Method and system for judging corrosion condition of electrode foil

Abstract

The invention belongs to the technical field of condition judgment, and particularly relates to an electrode foil corrosion condition judgment method and system, and the method comprises the following steps: S1, obtaining a grayscale image of a to-be-analyzed electrode foil; constructing a structure tensor based on the composite gradient reflecting the local brightness and texture information of the pixel points, determining an anisotropic diffusion coefficient according to the structure tensor, and performing iterative anisotropic diffusion filtering processing on the grayscale image by using the anisotropic diffusion coefficient to obtain a filtered image; and S2, fusing the anisotropic diffusion coefficient determined for each pixel point with the pixel intensity of the filtered image to obtain a high-contrast corrosion significance map. According to the method, the contour edge information of the corrosion area can be kept while the complex texture and noise of the background area of the electrode foil image are smoothed, the contradiction between denoising and edge protection of a traditional filtering method is solved, the result of the corrosion condition is more reliable, and the automation level of electrode foil product quality detection is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of situation judgment, and particularly relates to an electrode foil corrosion situation judgment method and system. BACKGROUND

[0002] The electrode foil is a basic material for manufacturing aluminum electrolytic capacitors, and the performance of the electrode foil determines the capacity, loss, leakage current and service life of the capacitors. In the production process of the electrode foil, a complex hole structure is formed on the surface of the aluminum foil through an electrochemical corrosion process, so as to increase the surface area and improve the unit volume capacitance of the capacitor. The uniformity, depth and form of the corrosion are crucial to the quality of the electrode foil, and excessive or insufficient corrosion will lead to performance degradation or even scrap of the product. Therefore, detection and judgment of the corrosion situation of the electrode foil is a core link of quality control in the production process. The industry mainly relies on manual visual inspection or microscopic observation for sampling observation, and such a mode is low in efficiency and the detection result is easily affected by factors such as experience, fatigue state and external environment of the operator.

[0003] The existing automatic detection method usually includes steps of image acquisition, preprocessing, segmentation and feature analysis. In actual application, due to the complex texture of the electrode foil surface itself, the imaging process is easily disturbed by factors such as uneven illumination and noise, resulting in low contrast and blurred edges of the corrosion area and the background matrix. The traditional image preprocessing method, such as Gaussian filtering and median filtering, will blur the outline edges of the corrosion area while smoothing the background noise, which affects the subsequent image segmentation. In addition, the existing evaluation method mainly depends on a single feature parameter, and cannot make a comprehensive judgment on the corrosion grade. SUMMARY

[0004] The application provides an electrode foil corrosion situation judgment method and system to solve the technical problems that the electrode foil corrosion detection in the prior art is low in manual efficiency and unstable in result, the existing automatic detection is disturbed by imaging, the preprocessing blurs the outline, and the evaluation parameter is single, which is difficult to meet the quality control demand.

[0005] In a first aspect, the application provides an electrode foil corrosion situation judgment method, which comprises the following steps: S1, a gray-scale image of an electrode foil to be analyzed is acquired; a structure tensor is constructed based on a composite gradient reflecting local brightness and texture information of a pixel point, and an anisotropic diffusion coefficient is determined according to the structure tensor, and the gray-scale image is subjected to iterative anisotropic diffusion filtering treatment by using the anisotropic diffusion coefficient, so as to obtain a filtered image; S2, fusing the anisotropic diffusion coefficient determined for each pixel point and the pixel intensity of the filtered image to obtain a high-contrast corrosion saliency map; constructing a mapping function based on the gray scale distribution of the corrosion saliency map, and inputting each pixel of the saliency map into the mapping function to obtain a segmentation threshold map of the same size as the filtered image; S3, using an automatic threshold segmentation method on the filtered image to segment all corrosion regions; calculating the area, perimeter and shape compactness of each corrosion region as morphological feature parameters; performing multi-dimensional matching between the statistical distribution of the morphological feature parameters of all corrosion regions and a standard corrosion level database to output a corrosion condition judgment result of the electrode foil.

[0006] Further, a structure tensor is constructed based on a composite gradient reflecting the local brightness and texture information of a pixel point, and an anisotropic diffusion coefficient is determined according to the structure tensor, and an iterative anisotropic diffusion filtering process is performed on the gray scale image using the anisotropic diffusion coefficient to obtain a filtered image, including: The composite gradient of each pixel point in the gray scale image is calculated, and the composite gradient is the weighted sum of the brightness gradient and the texture gradient, and the weight coefficients are both 0.5; A 3x3 structure tensor is constructed based on the composite gradient, and the eigenvalues of the structure tensor are calculated, and the anisotropic diffusion coefficient is determined according to the eigenvalues; An anisotropic diffusion filtering is performed, and a bimodal separation index of the gray scale histogram of the current gray scale image is calculated; The anisotropic diffusion filtering is repeated until the variation of the bimodal separation index calculated by continuous three iterations is less than 0.001, and then the iteration is terminated.

