Multi-source remote sensing vegetation coverage inversion product fusion method, device and equipment

Abstract

The invention provides a multi-source remote sensing vegetation coverage inversion product fusion method, device and equipment, and relates to the technical field of vegetation coverage monitoring, and the method comprises the steps: obtaining a multi-source remote sensing vegetation coverage inversion product; carrying out authenticity verification on the remote sensing vegetation coverage inversion products so as to screen out an optimal remote sensing vegetation coverage inversion product corresponding to each resolution; the optimal remote sensing vegetation coverage inversion product corresponding to the first resolution is used as a true value, the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution except the first resolution is used as input data, a training data set is constructed, and the first resolution is higher than the second resolution; and training the neural network by using the training data set to obtain a target neural network corresponding to the second resolution. According to the method, the space-time resolution and the accuracy of a remote sensing vegetation coverage inversion product can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing data fusion and vegetation coverage monitoring, and in particular to a method, device and equipment for fusing multi-source remote sensing vegetation coverage inversion products. Background Art

[0002] At present, the monitoring methods of vegetation coverage are divided into two categories: ground measurement and remote sensing monitoring. Ground measurement is of great value in small-scale, high-precision vegetation coverage research, but due to environmental conditions, time and cost constraints, the spatial coverage of ground measurement is limited, and it is difficult to quickly cover a large area. Therefore, for large-scale research, satellite remote sensing monitoring has become an irreplaceable and important means. Medium-resolution satellite data represented by the Fengyun-3 series and MODIS (Moderate-resolution Imaging Spectroradiometer) series have high update frequency and wide observation range, which are suitable for large-scale, rapidly changing vegetation dynamic monitoring, but their spatial resolution is low (250m-1km), which limits their application in small-scale detail analysis. High-resolution satellite data represented by the Gaofen series, Landsat series and Sentinel series have high spatial resolution (2m-30m) and are suitable for refined vegetation monitoring, but the repetition period is long, and there are certain limitations in monitoring rapidly changing vegetation dynamics. Therefore, integrating the temporal and spatial advantages of multi-source satellites to obtain vegetation coverage monitoring results with high temporal and spatial resolution has become a key means to solve this problem.

[0003] In recent years, remote sensing vegetation cover fusion technology has made significant progress, mainly including spatiotemporal fusion models, machine learning models and deep learning models. Spatiotemporal fusion models such as STARFM (Star Domain Factor Method), ESTARFM (Enhanced STARFM Algorithm) and FSDAF (Flexible Spatiotemporal Data Fusion) use MODIS and Landsat data to generate high-resolution images through spatial and temporal weighting, but they are still insufficient in processing high-frequency vegetation cover data. Machine learning models such as random forests and support vector machines have achieved certain results in multi-source remote sensing data fusion, but there are performance bottlenecks when processing high-dimensional and complex scenes. Deep learning models such as convolutional neural networks (CNN) and generative adversarial networks (GAN) perform well in remote sensing image processing and data fusion, but the training complexity is high and unstable, making it difficult to meet the needs of actual applications.

[0004] Most of the above studies are image-level fusion, and vegetation coverage monitoring is performed based on fused satellite remote sensing image data. However, this method fundamentally changes the observed spectral data and may lead to inaccurate vegetation coverage monitoring results. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a multi-source remote sensing vegetation cover inversion product fusion method, device and equipment, which can significantly improve the temporal and spatial resolution and accuracy of remote sensing vegetation cover inversion products.

[0006] In a first aspect, an embodiment of the present invention provides a method for fusing multi-source remote sensing vegetation coverage inversion products, including:

[0007] Obtain multi-source remote sensing vegetation coverage inversion products, which include multiple remote sensing vegetation coverage inversion products corresponding to different resolutions;

[0008] Verify the authenticity of remote sensing vegetation coverage inversion products, and select the optimal remote sensing vegetation coverage inversion product corresponding to each resolution from multiple remote sensing vegetation coverage inversion products corresponding to each resolution;

[0009] The optimal remote sensing vegetation coverage inversion product corresponding to the first resolution is used as the true value, and the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution other than the first resolution is used as input data to construct a training data set, wherein the first resolution is higher than the second resolution;

[0010] Using the training data set to train the neural network, a target neural network corresponding to the second resolution is obtained;

[0011] Among them, the target neural network is used to generate an optimization result of the inversion product corresponding to the first resolution based on the remote sensing vegetation cover inversion product to be processed corresponding to the second resolution, and the optimization result of the inversion product is used to fuse with the remote sensing vegetation cover inversion product to be processed corresponding to the first resolution to obtain a vegetation coverage fusion product.

