of the right neighboring column pixels of the corresponding column pixel index is greater than the left and right standard deviation thresholds, it is determined that the column pixel corresponding to the column pixel index is a false edge caused by stripes introduced by the roller surface.

[0167] In some embodiments, a lithium battery pole piece edge detection system is also provided, including an image acquisition device, a processor and a storage device, wherein the image acquisition device is used to acquire original lithium battery pole piece images, the storage device stores multiple instructions, and the processor is used to read the instructions and execute the above method.

[0168] The lithium battery pole piece edge detection method, device and system provided in the above embodiments have at least the following beneficial effects:

[0169] (1) Through opening operation, calculation of lateral second-order gradient, determination of unique edge index in the neighborhood, and judgment of standard deviation of left and right neighborhood means, the edge detection of lithium battery can be adaptively performed when the components of lithium battery pole pieces are unknown, the widths of each part are unknown, there is defect interference, and the grayscale is uneven. There is no need to perform targeted optimization on each lithium battery pole piece image; parameters such as the width, width and position error of each part of the lithium battery pole piece can also be further obtained;

[0170] (2) Compared with edge detection methods such as Canny, Sobel, and Prewitt, it has stronger anti-interference ability and more accurate detection results;

[0171] (3) The proposed algorithm only involves basic operations and simple logical judgments. The principle is simple and easy to burn into the hardware circuit of the lithium battery electrode edge detection device;

[0172] (4) It is easy to reproduce and is applicable to different lithium battery manufacturers and different image acquisition methods, with strong versatility.

[0173] Although preferred embodiments of the present invention have been described, additional changes and modifications may be made to these embodiments by those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A lithium battery pole piece edge detection method, characterized in that: include: Receive the original lithium battery electrode image; Performing an opening operation on the original lithium battery pole piece image according to a set direction to obtain an opening operation image; Calculating the second-order gradient of the set direction of the open operation image to obtain a second-order gradient image; Calculate the mean and standard deviation of pixels and make conditional judgments based on the second-order gradient image and the original lithium battery pole piece image to obtain a lithium battery pole piece edge detection result; Based on the second-order gradient image and the original lithium battery pole piece image, the mean and standard deviation of pixels are calculated and the condition is judged to obtain the lithium battery pole piece edge detection result, including: Initialize the first edge set, calculate the mean and standard deviation of pixels based on the second-order gradient image, and add 1 to the column pixel index that meets the first preset condition according to the calculation result and store it in the first edge set: traverse and calculate the first mean and standard deviation of each column pixel in the second-order gradient image, add 1 to the column pixel index whose first mean is greater than the preset mean and whose standard deviation is less than the preset standard deviation and store it in the first edge set; Initialize the second edge set, perform mean calculation based on the pixel indices of columns of pixels in the first edge set minus 1 corresponding to the pixels in the second-order gradient image, and store the pixel indices of columns of pixels that meet the second preset condition in the second edge set according to the calculation result: calculate the second mean of the pixel indices of columns of consecutive columns of pixels in the first edge set minus 1 corresponding to the pixels in the second-order gradient image, and store the pixel indices of columns of pixels corresponding to the maximum value of the second mean obtained in the same boundary in the second edge set; Initialize a third edge set, perform left and right neighborhood mean standard deviation judgment based on the column pixel indexes in the second edge set corresponding to the pixels in the original lithium battery pole piece image, store the column pixel indexes that meet the third preset condition into the third edge set according to the judgment result, and obtain the third edge set as the column indexes of all real edges in the original lithium battery pole piece image; The conditions for a true edge are: !{[abs(lmm-rmm) <subMeanT & abs(set_2[i]-num_2 / 2) <=camT] || [min(lvm,rvm)>VarT]}; Among them, lmm represents the average of the left neighbor column pixel means, rmm represents the average of the right neighbor column pixel means, subMeanT represents the left and right mean difference threshold, lvm represents the average of the left neighbor column pixel standard deviation, rvm represents the average of the right neighbor column pixel standard deviation, VarT represents the left and right standard deviation thresholds, set_2[i] represents the corresponding column pixel index in the second edge set, num_2 represents the number of column pixel indices in the second edge set, and camT represents the left and right deviation thresholds.

2. The method according to claim 1, characterized in that The direction perpendicular to the extension direction of the dressing in the original lithium battery pole piece image is taken as the set direction.

3. The method according to claim 1, characterized in that Calculating the second-order gradient of the set direction of the opening operation image to obtain a second-order gradient image includes: The absolute value of the difference between the pixel value of the i-th row and j-th column in the opening operation image and the pixel value of the i-th row and j+2-th column in the second-order gradient image is calculated as the pixel value of the i-th row and j-th column in the second-order gradient image.

