Photovoltaic cell defect detection method and system based on image processing

Abstract

The present application relates to the field of image processing technology, and in particular to a photovoltaic cell defect detection method and system based on image processing, the method comprising: performing bit stratification on the grayscale image of the photovoltaic cell to obtain eight bit images; calculating the amount of information of any pixel in each bit image to obtain an information amount sequence, and clustering the pixel points according to the information amount sequence to obtain multiple target clustering clusters; calculating the recombination vector of each target clustering cluster, the recombination vector including the recombination coefficient of each bit image; weighted summing the bit image according to the recombination vector to obtain a recombined image; and comparing the grayscale image and the recombined image to obtain the defect detection result. The technical solution of the present application can improve the accuracy of photovoltaic cell defect detection.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a photovoltaic cell defect detection method and system based on image processing. Background Art

[0002] As a clean energy, solar energy is green, environmentally friendly and renewable. It can effectively alleviate the dependence of the economy and society on fossil energy and solve the increasing shortage of traditional fossil energy and the series of environmental problems it causes. Photovoltaic cells can convert solar energy into electrical energy and provide power for power equipment. The quality of photovoltaic cells directly affects the efficiency of converting solar energy into electrical energy.

[0003] At present, the patent application document with publication number CN112991264A discloses a method for detecting crack defects in monocrystalline silicon photovoltaic cells, the method comprising: performing morphological opening operations and homomorphic filtering processing on the acquired defect-free monocrystalline silicon photovoltaic cell image in sequence to obtain a sample image; extracting edge points on the sample image to obtain multiple edge points; extracting features on each edge point to obtain a set of feature vectors corresponding to each edge point; performing extreme value clustering on multiple sets of feature vectors to obtain clustering results; taking the average value of all feature vectors contained in each class in the clustering results as the cluster center of the class; calculating the Euclidean distance from each edge point in each class to the cluster center, and calculating the distance threshold corresponding to the cluster according to the Euclidean distance; obtaining multiple sets of feature vectors corresponding to the image to be detected, and calculating the Euclidean distance between each set of feature vectors and all cluster centers corresponding to the sample image, to obtain multiple Euclidean distance values; comparing each obtained Euclidean distance value with all distance thresholds, and determining the type of edge points corresponding to each set of feature vectors according to the comparison results, thereby obtaining the defective part of the image to be detected.

[0004] The above method clusters the edge point features of defect-free monocrystalline silicon photovoltaic cell images, and finds defect points by judging whether the edge points of the image to be inspected are in all clusters; however, the above method requires obtaining defect-free monocrystalline silicon photovoltaic cell images in advance, and the accuracy of defect-free monocrystalline silicon photovoltaic cell images directly affects the detection accuracy, resulting in low defect detection accuracy. Summary of the invention

[0005] In order to solve the technical problem of low accuracy in photovoltaic cell defect detection, the present application provides a photovoltaic cell defect detection method and system based on image processing, which can obtain defect detection results without pre-acquiring defect-free photovoltaic cell images, thereby improving the accuracy of photovoltaic cell defect detection.

[0006] In a first aspect, the present application provides a photovoltaic cell defect detection method based on image processing, the method comprising: performing bit stratification on the grayscale image of the photovoltaic cell to obtain eight bit images; calculating the amount of information of any pixel in each bit image to obtain an information amount sequence, and clustering the pixel points according to the information amount sequence to obtain multiple target clustering clusters; calculating the recombination vector of each target clustering cluster, the recombination vector including the recombination coefficient of each bit image, the target clustering cluster Medium bit image The recombination coefficient for: , The target cluster Each pixel in the bit image The information variance of The target cluster The information variance of each pixel in all bit images is summed; the bit images are weighted and summed according to the reorganization vector to obtain the reorganized pixel value of each pixel in the corresponding target cluster; the reorganized image is constructed according to the reorganized pixel values ​​of all pixels in the grayscale image; the defect detection result is obtained by comparing the grayscale image and the reorganized image.

