Power grid multi-source remote sensing data fusion and change detection method and related equipment

Abstract

The invention relates to the technical field of multi-source remote sensing data processing, and discloses a power grid multi-source remote sensing data fusion and change detection method and related equipment, and the method comprises the steps: carrying out the fusion processing of target multi-source remote sensing data corresponding to a target power grid, and obtaining fused remote sensing data; and performing change detection according to the fused remote sensing data in a preset time window to obtain a change detection result corresponding to the target power grid. Through fusion of multi-source remote sensing data, advantages of different data sources can be comprehensively utilized, information of a power transmission area corresponding to a target power grid is more comprehensively acquired, and the problems of large ground feature scale difference, complex and diversified ground feature types and unbalanced ground feature category distribution are effectively solved. A basis is provided for improving the accuracy and integrity of a change detection result; change detection is carried out by means of fused remote sensing data in a preset time window, the change trend of the power grid can be dynamically tracked, and valuable reference basis is provided for long-term planning and decision making.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-source remote sensing data processing, and in particular to a method for fusion and change detection of multi-source remote sensing data of a power grid and related equipment. Background Art

[0002] With the continuous development of science and technology, satellite, drone and other technologies are becoming more and more mature, providing us with rich remote sensing image data. In the field of power grid, it is of great significance to use remote sensing image data for monitoring and analysis. At present, the power grid field uses a single remote sensing image data for monitoring and analysis. Faced with the large differences in the scale of objects in the power transmission area of ​​the power grid, the complex and diverse types of objects, and the unbalanced distribution of object categories, a single remote sensing image data for monitoring and analysis is difficult to meet the needs of accurate monitoring and change detection of the power grid. Summary of the invention

[0003] Based on this, it is necessary to address the technical problems in the existing technology that the scales of objects in the transmission area of ​​the power grid are large, the types of objects are complex and diverse, and the distribution of object categories is unbalanced. Monitoring and analysis of single remote sensing image data is difficult to meet the needs of accurate monitoring and change detection of the power grid. A power grid multi-source remote sensing data fusion and change detection method and related equipment are proposed.

[0004] In a first aspect, a method for fusion and change detection of multi-source remote sensing data of a power grid is provided, the method comprising:

[0005] Obtain target multi-source remote sensing data corresponding to the target power grid;

[0006] Performing fusion processing on the target multi-source remote sensing data to obtain fused remote sensing data;

[0007] Change detection is performed based on each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid.

[0008] In a second aspect, a power grid multi-source remote sensing data fusion and change detection device is provided, the device comprising:

[0009] A data acquisition module is used to acquire target multi-source remote sensing data corresponding to the target power grid;

[0010] A fusion processing module is used to perform fusion processing on the target multi-source remote sensing data to obtain fused remote sensing data;

[0011] The change detection module is used to perform change detection according to each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid.

[0012] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned power grid multi-source remote sensing data fusion and change detection method when executing the computer program.

[0013] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned power grid multi-source remote sensing data fusion and change detection method are implemented.

[0014] The present application discloses a method and related equipment for fusion and change detection of multi-source remote sensing data of a power grid, wherein the method obtains target multi-source remote sensing data corresponding to the target power grid, fuses the target multi-source remote sensing data, obtains fused remote sensing data, performs change detection based on each of the fused remote sensing data within a preset time window, and obtains change detection results corresponding to the target power grid. By fusing multi-source remote sensing data, it is possible to comprehensively utilize the advantages of different data sources, more comprehensively obtain information on the transmission area corresponding to the target power grid, effectively address the problems of large differences in ground object scales, complex and diverse ground object types, and unbalanced distribution of ground object categories, and thus provide a basis for improving the accuracy and completeness of change detection results; by performing change detection with the help of fused remote sensing data within a preset time window, it is possible to dynamically track the changing trend of the power grid, and provide a valuable reference for long-term planning and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] in:

[0017] Figure 1 A diagram of an application environment of a method for fusion and change detection of multi-source remote sensing data of a power grid in an embodiment;

[0018] Figure 2 A flowchart of a method for fusion and change detection of multi-source remote sensing data of a power grid in one embodiment;

[0019] Figure 3 It is a structural block diagram of a power grid multi-source remote sensing data fusion and change detection device in one embodiment;

[0020] Figure 4 FIG. 4 is a structural block diagram of a computer device in one embodiment. DETAILED DESCRIPTION

[0021] 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] The power grid multi-source remote sensing data fusion and change detection method provided by the embodiment of the present invention can be applied in the following aspects: Figure 1 In the application environment, the client 110 communicates with the server 120 through the network.

[0023] The server 120 can obtain the target multi-source remote sensing data corresponding to the target power grid through the client 110. The server 120 is used to: fuse the target multi-source remote sensing data to obtain fused remote sensing data; perform change detection based on each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid. By fusing multi-source remote sensing data, it is possible to comprehensively utilize the advantages of different data sources, more comprehensively obtain information on the transmission area corresponding to the target power grid, and effectively deal with the problems of large differences in land scales, complex and diverse land types, and unbalanced distribution of land types, thereby providing a basis for improving the accuracy and completeness of change detection results; using the fused remote sensing data within a preset time window for change detection, it is possible to dynamically track the changing trend of the power grid and provide a valuable reference for long-term planning and decision-making.

[0024] The client 110 may be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, and portable wearable devices. The server 120 may be implemented by an independent server or a server cluster consisting of multiple servers.

[0025] The present invention is described in detail below through specific embodiments.

[0026] See also Figure 2 As shown, Figure 2 A flow chart of a method for fusion and change detection of multi-source remote sensing data of a power grid provided by an embodiment of the present invention includes the following steps:

[0027] S1: Acquire target multi-source remote sensing data corresponding to the target power grid;

[0028] The target grid is the grid that you want to analyze.

