CN117037042A - Automatic numbering method and related device for photovoltaic strings based on visual recognition model - Google Patents

Abstract

The invention discloses a photovoltaic string automatic numbering method based on a visual identification model and a related device, wherein the method comprises the following steps: acquiring an image to be numbered, wherein the image to be numbered comprises at least one photovoltaic group string; partitioning the images to be numbered to obtain at least one partitioned image; performing background separation on each partition image to obtain a background contour of the partition and a photovoltaic group string contour; inputting the outline of each photovoltaic group string to a photovoltaic group string vision recognition model to obtain a segmentation result of each photovoltaic plate; and numbering the photovoltaic panels in the image to be numbered based on the segmentation result of each photovoltaic panel. According to the method, the image to be numbered is partitioned and separated from the background, and the outline of the photovoltaic string is input into the visual identification model to automatically number the photovoltaic string, so that the coding efficiency and accuracy of the photovoltaic module are improved.

Description

Photovoltaic string automatic numbering method and related devices based on visual recognition model

Technical field

The invention relates to the technical field of photovoltaic panels, and specifically relates to an automatic numbering method for photovoltaic strings based on a visual recognition model and related devices.

Background technique

Photovoltaic is the abbreviation of solar photovoltaic power generation system. Photovoltaic panel is a new type of power generation device that uses the photovoltaic effect of semiconductor materials to directly convert solar radiation energy into electrical energy. In a photovoltaic power generation system, several photovoltaic modules are connected in series to form a circuit unit with a certain DC output, which is referred to as a photovoltaic module string or photovoltaic group string. At present, the numbering of photovoltaic strings in photovoltaic sites mainly uses on-site pictures taken by drones, and the images are numbered through traditional OpenCV, or the strings are corresponding one-to-one according to CAD drawings. Since photovoltaic sites are usually large, image segmentation obtained using OpenCV has problems such as low numbering efficiency and difficulty, and manual coding takes a long time, resulting in low detection and maintenance efficiency.

Contents of the invention

In view of the above problems, the present invention is proposed to provide an automatic photovoltaic string numbering method and device, computing equipment and computer storage medium based on a visual recognition model that overcomes the above problem of low photovoltaic string coding efficiency.

According to one aspect of the present invention, an automatic numbering method for photovoltaic strings based on a visual recognition model is provided, including:

Step S1, obtain an image to be numbered, wherein the image to be numbered includes at least one photovoltaic string;

Step S2, partition the image to be numbered to obtain at least one partitioned image;

Step S3, perform background separation on each partition image to obtain the background outline of the partition and the photovoltaic string outline;

Step S4: Input the outline of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel;

Step S5: Number the photovoltaic panels in the image to be numbered based on the segmentation results of each photovoltaic panel. In an optional manner, the first user's operation on the 3D item design model to be evaluated includes at least one of the following: model component disassembly, model explosion view scaling, and component-based rotation display. , wire frame diagram hidden line display switch, rendering switch and plan view display from each perspective.

In an optional manner, inputting the contours of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel further includes:

A photovoltaic string visual recognition model is constructed in advance, and the photovoltaic string outline image and the corresponding labeled slices are fed into the photovoltaic string visual recognition model for training, where the photovoltaic string outline image includes the arrangement of the photovoltaic strings. occlusion map;

Input the outline image of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel.

In an optional way, the photovoltaic string visual recognition model is a multi-modal image segmentation network, which includes an encoding module and a decoding module;

The encoding module and the decoding module each include at least one residual module, and the residual module includes a convolution layer, a BN layer and a LeakyReLU layer connected in sequence;

An attention module is provided between the residual module of the encoding module and the residual module of the decoding module. The attention module includes a channel attention CA module and a spatial attention SA module.

In an optional manner, partitioning the image to be numbered to obtain at least one partitioned image further includes:

Step S1: For any pixel of the image to be numbered, determine the pixel coordinates and RGB color mean of four adjacent pixels around it;

Step S2: For any adjacent pixel, determine the pixel whose RGB color mean is less than the first preset threshold as a weak edge pixel, and determine the pixel whose RGB color mean is greater than the second preset threshold as a strong edge pixel. point;

Step S3, connect the strong edge pixel points into edges. When connected to the endpoint of the edge, re-determine the weak edge pixel points as new edge points in the neighborhood pixel points of the strong edge pixel point, and continue to detect and connect the weak edge pixel points. New edge points until the contour is closed to obtain a partitioned image;

Step S4: For any pixel of the image to be numbered outside the partition image, repeat step S1 to obtain multiple partition images in sequence.

