Foil cutting abnormality monitoring method based on machine vision

By adjusting the pyramid layer number and scale data, edge detection and feature point screening, the problem that light changes in foil image processing affect feature point extraction is solved, and the stability and accuracy of feature point under different lighting conditions is achieved, and the improvement of feature point stability and accuracy is achieved. The accuracy and robustness of foil cutting abnormality monitoring.

CN118628495BActive Publication Date: 2024-11-05广州优箔良材科技有限公司
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411104360.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-13
Publication Date
2024-11-05
Estimated Expiration
2044-08-13

AI Technical Summary

Technical Problem

The prior art in the foil image processing due to light changes and reflection problems, filtering preprocessing affects the accuracy of feature point extraction, and thus affects the false alarm rate of abnormal monitoring, making it difficult to maintain the stability of feature point under different lighting conditions and accuracy.

Method used

By adjusting the number of pyramid layers and the scale data of each layer, adapting to different lighting conditions, performing edge detection and feature point screening, calculating the linear characteristics and approximate connectivity domain feature performance of the target pixel point, and using SURF feature extraction and analysis to judge foil cutting Whether there is an exception.

Benefits of technology

Effectively detect abnormal situations during the foil cutting process, improve product quality, enhance the robustness and accuracy of the method, adapt to different lighting conditions and foil image scales, and reduce misjudgment of feature points.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118628495B_ABST
    Figure CN118628495B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of image processing, and more specifically, to a method for monitoring foil cutting anomalies based on machine vision, the method comprising: obtaining a foil image after cutting and a standard foil image; adjusting the number of pyramid layers based on the degree of illumination change of the foil image, performing edge detection on the foil image corresponding to each pyramid layer, obtaining and screening the edge detection results, obtaining target pixel points, calculating the degree of expression of linear features corresponding to the target pixel points and the degree of expression of approximate connected domain features corresponding to the target pixel points, and obtaining scale data of each pyramid layer when extracting feature points from the foil image; performing SURF feature extraction on the foil image according to the number of pyramid layers and the scale data corresponding to each pyramid layer, and judging whether there is anomaly in foil cutting. The present invention adapts to the feature extraction requirements under different illumination conditions by adjusting the number of pyramid layers or the scale of each layer to ensure more accurate capture of local features in the image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of image processing, and more specifically, to a method for monitoring foil cutting abnormalities based on machine vision. Background Art

[0002] With the rise of electric vehicles, renewable energy and portable electronics, the demand for batteries has increased dramatically. Therefore, efficiency and quality issues in battery production have become particularly critical. The positive and negative electrodes of the battery are made of foil (aluminum foil or copper foil), and the quality of the foil directly affects the performance and life of the battery. The cutting process of the foil needs to be precisely controlled to ensure dimensional consistency and surface integrity.

[0003] The existing Chinese patent application document with publication number CN116051549A discloses a method, system, medium and equipment for defect segmentation of solar cell. The method comprises the following steps: obtaining a defect image of a solar cell and performing preprocessing; cutting and encoding the preprocessed defect image based on a hierarchical encoder, extracting feature information of multiple levels of the preprocessed defect image, and obtaining a feature map with different resolutions from the original image, namely, a defect multi-level feature map; decoding the defect multi-level feature map by a decoder through linear operations and upsampling operations to obtain a decoded defect multi-level feature map; fusing high-resolution information with low-resolution information in the decoded defect multi-level feature map by a hierarchical residual method to obtain a defect image to be segmented; and performing defect segmentation based on the defect image to be segmented to obtain a segmented defect image of the solar cell.

[0004] In the above scheme, image processing is performed by machine vision to obtain the segmented battery cell defect image. However, in the process of foil image acquisition, since the foil surface has undergone special treatment, such as polishing or coating, reflection or illumination changes may occur in the foil image. Usually, the image is pre-processed by filtering to blur the illuminated area. However, the filtering effect will also affect the accuracy of the final feature point extraction result, thereby affecting the false alarm rate of abnormal monitoring. Summary of the invention

[0005] In order to solve the problem of reflection or illumination change in foil image, filtering preprocessing is used to blur the illumination area, but the filtering effect still affects the accuracy of feature point extraction results. The present invention provides solutions in the following aspects.