[0029] The target multi-source remote sensing data is the multi-source remote sensing data obtained by photographing the transmission area corresponding to the target power grid. The transmission area refers to the area where the transmission line is located. The transmission area includes towers, transmission lines, and objects in the surrounding environment of the towers.

[0030] Multi-source remote sensing data refers to remote sensing data obtained through different sensors and platforms (such as satellites, aircraft, drones, etc.). These data include information in different bands such as visible light, infrared, microwave, etc., and can provide comprehensive and multi-angle surface information. Multi-source remote sensing data has different spatial resolutions, spectral resolutions, and temporal resolutions. These characteristics make the application of remote sensing in different fields more comprehensive and in-depth.

[0031] Specifically, the target multi-source remote sensing data corresponding to the target power grid input by the user can be obtained, the target multi-source remote sensing data corresponding to the target power grid can be obtained from a preset storage space, and the target multi-source remote sensing data corresponding to the target power grid can also be obtained from a third-party application.

[0032] S2: fusing the target multi-source remote sensing data to obtain fused remote sensing data;

[0033] Specifically, a suitable data fusion algorithm (eg, a feature-based fusion algorithm or a pixel-based fusion algorithm) is used to perform fusion processing on the target multi-source remote sensing data, and the data obtained by the fusion processing is used as fused remote sensing data.

[0034] For feature-based fusion algorithms, feature information of objects in different data sources (the remote sensing data corresponding to each shooting method is regarded as a data source), such as edges, contours, textures, etc., can be extracted and fused at the feature level.

[0035] The pixel-based fusion algorithm directly fuses the values ​​of each pixel from different data sources, and can determine the value of the fused pixel by weighted average or other methods. In the fusion process, the weight distribution of different data sources must also be considered, where the weight distribution is reasonably set based on factors such as the reliability and resolution of the data source.

[0036] S3: Perform change detection according to each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid.

[0037] The preset time window includes: start time and end time.

[0038] Specifically, firstly, all the fused remote sensing data with data time (data time can be the acquisition time of remote sensing data or the generation time of fused remote sensing data) within a preset time window are extracted, and then, a model (e.g., a classification model) obtained based on artificial intelligence training is used to perform change detection based on all the extracted fused remote sensing data to obtain the change detection result corresponding to the target power grid. In other words, the change detection result of the target power grid within the preset time window is obtained.

[0039] By fusing multi-source remote sensing data, this embodiment can comprehensively utilize the advantages of different data sources, more comprehensively obtain information on the transmission area corresponding to the target power grid, and effectively deal with the problems of large differences in ground object scales, complex and diverse ground object types, and unbalanced distribution of ground object categories, thereby providing a basis for improving the accuracy and completeness of change detection results; by using the fused remote sensing data within a preset time window to perform change detection, it is possible to dynamically track the changing trend of the power grid and provide a valuable reference for long-term planning and decision-making.

[0040] In one embodiment, the step of acquiring target multi-source remote sensing data corresponding to the target power grid includes:

[0041] S11: Acquire initial multi-source remote sensing data corresponding to the target power grid;

[0042] The initial multi-source remote sensing data is the multi-source remote sensing data obtained by photographing the transmission area corresponding to the target power grid. In other words, the initial multi-source remote sensing data is the multi-source remote sensing data that has not been preprocessed.

[0043] Specifically, the initial multi-source remote sensing data corresponding to the target power grid input by the user can be obtained, the initial multi-source remote sensing data corresponding to the target power grid can be obtained from a preset storage space, and the initial multi-source remote sensing data corresponding to the target power grid can also be obtained from a third-party application.

[0044] S12: performing data cleaning, calibration, and registration on the initial multi-source remote sensing data to obtain the target multi-source remote sensing data.

[0045] Specifically, the initial multi-source remote sensing data is sequentially cleaned, calibrated, and registered to achieve pre-processing before fusion, and the registered data is used as the target multi-source remote sensing data.

[0046] Through data cleaning, this embodiment effectively removes noise, outliers, invalid data, etc. in the initial multi-source remote sensing data, improves the purity and quality of the data, and provides a more reliable basis for subsequent processing; the calibration operation ensures the accuracy and consistency of each remote sensing data source, so that data from different sources are unified in measurement standards and errors caused by data deviation are reduced; the registration enables remote sensing data from different sources to be accurately aligned in space, eliminating the analysis difficulties caused by position differences and ensuring the accuracy of subsequent fusion and analysis; this embodiment improves the availability and credibility of multi-source remote sensing data as a whole, so that the fused remote sensing data can more accurately reflect the actual situation of the target power grid.

[0047] In one embodiment, the step of performing data cleaning, calibration, and registration on the initial multi-source remote sensing data to obtain the target multi-source remote sensing data includes:

[0048] S121: Using a median filtering algorithm, filtering the initial multi-source remote sensing data to obtain first remote sensing data;

[0049] Specifically, a median filtering algorithm is used to traverse each pixel point of the initial multi-source remote sensing data, sort the pixel values ​​in the neighborhood of the calculated pixel point, and take the middle value of each sorted pixel value as the new pixel value of the pixel point.

[0050] S122: Using a mean filtering algorithm to filter the first remote sensing data to obtain second remote sensing data;

[0051] Specifically, a mean filtering algorithm is used to traverse each pixel point of the initial multi-source remote sensing data, and the average value of the pixel values ​​in the neighborhood of the pixel point is calculated as the new pixel value of the pixel point. After the first remote sensing data is filtered using the mean filtering algorithm, the filtered first remote sensing data is used as the second remote sensing data.

[0052] S123: performing radiation calibration on the second remote sensing data to obtain third remote sensing data;

[0053] The linear fitting algorithm is a method to find the best estimated value of the parameter through the observed value, which is used to find the best theoretical curve. Linear fitting is particularly suitable for compressing data in real-time databases to reduce storage capacity. The revolving door algorithm is one of the efficient linear fitting algorithms, which has the advantages of fast execution speed, high compression ratio, simple implementation and controllable error.