[0007] Further, in S3, the global gray scale variance of the corrosion saliency map is calculated, and when the global gray scale variance is less than a preset low fluctuation threshold, an automatic threshold segmentation method is used on the filtered image to segment all corrosion regions, including: The Otsu method is used to calculate a global segmentation threshold, and the segmentation threshold makes the inter-class variance of the corrosion region pixel class and the background region pixel class maximum; the pixel points in the filtered image with a gray scale value greater than the segmentation threshold are classified as corrosion regions, and the pixel points with a gray scale value less than or equal to the segmentation threshold are classified as background regions.

[0008] Further, the area, perimeter and shape compactness of each corrosion region are calculated as morphological feature parameters, including: The total number of pixel points contained in each segmented corrosion region is counted as the area A of the corrosion region; An edge tracking algorithm is used to determine the boundary of each corrosion region, and the total number of pixel points on the boundary is counted as the perimeter P of the corrosion region; The shape compactness C of each corrosion region is calculated according to the formula The shape compactness C of each corrosion region is calculated according to the formula

[0009] Further, the statistical distribution of the morphological feature parameters of all corrosion areas is multi-dimensionally matched with a standard corrosion grade database, and a corrosion condition judgment result of the electrode foil is output, including: The mean and standard deviation of the area, perimeter, and shape compactness of all corrosion areas of the electrode foil to be analyzed are respectively counted, to form a 6-dimensional feature vector; The standard 6-dimensional feature vector corresponding to each grade is retrieved from the standard corrosion grade database; The weighted Euclidean distance between the 6-dimensional feature vector of the electrode foil to be analyzed and the standard feature vector of each grade in the database is calculated; The standard grade with the minimum weighted Euclidean distance is output as the corrosion condition judgment result of the electrode foil.

[0010] Further, in S1, the luminance gradient component of the gray-scale image is calculated using a Sobel operator, a set of multi-directional Gabor filters is used to extract the texture features of the gray-scale image and calculate the gradient component thereof, and the two are weighted and fused to obtain a composite gradient.

[0011] Further, in S2, a mapping function is constructed based on the gray-scale distribution of the corrosion saliency map, and the mapping function expression is as follows: T(x,y)=μ(x,y)×[1-k×(1-S(x,y))], wherein T(x,y) is a threshold value, μ(x,y) is a local mean value, k is a preset modulation coefficient, and S(x,y) is a pixel saliency.

[0012] Further, in S3, a plurality of standard grades are pre-stored in the standard corrosion grade database, and each grade corresponds to a standard three-dimensional feature distribution.

[0013] Further, in S1, a color image is obtained by photographing the electrode foil sample to be analyzed, and the color image is converted into an 8-bit gray-scale image.

[0014] In a second aspect, the present application provides an electrode foil corrosion condition judgment system, comprising a memory and a processor, and the memory stores computer program instructions, which, when executed by the processor, implement the above-mentioned electrode foil corrosion condition judgment method.

[0015] The beneficial effects are as follows: By constructing a composite gradient structure tensor reflecting the local brightness and texture information of pixels and performing anisotropic diffusion filtering, it is possible to smooth the complex texture and noise in the background area of ​​the electrode foil image while preserving the contour edge information of the etched area, thus resolving the contradiction between noise reduction and edge preservation in traditional filtering methods. This invention performs multidimensional matching of statistical distribution with a standard corrosion level database, achieving a comprehensive evaluation of the corrosion condition. This makes the corrosion results more reliable and improves the automation level of electrode foil product quality inspection. Attached Figure Description

[0016] Figure 1 This is a flowchart of a method for judging the corrosion status of electrode foil. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] An embodiment of the electrode foil corrosion determination method provided by the present invention: like Figure 1 As shown, the method for judging the corrosion condition of electrode foil includes the following steps: S1. Obtain the grayscale image of the electrode foil to be analyzed; construct a structure tensor based on the composite gradient reflecting the local brightness and texture information of the pixel, determine the anisotropic diffusion coefficient according to the structure tensor, and use the anisotropic diffusion coefficient to perform iterative anisotropic diffusion filtering on the grayscale image to obtain the filtered image.

