A lithium battery tab defect detection method combining region growth and ring correction

Abstract

The present invention relates to a lithium battery tab defect detection method combining regional growth and annular correction, 1) obtaining a lithium battery image to be detected, and preprocessing the lithium battery image to be detected; 2) segmenting the tab and coating in the preprocessed lithium battery image to be detected, and obtaining the tab defect area and the coating area; 3) removing snowflake interference from the segmented coating area; 4) detecting the contour information of the tab defect area, and circumscribing the rectangle for contour positioning, and calculating the key parameters of the tab defect; 5) on the basis of calculating the key parameters of the tab defect, correcting the key parameters of the tab defect, and then realizing accurate detection of defects. Compared with the prior art, the present invention has the advantages of improving detection accuracy and high detection efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of machine vision automated detection, and in particular to a lithium battery tab defect detection method combining regional growth and ring correction. Background Art

[0002] Lithium-ion batteries have the advantages of low self-discharge, high energy density, no memory effect, green environmental protection and long cycle life. However, during the production and processing of lithium battery tabs, they are easily affected by the tab cutting machine, resulting in a variety of defects such as missing tabs, tab wrinkles, and tab joints. At present, the defect detection efficiency of many lithium battery manufacturers is low, the labor intensity is high, and some minor defects are easily missed.

[0003] The defect detection method based on machine vision is a non-contact, non-destructive automatic detection method. When using machine vision technology for defect detection, there are mainly steps such as image acquisition, image preprocessing, feature extraction, and calculation of key parameters of defect areas. Image acquisition uses industrial cameras to obtain pictures. Image preprocessing usually suppresses background areas and enhances defect areas. Commonly used methods include Gaussian filtering, mean filtering, and median filtering. Feature extraction usually uses Sobel edge detection algorithm, OTSU algorithm, and Canny edge detection algorithm. The key parameters of defect areas are calculated using an image gradient calculation method. However, there is currently little research on the detection of lithium battery tab defects after tab cutting machines, and there is no technology that uses machine vision technology to detect key parameters of tab defects and locate tab defects. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a lithium battery tab defect detection method combining regional growth and ring correction.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A lithium battery tab defect detection method combining regional growth and ring correction, the method comprising the following steps:

[0007] S1: Obtain the image of the lithium battery to be detected and pre-process the image of the lithium battery to be detected. The pre-processing is: using piecewise linear transformation and combining the method of determining pixel nodes with three-dimensional stereo images to process the image of the lithium battery to be detected. The specific contents are:

[0008] The pixel nodes a and b of the original lithium battery image to be detected are determined by the three-dimensional stereo and front of the lithium battery tab; the pixel values ​​of the lithium battery image to be detected after the piecewise linear transformation are obtained by using the piecewise linear transformation formula. The piecewise linear transformation formula is:

[0009]

[0010] Wherein, f(x, y) is the pixel value of the original lithium battery image to be detected, g(x, y) is the pixel value of the image after piecewise linear transformation; a and b are the pixel nodes of the original lithium battery image to be detected, and c and d are the pixel nodes of the image after piecewise linear transformation.

[0011] S2: Segment the tab and coating in the preprocessed lithium battery image to be inspected to obtain the tab defect area and the coating area.

[0012] Furthermore, the lugs and coatings in the preprocessed lithium battery image to be inspected are segmented using a region growing algorithm.

[0013] S3: Remove snowflake interference from the segmented coating area.

[0014] Furthermore, the open operator in morphological processing is used to remove snowflake interference from the segmented coating area.

[0015] S4: Detect the contour information of the tab defect area, use a circumscribed rectangle to locate the contour, and calculate the key parameters of the tab defect.

[0016] Furthermore, the Canny edge detection algorithm is used to detect the contour information of the defect. The expression for detecting the contour information of the defect using the Canny edge detection algorithm is:

[0017]

[0018] In the formula, f x (i,j) is the horizontal difference, f y (i, j) is the difference in the vertical direction, M(i, j) is the gradient amplitude of Canny edge detection, θ(i, j) is the gradient direction of Canny edge detection, and (i, j) is the coordinates of all points in the entire image.

[0019] Furthermore, the image gradient is used to calculate the relative contour area and relative perimeter of the tab defect. The calculation expressions of the relative contour area and relative perimeter are:

[0020]

[0021]

[0022] Where S is the area of ​​the tab defect, A is the tab defect region, (i, j) is the coordinate within the defect region; C is the circumference of the tab defect, and P is the defect edge detected by the Canny operator.