In an optional manner, said background separation of each of the partition images to obtain the background outline of the partition and the photovoltaic string outline further includes:

Step S1, remove the color information of the partition image, obtain a grayscale image and normalize it;

Step S2: Calculate the binary segmentation threshold of the grayscale image using the maximum inter-class variance method;

Step S3: Set the area in the grayscale image where the pixel value is less than the preset segmentation threshold to 0, and use black to segment the photovoltaic string outline of the partition; set the area where the pixel value is greater than the preset segmentation threshold to 1, Use white to outline the background of the partition.

In an optional manner, before inputting each of the photovoltaic string outline images to the photovoltaic string visual recognition model and obtaining the segmentation results of each photovoltaic panel, the method further includes:

Convert the photovoltaic string outline image into an HSV color model, and perform grayscale processing on it using a weighted average method;

Obtain the coordinates of each contour in the photovoltaic string contour image, use the DBSCAN clustering algorithm to perform cluster analysis on the coordinates of each contour, and obtain the Dunn index of each contour;

Set the area of the contour corresponding to the Dunn index smaller than the preset threshold to 1.

In an optional manner, the method further includes:

Use a template matching method to identify the image to be numbered to obtain the photovoltaic string;

Arrange the photovoltaic strings in a matrix according to a preset order, and arrange all the photovoltaic strings in the same row or column to form rows or columns to be filled;

Fill the rows or columns to be filled with the photovoltaic strings to form complete rows or columns of components;

The complete rows of components are mapped to all rows of the photovoltaic strings, or the complete columns of components are mapped to all columns of the photovoltaic strings.

According to another aspect of the present invention, an automatic numbering device for photovoltaic strings based on a visual recognition model is provided, including:

An image acquisition module, configured to acquire an image to be numbered, wherein the image to be numbered includes at least one photovoltaic string;

A partitioning module, used to partition the image to be numbered to obtain at least one partitioned image;

A background separation module, used to separate the background of each partition image to obtain the background outline of the partition and the photovoltaic string outline;

A segmentation module, used to input the outline of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel;

An encoding module, configured to number the photovoltaic panels in the image to be numbered based on the segmentation results of each photovoltaic panel.

According to another aspect of the present invention, a computing device is provided, including: a processor, a memory, a communication interface, and a communication bus. The processor, the memory, and the communication interface complete mutual communication through the communication bus. communication;

The memory is used to store at least one executable instruction, and the executable instruction causes the processor to perform operations corresponding to the above-mentioned automatic numbering method of photovoltaic strings based on the visual recognition model.

According to yet another aspect of the present invention, a computer storage medium is provided. At least one executable instruction is stored in the storage medium. The executable instruction causes the processor to perform the automatic numbering of photovoltaic strings based on the visual recognition model as described above. The operation corresponding to the method.

According to the solution provided by the present invention, an image to be numbered is obtained, wherein the image to be numbered includes at least one photovoltaic string; the image to be numbered is partitioned to obtain at least one partition image; and each of the partition images is background separated , obtain the background outline of the partition and the photovoltaic string outline; input each photovoltaic string outline into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel; treat the numbers based on the segmentation results of each photovoltaic panel The photovoltaic panels in the image are numbered. The present invention partitions and separates the background of the image to be numbered, and inputs the photovoltaic string outline into the visual recognition model to automatically number the photovoltaic strings, thereby improving the efficiency and accuracy of photovoltaic component coding.

The above description is only an overview of the technical solution of the present invention. In order to have a clearer understanding of the technical means of the present invention, it can be implemented according to the content of the description, and in order to make the above and other objects, features and advantages of the present invention more obvious and understandable. , the specific embodiments of the present invention are listed below.

Description of the drawings

Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are for the purpose of illustrating preferred embodiments only and are not to be construed as limiting the invention. Also throughout the drawings, the same reference characters are used to designate the same components. In the attached picture:

Figure 1 shows a schematic flow chart of the automatic numbering method of photovoltaic strings based on the visual recognition model according to the embodiment of the present invention;

Figure 2 shows a schematic flow chart of partitioning images to be numbered according to an embodiment of the present invention;

Figure 3 shows a schematic flow chart of using a template matching method to identify images to be numbered according to an embodiment of the present invention;

Figure 4 shows a schematic diagram 1 of the photovoltaic string visual recognition model according to the embodiment of the present invention;

Figure 5 shows a schematic diagram two of the photovoltaic string visual recognition model according to the embodiment of the present invention;

Figure 6 shows a schematic diagram of a photovoltaic string being shielded by a photovoltaic panel according to an embodiment of the present invention;

Figure 7 shows a schematic diagram of photovoltaic string identification according to an embodiment of the present invention;

Figure 8 shows a schematic structural diagram of a photovoltaic string automatic numbering device based on a visual recognition model according to an embodiment of the present invention;

Figure 9 shows a schematic structural diagram of a computing device according to an embodiment of the present invention.