[0007] Firstly, the grayscale image of the photovoltaic cell is split into eight bit images according to the bit layering algorithm, the information content of each pixel in each bit image is calculated, and the information content sequence of each pixel is constructed. Then, each pixel is clustered according to the information content sequence to obtain multiple target clustering clusters; the information content variance of each bit image in the target clustering cluster is calculated. If the information content variance of the bit image is large, it means that for each pixel in the target clustering cluster, the bit image has defect information or noise information. Therefore, the recombination coefficient of the bit image in the recombination vector of the target clustering cluster is small, ensuring that the recombined pixel value of each pixel in the target clustering cluster does not contain defect information or noise information; each target clustering cluster realizes adaptive reconstruction of the grayscale image according to the corresponding recombination vector, so that the recombined image does not contain defect information or noise information; accurate defect detection results can be obtained by comparing the grayscale image and the recombined image, thereby improving the accuracy of defect detection of photovoltaic cells.

[0008] Preferably, the pixel The reconstructed pixel value for: , is the number of bit images, Pixel Target cluster The bit image in the reorganized vector The recombination coefficient, Bit Image Medium pixel The pixel value of .

[0009] Based on pixels The reorganization vector of the target cluster weights the information of each bit image to obtain the pixel point The reconstructed pixel value of There are defective information or noise information, bit image The recombination coefficient Small, ensure pixel The reconstructed pixel value Does not contain defect information or noise information.

[0010] Preferably, the grayscale image of the photovoltaic cell is bit-layered to obtain eight bit images, including: obtaining a binary representation of the grayscale value of any pixel in the grayscale image, the binary representation including 8 bits, each bit corresponding to a bit image; and using the value of each bit in the binary representation as the pixel value of the pixel in the corresponding bit image.

[0011] Preferably, the calculating the amount of information of any pixel point in each bit image to obtain an information amount sequence includes: calculating the image entropy within the neighborhood range of the pixel point in the bit image to obtain the amount of information of each pixel point in the bit image; arranging the amount of information of any pixel point in each bit image in a preset order to obtain the information amount sequence of the pixel point.

[0012] By calculating the amount of information of the pixels of the grayscale image in each bit image, an information sequence of each pixel is constructed. This information sequence can reflect the change in the amount of information of the pixel in each bit image, providing a data basis for the subsequent clustering process and the calculation of the recombination coefficient.

[0013] Preferably, the image entropy is one-dimensional image entropy or two-dimensional image entropy; the pixel point neighborhood range is a rectangular range of a preset size centered on the pixel point.

[0014] Preferably, the pixel points are clustered according to the information sequence to obtain multiple target clusters, including: performing Kmeans clustering on each pixel point according to the initial K value to obtain an initial cluster, and calculating the contour coefficient of the initial K value, wherein the clustering distance in the Kmeans clustering process is positively correlated with the Euclidean distance of the information sequence between the pixel points; adjusting the initial K value to draw an effect curve, wherein the abscissa of the effect curve is the initial K value, and the ordinate is the contour coefficient corresponding to the initial K value; taking the initial K value corresponding to the inflection point in the effect curve as the target K value, and taking the initial cluster of the target K value as the target cluster.

[0015] There are multiple regions in the grayscale image of photovoltaic cells, such as the grid line region and the cell region, and the information sequence of the pixels in the same region is similar. Therefore, clustering the pixels according to the information sequence can realize unsupervised classification of each pixel in the grayscale image, and one target cluster corresponds to one region.

[0016] Preferably, the pixel and pixels The clustering distance for:

[0017] , and Pixel and pixels The information sequence of and Pixels and pixels The coordinate position in the grayscale image, is the balance coefficient.

[0018] Preferably, the comparing the grayscale image and the recombined image to obtain the defect detection result comprises: calculating a difference map between the grayscale image and the recombined image; and performing threshold segmentation on the difference map after filtering to obtain the defect detection result.

[0019] The reconstructed image does not contain defect information and noise information. By comparing the grayscale image and the reconstructed image, the defect information and noise information in the grayscale image can be located. The difference image contains both defect information and noise information. After removing the noise information in the difference image through Gaussian filtering or median filtering, the defect information in the grayscale image can be accurately located to obtain accurate defect detection results.