[0012] In one embodiment, the remote sensing vegetation coverage inversion product is verified for authenticity, so as to screen out the best remote sensing vegetation coverage inversion product corresponding to each resolution from multiple remote sensing vegetation coverage inversion products corresponding to each resolution, including:

[0013] Obtain the latitude and longitude information and measured values ​​of vegetation coverage corresponding to the ground station;

[0014] Use the latitude and longitude information to match the remote sensing vegetation coverage inversion product with the measured vegetation coverage value in time and space, and use the measured vegetation coverage value matched in time and space to verify the authenticity of the remote sensing vegetation coverage inversion product, and obtain the comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product;

[0015] For each resolution, based on the comprehensive quality evaluation score, the optimal remote sensing vegetation cover inversion product corresponding to the resolution is screened out from multiple remote sensing vegetation cover inversion products corresponding to the resolution.

[0016] In one embodiment, the remote sensing vegetation coverage inversion product is verified for authenticity using the spatiotemporally matched measured vegetation coverage values ​​to obtain a comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product, including:

[0017] Based on the measured values ​​of vegetation coverage matched in time and space, multiple evaluation indicators corresponding to the remote sensing vegetation coverage inversion products are determined;

[0018] The evaluation indicators are standardized to obtain multiple standardized evaluation indicators corresponding to the remote sensing vegetation coverage inversion products;

[0019] Determine the weight value corresponding to each standardized evaluation index, and use the weight value to fuse the standardized evaluation indicators to obtain the comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product.

[0020] In one implementation, the evaluation index includes a root mean square error index, a correlation coefficient index, and a deviation index, and the weight values ​​corresponding to the root mean square error index and the deviation index are higher than the weight value corresponding to the correlation coefficient index.

[0021] In one embodiment, the comprehensive quality evaluation score is negatively correlated with the quality of the remote sensing vegetation coverage inversion product; based on the comprehensive quality evaluation score, the optimal remote sensing vegetation coverage inversion product corresponding to the resolution is screened out from multiple remote sensing vegetation coverage inversion products corresponding to the resolution, including:

[0022] According to the order of comprehensive quality evaluation scores from low to high, the optimal remote sensing vegetation cover inversion product corresponding to the resolution is screened out from multiple remote sensing vegetation cover inversion products corresponding to the resolution.

[0023] In one implementation, the neural network adopts a U-Net network; the neural network is trained using a training data set to obtain a target neural network corresponding to the second resolution, including:

[0024] Through the U-Net network, based on the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution, climate element data, terrain element data and social element data, the optimization result of the inversion product corresponding to the first resolution is generated;

[0025] The loss value is determined based on the optimal remote sensing vegetation cover inversion product corresponding to the first resolution and the inversion product optimization result, and the network parameters of the U-Net network are adjusted using the loss value to obtain the target neural network corresponding to the second resolution.

[0026] In one embodiment, the optimization result of the inversion product corresponding to the first resolution is generated based on the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution, climate element data, terrain element data, and social element data through the U-Net network, including:

[0027] Through the encoder in the U-Net network, convolution and pooling operations are performed on the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution, climate factor data, terrain factor data, and social factor data to obtain a vegetation coverage feature map;

[0028] Through the decoder in the U-Net network, deconvolution operations and jump connection operations are performed on the vegetation coverage feature map to obtain the optimized result of the inversion product corresponding to the first resolution.