[0023] S5: Based on the calculation of the key parameters of the tab defect, the key parameters of the tab defect are corrected to achieve accurate detection of the defect.

[0024] Furthermore, a ring correction method is used to correct key parameters of the tab defect.

[0025] The lithium battery tab defect detection method combining regional growth and ring correction provided by the present invention has at least the following beneficial effects compared with the prior art:

[0026] 1) In the process of image preprocessing, the lithium battery tab defect detection method proposed in the present invention adopts a piecewise linear transformation method to maximize the enhancement of the defect area and suppress the background area;

[0027] 2) The region growing algorithm can be used to quickly segment the tabs and coatings, and the coatings become the background, making the tabs stand out;

[0028] 3) The ring correction method is used to correct the lug circumference and area parameters. Through the detection effect of the standard image, it is proved that the ring correction method can improve the detection accuracy of the defective area of ​​the lithium battery lug;

[0029] 4) The detection method of the present invention can effectively detect various defects of lithium battery tabs, such as missing tabs, tab wrinkles, and tab joints. It has high detection efficiency and can meet the needs of enterprise production. It is of great significance to realize the automated detection of lithium battery tab defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The types of lithium battery tab defect detection, where sub-image (a) is tab missing, sub-image (b) is tab wrinkle, and sub-image (c) is tab joint;

[0031] Figure 2 It is the flow chart of defect detection algorithm;

[0032] Figure 3 is the piecewise linear transformation of the lithium battery tab, where sub-figure (a) is the three-dimensional image of the lithium battery tab, sub-figure (b) is the front view of the lithium battery tab, and sub-figure (c) is the piecewise linear transformation result;

[0033] Figure 4 is a region growing algorithm; wherein, sub-image (a) is the growth result of the tab missing region, sub-image (b) is the growth result of the tab wrinkle region, and sub-image (c) is the growth result of the tab joint region;

[0034] Figure 5 The comparison diagram of the tab connection after morphological processing, where sub-image (a) is the tab connection diagram enhanced with snowflake noise, and sub-image (b) is the tab connection diagram after morphological processing;

[0035] Figure 6 It is the Canny edge detection map;

[0036] Figure 7 This is the effect diagram of the tab connection detection;

[0037] Figure 8(a) shows the correction principle of parameters S and C;

[0038] Figure 8(b) shows the correction principle of parameters TL, BR, W and H;

[0039] Fig. 9 It is a comparison diagram of the detection standard defect, where sub-image (a) is the detection result diagram without ring correction, and sub-image (b) is the detection result diagram after ring correction;

[0040] Fig.10 This is the detection result of the tab connection with ring correction;

[0041] Fig.11 The figures show the detection results of various tab defects, where sub-figures (a) and (b) are the tab missing detection results of different production lines under coating areas of different sizes; sub-figures (c) and (d) are the tab wrinkle detection results, where sub-figure (c) has two tab wrinkle defects and sub-figure (d) has one tab wrinkle defect; sub-figures (e) and (f) are the tab joint detection results, where sub-figure (e) is a twisted tab joint and sub-figure (f) is a flat tab joint. DETAILED DESCRIPTION

[0042] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work should belong to the scope of protection of the present invention.

[0043] Example

[0044] The present invention relates to a lithium battery tab defect detection method combining regional growth and ring correction, such as Figure 2 As shown, the method comprises the following steps:

[0045] Step 1: Obtain the image of the lithium battery to be inspected and preprocess the image. Specifically, the preprocessing is: using piecewise linear transformation and combining the method of determining pixel nodes with three-dimensional stereo images to achieve the image coating area unchanged, defect area enhancement and background area suppression.

[0046] The background area of ​​the lithium battery tab image interferes with the subsequent tab defect segmentation, so the background area needs to be suppressed. When the tab is wrinkled, there may be a dark shadow with a very low pixel value, such as Figure 1As shown, the defective area needs to be enhanced. The present invention utilizes piecewise linear transformation and combines a method for determining pixel nodes with a three-dimensional image to achieve image coating area unchanged, defective area enhancement, and background area suppression.

[0047] The formula of piecewise linear transformation is shown in formula (1).

[0048]

[0049] In formula (1), f(x, y) is the pixel value of the original image, and g(x, y) is the pixel value of the image after piecewise linear transformation. a and b are the pixel nodes of the original image, and c and d are the pixel nodes of the image after piecewise linear transformation. The pixel nodes a and b of the original image are determined by observing the pixel value range of the three-dimensional background area, the defect area and the coating area of ​​the front side of the lithium battery tab, as follows: Figure 3 Neutron images (a) and (b), the input image of the lithium battery tab is transformed by piecewise linear transformation according to formula (1) as shown below: Figure 3 Neutron image (c) shows the comparison of lithium battery tab connection Figure 1 The neutron image (c) shows that the defect area is significantly enhanced.