Detailed ways

Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a thorough understanding of the invention, and to fully convey the scope of the invention to those skilled in the art.

Figure 1 shows a schematic flow chart of the automatic numbering method of photovoltaic strings based on the visual recognition model according to an embodiment of the present invention. This method partitions and separates the background of the image to be numbered, and inputs the photovoltaic string outline into the visual recognition model to automatically number the photovoltaic strings. Specifically, it includes the following steps:

Step S101: Obtain an image to be numbered, where the image to be numbered includes at least one photovoltaic string.

In a photovoltaic power generation system, several photovoltaic modules are connected in series to form a circuit unit with a certain DC output, referred to as a module string or group string. A drone can be used to photograph the sample photovoltaic panels in a preset direction to obtain an image to be numbered. The preset direction is a direction in which the sample photovoltaic panels do not tilt in the image. For example, the drone can be used to photograph sample photovoltaic panels with the following settings:

200m≤flight height≤300m.

Flight mode: parallel or vertical photovoltaic panels, flying in point order.

Flight point distribution: According to the principle of covering all photovoltaic panels, repeated shooting of photovoltaic panels is allowed.

Video shooting angle: vertical to the ground.

Video resolution: 1920×1080 or higher, with clarity as the principle.

Optionally, while photographing the sample photovoltaic panels, obtain the longitude and latitude of the shooting point and other information, and then after the shooting is completed, import the photographed picture and the longitude and latitude of the shooting point and other information into the preset software to automatically reconstruct the model, so as to Based on the longitude and latitude coordinates and other information, a full-view area map containing all photovoltaic panels is constructed, and the full-view area map is the image to be numbered (image a as shown in Figure 7). This embodiment can also identify the shielded photovoltaic panel image as shown in Figure 6 (there is a tilted direction in the image).

Step S102, partition the image to be numbered to obtain at least one partitioned image.

The image to be numbered may include multiple patches or partitions (incoherent areas), and each partition needs to be identified in advance.

In an optional manner, partitioning the image to be numbered to obtain at least one partitioned image further includes:

Step S1: For any pixel of the image to be numbered, determine the pixel coordinates and RGB color mean of four adjacent pixels around it;

Step S2: For any adjacent pixel, determine the pixel whose RGB color mean is less than the first preset threshold as a weak edge pixel, and determine the pixel whose RGB color mean is greater than the second preset threshold as a strong edge pixel. point;

Step S3, connect the strong edge pixel points into edges. When connected to the endpoint of the edge, re-determine the weak edge pixel points as new edge points in the neighborhood pixel points of the strong edge pixel point, and continue to detect and connect the weak edge pixel points. New edge points until the contour is closed to obtain a partitioned image;

Step S4: For any pixel of the image to be numbered outside the partition image, repeat step S1 to obtain multiple partition images in sequence.

Step S103: perform background separation on each partition image to obtain the background outline and photovoltaic string outline of the partition.

In this embodiment, the main principle of partitioning the image to be numbered is to identify pixels with obvious changes in color or brightness in the image. The significant changes in these pixels often represent important changes in the attributes of this part of the image. These include discontinuities in depth, discontinuity in direction, discontinuity in brightness, etc. When detecting the image to be numbered, first detect the pixels of the area contour, and for any adjacent pixel, determine the pixel whose RGB color mean is less than the first preset threshold as a weak edge pixel, and determine its RGB color Pixels whose mean value is greater than the second preset threshold are determined as strong edge pixels. Connect the strong edge pixel point into an edge. When connected to the endpoint of the edge, re-determine the weak edge pixel point as a new edge point in the neighborhood pixel points of the strong edge pixel point, and continue to detect and connect the new edge point. Until the contour is closed, the coordinates of the contour are output. The boundary of the image area to be numbered can also be detected through image edge detection methods such as Roberts operator, Sobel operator, and second-order Laplacian operator, which is not limited in this article.