[0054] Specifically, before performing radiation calibration on the second remote sensing data, first, a large number of measurement data points of a known radiation standard field are obtained, and these data points include different radiation values ​​and corresponding image grayscale values. Then, these data points are plotted in a coordinate system, and then a linear fitting algorithm (such as the least squares method) is used to fit a straight line that best represents the trend of these data points. The equation of the straight line can be expressed as y=kx+b (where y represents the radiation value, x represents the image grayscale value, k is the slope, and b is the intercept).

[0055] The specific implementation method of the radiation calibration of the second remote sensing data is: first, for each pixel in the second remote sensing data, the grayscale value is substituted into the equation of the fitted straight line to calculate the radiation value corresponding to the grayscale value; then, according to the set calibration target, the calculated radiation value may need to be adjusted or transformed, such as multiplying by a specific coefficient or performing other correction operations; finally, the calibrated radiation value is reallocated to the corresponding pixel point to obtain the calibrated remote sensing data, and the calibrated remote sensing data is used as the third remote sensing data.

[0056] S124: Based on the SI FT algorithm and the bilinear interpolation algorithm, perform image registration on the third remote sensing data to obtain the target multi-source remote sensing data.

[0057] The SI FT algorithm is a scale-invariant feature extraction method that constructs a scale space through a Gaussian pyramid, detects the extreme points of the DoG space (Difference of Gauss i an) as key points, removes unstable points, assigns directions, and generates descriptors with rotation, scale, and illumination invariance.

[0058] Specifically, the feature points are extracted from the third remote sensing data using the SI FT algorithm, and matched using feature descriptors; the transformation matrix is ​​calculated based on the matching point pairs, and the image is resampled using the bilinear interpolation algorithm to achieve registration, and the registered data is used as the target multi-source remote sensing data, thereby accurately aligning images from different sources.

[0059] This embodiment effectively removes noise, outliers, invalid data, etc. in the initial multi-source remote sensing data through data cleaning, improves the purity and quality of the data, and provides a more reliable basis for subsequent processing; the calibration operation ensures the accuracy and consistency of each remote sensing data source, so that data from different sources are unified in measurement standards and errors caused by data deviation are reduced; the registration enables remote sensing data from different sources to be accurately aligned in space, eliminating the analysis difficulties caused by position differences and ensuring the accuracy of subsequent fusion and analysis; overall, the availability and credibility of multi-source remote sensing data are improved, so that the fused remote sensing data can more accurately reflect the actual situation of the power grid.

[0060] In one embodiment, the step of fusing the target multi-source remote sensing data to obtain fused remote sensing data includes:

[0061] S21: Inputting the target multi-source remote sensing data into a preset fusion mode model to perform classification prediction of the fusion mode of each pixel point to obtain a fusion mode prediction result;

[0062] Specifically, the target multi-source remote sensing data are spliced ​​and input into a preset fusion mode model to perform classification prediction of the fusion mode of each pixel point, and the vector element with the largest element value (that is, probability value) is extracted from each vector element corresponding to each pixel point in the predicted vector, and the classification category corresponding to the extracted vector element (expressing the pixel point position and fusion mode identifier) ​​is used as the single pixel result corresponding to the pixel point; and each single pixel result corresponding to the target multi-source remote sensing data is used as the fusion mode prediction result.

[0063] The fusion mode identifier may be data that uniquely identifies a fusion mode, such as a fusion mode name and a fusion mode ID.

[0064] Fusion methods include: weighted fusion and principal component fusion.

[0065] The preset fusion mode model is a pre-trained multi-classification model. The model structure and model training method of the preset fusion mode model can be selected from the existing technology and will not be described in detail here.

[0066] S22: In the fusion mode prediction result, if the pixel value of the designated pixel point is the first value, determining that the single pixel point fusion mode corresponding to the designated pixel point is weighted fusion; if the pixel value of the designated pixel point is the second value, determining that the single pixel point fusion mode corresponding to the designated pixel point is principal component fusion, wherein the designated pixel point is any one pixel point among all the pixel points corresponding to the fusion mode prediction result;

[0067] Specifically, in the fusion mode prediction result, if the pixel value of the designated pixel is the first value, this means that the fusion mode of the designated pixel is weighted fusion, therefore, the single pixel fusion mode corresponding to the designated pixel is determined to be weighted fusion; in the fusion mode prediction result, if the pixel value of the designated pixel is the second value, this means that the fusion mode of the designated pixel is principal component fusion, therefore, the single pixel fusion mode corresponding to the designated pixel is determined to be principal component fusion. For all pixels corresponding to the fusion mode prediction result, the step of determining that in the fusion mode prediction result, if the pixel value of the designated pixel is the first value, the single pixel fusion mode corresponding to the designated pixel is weighted fusion, and if the pixel value of the designated pixel is the second value, the single pixel fusion mode corresponding to the designated pixel is principal component fusion is performed, and the single pixel fusion mode of each pixel corresponding to the fusion mode prediction result can be determined.

[0068] Weighted fusion is fusion based on weighted summation.

[0069] Principal component fusion is based on the principal component analysis algorithm. The principal component analysis algorithm is to perform principal component analysis, extract the main component image, select the first few main components for fusion, and convert the fused main component image back to the original image space.

[0070] S23: fusing each pixel value corresponding to the designated pixel point in the target multi-source remote sensing data according to the single pixel fusion mode corresponding to the designated pixel point, to obtain a fusion value corresponding to the designated pixel point;

[0071] Specifically, according to the fusion algorithm corresponding to the single pixel fusion method corresponding to the designated pixel, each pixel value corresponding to the designated pixel in the target multi-source remote sensing data is fused, and the fused value is used as the fusion value corresponding to the designated pixel, thereby realizing the fusion of multiple data (from different data sources) for the same point into one value.