[0006] The method for monitoring foil cutting abnormality based on machine vision comprises: obtaining a foil image after cutting and a standard foil image; adjusting the number of pyramid layers based on the degree of illumination change of the foil image, wherein the number of pyramid layers refers to the number of layers into which the foil image is decomposed; performing edge detection on the foil image corresponding to each pyramid layer, obtaining and screening the edge detection results to obtain target pixel points, calculating the expression degree of linear features corresponding to the target pixel points and the expression degree of approximate connected domain features corresponding to the target pixel points, and obtaining scale data of each pyramid layer when extracting feature points of the foil image; performing SURF feature extraction on the foil image according to the number of pyramid layers and the scale data corresponding to each pyramid layer, and analyzing according to the feature points to determine whether there is an abnormality in foil cutting; wherein the scale data of each pyramid layer , where Indicates the scale data corresponding to each layer of the pyramid when extracting feature points from the foil image. Indicates the degree of representation of the target pixel corresponding to the approximate connected domain. Indicates the degree of linear feature expression corresponding to the target pixel point. Indicates The exponential function with base , To preset hyperparameters, is the ceiling function.

[0007] The effect is that by obtaining the expression degree of the linear features and approximate connected domain features corresponding to the target pixel points, the scale data of each layer of the pyramid is obtained, so as to perform SURF feature extraction and feature point analysis, which can effectively detect abnormal situations in the foil cutting process, improve product quality, adapt to different lighting conditions and foil image scales, and improve the robustness, adaptability and accuracy of the method.

[0008] Preferably, adjusting the number of pyramid layers based on the degree of illumination change of the foil image includes:

[0009] The grayscale difference value of each row in the obtained foil material image is accumulated and summed as the row difference of the surface image, and the grayscale difference value of each column in the obtained foil material image is accumulated and summed as the column difference of the surface image;

[0010] The difference between the row difference of each row in the foil material image and the row difference of two adjacent rows is used as the row difference of the corresponding row, and the difference between the column difference of each column in the foil material image and the column difference of two adjacent columns is used as the column difference of the corresponding column;

[0011] A ratio of the row difference cumulative sum of all rows to the column difference cumulative sum of all columns in the foil image is calculated, and in response to a larger ratio, the number of pyramid levels is smaller.

[0012] The effects are as follows: by calculating the grayscale difference value of each row and column in the foil image, as well as the difference value between the row difference and the column difference, the degree of illumination change of the image can be evaluated; by comparing the ratio of the cumulative sum of row differences to the cumulative sum of column differences, the size of the pyramid layers can be determined; it is beneficial to dynamically adjust the number of pyramid layers according to the illumination conditions of the image, thereby improving the efficiency and accuracy of image processing.

[0013] Preferably, the number of pyramid layers satisfies the following relationship:

[0014] ,

[0015] In the formula, Indicates the number of pyramid layers when extracting feature points from foil images. Indicates The row-by-row differences, Indicates Column by column differences, represents the length of the foil image, Indicates the width of the foil image, Indicates The exponential function with base , represents the preset hyperparameters, is the ceiling function.

[0016] Preferably, obtaining edge detection results and screening to obtain target pixel points includes:

[0017] The edge detection result is screened to remove edge pixel points located at the edge of the foil material image, and only edge pixel points inside the foil material image are retained, and the edge pixel points inside the foil material image are used as target pixel points.

[0018] Preferably, calculating the expression degree of the linear feature corresponding to the target pixel point includes:

[0019] Select any target pixel point as the first target pixel point, obtain the target pixel point closest to the first target pixel point and record it as the second target pixel point;

[0020] Obtain the gradient direction difference between the first target pixel and the second target pixel, repeat this step to obtain the gradient direction difference between the second target pixel and the corresponding nearest target pixel, until all target pixels are traversed, obtain the mean of all gradient direction differences, and use the mean as the degree of expression of the linear feature corresponding to the target pixel.

[0021] The effect is that by analyzing the difference in gradient direction between the target pixel and its nearest neighboring pixel, it can not only reveal subtle changes and local details in the image, such as edges, textures, and shapes, but also enhance the acquisition of the overall structure of the image, playing a key role in computer vision tasks such as image recognition, classification, and segmentation.