[0050] Step 2: Use the region growing algorithm to quickly segment the tabs and the coating, and the coating becomes the background, making the tabs stand out.

[0051] Because there is a very clear dividing line between the defective area of ​​the tab and the coating area, the region growing algorithm is used for segmentation. Region growth refers to starting from the seed point pixel, gradually adding neighboring pixels according to certain criteria, and when certain conditions are met, the region growth terminates. The seed point selected by the present invention is the center point of the image. If the grayscale value difference between the seed point and the surrounding 8 neighborhood growth points is less than 10, it is judged as a similar point. Similar points are used as seed points for the next growth. When there are no pixels that meet the growth criteria, the growth stops. The region growing algorithm can quickly segment the tab from the coating, and the coating becomes the background, allowing the tab to stand out. The result after segmentation by the region growing algorithm is as follows Figure 4 shown.

[0052] Step 3: Since some points with high pixel values ​​in the coating area are not removed, a lot of snowflake noise interference will appear. Therefore, the present invention selects the open operator in morphological processing to remove the snowflake interference.

[0053] Comparison of morphologically processed tabs Figure 5 The effect after morphological opening operator processing is shown in Figure 5(b), where it can be seen that the snowflake noise interference is removed.

[0054] Step 4: Use the Canny edge detection algorithm to detect the contour information of the defect, locate the contour with the circumscribed rectangle, and calculate the key parameters such as the area and perimeter of the tab based on the image gradient. This step uses the Canny edge detection algorithm to process the data after the morphological opening operation, which is Figure 5 Sub-figure (b).

[0055] The Canny operator has strong anti-noise capability and accurate positioning. The present invention uses a 3*3 neighborhood to calculate the gradient amplitude, and the calculation formula of the gradient amplitude M(i, j) and the gradient direction θ(i, j) of the Canny edge detection is shown in formula (2).

[0056]

[0057] The horizontal axis of the image is the horizontal direction, and the vertical axis of the image is the vertical direction. In formula (2), f x (i,j) is the horizontal difference, f y (i, j) is the vertical difference. The Canny edge detection result is as follows Figure 6 As shown in Figure 2, compared with the ear image after morphological processing, the Canny operator can accurately detect its edge. (i, j) in formula (2) represents the coordinates of all points in the entire image.

[0058] The defects of lithium battery tabs are mostly irregular polygons, so it is difficult to calculate the absolute area and perimeter. Therefore, the relative area and relative perimeter can be calculated (in the image processing process, the area and perimeter of irregular shapes can only be calculated by accumulating pixels, and the numerical value is the relative area and relative perimeter). The area and perimeter of the tab defect contour are calculated based on the image gradient. In essence, it is the accumulation of pixels. The calculation unit is pixel. The mathematical expressions are shown in (3) and (4) respectively.

[0059]

[0060]

[0061] In formula (3), S represents the area of ​​the defect, A represents the defect area, and (i, j) represents the coordinates within the defect area. In formula (4), C represents the perimeter of the defect, and P represents the defect edge detected by the Canny operator. (i, j) in formulas (3) and (4) only represents the coordinates within the defect area. In order to facilitate the observation of the effect of tab defect detection, the defect contour is drawn on the original image by polygonal fitting, and the circumscribed rectangle of the defect contour is drawn to realize the positioning of the defect contour. The effect of tab defect detection is shown in Figure 2. Figure 7 shown.

[0062] Step 5: Based on the calculation of the key parameters of the tab defect, a ring correction method is proposed to correct the key parameters to achieve accurate defect detection.

[0063] The tab defect area is detected as a whole. Since the morphological processing and Canny edge detection in the algorithm affect the size of S and C, the line width of the circumscribed rectangle affects the TL, BR, W, H and other related parameters of the tab defect area, including the maximum width W and the maximum height H of the tab defect area, the combination of the minimum horizontal axis value and the minimum vertical axis value of the tab defect area TL, and the combination of the maximum horizontal axis value and the maximum vertical axis value of the tab defect area BR. The parameters TL and BR realize the positioning of the tab defect area. The detection accuracy fails to meet the requirements of actual applications, and the relevant parameters need to be further corrected. The principle of the tab defect area parameters is shown in Figure 8(a) and Figure 8(b), and the correction formula of the lithium battery tab defect area parameters is shown in Formula (5).