[0029] In a second aspect, an embodiment of the present invention further provides a multi-source remote sensing vegetation coverage inversion product fusion device, comprising:

[0030] A product acquisition module is used to obtain multi-source remote sensing vegetation coverage inversion products, which include multiple remote sensing vegetation coverage inversion products corresponding to different resolutions;

[0031] The product screening module is used to verify the authenticity of the remote sensing vegetation coverage inversion product, so as to screen out the optimal remote sensing vegetation coverage inversion product corresponding to each resolution from multiple remote sensing vegetation coverage inversion products corresponding to each resolution;

[0032] A training set construction module is used to construct a training data set using an optimal remote sensing vegetation coverage inversion product corresponding to a first resolution as a true value and an optimal remote sensing vegetation coverage inversion product corresponding to a second resolution other than the first resolution as input data, wherein the first resolution is higher than the second resolution;

[0033] A model training module, used to train the neural network using the training data set to obtain a target neural network corresponding to the second resolution;

[0034] Among them, the target neural network is used to generate an optimization result of the inversion product corresponding to the first resolution based on the remote sensing vegetation cover inversion product to be processed corresponding to the second resolution, and the optimization result of the inversion product is used to fuse with the remote sensing vegetation cover inversion product to be processed corresponding to the first resolution to obtain a vegetation coverage fusion product.

[0035] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement any one of the methods provided in the first aspect.

[0036] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement any one of the methods provided in the first aspect.

[0037] The multi-source remote sensing vegetation coverage inversion product fusion method, device and equipment provided by the embodiment of the present invention first obtain multi-source remote sensing vegetation coverage inversion products, which include multiple remote sensing vegetation coverage inversion products corresponding to different resolutions; then verify the authenticity of the remote sensing vegetation coverage inversion products to screen out the optimal remote sensing vegetation coverage inversion product corresponding to each resolution from the multiple remote sensing vegetation coverage inversion products corresponding to each resolution; then use the optimal remote sensing vegetation coverage inversion product corresponding to the first resolution as the true value, and use the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution other than the first resolution as input data to construct a training data set, wherein the first resolution is higher than the second resolution; finally, use the training data set to train a neural network to obtain a target neural network corresponding to the second resolution, the target neural network is used to generate an optimization result of the inversion product corresponding to the first resolution based on the remote sensing vegetation coverage inversion product to be processed corresponding to the second resolution, and the inversion product optimization result is used to fuse with the remote sensing vegetation coverage inversion product to be processed corresponding to the first resolution to obtain a vegetation coverage fusion product. The above method selects the optimal remote sensing vegetation cover inversion product from remote sensing vegetation cover inversion products of different resolutions through authenticity verification, and uses it to construct a training data set so that the trained target neural network can optimize the lower resolution remote sensing vegetation cover inversion product to be processed, and fuse it with the higher resolution remote sensing vegetation cover inversion product to be processed. Compared with the prior art, the observed spectral data is fundamentally changed. The embodiment of the present invention realizes remote sensing vegetation cover monitoring from the product level, and significantly improves the spatiotemporal resolution and accuracy of the remote sensing vegetation cover inversion product.

[0038] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0039] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 A schematic diagram of a flow chart of a multi-source remote sensing vegetation coverage inversion product fusion method provided in an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of the overall process of a multi-source remote sensing vegetation coverage inversion product fusion method provided by an embodiment of the present invention;

[0043] Figure 3 A schematic diagram of the structure of a multi-source remote sensing vegetation coverage inversion product fusion device provided in an embodiment of the present invention;

[0044] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described in combination with the embodiments below. 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.

[0046] At present, the existing technology fundamentally changes the observed spectral data, which may lead to inaccurate vegetation coverage monitoring results. Based on this, the present invention implements a multi-source remote sensing vegetation coverage inversion product fusion method, device and equipment, which can significantly improve the temporal and spatial resolution and accuracy of remote sensing vegetation coverage inversion products.

[0047] To facilitate understanding of this embodiment, a multi-source remote sensing vegetation coverage inversion product fusion method disclosed in an embodiment of the present invention is first described in detail. Figure 1 The flowchart of a multi-source remote sensing vegetation coverage inversion product fusion method shown in FIG. 1 mainly includes the following steps S102 to S108:

[0048] Step S102, obtaining a multi-source remote sensing vegetation coverage inversion product, where the multi-source remote sensing vegetation coverage inversion product includes a plurality of remote sensing vegetation coverage inversion products corresponding to different resolutions.