4. The method according to claim 1, characterized in that: Based on the column pixel index in the second edge set corresponding to the pixel in the original lithium battery pole piece image, the left and right neighborhood mean standard deviation is judged, and the column pixel index that meets the third preset condition is stored in the third edge set according to the judgment result, including: Set the left and right mean difference threshold, left and right deviation threshold, and left and right standard deviation threshold; Traversing the second edge set, calculating the average of the left neighboring column pixel means, the average of the left neighboring column pixel standard deviations, the average of the right neighboring column pixel means, and the average of the right neighboring column pixel standard deviations corresponding to the column pixel index in the second edge set in the original lithium battery pole piece image; If the column pixel index in the second edge set corresponds to the column pixel in the original lithium battery electrode image and satisfies one of the following conditions, the corresponding column pixel index is stored in the third edge set: The absolute value of the difference between the average value of the left neighbor column pixel means and the average value of the right neighbor column pixel means is greater than or equal to the left and right mean difference threshold, and the minimum value of the average value of the left neighbor column pixel standard deviation and the average value of the right neighbor column pixel standard deviation is less than or equal to the left and right standard deviation threshold; or, The absolute value of the difference between the corresponding column pixel index and half of the number of column pixel indices in the second edge set is greater than the left and right deviation thresholds, and the minimum value of the average value of the left neighboring column pixel standard deviation and the average value of the right neighboring column pixel standard deviation is less than or equal to the left and right standard deviation thresholds.

5. The method according to claim 4, characterized in that If the absolute value of the difference between the average value of the left neighboring column pixel means and the average value of the right neighboring column pixel means of the corresponding column pixel index is less than the left and right mean difference threshold, and the absolute value of the difference between the corresponding column pixel index and half of the number of column pixel indices in the second edge set is less than or equal to the left and right deviation threshold, it is determined that the column pixel corresponding to the column pixel index is a false edge caused by image splicing; If the minimum value of the average value of the standard deviation of the left neighboring column pixels and the average value of the standard deviation of the right neighboring column pixels of the corresponding column pixel index is greater than the left and right standard deviation thresholds, it is determined that the column pixel corresponding to the column pixel index is a false edge caused by stripes introduced by the roller surface.

6. A lithium battery pole piece edge detection device, characterized in that: include: A receiving module, used for receiving an original lithium battery electrode image; An opening operation calculation module, used to perform an opening operation on the original lithium battery electrode image according to a set direction to obtain an opening operation image; A second-order calculation module, used to calculate the second-order gradient of the set direction of the open operation image to obtain a second-order gradient image; A judgment module, used to calculate the mean and standard deviation of pixels and make conditional judgments based on the second-order gradient image and the original lithium battery pole piece image to obtain a lithium battery pole piece edge detection result; The judgment module calculates the mean and standard deviation of pixels and makes conditional judgment based on the second-order gradient image and the original lithium battery pole piece image to obtain the lithium battery pole piece edge detection result, including: Initialize the first edge set, calculate the mean and standard deviation of pixels based on the second-order gradient image, and add 1 to the column pixel index that meets the first preset condition according to the calculation result and store it in the first edge set: traverse and calculate the first mean and standard deviation of each column pixel in the second-order gradient image, add 1 to the column pixel index whose first mean is greater than the preset mean and whose standard deviation is less than the preset standard deviation and store it in the first edge set; Initialize the second edge set, perform mean calculation based on the pixel indices of columns of pixels in the first edge set minus 1 corresponding to the pixels in the second-order gradient image, and store the pixel indices of columns of pixels that meet the second preset condition in the second edge set according to the calculation result: calculate the second mean of the pixel indices of columns of consecutive columns of pixels in the first edge set minus 1 corresponding to the pixels in the second-order gradient image, and store the pixel indices of columns of pixels corresponding to the maximum value of the second mean obtained in the same boundary in the second edge set; Initialize a third edge set, perform left and right neighborhood mean standard deviation judgment based on the column pixel indexes in the second edge set corresponding to the pixels in the original lithium battery pole piece image, store the column pixel indexes that meet the third preset condition into the third edge set according to the judgment result, and obtain the third edge set as the column indexes of all real edges in the original lithium battery pole piece image; The conditions for a true edge are: !{[abs(lmm-rmm) <subMeanT & abs(set_2[i]-num_2 / 2) <=camT] || [min(lvm,rvm)>VarT]}; Among them, lmm represents the average of the left neighbor column pixel means, rmm represents the average of the right neighbor column pixel means, subMeanT represents the left and right mean difference threshold, lvm represents the average of the left neighbor column pixel standard deviation, rvm represents the average of the right neighbor column pixel standard deviation, VarT represents the left and right standard deviation thresholds, set_2[i] represents the corresponding column pixel index in the second edge set, num_2 represents the number of column pixel indices in the second edge set, and camT represents the left and right deviation thresholds.

7. A lithium battery pole piece edge detection system, characterized in that: It comprises an image acquisition device, a processor and a storage device, wherein the image acquisition device is used to acquire an original lithium battery pole piece image, the storage device stores a plurality of instructions, and the processor is used to read the instructions and execute any method according to claims 1-5.

Images