In an optional manner, said background separation of each of the partition images to obtain the background outline of the partition and the photovoltaic string outline further includes:

Step S1: Remove the color information of the partition image to obtain a grayscale image and normalize it.

Step S2: Calculate the binary segmentation threshold of the grayscale image using the maximum inter-class variance method. The maximum inter-class variance method (OTSU algorithm, also known as Otsu method) is an efficient algorithm for binarizing images. It uses thresholds to divide the original image into foreground and background images. According to the grayscale characteristics of the image, the original image is divided into two images: foreground and background. The image is divided into two parts, the background and the target. The greater the inter-class variance between the background and the target, the greater the difference between the two parts that make up the image. When part of the target is mistakenly divided into the background or part of the background is mistakenly divided into the target, it will result in two parts. The difference becomes smaller. Therefore, the segmentation that maximizes the between-class variance means the smallest probability of misclassification.

Step S3: Set the area in the grayscale image where the pixel value is less than the preset segmentation threshold to 0, and use black to segment the photovoltaic string outline of the partition; set the area where the pixel value is greater than the preset segmentation threshold to 1, Use white to outline the background of the partition. For example, according to a large number of experimental statistics, the binary segmentation threshold is set to 0.9 to binarize the grayscale image, set the area with a pixel value less than 0.9 in the grayscale image to 0, and use black to represent the segmented foreground contour area; and Areas with pixel values greater than 0.9 are set to 1, and white represents the segmented background area.

Step S104: Input the outline of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel.

A photovoltaic string visual recognition model is constructed in advance, and the photovoltaic string outline image and the corresponding labeled slices are fed (inputted) to the photovoltaic string visual recognition model for training, wherein the photovoltaic string outline image includes the photovoltaic group The arrangement and occlusion diagram of strings (shown in Figure 6). Input the outline image of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel.

In an optional way, as shown in Figure 4, the photovoltaic string visual recognition model is a multi-modal image segmentation network, which includes an encoding module and a decoding module;

The encoding module and the decoding module each include at least one residual module, and the residual module includes a convolution layer, a BN layer and a LeakyReLU layer connected in sequence;

An attention module is provided between the residual module of the encoding module and the residual module of the decoding module. The attention module includes a channel attention CA module and a spatial attention SA module.

The existing multi-modal image segmentation network (MultiResUNet network) has problems such as too small receptive fields. In this embodiment, the encoding module and the decoding module each include at least one residual module, and the residual module includes sequential connections. The convolution layer, BN layer and LeakyReLU layer can expand the receptive field to segment the target image, helping to segment targets of different sizes.

In order to further solve the problem of unclear channel and spatial relationships existing in existing multi-modal image segmentation networks, embodiments of the present invention add an attention module with a dual attention mechanism to the network to reintegrate channel and spatial features. Weights.

The attention module includes a channel attention CA module and a spatial attention SA module, and the outputs of the channel attention CA module and the channel attention SA module are added together. Optionally, the channel attention CA module includes two CA units with the same structure, and each of the CA units includes an adaptive global average pooling layer and a convolution layer connected in sequence. After adding the outputs of the two CA units, the channel weight is obtained through the Softmax activation function. The low-level features from the encoding module in the input data are multiplied by the channel weight and then combined with the high-level features from the decoding module in the input data. The addition results in output features, which are recorded as the output of the CA module.

As shown in Figure 5, the encoding module includes five residual modules from the shallow layer of the network to the deep layer of the network, numbered 1, 2, 3, 4, and 5. A maximum pooling layer is used between each two residual modules for downsampling. ;The decoding module includes four residual modules from the shallow layer of the network to the deep layer of the network, numbered 6, 7, 8, and 9. There is an upsampling layer between each two residual modules, and the output of the encoding module passes through an upsampling layer. Input decoding module; the output features of the residual modules numbered 5, 6, 7, and 8 are first subjected to a 2 times upsampling operation, and then passed through the attention module with the output features from the residual modules numbered 4, 3, 2, and 1 respectively. The re-weighted features are fused through the Concat connection operation; among them, the output features from the residual modules numbered 4, 3, 2, and 1 need to pass through the residual modules (1, 2, 3 respectively) before passing through the attention module. ,4).