[0019] The electrode foil sample to be analyzed is vertically photographed under uniform illumination, and the color image is converted into an 8-bit gray scale image. The luminance gradient components Ix and Iy of the gray scale image are calculated using a Sobel operator, while a set of multi-directional Gabor filters are used to extract the texture features of the gray scale image and calculate the gradient components Tx and Ty, and then the two are weighted and fused to obtain the composite gradients Gx and Gy. The structure tensor J of each pixel point is constructed based on the composite gradients. Then the structure tensor J is subjected to eigenvalue decomposition to obtain the eigenvalues λ1 and λ2. The diffusion coefficient matrix D is determined according to the size and difference of the eigenvalues, when λ1 is much larger than λ2, it indicates that this is a strong edge, then the diffusion coefficient perpendicular to the edge direction is set to be close to zero, and the diffusion coefficient along the edge direction is set to be 1; when λ1 and λ2 are both very small and close, it indicates a smooth area, then the diffusion coefficients in both directions are set to be 1. The gray scale image is iteratively updated based on the diffusion coefficient. After each iteration, the inter-class variance of the gray scale histogram of the current gray scale image is calculated as a bimodal separation index, when the variation of the inter-class variance value calculated by 3 consecutive iterations is less than a preset small threshold, for example, 0.001, it is determined that the filtering has converged, and the iteration is terminated.

[0020] In an optional embodiment, the structure tensor is constructed based on the composite gradient reflecting the local luminance and texture information of the pixel points, and the anisotropic diffusion coefficient is determined according to the structure tensor, and the gray scale image is iteratively anisotropically diffused and filtered using the anisotropic diffusion coefficient to obtain the filtered image, including: The composite gradient of each pixel point in the gray scale image is calculated, and the composite gradient is the weighted sum of the luminance gradient and the texture gradient, and the weight coefficients are both 0.5; The 3x3 structure tensor is constructed based on the composite gradient, and the eigenvalues of the structure tensor are calculated, and the anisotropic diffusion coefficient is determined according to the eigenvalues; An anisotropic diffusion filtering is performed, and the bimodal separation index of the gray scale histogram of the current gray scale image is calculated; The anisotropic diffusion filtering is repeated until the variation of the bimodal separation index calculated by 3 consecutive iterations is less than 0.001, and then the iteration is terminated.

[0021] The filtering process aims to smooth the noise and irrelevant texture in the background area of the grayscale image, while sharpening and preserving the real edges of the corrosion area. For example, for a pixel point with coordinates (100, 120) in the grayscale image, the luminance gradient is 30 calculated by the Sobel operator, and the texture gradient is 18 calculated by a set of Gaussian filters, and the composite gradient is 24. A 3x3 structure tensor is constructed using the composite gradient values in the neighborhood of the pixel point. The calculated eigenvalues can be λ1=400 and λ2=60. Since the two eigenvalues differ greatly, it indicates that there is a strong edge here, and the diffusion coefficient will be smaller to inhibit the smoothing across the edge. For example, the bimodal separation index calculated after the 8th iteration is 0.962, the 9th is 0.965, and the 10th is 0.967. Continue to iterate, the indexes calculated in the 11th, 12th and 13th times are 0.9675, 0.9679 and 0.9682 respectively. The change amount between the 11th and the 10th is 0.0005, the change amount between the 12th and the 11th is 0.0004, and the change amount between the 13th and the 12th is 0.0003. Since the change amounts of the last three iterations are 0.0005, 0.0004 and 0.0003, which are all less than the preset threshold 0.001, it indicates that the quality of the grayscale image has tended to be stable, and the separation degree of the corrosion and the background area has reached the best. At this time, the iteration is terminated to prevent over-smoothing leading to the loss of small corrosion features.

[0022] S2, the anisotropic diffusion coefficient determined for each pixel point is fused with the pixel intensity of the filtered image to obtain a high-contrast corrosion significance map; a mapping function is constructed based on the grayscale distribution of the corrosion significance map, and each pixel of the significance map is input into the mapping function to obtain a segmentation threshold map with the same size as the filtered image.