[0020] In a second aspect of the present application, there is also provided a photovoltaic cell defect detection system based on image processing, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the photovoltaic cell defect detection method based on image processing according to the first aspect of the present application is implemented.

[0021] The technical solution of this application has the following beneficial technical effects:

[0022] Firstly, the grayscale image of the photovoltaic cell is split into eight bit images according to the bit layering algorithm, the information content of each pixel in each bit image is calculated, and the information content sequence of each pixel is constructed. Then, each pixel is clustered according to the information content sequence to obtain multiple target clusters, so that the difference in the information content sequence of each pixel in a target cluster is small, and the difference in the information content sequence of each pixel between different target clusters is large; each pixel in a target cluster belongs to the same area of ​​the photovoltaic cell, and the information content variance of each bit image in the target cluster is calculated. If the information content variance of the bit image is large, it means that the target cluster is correct. For each pixel point in the target cluster, the bit image contains defect information or noise information. Therefore, the recombination coefficient of the bit image in the recombination vector of the target cluster is small, ensuring that the recombined pixel value of each pixel point in the target cluster does not contain defect information or noise information; in this way, each target cluster realizes adaptive reconstruction of the grayscale image according to the corresponding recombination vector, so that the recombined image does not contain defect information or noise information; by comparing the grayscale image and the recombined image to locate the defect information or noise information, and after filtering and denoising, accurate defect detection results can be obtained, thereby improving the accuracy of defect detection of photovoltaic cells. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present application will become easy to understand. In the accompanying drawings, several embodiments of the present application are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0024] Figure 1 is a flow chart of a photovoltaic cell defect detection method based on image processing according to an embodiment of the present application;

[0025] Figure 2 It is a structural block diagram of a photovoltaic cell defect detection system based on image processing according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0027] It should be understood that when the terms "first", "second", etc. are used in the claims, specification and drawings of the present application, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the specification and claims of the present application indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their collections.

[0028] According to a first aspect of the present application, the present application provides a photovoltaic cell defect detection method based on image processing. Figure 1 1 is a flow chart of a photovoltaic cell defect detection method based on image processing according to an embodiment of the present application. Figure 1 As shown, the photovoltaic cell defect detection method based on image processing includes steps S101 to S105, which are described in detail below.

[0029] S101, performing bit stratification on the grayscale image of the photovoltaic cell to obtain eight bit images.

[0030] In one embodiment, a grayscale image of a photovoltaic cell is collected, and a bit layering algorithm is used to perform bit layering on the grayscale image to obtain eight bit images.

[0031] Specifically, the grayscale image of the photovoltaic cell is bit-layered to obtain eight bit images, including: obtaining a binary representation of the grayscale value of any pixel in the grayscale image, the binary representation includes 8 bits, and each digit corresponds to a bit image; and using the value of each digit in the binary representation as the pixel value of the pixel in the corresponding bit image.

[0032] For example, the pixel point in the grayscale image The gray value of is 121, and its binary representation is 11110001. Bit image 8 corresponds to the first digit in the binary table, and bit image 1 corresponds to the last digit in the binary table. In bit image 1 to bit image 8, the pixel point The pixel values ​​are 1, 0, 0, 0, 1, 1, 1, 1.

[0033] At this point, the grayscale image of the photovoltaic cell is split into eight bit images, and each bit image stores a different grayscale value range.

[0034] It should be noted that, in order to improve the accuracy of photovoltaic cell defect detection, before bit stratification of the grayscale image of the photovoltaic cell, the method further includes: performing illumination homogenization processing on the grayscale image using a Retinex algorithm.

[0035] The Retinex algorithm is a single-scale Retinex algorithm, a multi-scale Retinex algorithm or a multi-scale adaptive Retinex algorithm. The Retinex algorithm can eliminate the influence of uneven illumination on image features and is a well-known technology for those skilled in the art.

[0036] S102, calculating the information amount of any pixel point in each bit image to obtain an information amount sequence, and clustering the pixel points according to the information amount sequence to obtain a plurality of target clusters.