[0049] Exemplarily, it includes remote sensing vegetation coverage inversion products corresponding to at least two resolutions, such as medium resolution and high resolution. The remote sensing vegetation coverage inversion products corresponding to medium resolution include FY-3D, FY-3F, MODIS and other products, and the remote sensing vegetation coverage inversion products corresponding to high resolution include Landsat8 / 9, Sentinel-2A / B, GF-1 / 6 and other products.

[0050] Step S104, verifying the authenticity of the remote sensing vegetation coverage inversion product, so as to screen out the best remote sensing vegetation coverage inversion product corresponding to each resolution from multiple remote sensing vegetation coverage inversion products corresponding to each resolution.

[0051] In one example, by verifying the authenticity of the remote sensing vegetation coverage inversion product, the comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product can be obtained. The comprehensive quality evaluation score is negatively correlated with the quality of the remote sensing vegetation coverage inversion product. Therefore, for each resolution, the remote sensing vegetation coverage inversion product corresponding to the resolution with the smallest comprehensive quality evaluation score can be used as the optimal remote sensing vegetation coverage inversion product. For example, the optimal remote sensing vegetation coverage inversion product corresponding to the medium resolution and the optimal remote sensing vegetation coverage inversion product corresponding to the high resolution can be screened out respectively.

[0052] Step S106, taking the optimal remote sensing vegetation coverage inversion product corresponding to the first resolution as the true value, and taking the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution other than the first resolution as input data, to construct a training data set.

[0053] The first resolution is higher than the second resolution. For example, the first resolution may be the aforementioned high resolution, and the second resolution may be the aforementioned medium resolution. The input data of the training data set includes the optimal remote sensing vegetation coverage inversion product corresponding to the medium resolution, climate element data, terrain element data, and social element data, and the true value is the optimal remote sensing vegetation coverage inversion product corresponding to the high resolution.

[0054] Step S108, training the neural network using the training data set to obtain a target neural network corresponding to the second resolution.

[0055] Among them, the target neural network can adopt a U-Net network. In one example, the aforementioned input data is input into the U-Net network to obtain the optimization result of the inversion product generated by the U-Net network for the input data, and the inversion product optimization result is used to generate a loss function with the optimal remote sensing vegetation coverage inversion product corresponding to the high resolution to adjust the parameters of the U-Net network to obtain a trained target neural network, and the target neural network is used to generate the optimization result of the inversion product corresponding to the first resolution based on the remote sensing vegetation coverage inversion product to be processed corresponding to the second resolution, and the inversion product optimization result is used to fuse with the remote sensing vegetation coverage inversion product to be processed corresponding to the first resolution to obtain a vegetation coverage fusion product.

[0056] The multi-source remote sensing vegetation cover inversion product fusion method provided in the embodiment of the present invention selects the optimal remote sensing vegetation cover inversion product from remote sensing vegetation cover inversion products of different resolutions through authenticity verification, and uses it to construct a training data set so that the trained target neural network can optimize the lower resolution remote sensing vegetation cover inversion product to be processed, and fuse it with the higher resolution remote sensing vegetation cover inversion product to be processed. Compared with the prior art, the observed spectral data is fundamentally changed. The embodiment of the present invention realizes remote sensing vegetation cover monitoring from the product level, and significantly improves the spatiotemporal resolution and accuracy of the remote sensing vegetation cover inversion product.

[0057] To facilitate understanding, the embodiment of the present invention provides a specific implementation method for the fusion of multi-source remote sensing vegetation cover inversion products. The embodiment of the present invention aims to obtain a high temporal resolution, high spatial resolution, and high-precision vegetation cover fusion product by fusing vegetation cover products of different temporal and spatial resolutions from multiple sources at home and abroad, such as FY-3F, MODIS, Landsat 8 / 9, Sentinel-2, and Gaofen 1 / 6 satellites, so as to provide strong data support for the refined monitoring of regional vegetation.