[0072] S24: Determine the fused remote sensing data according to each of the fusion values ​​corresponding to the target multi-source remote sensing data.

[0073] Specifically, according to the position distribution relationship of the pixel points corresponding to the target multi-source remote sensing data, the fusion values ​​corresponding to the target multi-source remote sensing data are arranged to obtain the fused remote sensing data. In other words, the pixel points at the same position of the target multi-source remote sensing data and the fused remote sensing data correspond to the same detection point of the power transmission area of ​​the target area.

[0074] By fusing multi-source remote sensing data, this embodiment can comprehensively utilize the advantages of different data sources, more comprehensively obtain information on the transmission area corresponding to the target power grid, effectively deal with the problems of large differences in ground object scales, complex and diverse ground object types, and unbalanced distribution of ground object categories, thereby providing a basis for improving the accuracy and completeness of change detection results.

[0075] In one embodiment, after the step of inputting the target multi-source remote sensing data into a preset fusion mode model to perform classification prediction of the fusion mode of each pixel point and obtaining the fusion mode prediction result, the step further includes:

[0076] S251: if there is a pixel value of the first value in the prediction result of the fusion method, determine weighted weight data according to the complexity level data of the ground object corresponding to the target power grid and the target multi-source remote sensing data;

[0077] Specifically, if there is a pixel value that is the first value in the prediction result of the fusion method, it means that weighted fusion is required. Therefore, the complexity level data of the target power grid and the target multi-source remote sensing data are spliced ​​and input into the pre-trained weight prediction model for classification prediction. The vector element with the largest element value (that is, probability value) is selected from the predicted vector, and the classification category (weight configuration identifier) ​​corresponding to the selected vector element is used as the first identifier, and the configuration data corresponding to the first identifier is used as the weighted weight data.

[0078] The feature complexity level data describes the corresponding relationship between the location of the feature and the complexity level of the feature.

[0079] The pre-trained weight prediction model is a pre-trained multi-classification model. The model structure and model training method of the pre-trained weight prediction model can be selected from the existing technology and will not be described in detail here.

[0080] S252: if there is a pixel value of the second value in the prediction result of the fusion method, the target multi-source remote sensing data is divided into blocks according to regions to obtain each remote sensing data block, and the remote sensing data block is input into a pre-trained principal component prediction model to perform classification prediction of the principal component image to obtain a principal component image prediction result corresponding to the remote sensing data block;

[0081] Specifically, if there is a pixel value of the second value in the prediction result of the fusion method, it means that principal component fusion is required. The target multi-source remote sensing data is divided into blocks by region using the preset regional block method, and the data block corresponding to each block is used as a remote sensing data block. The remote sensing data block is input into the pre-trained principal component prediction model for classification prediction of the principal component image, and the vector element with the largest element value (that is, the probability value) is selected from the predicted vector. The classification category corresponding to the selected vector element is used as the second identifier, and the principal component configuration corresponding to the second identifier is used as the principal component image prediction result corresponding to the remote sensing data block. The remote sensing data block is all the data corresponding to a region in the target multi-source remote sensing data, and the region refers to the sub-region of the transmission area corresponding to the target power grid.

[0082] The principal component configuration describes the corresponding relationship between the data source identifier and the fusion ratio. The data source identifier can be an ID, name, or other data that uniquely identifies the source of a remote sensing data.

[0083] The remote sensing data block includes all data corresponding to the area corresponding to the block in the target multi-source remote sensing data.

[0084] The pre-trained principal component prediction model is a pre-trained multi-classification model. The model structure and model training method of the pre-trained principal component prediction model can be selected from the existing technology and will not be described in detail here.

[0085] The step of fusing the pixel values ​​corresponding to the designated pixel in the target multi-source remote sensing data according to the single pixel fusion mode corresponding to the designated pixel to obtain the fusion value corresponding to the designated pixel includes:

[0086] S231: If the single pixel fusion mode corresponding to the designated pixel is weighted fusion, then, according to the weighted weight data, each pixel value corresponding to the designated pixel in the target multi-source remote sensing data is fused to obtain the fusion value corresponding to the designated pixel;

[0087] Specifically, if the single pixel fusion method corresponding to the designated pixel point is weighted fusion, it means that the designated pixel point needs to be weighted summed. Therefore, according to the weighted weight data, the pixel values ​​corresponding to the designated pixel point in the target multi-source remote sensing data are weighted summed to achieve fusion, and the data obtained by weighted sum is used as the fusion value corresponding to the designated pixel point.

[0088] S232: If the single pixel fusion mode corresponding to the designated pixel is principal component fusion, the principal component image prediction result corresponding to the remote sensing data block corresponding to the designated pixel is used to fuse each pixel value corresponding to the designated pixel in the target multi-source remote sensing data to obtain the fusion value corresponding to the designated pixel.

[0089] Specifically, if the single pixel fusion mode corresponding to the designated pixel is principal component fusion, it means that the designated pixel needs to be fused by principal component analysis. Therefore, first, the data source identifier of the principal component image prediction result corresponding to the remote sensing data block corresponding to the designated pixel is used to extract pixel values ​​from each pixel value corresponding to the designated pixel in the target multi-source remote sensing data as the value to be processed. Then, the data source identifier and fusion ratio of the principal component image prediction result corresponding to the remote sensing data block corresponding to the designated pixel are used to perform weighted summation on all the values ​​to be processed, and the data obtained by the weighted sum is used as the fusion value corresponding to the designated pixel.