Before inputting the contour image of each photovoltaic string into the photovoltaic string visual recognition model and obtaining the segmentation results of each photovoltaic panel, the method further includes:

The photovoltaic string profile image is converted into an HSV color model (the chromaticity map H, saturation map S and brightness map V of the HSV model are obtained), that is, the histogram of the H component is extracted as a color feature vector to characterize the color characteristics of the image. Hue contains the color information of the image, which is a significant feature for distinguishing materials of different colors. Taking the H component in the HSV model as the target feature, the histogram statistical method is used to extract the distribution characteristics of the H component as the color feature, which is not limited to the number of underlying features. characteristics and the ability to distinguish features independent of specific feature types. Use the weighted average method to perform grayscale processing and normalize it to obtain a 256-dimensional feature vector.

The coordinates of each contour in the photovoltaic string contour image are obtained, and the DBSCAN clustering algorithm is used to perform cluster analysis on the coordinates of each contour to obtain the Dunn index of each contour. Among them, the DBSCAN clustering algorithm is an unsupervised machine learning algorithm based on density clustering (DBSCAN). If the mutual distance of data points is less than or equal to the specified epsilon, then it is the same class and has minPts number of neighbors within the radius of a neighborhood. A domain is considered a cluster. DBSCAN determines the distance between two points if they are similar and belong to the same class. Dunn's index refers to the minimum value of the nearest distance between any two clusters, divided by the maximum value of the distance between the two furthest points in any cluster. If the minimum value of the nearest distance between any two clusters is larger (that is, the samples between clusters are far away from each other), the Dunn index will be larger; if the maximum value of the distance between the two furthest points in any cluster is The smaller it is (that is, the samples in the cluster are very close), the larger the Dunn index is.

Set the area of the contour corresponding to the Dunn index smaller than the preset threshold to 1, and use white to represent the segmented background area, which can eliminate false contours and noise interference (such as eliminating combiner boxes).

Step S105: Number the photovoltaic panels in the image to be numbered based on the segmentation results of each photovoltaic panel.

For example, if there are multiple inverters, number the inverters first, and the photovoltaic strings are numbered according to the inverters. The input terminals of the inverters have multiple sets of input terminals, and the photovoltaic strings are numbered in order, such as 2-3 -8, 2 represents the No. 2 inverter, 3 represents the third string in the No. 2 inverter, 8 represents the 8th photovoltaic panel in the third string photovoltaic group, and so on. This number is easy to remember and easy to find. .

In an optional manner, the method further includes:

A template matching method is used to identify the image to be numbered to obtain a photovoltaic string. The template matching method is to select a picture of a photovoltaic string and perform comparative analysis to divide the photovoltaic modules.

The photovoltaic strings are arranged in a matrix according to a preset order, and all the photovoltaic strings are arranged in the same row or column to form rows or columns to be filled.

The rows or columns to be filled are filled with the photovoltaic strings to form complete rows or columns of components.

Arrange all the photovoltaic modules identified on the photovoltaic string in the same row or column according to the length or width of the photovoltaic string to form rows or columns to be filled to complete the logical filling of the components. Arrange the photovoltaic modules in all rows in the same row. For example, you can keep the abscissa of the upper left point of the photovoltaic string frame unchanged and the ordinate of the photovoltaic modules unified, so that they can be arranged in the same row. Modules within the same row can be filled according to the width of the PV string. Similarly, you can also choose to arrange the photovoltaic strings in the same column. For example, the vertical coordinate of the upper left point of the photovoltaic string frame can be unchanged and the horizontal coordinate can be unified, so that the photovoltaic strings can be arranged in the same column. Map complete PV strings to all rows of PV strings, or map columns of PV strings to all columns of PV strings.

Arrange all the identified photovoltaic strings in the same row to form a row to be filled. This can reduce the number of logical fillings and improve the cutting accuracy of the overall photovoltaic strings.

According to the solution provided by the present invention, an image to be numbered is obtained, wherein the image to be numbered includes at least one photovoltaic string; the image to be numbered is partitioned to obtain at least one partition image; and each of the partition images is background separated , obtain the background outline of the partition and the photovoltaic string outline; input each photovoltaic string outline into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel; treat the numbers based on the segmentation results of each photovoltaic panel The photovoltaic panels in the image are numbered. The present invention partitions and separates the background of the image to be numbered, and inputs the photovoltaic string outline into the visual recognition model to automatically number the photovoltaic strings, thereby improving the efficiency and accuracy of photovoltaic component coding.