[0022] Preferably, calculating the expression degree of the linear feature corresponding to the target pixel point further includes:

[0023] Calculate the gradients in the horizontal and vertical directions of the foil image, set a local window for the foil image, and construct a structure tensor by calculating the sum of the grayscale values ​​of all pixels in the local window and the sum of the squares of the grayscale values ​​of all pixels in the local window; wherein the structure tensor is a 2×2 matrix;

[0024] Obtain the eigenvalue and eigenvector of the structure tensor corresponding to each target pixel, where the eigenvalue reflects the strength of the structure information and the eigenvector represents the information direction of the structure;

[0025] The expression degree of the linear feature corresponding to the target pixel point is obtained according to the feature value and the information direction.

[0026] The effect is that by analyzing the eigenvalues ​​and eigenvectors of the structure tensor, structural information in the image, such as edges, textures, etc., can be effectively extracted.

[0027] Preferably, obtaining the degree of expression of the target pixel point corresponding to the approximate connected domain feature includes:

[0028] Select any target pixel point as the first target pixel point, and in response to the existence of other target pixel points in the gradient direction of the first target pixel point, repeat this step until there are no other target pixel points in the gradient direction, and then record the target pixel point as the last target pixel point;

[0029] The distance between the first target pixel and the last target pixel is taken as the degree of representation of the target pixel to the approximate connected domain feature.

[0030] The effect is that by evaluating the connectivity and directional characteristics of target pixels in the image, it helps to identify significant structural changes in the foil image, thereby improving the accuracy of feature point detection and further improving detection efficiency.

[0031] Preferably, the analyzing according to the characteristic points to determine whether there is an abnormality in the foil cutting includes:

[0032] Analyze the feature points in the foil image with the feature points in the standard template, and compare the parameters of the position and quantity of the feature points to determine whether the current cutting state is abnormal;

[0033] In response to the average of the accumulated sum of the position differences between the feature points in the foil image and the corresponding feature points in the standard template exceeding the abnormal monitoring threshold, the alarm mechanism is immediately triggered.

[0034] The present invention has the following effects:

[0035] 1. The present invention adapts to the feature extraction requirements under different lighting conditions by adjusting the number of pyramid layers or the scale of each layer to ensure more accurate capture of local features in the image and retain more detail information. It can also maintain the stability and accuracy of feature points under large changes in lighting and avoid excessive information loss.

[0036] 2. The present invention analyzes the illumination changes of the foil image and adjusts the appropriate number of pyramid layers and scale data, so as to avoid the influence of illumination in the process of extracting image feature points, thereby reducing the misjudgment of feature points and further improving the accuracy of the final foil cutting abnormality monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0038] Figure 1 It is a method flow chart of steps S1 to S4 in the method for monitoring foil cutting abnormality based on machine vision in an embodiment of the present invention.

[0039] Figure 2 It is a method flow chart of steps S20 to S22 in the method for monitoring foil cutting abnormality based on machine vision in an embodiment of the present invention.

[0040] Figure 3 It is a method flow chart of steps S30 to S34 in the method for monitoring foil cutting abnormality based on machine vision in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0042] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0043] It should be noted that the specific scenario targeted by this invention is: foil cutting detection for the production of positive and negative electrodes of lithium-ion batteries, and foil quality detection and screening to ensure the performance and life of the battery.

[0044] Reference Figure 1 The foil cutting abnormality monitoring method based on machine vision includes steps S1 to S4, which are as follows:

[0045] S1: Obtain the cut foil image and the standard foil image.

[0046] Further, a number of pieces of cut foil are randomly selected, and images of each piece of foil are collected separately. At the same time, a standard template of the same size as the collected foil is obtained under the same shooting environment and used as the standard foil image.

[0047] It should be noted that the present technical solution combines machine vision for foil image detection, which improves detection efficiency compared to traditional detection methods. However, when detecting through machine vision, feature point extraction is required. Since the edge of the illuminated area may be extracted as a feature point, it will affect the judgment of whether there is an abnormality in foil cutting when matching the feature point with the feature point of the standard template; and the change of illumination will cause the grayscale distribution in the image to change. In the case of drastic changes in illumination, fewer pyramid layers indicate stable features, that is, the retention of features in the non-illuminated area. At the same time, compared with the changes in the non-illuminated area, the illumination change is slightly more complex, and the scale data in each layer of the pyramid determines how many different features can be extracted at that scale. In other words, smaller scale data can avoid extracting complex types of features at the same scale, and the characteristics of the foil itself are retained. Therefore, it is necessary to adjust the number of pyramid layers according to the degree of illumination change of the foil image.