[0090] In the case where the pixel value is the first value, this embodiment determines the weighted weight data according to the complexity level data of the target power grid and the target multi-source remote sensing data, which can more accurately consider the characteristic differences of different regions, make the fusion result more in line with the actual scene, improve the accuracy and adaptability of the fusion, and achieve fine control of the fusion process. The weight can be dynamically adjusted according to the specific situation to improve the fusion quality; in the case where the pixel value is the second value, the target multi-source remote sensing data is divided into blocks by region and the principal component image prediction is performed using a pre-trained model, which can process the data more carefully and effectively extract important features, which helps to better deal with complex ground conditions and improve the pertinence and accuracy of the fusion effect through principal component fusion. In other words, the classification processing and targeted fusion methods of this embodiment enhance the flexibility and adaptability of the fusion process and can better deal with different scenes and data characteristics.

[0091] In one embodiment, the step of performing change detection according to each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid includes:

[0092] S31: Divide each of the fused remote sensing data in the preset time window into blocks according to regions, and obtain remote sensing data packets corresponding to each block;

[0093] Specifically, each of the fused remote sensing data within the preset time window is divided into blocks according to regions, and all data corresponding to the same region in each of the fused remote sensing data within the preset time window is taken as a remote sensing data packet.

[0094] S32: extracting features from the remote sensing data packet to obtain single-packet feature data;

[0095] Specifically, a model based on gray-level co-occurrence matrix algorithm is used to extract texture features of the remote sensing data packet, such as contrast, correlation and other features, as first feature data; a model based on neural network training is used to extract spectral features of the remote sensing data packet, such as band reflectance, vegetation coefficient and other features, as second feature data; the first feature data and the second feature data corresponding to the same remote sensing data packet are used as single-packet feature data.

[0096] S33: inputting the preset difference threshold, the preset ratio threshold and the single-packet feature data into a pre-trained change prediction model to perform classification prediction of the change type, and obtain single-block change data;

[0097] The preset difference threshold is a preset difference threshold, which is a constant. The preset ratio threshold is a preset ratio threshold, which is a constant.

[0098] Specifically, the preset difference threshold, the preset ratio threshold and the single package feature data are spliced ​​and input into the pre-trained change prediction model to perform classification prediction of the change type, and the vector element with the largest element value is extracted from the predicted vector, and the classification category corresponding to the extracted vector element is used as a single block of change data.

[0099] The pre-trained change prediction model is a multi-classification model.

[0100] S34: Determine the change detection result corresponding to the target power grid according to each of the single block change data.

[0101] Specifically, a preset combination method is adopted to combine the single-block change data to obtain the change detection result corresponding to the target power grid.

[0102] Optionally, the preset combination mode is a preset splicing order.

[0103] This embodiment targets each of the fused remote sensing data within a preset time window, divides them into blocks by region, and then classifies and predicts the change type for each block, thereby performing change detection with the help of the fused remote sensing data within the preset time window, and can dynamically track the change trend of the power grid, providing valuable reference for long-term planning and decision-making.

[0104] In one embodiment, before the step of inputting the preset difference threshold, the preset ratio threshold and the single packet feature data into the pre-trained change prediction model to perform classification prediction of the change type to obtain the single block change data, the step further includes:

[0105] S41: Obtain a first initial model, a second initial model, a third initial model, a training set, a validation set, and a test set, wherein the first initial model is a classification model constructed based on a Siam Network-UNet twin neural network model, the second initial model is a classification model constructed based on a DeepLabV3+ model and a twin neural network model, and the third initial model is a classification model constructed based on an HRNet network;

[0106] The training set includes multiple training samples. The training samples include: sample data and training labels.

[0107] The validation set includes multiple validation samples, which include sample data and validation labels.

[0108] The test set includes multiple test samples. The test samples include: sample data and test labels.

[0109] The sample data is obtained by fusing the multi-source remote sensing data corresponding to the power grid.

[0110] Specifically, the first initial model includes: a first feature extraction unit and a first classification layer. The first feature extraction unit adopts a Siam Network-UNet twin neural network model. The first classification layer is used to perform classification prediction based on the features extracted by the first feature extraction unit. The first classification layer includes: a fully connected layer and an activation layer connected in sequence. The activation layer adopts a softmax activation function. The softmax function refers to a normalized exponential function.

[0111] Siam Network (Twin Neural Network) is a neural network architecture for similarity learning tasks. It usually consists of two sub-networks with the same structure and shared parameters. The two sub-networks receive two inputs (such as two images) respectively, and then judge the similarity between the two by comparing and analyzing their features. Siam Network-UNet combines the twin neural network with the UNet architecture. UNet is a network model commonly used for image segmentation, with good feature extraction and spatial information processing capabilities. This combined model can use the twin neural network to perform similarity analysis on two inputs, while also having the advantages of UNet in image feature processing and segmentation. It can be applied to some tasks that require simultaneous consideration of similarity judgment and image precision processing, such as tracking and segmentation of specific targets.

[0112] The second initial model includes: a second feature extraction unit and a second classification layer, the second feature extraction unit includes a DeepLabV3+ model and a twin neural network model in sequence, and the second classification layer is used to perform classification prediction according to the features extracted by the second feature extraction unit. The second classification layer includes: a fully connected layer and an activation layer connected in sequence.

[0113] The DeepLabV3+ model is a semantic segmentation model that performs well in image segmentation tasks and can accurately divide and label different areas in an image.

[0114] The third initial model includes: a third feature extraction unit and a third classification layer, the third feature extraction unit adopts the HRNet network, and the third classification layer is used to perform classification prediction according to the features extracted by the third feature extraction unit. The third classification layer includes: a fully connected layer and an activation layer connected in sequence.

[0115] HRNet (High-Resolution Network) is a network model that performs well in computer vision tasks, especially in image classification and semantic segmentation. The main feature of HRNet is that it always maintains high-resolution feature representation throughout the network, and by connecting subnetworks of different resolutions in parallel and performing multi-scale feature fusion, it can better capture the details and spatial information in the image to achieve more accurate prediction and recognition effects.