[0037] In one embodiment, the size of the eight bit images is the same as the grayscale image of the photovoltaic cell, that is, the pixel points in the grayscale image are The corresponding pixel point can be found in each bit image.

[0038] The method of calculating the amount of information of any pixel in each bit image to obtain an information amount sequence includes: calculating the image entropy within the neighborhood range of the pixel in the bit image to obtain the amount of information of each pixel in the bit image; arranging the amount of information of any pixel in each bit image in a preset order to obtain the information amount sequence of the pixel.

[0039] The information amount sequence can reflect the information amount change of the corresponding pixel in different bit images, the image entropy is one-dimensional image entropy or two-dimensional image entropy, and the pixel neighborhood range is a rectangular range with a preset size centered on the pixel, and the preset size is 3×3, 5×5 or 7×7. Exemplarily, the image entropy is one-dimensional image entropy, then the pixel The amount of information in a bit image 1 for:

[0040] , is the pixel in bit image 1 The probability that a pixel value within the neighborhood is 1.

[0041] In one embodiment, there are multiple regions such as grid line regions and cell regions in the grayscale image of a photovoltaic cell, and the information sequence of pixels in the same region is similar. Therefore, by clustering the pixels according to the information sequence, unsupervised classification of the pixels can be achieved, and one target cluster corresponds to one region.

[0042] Specifically, the pixel points are clustered according to the information sequence to obtain multiple target clustering clusters, including: performing Kmeans clustering on each pixel point according to the initial K value to obtain the initial clustering cluster, and calculating the contour coefficient of the initial K value, wherein the clustering distance in the Kmeans clustering process is positively correlated with the Euclidean distance of the information sequence between the pixel points; adjusting the initial K value to draw an effect curve, wherein the abscissa of the effect curve is the initial K value, and the ordinate is the contour coefficient corresponding to the initial K value; taking the initial K value corresponding to the inflection point in the effect curve as the target K value, and taking the initial clustering cluster of the target K value as the target clustering cluster.

[0043] The initial K value is 2, and each adjustment of the initial K value is a process of adding 1 to the initial K value. The cluster distance is equal to the Euclidean distance of the information sequence.

[0044] In other embodiments, the pixel point and pixels The clustering distance for:

[0045] , and Pixel and pixels The information sequence of and Pixels and pixels The coordinate position in the grayscale image, is the balance coefficient. In general, the value of the balance coefficient is greater than 0.5. In this embodiment, .

[0046] In this way, each pixel in the grayscale image is clustered according to the information amount sequence of the pixel in different bit images, and the grayscale image of the photovoltaic cell is divided into multiple regions, and one target cluster corresponds to one region; the regions include cell regions, gate line regions, etc. In the absence of defects or noise, the grayscale values ​​in the same region should be consistent.

[0047] S103, calculating a recombination vector of each target cluster, wherein the recombination vector includes a recombination coefficient of each bit image, and the recombination coefficient is negatively correlated with the information amount variance of each pixel point in the target cluster in the corresponding bit image.

[0048] In one embodiment, all pixels in a target cluster belong to the same region. In the absence of defects or noise, the information sequence of all pixels in a target cluster is exactly the same. Therefore, by calculating the information variance of all pixels in the target cluster in any bit image, the defect information and noise information contained in the corresponding bit image can be determined. When the information variance of bit image 1 in the target cluster is equal to 0, it means that the information of all pixels in the target cluster in bit image 1 is the same, indicating that there is no defect information and noise information in bit image 1.

[0049] Specifically, the target cluster Medium bit image The recombination coefficient for: , The target cluster Each pixel in the bit image The information variance of The target cluster The sum of the information variance of each pixel in all bit images.

[0050] Among them, the target cluster Each pixel in the bit image The information variance The calculation formula is: , The target cluster The number of pixels in The target cluster Medium pixel In bit image The amount of information, The target cluster Each pixel in the bit image The average amount of information.