[0058] Specifically, the overall idea of ​​the embodiment of the present invention is: collect publicly available measured values ​​of vegetation coverage; collect vegetation coverage inversion products from FY-3D, FY-3F, MODIS, Landsat8 / 9, Sentinel-2A / B, and GF-1 / 6; normalize multi-source vegetation coverage inversion products; use the measured values ​​of vegetation coverage as true values ​​to verify the authenticity of multi-source satellite vegetation coverage inversion products; select one of the medium-resolution satellites (FY-3D, FY-3F, MODIS) and high-resolution satellites (Landsat8 / 9, Sentinel-2A / B, GF-1 / 6) with the highest accuracy. The highest satellite vegetation coverage inversion product participates in subsequent fusion; taking the best high-resolution satellite vegetation coverage inversion product as the standard, supplemented by climate factors, terrain factors and social factors related to vegetation growth, a U-Net network model is established between the best high-resolution satellite vegetation coverage inversion product (that is, the best vegetation coverage inversion product corresponding to the first resolution mentioned above) and the best medium-resolution satellite vegetation coverage inversion product (that is, the best vegetation coverage inversion product corresponding to the second resolution mentioned above), and the model is applied to the best medium-resolution satellite vegetation coverage inversion product to obtain a daily high-resolution and high-precision vegetation coverage fusion product.

[0059] Based on this, the embodiment of the present invention provides a specific implementation of the aforementioned step S104. The specific process of authenticity verification of multi-source satellite vegetation coverage inversion data is shown in the following steps 1.1 to 1.3:

[0060] Step 1.1, obtain the latitude and longitude information and measured value of vegetation coverage corresponding to the ground station.

[0061] In one example, ground-measured data on vegetation coverage were downloaded from public websites such as the National Ecological Data Center Resource Sharing Service Platform and the National Glacier, Frozen and Desert Science Data Center. The data were analyzed, and quality control processing such as outlier removal was performed to obtain the latitude and longitude information of the ground station and the measured values ​​of vegetation coverage.

[0062] Furthermore, before performing spatiotemporal matching, the remote sensing vegetation coverage inversion products need to be preprocessed. The preprocessing process includes: collecting multi-source satellite vegetation coverage inversion products, including medium-resolution satellite (FY-3D, FY-3F, MODIS) and high-resolution satellite (Landsat8 / 9, Sentinel-2A / B, GF-1 / 6) vegetation coverage data. Parse the data to obtain the observation range, resolution, observation time and other parameters of the satellite-inverted vegetation coverage data. Normalize the vegetation coverage inversion products generated by different satellites to ensure that data from different sources have the same numerical range and scale, and unify the multi-source satellite products into equal longitude and latitude projections.

[0063] Step 1.2: Use the latitude and longitude information to perform spatiotemporal matching between the remote sensing vegetation coverage inversion product and the measured vegetation coverage value, and use the spatiotemporal matching measured vegetation coverage value to verify the authenticity of the remote sensing vegetation coverage inversion product, and obtain the comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product.

[0064] In one example, the process of spatiotemporal matching is as follows: the measured value of vegetation coverage is taken as the true value, and the row and column numbers of the central longitude and latitude grid closest to the ground station are calculated based on the spatiotemporal information of the ground station and multi-source satellites. The vegetation coverage inversion product within the 3*3 range is extracted according to the grid row and column numbers, and the average is taken as the satellite estimation result, which is matched with the measured value of vegetation coverage.

[0065] In one example, after matching the measured values ​​of vegetation coverage with the satellite estimation results in time and space, the evaluation index is calculated to evaluate the quality of the vegetation coverage product. The authenticity verification process is as follows:

[0066] (1) Based on the measured values ​​of vegetation coverage with temporal and spatial matching, multiple evaluation indicators corresponding to the remote sensing vegetation coverage inversion products are determined, including the root mean square error indicator, the correlation coefficient indicator, and the deviation indicator.