[0116] S42: using the training set to train the first initial model and using the verification set to verify the trained first initial model, and using the first initial model with a passing result of the verification as the first model to be screened;

[0117] Specifically, the first initial model is trained using the training set and the first initial model after training is verified using the verification set. If the verification result is not passed, the training set is first randomly sorted, and then the step of training the first initial model using the training set and verifying the first initial model after training using the verification set is jumped to and continued until the verification result is passed. If the first initial model whose verification result is passed is a model whose performance meets the training requirements, the first initial model whose verification result is passed is used as the first model to be screened.

[0118] S43: using the training set to train the second initial model and using the verification set to verify the trained second initial model, and using the second initial model with a verification result of passing as the second model to be screened;

[0119] Specifically, the second initial model is trained using the training set and the trained second initial model is verified using the verification set. If the verification result is not passed, the training set is first randomly sorted, and then the step of training the second initial model using the training set and verifying the trained second initial model using the verification set is continued until the verification result is passed. If the second initial model with a passed verification result is a model whose performance meets the training requirements, the second initial model with a passed verification result is used as the second model to be screened.

[0120] S44: using the training set to train the third initial model and using the verification set to verify the trained third initial model, and using the third initial model with a passing result of the verification as the third model to be screened;

[0121] Specifically, the third initial model is trained using the training set and the third initial model after training is verified using the verification set. If the verification result is not passed, the training set is first randomly sorted, and then the step of training the third initial model using the training set and verifying the third initial model after training using the verification set is continued until the verification result is passed. If the third initial model whose verification result is passed is a model whose performance meets the training requirements, the third initial model whose verification result is passed is used as the third model to be screened.

[0122] S45: According to the test set, a model with the highest accuracy is selected from the first model to be screened, the second model to be screened, and the third model to be screened as the change prediction model.

[0123] Specifically, the sample data of each test sample in the test set is input into the first screening model for prediction, all prediction results corresponding to the first screening model are compared with the test labels in the test set of the test samples in the test set to obtain a first comparison result, and the number of correct first comparison results is divided by the number of test samples in the test set to obtain a first accuracy rate; the sample data of each test sample in the test set is input into the second screening model for prediction, all prediction results corresponding to the second screening model are compared with the test labels in the test set of the test samples in the test set to obtain a second comparison result, and the number of correct second comparison results is divided by the number of test samples in the test set to obtain a second accuracy rate; the sample data of each test sample in the test set is input into the third screening model for prediction, all prediction results corresponding to the third screening model are compared with the test labels in the test set of the test samples in the test set to obtain a third comparison result, and the number of correct third comparison results is divided by the number of test samples in the test set to obtain a third accuracy rate. The maximum value is selected from the first accuracy rate, the second accuracy rate, and the third accuracy rate, and the model corresponding to the maximum value (one of the first model to be screened, the second model to be screened, and the third model to be screened) is used as the change prediction model.

[0124] It can be understood that the change prediction model of the present application can also be a classification model constructed by other network models, for example, a classification model constructed based on the Bert model (Bidirectional Encoder Representations from Transformers).

[0125] This embodiment selects the model with the highest accuracy as the change prediction model after training three models, thereby improving the prediction performance of the change prediction model and further improving the accuracy of the change detection result.

[0126] See also Figure 3 As shown, in one embodiment, a power grid multi-source remote sensing data fusion and change detection device is provided, the device comprising:

[0127] The data acquisition module 801 is used to acquire target multi-source remote sensing data corresponding to the target power grid;

[0128] A fusion processing module 802 is used to perform fusion processing on the target multi-source remote sensing data to obtain fused remote sensing data;

[0129] The change detection module 803 is used to perform change detection according to each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid.

[0130] By fusing multi-source remote sensing data, this embodiment can comprehensively utilize the advantages of different data sources, more comprehensively obtain information on the transmission area corresponding to the target power grid, and effectively deal with the problems of large differences in ground object scales, complex and diverse ground object types, and unbalanced distribution of ground object categories, thereby providing a basis for improving the accuracy and completeness of change detection results; by using the fused remote sensing data within a preset time window to perform change detection, it is possible to dynamically track the changing trend of the power grid and provide a valuable reference for long-term planning and decision-making.

[0131] In one embodiment, the step of acquiring target multi-source remote sensing data corresponding to the target power grid in the data acquisition module 801 includes:

[0132] Acquiring initial multi-source remote sensing data corresponding to the target power grid;

[0133] The initial multi-source remote sensing data is cleaned, calibrated and registered to obtain the target multi-source remote sensing data.

[0134] In one embodiment, the step of performing data cleaning, calibration, and registration on the initial multi-source remote sensing data to obtain the target multi-source remote sensing data in the data acquisition module 801 includes:

[0135] Using a median filtering algorithm to filter the initial multi-source remote sensing data to obtain first remote sensing data;

[0136] Using a mean filtering algorithm to filter the first remote sensing data to obtain second remote sensing data;

[0137] performing radiation calibration on the second remote sensing data to obtain third remote sensing data;

[0138] Based on the SI FT algorithm and the bilinear interpolation algorithm, image registration is performed on the third remote sensing data to obtain the target multi-source remote sensing data.

[0139] In one embodiment, the step of fusing the target multi-source remote sensing data to obtain fused remote sensing data in the fusion processing module 802 includes:

[0140] Inputting the target multi-source remote sensing data into a preset fusion mode model to perform classification prediction of the fusion mode of each pixel point to obtain a fusion mode prediction result;

[0141] In the fusion mode prediction result, if the pixel value of the designated pixel point is the first value, then the single pixel point fusion mode corresponding to the designated pixel point is determined to be weighted fusion; if the pixel value of the designated pixel point is the second value, then the single pixel point fusion mode corresponding to the designated pixel point is determined to be principal component fusion, wherein the designated pixel point is any pixel point among all the pixel points corresponding to the fusion mode prediction result;

[0142] According to the single pixel fusion mode corresponding to the designated pixel point, each pixel value corresponding to the designated pixel point in the target multi-source remote sensing data is fused to obtain a fusion value corresponding to the designated pixel point;

[0143] The fused remote sensing data is determined according to each of the fusion values ​​corresponding to the target multi-source remote sensing data.