Figure 8 shows a schematic structural diagram of an automatic numbering device for photovoltaic strings based on a visual recognition model according to an embodiment of the present invention. The photovoltaic string automatic numbering device 800 based on the visual recognition model includes: an image acquisition module 810, a partition module 820, a background separation module 830, a segmentation module 840 and an encoding module 850.

The image acquisition module 810 is used to acquire an image to be numbered, wherein the image to be numbered includes at least one photovoltaic string;

The partition module 820 is used to partition the image to be numbered to obtain at least one partitioned image;

The background separation module 830 is used to separate the background of each partition image to obtain the background outline of the partition and the photovoltaic string outline;

The segmentation module 840 is used to input the outline of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel;

The encoding module 850 is configured to number the photovoltaic panels in the image to be numbered based on the segmentation results of each photovoltaic panel.

In an optional manner, the segmentation module 840 is further used to:

A photovoltaic string visual recognition model is constructed in advance, and the photovoltaic string outline image and the corresponding labeled slices are fed into the photovoltaic string visual recognition model for training, where the photovoltaic string outline image includes the arrangement of the photovoltaic strings. occlusion map;

Input the outline image of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel.

In an optional manner, the segmentation module 840 is further used to:

The photovoltaic string visual recognition model is a multi-modal image segmentation network, which includes an encoding module and a decoding module;

The encoding module and the decoding module each include at least one residual module, and the residual module includes a convolution layer, a BN layer and a LeakyReLU layer connected in sequence;

An attention module is provided between the residual module of the encoding module and the residual module of the decoding module. The attention module includes a channel attention CA module and a spatial attention SA module.

In an optional manner, the background separation module 830 is further used to:

Step S1: For any pixel of the image to be numbered, determine the pixel coordinates and RGB color mean of four adjacent pixels around it;

Step S2: For any adjacent pixel, determine the pixel whose RGB color mean is less than the first preset threshold as a weak edge pixel, and determine the pixel whose RGB color mean is greater than the second preset threshold as a strong edge pixel. point;

Step S3, connect the strong edge pixel points into edges. When connected to the endpoint of the edge, re-determine the weak edge pixel points as new edge points in the neighborhood pixel points of the strong edge pixel point, and continue to detect and connect the weak edge pixel points. New edge points until the contour is closed to obtain a partitioned image;

Step S4: For any pixel of the image to be numbered outside the partition image, repeat step S1 to obtain multiple partition images in sequence.

In an optional manner, the segmentation module 840 is further used to:

Step S1, remove the color information of the partition image, obtain a grayscale image and normalize it;

Step S2: Calculate the binary segmentation threshold of the grayscale image using the maximum inter-class variance method;

Step S3: Set the area in the grayscale image where the pixel value is less than the preset segmentation threshold to 0, and use black to segment the photovoltaic string outline of the partition; set the area where the pixel value is greater than the preset segmentation threshold to 1, Use white to outline the background of the partition.

In an optional manner, a contour clustering module is also included, and the contour clustering module is further used to:

Convert the photovoltaic string outline image into an HSV color model, and perform grayscale processing on it using a weighted average method;

Obtain the coordinates of each contour in the photovoltaic string contour image, use the DBSCAN clustering algorithm to perform cluster analysis on the coordinates of each contour, and obtain the Dunn index of each contour;

Set the area of the contour corresponding to the Dunn index smaller than the preset threshold to 1.

In an optional manner, a template matching module is also included, and the template matching module is further used to:

Use a template matching method to identify the image to be numbered to obtain the photovoltaic string;

Arrange the photovoltaic strings in a matrix according to a preset order, and arrange all the photovoltaic strings in the same row or column to form rows or columns to be filled;

Fill the rows or columns to be filled with the photovoltaic strings to form complete rows or columns of components;

The complete rows of components are mapped to all rows of the photovoltaic strings, or the complete columns of components are mapped to all columns of the photovoltaic strings.

According to the solution provided by the present invention, an image to be numbered is obtained, wherein the image to be numbered includes at least one photovoltaic string; the image to be numbered is partitioned to obtain at least one partition image; and each of the partition images is background separated , obtain the background outline of the partition and the photovoltaic string outline; input each photovoltaic string outline into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel; treat the numbers based on the segmentation results of each photovoltaic panel The photovoltaic panels in the image are numbered. The present invention partitions and separates the background of the image to be numbered, and inputs the photovoltaic string outline into the visual recognition model to automatically number the photovoltaic strings, thereby improving the efficiency and accuracy of photovoltaic component coding.