[0048] S2: adjusting the number of pyramid levels based on the degree of illumination change of the foil image, wherein the number of pyramid levels refers to the number of levels into which the foil image is decomposed.

[0049] To further explain, the pyramid in the image is a well-known technology in the art and will not be described in detail.

[0050] Specifically, refer to Figure 2 ,The steps for obtaining the number of pyramid layers when extracting features from foil images are as follows:

[0051] S20: accumulating and summing the grayscale difference value of each row in the obtained foil material image as the row difference of the surface image, and accumulating and summing the grayscale difference value of each column in the obtained foil material image as the column difference of the surface image;

[0052] Specifically, the row difference of the foil image satisfies the following relationship:

[0053] ,

[0054] In the formula, Indicates Row-wise differences of rows, Indicates the foil image Line The gray level of each pixel, Indicates the foil image Line The gray level of the pixel adjacent to the right of the pixel, Indicates the length of the foil image.

[0055] The row difference and column difference of the foil image are obtained in the same manner, which will not be described in detail here.

[0056] S21: obtaining a difference between a row difference of each row in the foil material image and a row difference between two adjacent rows as a row difference of the corresponding row, and obtaining a difference between a column difference of each column in the foil material image and a column difference between two adjacent columns as a column difference of the corresponding column;

[0057] Specifically, the row difference of the foil image satisfies the following relationship:

[0058] ,

[0059] In the formula, Indicates The row-by-row differences, Indicates Row-wise differences of rows, Indicates Row-wise differences of rows, Indicates Row-wise differences of rows.

[0060] The row difference and column difference of the foil material image are obtained in the same manner, which will not be described in detail herein.

[0061] S22: Calculate the ratio of the cumulative sum of row differences of all rows to the cumulative sum of column differences of all columns in the foil material image, and in response to the larger the ratio, the smaller the number of pyramid levels.

[0062] Specifically, the number of pyramid layers satisfies the following relationship:

[0063] ,

[0064] In the formula, Indicates the number of pyramid layers when extracting feature points from foil images. Indicates The row-by-row differences, Indicates Column by column differences, represents the length of the foil image, Indicates the width of the foil image, Indicates The exponential function with base , represents the preset hyperparameters, is the ceiling function.

[0065] In this embodiment , due to The result of the negative correlation exponential function with the base as , by presetting the hyperparameters Control the value result in , and the final upward rounding function ensures that the final number of layers is , to prevent the number of pyramid layers from being too many and affecting the operating efficiency of the monitoring system.

[0066] To further explain, since the scale data of each layer of the pyramid determines the diversity and accuracy of the captured features, the more obvious the illumination area in the image corresponding to this layer of the pyramid is, the smaller the scale data corresponding to this layer of the pyramid is, thus avoiding the extraction of complex illumination features at the same scale and retaining the characteristics of the foil itself.

[0067] Specifically, the illumination will have a distinct shape distribution on the image, some of which are in the shape of bands or approximate connected domains. In other words, in the image corresponding to the pyramid layer, the distribution of edge pixels corresponds to linear features or the approximate connected domain features are better, the more obvious the illumination area in the image corresponding to the pyramid layer is, and the smaller the scale data corresponding to the pyramid layer is.

[0068] S3: Perform edge detection on the foil image corresponding to each layer of the pyramid, obtain the edge detection results and screen them to obtain the target pixel points, calculate the expression degree of the linear features corresponding to the target pixel points and the expression degree of the approximate connected domain features corresponding to the target pixel points, and obtain the scale data of each layer of the pyramid when extracting feature points from the foil image.

[0069] In this embodiment, the Sobel operator is used to perform edge detection on the foil image.

[0070] Specifically, refer to Figure 3 , the steps for obtaining scale data of each layer of pyramid are as follows:

[0071] S30: screening the edge detection result, filtering out edge pixel points located at the edge of the foil material image, retaining only edge pixel points inside the foil material image, and taking the edge pixel points inside the foil material image as target pixel points.

[0072] S31: Select any target pixel point as a first target pixel point, and obtain the target pixel point closest to the first target pixel point as a second target pixel point;

[0073] S32: Obtain the gradient direction difference between the first target pixel and the second target pixel, repeat this step to obtain the gradient direction difference between the second target pixel and the corresponding nearest target pixel, until all target pixels are traversed, obtain the mean of all gradient direction differences, and use the mean as the expression degree of the linear feature corresponding to the target pixel.