[0051] Understandably, the target cluster Each pixel in belongs to the same area on the photovoltaic cell. In the absence of defects or noise, the target cluster The information variance of each bit image in should be 0. Therefore, if the target cluster In the bit image When the information variance is large, it means that the pixels in the same area are in the bit image. There are differences in the target cluster In terms of pixels in a bit image There is defect information or noise information in it, so it is a bit image Assign a smaller reconstruction coefficient so that the reconstructed image does not contain defect information and noise information.

[0052] In this way, the target cluster can be obtained by the same method The reorganization coefficient of each bit image in the target cluster is The recombination vectors of different target clusters are different. Subsequently, the reorganized pixel values ​​of the pixels in the corresponding target clusters are calculated according to the recombination vectors of the target clusters, realizing the adaptive recombination of different regions in the grayscale image, thereby accurately removing the defect information and noise information in different regions of the grayscale image.

[0053] S104, performing weighted summation on the bit image according to the reorganization vector to obtain the reorganized pixel value of each pixel in the corresponding target cluster, and constructing a reorganized image according to the reorganized pixel values ​​of all pixels in the grayscale image.

[0054] In one embodiment, since the recombination vectors of different target clusters are different, the reorganized pixel value of each pixel point in the corresponding target cluster can be obtained by weighted summing of the bit image according to the recombination vector.

[0055] Specifically, the pixel The reconstructed pixel value for:

[0056] , is the number of bit images, Pixel Target cluster The bit image in the reorganized vector The recombination coefficient, Bit Image Medium pixel The pixel value of .

[0057] In this way, the reconstructed pixel values ​​of all pixels in the grayscale image are calculated to obtain a reconstructed image, which can remove defect information and noise information in the grayscale image of the photovoltaic cell.

[0058] S105, comparing the grayscale image and the reconstructed image to obtain a defect detection result.

[0059] In one embodiment, the reconstructed image does not contain defect information and noise information. By comparing the grayscale image and the reconstructed image, the defect information and noise information in the grayscale image can be located.

[0060] Specifically, the comparing the grayscale image and the recombined image to obtain the defect detection result includes: calculating a difference map between the grayscale image and the recombined image; and performing threshold segmentation on the difference map after filtering to obtain the defect detection result.

[0061] The difference map includes the grayscale difference of each pixel point. Grayscale difference Satisfies the relationship: , Pixel The reconstructed pixel value of is the pixel in the grayscale image The difference image contains both defect information and noise information. In order to remove the noise information in the difference image, the difference image needs to be filtered. The filtering process is Gaussian filtering or median filtering.

[0062] The threshold segmentation can use the maximum inter-class variance method to determine the segmentation threshold, and the pixel points whose grayscale difference is greater than or equal to the segmentation threshold are set to 1, and the pixel points whose grayscale difference is less than the segmentation threshold are set to 0, to obtain a binary image. The area in the binary image where the pixel point value is 1 is the defect area, thereby obtaining the defect detection result of the photovoltaic cell.

[0063] The above introduces the technical principles and implementation details of the photovoltaic cell defect detection method based on image processing of the present application through specific embodiments. First, the grayscale image of the photovoltaic cell is split into eight bit images according to the bit layering algorithm, the amount of information of each pixel in each bit image is calculated, and the information sequence of each pixel is constructed, and then each pixel is clustered according to the information sequence to obtain multiple target clusters, so that the difference in the information sequence of each pixel in a target cluster is small, and the difference in the information sequence of each pixel between different target clusters is large; each pixel in a target cluster belongs to the same area of ​​the photovoltaic cell, and the information variance of each bit image in the target cluster is calculated. If the information variance of the bit image is large, it means that the information is not good. For each pixel point in the target cluster, the bit image contains defect information or noise information. Therefore, the recombination coefficient of the bit image in the recombination vector of the target cluster is small, ensuring that the recombined pixel value of each pixel point in the target cluster does not contain defect information or noise information; in this way, each target cluster realizes adaptive reconstruction of the grayscale image according to the corresponding recombination vector, so that the recombined image does not contain defect information or noise information; by comparing the grayscale image and the recombined image to locate the defect information or noise information, and after filtering and denoising, accurate defect detection results can be obtained, thereby improving the accuracy of defect detection of photovoltaic cells.