Figure 9 shows a schematic structural diagram of an embodiment of the computing device of the present invention. The specific embodiment of the present invention does not limit the specific implementation of the computing device.

As shown in FIG. 9 , the computing device may include: a processor 902 , a communications interface 904 , a memory 906 , and a communications bus 908 .

Among them: the processor 902, the communication interface 904, and the memory 906 complete communication with each other through the communication bus 908. The communication interface 904 is used to communicate with network elements of other devices such as clients or other servers. The processor 902 is configured to execute the program 910. Specifically, it can execute the relevant steps in the above embodiment of the automatic numbering method for photovoltaic strings based on the visual recognition model.

Specifically, program 910 may include program code including computer operating instructions.

The processor 902 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The one or more processors included in the computing device may be the same type of processor, such as one or more CPUs; or they may be different types of processors, such as one or more CPUs and one or more ASICs.

Memory 906 is used to store programs 910. The memory 906 may include high-speed RAM memory, and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.

Embodiments of the present invention provide a non-volatile computer storage medium that stores at least one executable instruction. The computer executable instruction can execute the photovoltaic group based on the visual recognition model in any of the above method embodiments. String automatic numbering method.

The algorithms or displays provided herein are not inherently associated with any particular computer, virtual system, or other device. Various general-purpose systems can also be used with teaching based on this. From the above description, the structure required to construct such a system is obvious. Furthermore, embodiments of the present invention are not directed to any specific programming language. It should be understood that a variety of programming languages may be utilized to implement the invention described herein, and that the above descriptions of specific languages are intended to disclose the best mode of carrying out the invention.

In the instructions provided here, a number of specific details are described. However, it is understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this description.

Similarly, it will be understood that in the above description of exemplary embodiments of the invention, various features of embodiments of the invention are sometimes grouped together into a single implementation in order to streamline the invention and assist in understanding one or more of the various inventive aspects. examples, diagrams, or descriptions thereof. This method of disclosure, however, is not to be interpreted as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the Detailed Description are hereby expressly incorporated into this Detailed Description, with each claim standing on its own as a separate embodiment of this invention.

Those skilled in the art will understand that modules in the devices in the embodiment can be adaptively changed and arranged in one or more devices different from that in the embodiment. The modules or units or components in the embodiments may be combined into one module or unit or component, and furthermore they may be divided into a plurality of sub-modules or sub-units or sub-components. All features disclosed in this specification (including accompanying claims, abstract and drawings) and any method so disclosed may be employed in any combination, except that at least some of such features and/or processes or units are mutually exclusive. All processes or units of the equipment are combined. Each feature disclosed in this specification (including accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.

Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features of different embodiments are meant to be within the scope of the invention. and form different embodiments. For example, in the following claims, any of the claimed embodiments may be used in any combination.

Various component embodiments of the present invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will understand that a microprocessor or a digital signal processor (DSP) may be used in practice to implement some or all functions of some or all components according to embodiments of the present invention. The invention may also be implemented as an apparatus or apparatus program (eg, computer program and computer program product) for performing part or all of the methods described herein. Such a program implementing the present invention may be stored on a computer-readable medium, or may be in the form of one or more signals. Such signals may be downloaded from an Internet website, or provided on a carrier signal, or in any other form.

It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the element claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, third, etc. does not indicate any order. These words can be interpreted as names. Unless otherwise specified, the steps in the above embodiments should not be understood as limiting the order of execution.

Claims (10)