[0144] In one embodiment, after the step of inputting the target multi-source remote sensing data into a preset fusion mode model to perform classification prediction of the fusion mode of each pixel point and obtaining the fusion mode prediction result in the fusion processing module 802, the step further includes:

[0145] If there is a pixel value of the first value in the prediction result of the fusion method, weighted weight data is determined according to the ground object complexity level data corresponding to the target power grid and the target multi-source remote sensing data;

[0146] If there is a pixel value of the second value in the prediction result of the fusion method, the target multi-source remote sensing data is divided into blocks according to regions to obtain each remote sensing data block, and the remote sensing data block is input into a pre-trained principal component prediction model to perform classification prediction of the principal component image to obtain a principal component image prediction result corresponding to the remote sensing data block;

[0147] The step of fusing the pixel values ​​corresponding to the designated pixel in the target multi-source remote sensing data according to the single pixel fusion mode corresponding to the designated pixel to obtain the fusion value corresponding to the designated pixel includes:

[0148] If the single pixel fusion mode corresponding to the designated pixel point is weighted fusion, then, according to the weighted weight data, each pixel value corresponding to the designated pixel point in the target multi-source remote sensing data is fused to obtain the fusion value corresponding to the designated pixel point;

[0149] If the single pixel fusion method corresponding to the designated pixel is principal component fusion, the principal component image prediction result corresponding to the remote sensing data block corresponding to the designated pixel is used to fuse each pixel value corresponding to the designated pixel in the target multi-source remote sensing data to obtain the fusion value corresponding to the designated pixel.

[0150] In one embodiment, the step of performing change detection according to each of the fused remote sensing data within a preset time window in the change detection module 803 to obtain a change detection result corresponding to the target power grid includes:

[0151] Dividing each of the fused remote sensing data in the preset time window into blocks by region to obtain remote sensing data packets corresponding to each block;

[0152] Extracting features from the remote sensing data packet to obtain single-packet feature data;

[0153] Inputting a preset difference threshold, a preset ratio threshold and the single-packet feature data into a pre-trained change prediction model to perform classification prediction of the change type, and obtain a single-block change data;

[0154] The change detection result corresponding to the target power grid is determined according to each of the single-block change data.

[0155] In one embodiment, the device further comprises: a model training module, wherein the model training module is used to:

[0156] Obtain a first initial model, a second initial model, a third initial model, a training set, a validation set, and a test set, wherein the first initial model is a classification model constructed based on the Siam Network-UNet twin neural network model, the second initial model is a classification model constructed based on the DeepLabV3+ model and the twin neural network model, and the third initial model is a classification model constructed based on the HRNet network;

[0157] The first initial model is trained by using the training set and the first initial model after training is verified by using the verification set, and the first initial model with a passing result of the verification is used as the first model to be screened;

[0158] The second initial model is trained by using the training set and the trained second initial model is verified by using the verification set, and the second initial model with a passing result of the verification is used as the second model to be screened;

[0159] The third initial model is trained by using the training set and the third initial model is verified by using the verification set, and the third initial model with a passing result of the verification is used as the third model to be screened;

[0160] According to the test set, the model with the highest accuracy is selected from the first model to be screened, the second model to be screened, and the third model to be screened as the change prediction model.

[0161] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the service side of a power grid multi-source remote sensing data fusion and change detection method.

[0162] In one embodiment, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the following steps are implemented:

[0163] Obtain target multi-source remote sensing data corresponding to the target power grid;

[0164] Performing fusion processing on the target multi-source remote sensing data to obtain fused remote sensing data;

[0165] Change detection is performed based on each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid.

[0166] By fusing multi-source remote sensing data, this embodiment can comprehensively utilize the advantages of different data sources, more comprehensively obtain information on the transmission area corresponding to the target power grid, and effectively deal with the problems of large differences in ground object scales, complex and diverse ground object types, and unbalanced distribution of ground object categories, thereby providing a basis for improving the accuracy and completeness of change detection results; by using the fused remote sensing data within a preset time window to perform change detection, it is possible to dynamically track the changing trend of the power grid and provide a valuable reference for long-term planning and decision-making.

[0167] In one embodiment, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0168] Obtain target multi-source remote sensing data corresponding to the target power grid;

[0169] Performing fusion processing on the target multi-source remote sensing data to obtain fused remote sensing data;

[0170] Change detection is performed based on each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid.

[0171] By fusing multi-source remote sensing data, this embodiment can comprehensively utilize the advantages of different data sources, more comprehensively obtain information on the transmission area corresponding to the target power grid, and effectively deal with the problems of large differences in ground object scales, complex and diverse ground object types, and unbalanced distribution of ground object categories, thereby providing a basis for improving the accuracy and completeness of change detection results; by using the fused remote sensing data within a preset time window to perform change detection, it is possible to dynamically track the changing trend of the power grid and provide a valuable reference for long-term planning and decision-making.

[0172] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can refer to the relevant descriptions on the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0173] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0174] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0175] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for fusion and change detection of multi-source remote sensing data of power grid, characterized in that: The method comprises: Obtain target multi-source remote sensing data corresponding to the target power grid; Performing fusion processing on the target multi-source remote sensing data to obtain fused remote sensing data; Change detection is performed based on each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid.