[0023] Specifically, the eigenvalue difference (λ1-λ2) of the anisotropic diffusion coefficient matrix is taken as the edge intensity response, and after normalization, it is pixel-level weighted and summed with the normalized inverted grayscale value (255-I) of the filtered image to generate the corrosion significance map S. The local mean μ(x, y) and standard deviation σ(x, y) of the filtered image in the N×N neighborhood are calculated. The mapping function T(x, y)=μ(x, y)×[1-k×(1-S(x, y))] is constructed, where k is a preset modulation coefficient. The mapping function adjusts the threshold value based on the local mean using the value of the corrosion significance map S. When the significance S(x, y) of a certain pixel is high, the corresponding threshold value T(x, y) is higher, thereby generating a segmentation threshold map.

[0024] S3, using an automatic threshold segmentation method on the filtered image, all corrosion areas are segmented out; the area, perimeter and shape compactness of each corrosion area are calculated as morphological feature parameters; the statistical distribution of the morphological feature parameters of all corrosion areas is multi-dimensionally matched with a standard corrosion grade database, and a corrosion condition judgment result of the electrode foil is output.

[0025] Specifically, each pixel point (x, y) of the filtered image is traversed to obtain a pixel intensity value I(x, y), which is compared with a threshold value T(x, y) at the corresponding position in the segmentation threshold map. If the pixel intensity value I(x, y) is less than the threshold value T(x, y), the pixel point is determined as a corrosion point and is assigned a value of 1; otherwise, it is determined as a background point and is assigned a value of 0. After the traversal is completed, a binary mask image is obtained, wherein the area with a logic value of 1 is the segmented corrosion area.

[0026] The global gray level variance of the corrosion significance map is calculated, and when the variance value is less than a preset low fluctuation threshold, the filtered image is not segmented by using the segmentation threshold map, but is directly segmented by using an automatic threshold segmentation method, and all corrosion areas are segmented out, including: The Otsu method is used to calculate a global segmentation threshold, and the segmentation threshold makes the inter-class variance of the corrosion area pixel class and the background area pixel class maximum; the pixel points with a gray level value greater than the segmentation threshold in the filtered image are classified as corrosion areas, and the pixel points with a gray level value less than or equal to the segmentation threshold are classified as background areas.

[0027] The Otsu method can automatically find a best segmentation point on the gray level histogram of the filtered image. Assuming that the gray level value range of a filtered image is 0 to 255, the Otsu method traverses all possible threshold values from 1 to 254. For each candidate threshold value, for example, T = 135, the algorithm divides all pixels into two groups: a background group with a gray level value less than or equal to 135 and a corrosion group with a gray level value greater than 135. Then, the square of the difference between the mean values of the gray levels of the two groups and the product of the proportion of the pixel number of each group are calculated, that is, the inter-class variance.

[0028] The above process continues until a threshold value, such as T = 168, is found, which makes the calculated inter-class variance reach a maximum value. The threshold value (T = 168) is the global segmentation threshold. The gray level value of each pixel point in the filtered image is compared with 168. The pixel point with a gray level value of 190 is a corrosion area pixel and is assigned a value of 255 (that is, white). The pixel point with a gray level value of 110 is a background pixel and is assigned a value of 0 (that is, black). A binary image containing only black and white is generated, wherein the white area represents the segmented corrosion area.

[0029] All pixels in the filtered image with gray value less than the global threshold are labeled as erosion regions, and the rest are labeled as background regions, thus generating a binary image. Connected component analysis is applied to the segmented binary image to label each independent erosion region. For each erosion region, its area is calculated by counting the total number of pixels it contains. The perimeter of the region is calculated by tracing its contour using an edge detection algorithm and accumulating the number or length of pixels on the contour. The shape compactness is preferably calculated by the formula 4 x π x area / perimeter squared, and the closer to 1 the value is, the closer the shape is to a circle. The area, perimeter, and shape compactness of all erosion regions in the current filtered image are generated into three histograms, respectively, to form a three-dimensional feature distribution representing the current erosion state. A standard erosion level database is pre-stored with several standard levels, such as level one, level two, and level three, each of which corresponds to a standard three-dimensional feature distribution. The matching process calculates the distance between the three-dimensional feature distribution of the sample to be tested and the three-dimensional feature distribution of each standard level in the database. The sample to be tested is determined to be the standard level with the smallest distance, and the level is output as the judgment result of the erosion condition.