[0064]

[0065] In formula (6), S′, C′, W′, H′, (TL′ x ,TL′ y ), (BR′ x BR′ y ) respectively represent the relevant parameters of the corrected tab defect area, α represents the ring width, β represents the random parameter, δ represents the fixed value, and ε represents the sum of the circumscribed rectangle line width and the ring width. Fig. 9 As shown in FIG. 1 , it is shown that the ring correction proposed by the present invention can improve the detection accuracy of the defective area of ​​the lithium battery tab. The detection result of the lithium battery tab joint after the ring correction algorithm is as follows: Fig.10 As shown, compared Figure 7 It can be seen that the key parameters of the lithium battery tab defect area are corrected, which relatively improves the detection accuracy of the lithium battery tab defect area. The effect of detecting various tab defects by the method of the present invention is shown as follows: Fig.11 As shown, Fig.11 Neutron images (a) and (b) are the results of the missing lug detection. The number of detected defects shown in the images is 0, indicating that the lug is missing. Fig.11 Sub-images (c) and (d) are the detection results of the tab wrinkles. The number of defects detected in sub-image (c) is 2, and the number of defects detected in sub-image (d) is 1. The algorithm can detect multiple defects at the same time. Fig.11 Sub-images (e) and (f) are the detection results of the tab joints. Sub-image (e) shows a twisted tab joint, and sub-image (f) shows a flat tab joint. This shows that the algorithm achieves good detection results for tab defects of different shapes.

[0066] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person familiar with the technical field can easily think of various equivalent modifications or substitutions within the technical scope disclosed by the present invention, and these modifications or substitutions should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A lithium battery tab defect detection method combining region growth and ring correction, characterized in that: The following steps are involved: 1) Obtaining an image of a lithium battery to be inspected and preprocessing the image of the lithium battery to be inspected; 2) Segmenting the tab and coating in the preprocessed lithium battery image to be inspected to obtain the tab defect area and coating area; 3) Remove snowflake interference from the segmented coating area; 4) Detect the contour information of the tab defect area, circumscribe the rectangle to locate the contour, and calculate the key parameters of the tab defect; 5) Based on the calculation of the key parameters of the tab defect, the key parameters of the tab defect are corrected to achieve accurate detection of the defect; In step 1), the preprocessing is: using piecewise linear transformation and combining a three-dimensional image to determine pixel nodes to process the lithium battery image to be inspected; The specific contents of processing the lithium battery image to be inspected by using piecewise linear transformation and combining the method of determining pixel nodes with three-dimensional stereo images are as follows: The pixel nodes a and b of the original lithium battery image to be detected are determined by the three-dimensional stereo and front of the lithium battery tab; the pixel values ​​of the lithium battery image to be detected after the piecewise linear transformation are obtained by using the piecewise linear transformation formula. The piecewise linear transformation formula is: Where, f(x, y) is the pixel value of the original lithium battery image to be detected, g(x, y) is the pixel value of the image after piecewise linear transformation; a and b are the pixel nodes of the original lithium battery image to be detected, and c and d are the pixel nodes of the image after piecewise linear transformation; In step 2), the lugs and coatings in the preprocessed lithium battery image to be inspected are segmented using a region growing algorithm; In step 3), the open operator in morphological processing is used to remove snowflake interference from the segmented coating area; In step 4), the Canny edge detection algorithm is used to detect the contour information of the defect; The expression of the contour information of the defect detected by the Canny edge detection algorithm is: In the formula, f x (i,j) is the horizontal difference, f y (i, j) is the vertical difference, M(i, j) is the gradient amplitude of Canny edge detection, θ(i, j) is the gradient direction of Canny edge detection, and (i, j) is the coordinates of all points in the entire image; In step 4), the relative contour area and relative perimeter of the tab defect are calculated using the image gradient; The calculation expressions of relative contour area and relative perimeter are: Where S is the area of ​​the tab defect, A is the tab defect region, (i, j) is the coordinate within the defect region; C is the perimeter of the tab defect, and P is the defect edge detected by the Canny operator; In step 5), the ring correction method is used to correct the key parameters of the tab defect, and the corresponding correction formula is: In the formula, S', C', W', H', (TL x ′,TL y ′)、(BR x ′,BR y ′) represent the relevant parameters of the corrected tab defect area, α represents the ring width, β represents the random parameter, δ represents the fixed value, and ε represents the sum of the circumscribed rectangle line width and the ring width.

Images