2. The method for fusion and change detection of multi-source remote sensing data of power grid according to claim 1, characterized in that: The step of obtaining target multi-source remote sensing data corresponding to the target power grid includes: Acquiring initial multi-source remote sensing data corresponding to the target power grid; The initial multi-source remote sensing data is cleaned, calibrated and registered to obtain the target multi-source remote sensing data.

3. The method for fusion and change detection of multi-source remote sensing data of power grid according to claim 2 is characterized in that: The step of performing data cleaning, calibration, and registration on the initial multi-source remote sensing data to obtain the target multi-source remote sensing data includes: Using a median filtering algorithm to filter the initial multi-source remote sensing data to obtain first remote sensing data; Using a mean filtering algorithm to filter the first remote sensing data to obtain second remote sensing data; performing radiation calibration on the second remote sensing data to obtain third remote sensing data; Based on the SIFT algorithm and the bilinear interpolation algorithm, image registration is performed on the third remote sensing data to obtain the target multi-source remote sensing data.

4. The method for fusion and change detection of multi-source remote sensing data of power grid according to claim 1, characterized in that: The step of fusing the target multi-source remote sensing data to obtain fused remote sensing data comprises: Inputting the target multi-source remote sensing data into a preset fusion mode model to perform classification prediction of the fusion mode of each pixel point to obtain a fusion mode prediction result; In the fusion mode prediction result, if the pixel value of the designated pixel point is the first value, then the single pixel point fusion mode corresponding to the designated pixel point is determined to be weighted fusion; if the pixel value of the designated pixel point is the second value, then the single pixel point fusion mode corresponding to the designated pixel point is determined to be principal component fusion, wherein the designated pixel point is any pixel point among all the pixel points corresponding to the fusion mode prediction result; According to the single pixel fusion mode corresponding to the designated pixel point, each pixel value corresponding to the designated pixel point in the target multi-source remote sensing data is fused to obtain a fusion value corresponding to the designated pixel point; The fused remote sensing data is determined according to each of the fusion values ​​corresponding to the target multi-source remote sensing data.

5. The method for fusion and change detection of multi-source remote sensing data of power grid according to claim 4, characterized in that: After the step of inputting the target multi-source remote sensing data into a preset fusion mode model to perform classification prediction of the fusion mode of each pixel point and obtaining the fusion mode prediction result, the method further includes: If there is a pixel value of the first value in the prediction result of the fusion method, weighted weight data is determined according to the ground object complexity level data corresponding to the target power grid and the target multi-source remote sensing data; If there is a pixel value of the second value in the prediction result of the fusion method, the target multi-source remote sensing data is divided into blocks according to regions to obtain each remote sensing data block, and the remote sensing data block is input into a pre-trained principal component prediction model to perform classification prediction of the principal component image to obtain a principal component image prediction result corresponding to the remote sensing data block; The step of fusing the pixel values ​​corresponding to the designated pixel in the target multi-source remote sensing data according to the single pixel fusion mode corresponding to the designated pixel to obtain the fusion value corresponding to the designated pixel includes: If the single pixel fusion mode corresponding to the designated pixel point is weighted fusion, then, according to the weighted weight data, each pixel value corresponding to the designated pixel point in the target multi-source remote sensing data is fused to obtain the fusion value corresponding to the designated pixel point; If the single pixel fusion method corresponding to the designated pixel is principal component fusion, the principal component image prediction result corresponding to the remote sensing data block corresponding to the designated pixel is used to fuse each pixel value corresponding to the designated pixel in the target multi-source remote sensing data to obtain the fusion value corresponding to the designated pixel.

6. The method for fusion and change detection of multi-source remote sensing data of power grid according to claim 1, characterized in that: The step of performing change detection according to each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid includes: Dividing each of the fused remote sensing data in the preset time window into blocks by region to obtain remote sensing data packets corresponding to each block; Extracting features from the remote sensing data packet to obtain single-packet feature data; Inputting a preset difference threshold, a preset ratio threshold and the single-packet feature data into a pre-trained change prediction model to perform classification prediction of the change type, and obtain a single-block change data; The change detection result corresponding to the target power grid is determined according to each of the single-block change data.

7. The method for fusion and change detection of multi-source remote sensing data of power grid according to claim 6, characterized in that: Before the step of inputting the preset difference threshold, the preset ratio threshold and the single-packet feature data into the pre-trained change prediction model to perform classification prediction of the change type to obtain the single-block change data, the step further includes: Obtain a first initial model, a second initial model, a third initial model, a training set, a validation set, and a test set, wherein the first initial model is a classification model constructed based on the Siam Network-UNet twin neural network model, the second initial model is a classification model constructed based on the DeepLabV3+ model and the twin neural network model, and the third initial model is a classification model constructed based on the HRNet network; The first initial model is trained by using the training set and the first initial model after training is verified by using the verification set, and the first initial model with a passing result of the verification is used as the first model to be screened; The second initial model is trained by using the training set and the trained second initial model is verified by using the verification set, and the second initial model with a passing result of the verification is used as the second model to be screened; The third initial model is trained by using the training set and the third initial model is verified by using the verification set, and the third initial model with a passing result of the verification is used as the third model to be screened; According to the test set, the model with the highest accuracy is selected from the first model to be screened, the second model to be screened, and the third model to be screened as the change prediction model.

8. A power grid multi-source remote sensing data fusion and change detection device, characterized in that: The device comprises: A data acquisition module is used to acquire target multi-source remote sensing data corresponding to the target power grid; A fusion processing module is used to perform fusion processing on the target multi-source remote sensing data to obtain fused remote sensing data; The change detection module is used to perform change detection according to each of the fused remote sensing data within a preset time window to obtain a change detection result corresponding to the target power grid.

9. A computer device, characterized in that: The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the power grid multi-source remote sensing data fusion and change detection method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the power grid multi-source remote sensing data fusion and change detection method according to any one of claims 1 to 7 are implemented.

Images