[0030] In an alternative embodiment, the area, perimeter, and shape compactness of each erosion region are calculated as morphological feature parameters, including: Counting the total number of pixels contained in each segmented erosion region as the area A of the erosion region; Determining the boundary of each erosion region using an edge tracking algorithm and counting the total number of pixels on the boundary as the perimeter P of the erosion region; Calculating the shape compactness C of each erosion region according to the formula Calculating the shape compactness C of each erosion region according to the formula

[0031] After obtaining the binary image, all independent white regions in the image are identified by a connected component labeling algorithm, and each region represents an erosion point. For one of the erosion regions labeled as No. 1, all white pixels belonging to it are counted by traversing the erosion region. If a total of 620 pixels are counted, the area A of the erosion region is 620. The edge tracking algorithm, such as the Moore neighborhood tracking method, is applied to the No. 1 erosion region, starting from one boundary pixel of the erosion region and moving along the outer boundary pixel by pixel until returning to the starting point. Assuming that a total of 110 boundary pixels are passed during this round of tracking, the perimeter P of the region is 110. Using the obtained area A = 620 and perimeter P = 110, the shape compactness C is calculated, and the result is about 0.64. The smaller the value, the more irregular or elongated the shape is.

[0032] In an optional embodiment, the statistical distribution of the morphological feature parameters of all corrosion areas is multi-dimensionally matched with a standard corrosion grade database, and a corrosion condition judgment result of the electrode foil is output, including: The mean and standard deviation of the area, perimeter, and shape compactness of all corrosion areas of the electrode foil to be analyzed are respectively counted, to form a 6-dimensional feature vector; The standard 6-dimensional feature vector corresponding to each grade is retrieved from the standard corrosion grade database; The weighted Euclidean distance between the 6-dimensional feature vector of the electrode foil to be analyzed and the standard feature vector of each grade in the database is calculated; The standard grade with the minimum weighted Euclidean distance is output as the corrosion condition judgment result of the electrode foil.

[0033] Suppose that 80 corrosion areas are detected on a filtered image to be measured. The area, perimeter, and compactness of the 80 areas are calculated and counted. The statistical results obtained are, for example, the average area is 95, the area standard deviation is 30; the average perimeter is 60, the perimeter standard deviation is 15; the average compactness is 0.7, and the compactness standard deviation is 0.1. Thus, a 6-dimensional feature vector of the sample to be measured is formed .

[0034] The feature vectors of the standard grades are extracted from the database, for example, the vector of grade one, slight corrosion , and the vector of grade two, moderate corrosion . The weighted Euclidean distance between the measured vector and the two standard vectors is calculated, and the weight coefficients are pre-set according to the importance of each feature parameter. After calculation, the distance between the measured sample and grade one is 45.8, and the distance between the measured sample and grade two is 12.3. Since 12.3 is the minimum distance, the features of the measured sample are closest to the standard of grade two, moderate corrosion. Therefore, the output judgment result is moderate corrosion.

[0035] An embodiment of an electrode foil corrosion condition judgment system provided by the present application: An electrode foil corrosion condition judgment system includes a processor and a memory, and the memory stores computer program instructions, which, when executed by the processor, implement the electrode foil corrosion condition judgment method described above.

[0036] The electrode foil corrosion condition judgment system further includes a communication interface and other components known to those skilled in the art, the settings and functions of which are known in the art, and thus will not be described here.

[0037] In this disclosure, a "storage medium" or "computer readable medium" can be any available medium that can be accessed by a general purpose or special purpose computer system. By way of example, and not limitation, such computer readable media can comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code means in the form of computer readable instructions or data structures and that can be accessed by a general purpose or special purpose computer system, or a combination thereof. When information is transferred or provided over a network or another communications connection (either hardwired, wireless, or combination thereof) to a computer, the computer properly views the connection as a computer readable medium. Thus, any such connection is properly termed a computer readable medium. Combinations of the above should also be included within the scope of the computer readable media.

[0038] The above are preferred embodiments of the present application, not to limit the protection scope of the present application, therefore: any equivalent changes made according to the structure, shape, principle of the present application should be covered within the protection scope of the present application.

Claims

1. A method for judging the corrosion condition of electrode foil, characterized in that, Includes the following steps: S1. Obtain the grayscale image of the electrode foil to be analyzed; construct a structure tensor based on the composite gradient reflecting the local brightness and texture information of the pixel, determine the anisotropic diffusion coefficient according to the structure tensor, and use the anisotropic diffusion coefficient to perform iterative anisotropic diffusion filtering on the grayscale image to obtain the filtered image. S2, the anisotropic diffusion coefficient determined for each pixel is fused with the pixel intensity of the filtered image to obtain a high-contrast erosion saliency map; a mapping function is constructed based on the gray-level distribution of the erosion saliency map, and each pixel of the saliency map is input into the mapping function to obtain a segmentation threshold map of the same size as the filtered image. S3. The filtered image is segmented using an automatic thresholding method to separate all corrosion areas; the area, perimeter, and shape compactness of each corrosion area are calculated as morphological feature parameters; the statistical distribution of the morphological feature parameters of all corrosion areas is matched with the standard corrosion level database in a multidimensional manner to output the corrosion status judgment result of the electrode foil.