[0074] Specifically, the degree of expression of the linear feature corresponding to the target pixel satisfies the following relationship:

[0075] ,

[0076] In the formula, Indicates the degree of linear feature expression corresponding to the target pixel point. Indicates the first The gradient direction of the target pixel, Indicates the first The gradient direction of the target pixel corresponding to the nearest target pixel is is the number of target pixels.

[0077] In addition, in another embodiment, the expression degree of the linear feature corresponding to the target pixel is obtained through the structure tensor of the foil image.

[0078] Calculate the gradients in the horizontal and vertical directions of the foil image, set a local window for the foil image, and construct a structure tensor by calculating the sum of the grayscale values ​​of all pixels in the local window and the sum of the squares of the grayscale values ​​of all pixels in the local window; wherein the structure tensor is a 2×2 matrix;

[0079] Further explanation, the size of the local window is set to 3×3.

[0080] Obtain the eigenvalue and eigenvector of the structure tensor corresponding to each target pixel, where the eigenvalue reflects the strength of the structure information and the eigenvector represents the information direction of the structure;

[0081] According to the eigenvalue and information direction, the expression degree of the linear feature corresponding to the target pixel point is obtained.

[0082] S33: Select any target pixel point as the first target pixel point, and in response to the existence of other target pixel points in the gradient direction of the first target pixel point, repeat this step until there are no other target pixel points in the gradient direction, and record the target pixel point as the last target pixel point;

[0083] S34: taking the distance between the first target pixel and the last target pixel as the degree of representation of the target pixel for the approximate connected domain feature.

[0084] Specifically, the degree of representation of the target pixel point corresponding to the approximate connected domain feature satisfies the following relationship:

[0085] ,

[0086] In the formula, Indicates the degree of representation of the target pixel corresponding to the approximate connected domain. Indicates the first target pixel The location coordinates of Represents the tail target pixel The location coordinates of .

[0087] Specifically, the scale data of each layer of the pyramid satisfies the following relationship:

[0088] ,

[0089] In the formula, Indicates the scale data corresponding to each layer of the pyramid when extracting feature points from the foil image. Indicates the degree of representation of the target pixel corresponding to the approximate connected domain. Indicates the degree of linear feature expression corresponding to the target pixel point. Indicates The exponential function with base , To preset hyperparameters, is the ceiling function.

[0090] In this embodiment, ,in, The results are used to control the scale data within , while avoiding the scale data being 0, it also avoids the scale data being too large, which affects the operating efficiency.

[0091] S4: According to the number of pyramid layers and the corresponding scale data of each pyramid layer, SURF features are extracted from the foil image, and analysis is performed based on the feature points to determine whether there is any abnormality in the foil cutting.

[0092] Analyze the feature points in the foil image with the feature points in the standard template, and compare the parameters of the position and quantity of the feature points to determine whether the current cutting state is abnormal;

[0093] In response to the average of the accumulated sum of the position differences between the feature points in the foil image and the corresponding feature points in the standard template exceeding the abnormal monitoring threshold, the alarm mechanism is immediately triggered.

[0094] Further explanation, the anomaly monitoring threshold is set to 10.

[0095] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0096] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for monitoring foil cutting anomalies based on machine vision, characterized in that: include: Acquire the cut foil image and the standard foil image; Adjusting the number of pyramid layers based on the degree of illumination change of the foil image, wherein the number of pyramid layers refers to the number of levels into which the foil image is decomposed; Perform edge detection on the foil image corresponding to each layer of the pyramid, obtain and screen the edge detection results to obtain target pixels, calculate the expression degree of the linear features corresponding to the target pixels and the expression degree of the approximate connected domain features corresponding to the target pixels, and obtain the scale data of each layer of the pyramid when extracting feature points from the foil image; According to the number of pyramid layers and the scale data corresponding to each pyramid layer, SURF feature extraction is performed on the foil image, and analysis is performed based on the feature points to determine whether there is an abnormality in the foil cutting; Among them, the scale data of each layer of the pyramid , where Indicates the scale data corresponding to each layer of the pyramid when extracting feature points from the foil image. Indicates the degree of representation of the target pixel corresponding to the approximate connected domain. Indicates the degree of linear feature expression corresponding to the target pixel point. Indicates The exponential function with base , To preset hyperparameters, is the ceiling function.