[0064] According to the second aspect of the present application, the present application also provides a photovoltaic cell defect detection system based on image processing. Figure 2 is a structural block diagram of a photovoltaic cell defect detection system based on image processing according to an embodiment of the present application. Figure 2As shown, the system 50 includes a processor and a memory, the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the photovoltaic cell defect detection method based on image processing according to the first aspect of the present application is implemented. The system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, whose settings and functions are known in the art, so they are not described here.

[0065] In the present application, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in this application may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0066] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0067] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.

Claims

1. A photovoltaic cell defect detection method based on image processing, characterized in that: The method comprises: The grayscale image of the photovoltaic cell is bit-layered to obtain eight bit images; Calculate the information content of any pixel in each bit image to obtain an information content sequence, and cluster the pixels according to the information content sequence to obtain multiple target clusters; Calculate the reorganization vector of each target cluster, the reorganization vector includes the reorganization coefficient of each bit image, the target cluster Medium bit image The recombination coefficient for: , The target cluster Each pixel in the bit image The information variance of The target cluster The sum of the information variance of each pixel in all bit images; The bit image is weighted and summed according to the reorganization vector to obtain the reorganized pixel value of each pixel in the corresponding target cluster, and the reorganized image is constructed according to the reorganized pixel values ​​of all pixels in the grayscale image; Compare the grayscale image and the reconstructed image to obtain the defect detection result; Pixels The reconstructed pixel value for: , is the number of bit images, Pixel Target cluster The bit image in the reorganized vector The recombination coefficient, Bit Image Medium pixel The pixel value of .

2. The photovoltaic cell defect detection method based on image processing according to claim 1, characterized in that: The grayscale image of the photovoltaic cell is subjected to bit layering to obtain eight bit images, including: Obtain a binary representation of the grayscale value of any pixel in the grayscale image, wherein the binary representation includes 8 bits, and each bit corresponds to a bit image; and use the value of each bit in the binary representation as the pixel value of the pixel in the corresponding bit image.

3. The photovoltaic cell defect detection method based on image processing according to claim 1, characterized in that: The calculating the information amount of any pixel in each bit image to obtain an information amount sequence includes: Calculate the image entropy within the neighborhood of a pixel point in the bit image to obtain the amount of information of each pixel point in the bit image; The information amounts of any pixel points in each bit image are arranged in a preset order to obtain an information amount sequence of the pixel points.

4. The photovoltaic cell defect detection method based on image processing according to claim 3, characterized in that: The image entropy is one-dimensional image entropy or two-dimensional image entropy; the pixel point neighborhood range is a rectangular range of a preset size with the pixel point as the center.

5. The photovoltaic cell defect detection method based on image processing according to claim 1, characterized in that: The pixel points are clustered according to the information amount sequence to obtain multiple target clusters, including: Kmeans clustering is performed on each pixel point according to the initial K value to obtain the initial clustering cluster, and the silhouette coefficient of the initial K value is calculated. In the Kmeans clustering process, the clustering distance is positively correlated with the Euclidean distance of the information sequence between the pixel points. Adjust the initial K value to draw an effect curve, wherein the abscissa of the effect curve is the initial K value and the ordinate is the contour coefficient corresponding to the initial K value; The initial K value corresponding to the inflection point in the effect curve is taken as the target K value, and the initial clustering cluster of the target K value is taken as the target clustering cluster.

6. The photovoltaic cell defect detection method based on image processing according to claim 5, characterized in that: Pixels and pixels The clustering distance for: , and Pixel and pixels The information sequence of and Pixels and pixels The coordinate position in the grayscale image, is the balance coefficient.

7. The photovoltaic cell defect detection method based on image processing according to claim 1, characterized in that: The defect detection result obtained by comparing the grayscale image and the recombined image includes: Calculate the difference map between the grayscale image and the reconstructed image; The difference image after filtering is subjected to threshold segmentation to obtain the defect detection result.

8. Photovoltaic cell defect detection system based on image processing, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the photovoltaic cell defect detection method based on image processing according to any one of claims 1 to 7 is implemented.

Images