2. The method for judging the corrosion status of electrode foil according to claim 1, characterized in that, A structure tensor is constructed based on a composite gradient reflecting the local brightness and texture information of pixels. An anisotropic diffusion coefficient is then determined based on the structure tensor. The grayscale image is then subjected to iterative anisotropic diffusion filtering using this coefficient to obtain the filtered image, including: Calculate the composite gradient of each pixel in the grayscale image. The composite gradient is a weighted sum of the brightness gradient and the texture gradient, with a weight coefficient of 0.5 for each. A 3x3 structure tensor is constructed based on the composite gradient, and the eigenvalues ​​of the structure tensor are calculated. The anisotropic diffusion coefficient is then determined based on the eigenvalues. Perform anisotropic diffusion filtering once and calculate the bimodal separation index of the gray-level histogram of the current gray-level image; Repeat the anisotropic diffusion filtering until the change in the bimodal separation index calculated in three consecutive iterations is less than 0.001, then terminate the iteration.

3. The method for judging the corrosion status of electrode foil according to claim 1, characterized in that, In S3, the global gray-level variance of the corrosion saliency map is calculated. When the global gray-level variance is less than a preset low-fluctuation threshold, an automatic thresholding segmentation method is used on the filtered image to segment all corrosion regions, including: The Otsu method is used to calculate the global segmentation threshold, which maximizes the inter-class variance between the pixel classes in the eroded region and the pixel classes in the background region. Pixels with gray values ​​greater than the segmentation threshold in the filtered image are classified as eroded regions, and pixels with gray values ​​less than or equal to the segmentation threshold are classified as background regions.

4. The method for judging the corrosion status of electrode foil according to claim 1, characterized in that, The area, perimeter, and shape compactness of each corroded region are calculated as morphological characteristic parameters, including: The total number of pixels contained in each segmented eroded region is counted, and this count is taken as the area A of the eroded region. An edge tracking algorithm is used to determine the boundary of each eroded region, and the total number of pixels on the boundary is counted as the perimeter P of the eroded region. According to the formula Calculate the shape compactness C of each eroded region.

5. The method for judging the corrosion status of electrode foil according to claim 1, characterized in that, The statistical distribution of morphological characteristic parameters of all corroded areas is matched in a multidimensional manner with a standard corrosion level database to output the corrosion status assessment results of the electrode foil, including: The mean and standard deviation of the area, perimeter, and shape compactness of all corrosion regions of the electrode foil to be analyzed are calculated to form a 6-dimensional feature vector. Retrieve the standard 6-dimensional feature vector corresponding to each corrosion level from the standard corrosion level database; Calculate the weighted Euclidean distance between the 6-dimensional eigenvector of the electrode foil to be analyzed and the standard eigenvector of each level in the database; The standard level with the smallest weighted Euclidean distance is output as the result of judging the corrosion status of the electrode foil.

6. The method for judging the corrosion status of electrode foil according to claim 1, characterized in that, In S1, the Sobel operator is used to calculate the brightness gradient component of the grayscale image, a set of multi-directional Gabor filters are used to extract the texture features of the grayscale image and calculate its gradient component, and the two are weighted and fused to obtain the composite gradient.

7. The method for judging the corrosion status of electrode foil according to claim 1, characterized in that, In S2, a mapping function is constructed based on the gray-level distribution of the erosion saliency map. The expression of the mapping function is as follows: T(x,y)=μ(x,y)×[1-k×(1-S(x,y))], where T(x,y) is the threshold, μ(x,y) is the local mean, k is the preset modulation coefficient, and S(x,y) is the pixel saliency.

8. The method for judging the corrosion status of electrode foil according to claim 1, characterized in that, In S3, the standard corrosion level database pre-stores several standard levels, each of which corresponds to a standard three-dimensional feature distribution.

9. The method for judging the corrosion status of electrode foil according to claim 1, characterized in that, In S1, a color image is captured of the electrode foil sample to be analyzed, and the color image is converted into an 8-bit grayscale image.

10. A system for judging the corrosion status of electrode foil, characterized in that, It includes a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the electrode foil corrosion determination method according to any one of claims 1-9 is implemented.

Images