2. The method for monitoring foil cutting abnormality based on machine vision according to claim 1, characterized in that: The step of adjusting the number of pyramid layers based on the degree of illumination change of the foil image includes: The grayscale difference value of each row in the obtained foil material image is accumulated and summed as the row difference of the surface image, and the grayscale difference value of each column in the obtained foil material image is accumulated and summed as the column difference of the surface image; The difference between the row difference of each row in the foil material image and the row difference of two adjacent rows is used as the row difference of the corresponding row, and the difference between the column difference of each column in the foil material image and the column difference of two adjacent columns is used as the column difference of the corresponding column; A ratio of the row difference cumulative sum of all rows to the column difference cumulative sum of all columns in the foil image is calculated, and in response to a larger ratio, the number of pyramid levels is smaller.

3. The method for monitoring foil cutting abnormality based on machine vision according to claim 2, characterized in that: The number of pyramid layers satisfies the following relationship: , In the formula, Indicates the number of pyramid layers when extracting feature points from foil images. Indicates The row-by-row differences, Indicates Column by column differences, represents the length of the foil image, Indicates the width of the foil image, Indicates The exponential function with base , represents the preset hyperparameters, is the ceiling function.

4. The method for monitoring foil cutting abnormality based on machine vision according to claim 1, characterized in that: Obtain edge detection results and filter them to obtain target pixels, including: The edge detection result is screened to remove edge pixel points located at the edge of the foil material image, and only edge pixel points inside the foil material image are retained, and the edge pixel points inside the foil material image are used as target pixel points.

5. The method for monitoring foil cutting abnormality based on machine vision according to claim 1, characterized in that: Calculating the expression degree of the linear feature corresponding to the target pixel point includes: Select any target pixel point as the first target pixel point, obtain the target pixel point closest to the first target pixel point and record it as the second target pixel point; Obtain the gradient direction difference between the first target pixel and the second target pixel, repeat this step to obtain the gradient direction difference between the second target pixel and the corresponding nearest target pixel, until all target pixels are traversed, obtain the mean of all gradient direction differences, and use the mean as the degree of expression of the linear feature corresponding to the target pixel.

6. The method for monitoring foil cutting abnormality based on machine vision according to claim 1, characterized in that: Calculating the expression degree of the linear feature corresponding to the target pixel point also includes: Calculate the gradients in the horizontal and vertical directions of the foil image, set a local window for the foil image, and construct a structure tensor by calculating the sum of the grayscale values ​​of all pixels in the local window and the sum of the squares of the grayscale values ​​of all pixels in the local window; wherein the structure tensor is a 2×2 matrix; Obtain the eigenvalue and eigenvector of the structure tensor corresponding to each target pixel, where the eigenvalue reflects the strength of the structure information and the eigenvector represents the information direction of the structure; The expression degree of the linear feature corresponding to the target pixel point is obtained according to the feature value and the information direction.

7. The method for monitoring foil cutting abnormality based on machine vision according to claim 1, characterized in that: Obtaining the degree of expression of the target pixel point corresponding to the approximate connected domain feature includes: Select any target pixel point as the first target pixel point, and in response to the existence of other target pixel points in the gradient direction of the first target pixel point, repeat this step until there are no other target pixel points in the gradient direction, and then record the target pixel point as the last target pixel point; The distance between the first target pixel and the last target pixel is taken as the degree of representation of the target pixel to the approximate connected domain feature.

8. The method for monitoring foil cutting abnormality based on machine vision according to claim 1, characterized in that: The analysis based on the characteristic points to determine whether there is an abnormality in foil cutting includes: Analyze the feature points in the foil image with the feature points in the standard template, and compare the parameters of the position and quantity of the feature points to determine whether the current cutting state is abnormal; In response to the average of the accumulated sum of the position differences between the feature points in the foil image and the corresponding feature points in the standard template exceeding the abnormal monitoring threshold, the alarm mechanism is immediately triggered.

Citation Information

Patent Citations

  • Solar cell defect segmentation method, system, medium and equipment

    CN116051549A

  • Vehicle detecting method based on Gauss difference multi-scale edge fusion

    CN103927526A

  • Intelligent detection method for surface defects of power adapter

    CN118014989A

  • Digital processing method and system for determination of optical flow

    GB201400875D0