1. An automatic numbering method for photovoltaic strings based on a visual recognition model, which is characterized by including: Step S1, obtain an image to be numbered, wherein the image to be numbered includes at least one photovoltaic string; Step S2, partition the image to be numbered to obtain at least one partitioned image; Step S3, perform background separation on each partition image to obtain the background outline of the partition and the photovoltaic string outline; Step S4: Input the outline of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel; Step S5: Number the photovoltaic panels in the image to be numbered based on the segmentation results of each photovoltaic panel. 2. The automatic numbering method of photovoltaic strings based on the visual recognition model according to claim 1, characterized in that the outline of each photovoltaic string is input into the photovoltaic string visual recognition model to obtain the segmentation of each photovoltaic panel. Results further include: A photovoltaic string visual recognition model is constructed in advance, and the photovoltaic string outline image and the corresponding labeled slices are fed into the photovoltaic string visual recognition model for training, where the photovoltaic string outline image includes the arrangement of the photovoltaic strings. occlusion map; Input the outline image of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel. 3. The photovoltaic string automatic numbering method based on the visual recognition model according to claim 2, characterized in that the photovoltaic string visual recognition model is a multi-modal image segmentation network, which includes an encoding module and a decoding module. ; The encoding module and the decoding module each include at least one residual module, and the residual module includes a convolution layer, a BN layer and a LeakyReLU layer connected in sequence; An attention module is provided between the residual module of the encoding module and the residual module of the decoding module. The attention module includes a channel attention CA module and a spatial attention SA module. 4. The automatic numbering method of photovoltaic strings based on the visual recognition model according to claim 1, characterized in that said partitioning the image to be numbered to obtain at least one partitioned image further includes: Step S1: For any pixel of the image to be numbered, determine the pixel coordinates and RGB color mean of four adjacent pixels around it; Step S2: For any adjacent pixel, determine the pixel whose RGB color mean is less than the first preset threshold as a weak edge pixel, and determine the pixel whose RGB color mean is greater than the second preset threshold as a strong edge pixel. point; Step S3, connect the strong edge pixel points into edges. When connected to the endpoint of the edge, re-determine the weak edge pixel points as new edge points in the neighborhood pixel points of the strong edge pixel point, and continue to detect and connect the weak edge pixel points. New edge points until the contour is closed to obtain a partitioned image; Step S4: For any pixel of the image to be numbered outside the partition image, repeat step S1 to obtain multiple partition images in sequence. 5. The photovoltaic string automatic numbering method based on the visual recognition model according to claim 1, characterized in that the background separation of each partition image is performed to obtain the background outline of the partition and the photovoltaic string outline. include: Step S1, remove the color information of the partition image, obtain a grayscale image and normalize it; Step S2: Calculate the binary segmentation threshold of the grayscale image using the maximum inter-class variance method; Step S3: Set the area in the grayscale image where the pixel value is less than the preset segmentation threshold to 0, and use black to segment the photovoltaic string outline of the partition; set the area where the pixel value is greater than the preset segmentation threshold to 1, Use white to outline the background of the partition. 6. The automatic numbering method of photovoltaic strings based on the visual recognition model according to claim 1, characterized in that the outline image of each photovoltaic string is input into the photovoltaic string visual recognition model to obtain the number of each photovoltaic panel. Before segmenting the results, the method also includes: Convert the photovoltaic string outline image into an HSV color model, and perform grayscale processing on it using a weighted average method; Obtain the coordinates of each contour in the photovoltaic string contour image, use the DBSCAN clustering algorithm to perform cluster analysis on the coordinates of each contour, and obtain the Dunn index of each contour; Set the area of the contour corresponding to the Dunn index smaller than the preset threshold to 1. 7. The photovoltaic string automatic numbering method based on the visual recognition model according to claim 1, characterized in that the method further includes: Use a template matching method to identify the image to be numbered to obtain the photovoltaic string; Arrange the photovoltaic strings in a matrix according to a preset order, and arrange all the photovoltaic strings in the same row or column to form rows or columns to be filled; Fill the rows or columns to be filled with the photovoltaic strings to form complete rows or columns of components; The complete rows of components are mapped to all rows of the photovoltaic strings, or the complete columns of components are mapped to all columns of the photovoltaic strings. 8. An automatic numbering device for photovoltaic strings based on a visual recognition model, characterized by including: An image acquisition module, configured to acquire an image to be numbered, wherein the image to be numbered includes at least one photovoltaic string; A partitioning module, used to partition the image to be numbered to obtain at least one partitioned image; A background separation module, used to separate the background of each partition image to obtain the background outline of the partition and the photovoltaic string outline; A segmentation module, used to input the outline of each photovoltaic string into the photovoltaic string visual recognition model to obtain the segmentation results of each photovoltaic panel; An encoding module, configured to number the photovoltaic panels in the image to be numbered based on the segmentation results of each photovoltaic panel. 9. A computing device, comprising: a processor, a memory, a communication interface and a communication bus, the processor, the memory and the communication interface completing communication with each other through the communication bus; The memory is used to store at least one executable instruction. The executable instruction causes the processor to perform operations corresponding to the photovoltaic string automatic numbering method based on the visual recognition model according to any one of claims 1-7. . 10. A computer storage medium, at least one executable instruction is stored in the storage medium, and the executable instruction causes the processor to execute the photovoltaic group based on the visual recognition model according to any one of claims 1-7. Operations corresponding to the string automatic numbering method.