[0067] 1) The deviation calculation formula is as follows:

[0068] Bias = x i -x oi ;

[0069] In the formula, Bias represents the average deviation; Xi represents the data to be tested, that is, the satellite estimation result; Xoi represents the test source data, that is, the measured value of vegetation coverage. The deviation is calculated for each matching sample pixel, and the obtained deviation value can be used to draw the deviation spatial distribution map.

[0070] 2) The root mean square error calculation formula is as follows:

[0071]

[0072] Where: RMSE represents the root mean square error; N represents the number of matching samples; Xi represents the data to be tested; Xoi represents the test source data.

[0073] 3) The correlation coefficient calculation formula is as follows:

[0074]

[0075] In the formula, Corr represents the correlation coefficient; N represents the number of matching samples; Xi represents the data to be tested; Xoi represents the test source data; Represents the sample mean of the data to be tested; Represents the sample mean of the test data.

[0076] (2) The evaluation indicators are standardized to obtain multiple standardized evaluation indicators corresponding to the remote sensing vegetation coverage inversion products.

[0077] In order to comprehensively evaluate the quality of multi-source satellite vegetation cover products, the following comprehensive evaluation method is used to comprehensively analyze the three indicators of root mean square error (RMSE), correlation coefficient (R), and bias.

[0078] The root mean square error (RMSE), correlation coefficient (R), and bias (Bias) are standardized to make their dimensions consistent. The following formula is used to standardize the indicators:

[0079] For RMSE and Bias, the smaller the standardized value, the better:

[0080]

[0081] For the correlation coefficient R, the larger the standardized value, the better:

[0082]

[0083] Among them, X is the original value, X norn is the standardized value, X min , X max are the minimum and maximum values ​​respectively.

[0084] (3) Determine the weight value corresponding to each standardized evaluation index, and use the weight value to fuse the standardized evaluation indicators to obtain the comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product.

[0085] In one example, the weight values ​​corresponding to the root mean square error index and the bias index are higher than the weight value corresponding to the correlation coefficient index. Generally, the weights of RMSE and Bias are higher because they directly reflect the error of the product; while the correlation coefficient reflects the correlation between the product and the ground measurement, and the weight can be slightly reduced. In the embodiment of the present invention, the weights of RMSE and Bias are both 0.4, and the weight of the correlation coefficient is 0.2.

[0086] In one example, the weighted summation method is used to calculate the standardized indicators:

[0087] Score = 0.4 × RMSE norm +0.4×Bias norm +0.2×R norm ;

[0088] Among them, Score is the comprehensive quality evaluation score.

[0089] Step 1.3, for each resolution, based on the comprehensive quality evaluation score, select the optimal remote sensing vegetation cover inversion product corresponding to the resolution from multiple remote sensing vegetation cover inversion products corresponding to the resolution.

[0090] In one example, the comprehensive quality evaluation score is negatively correlated with the quality of the remote sensing vegetation coverage inversion product, that is, the smaller the comprehensive quality evaluation score Score is, the better the overall quality of the product is. Therefore, the optimal remote sensing vegetation coverage inversion product corresponding to the resolution can be screened out from multiple remote sensing vegetation coverage inversion products corresponding to the resolution in the order of the comprehensive quality evaluation scores from low to high.

[0091] Specifically, according to the comprehensive quality evaluation score Score, the vegetation cover inversion product with the lowest comprehensive quality evaluation score Score is selected from the medium-resolution satellite products as the optimal medium-resolution vegetation cover inversion product; similarly, the vegetation cover inversion product with the lowest comprehensive quality evaluation score Score is selected from the high-resolution satellite products as the optimal high-resolution vegetation cover inversion product. The above two optimal products will participate in subsequent fusion calculations.

[0092] Furthermore, the neural network in the embodiment of the present invention adopts a U-Net network. The U-Net network model is selected for multi-source remote sensing vegetation coverage data fusion, mainly because of its powerful multi-scale feature extraction and fusion capabilities. U-Net is a fully convolutional network (FCN), which is mainly composed of two parts: an encoder (Contracting Path) and a decoder (Expanding Path). It effectively combines high-resolution and low-resolution features through skip connections, thereby enhancing the ability to retain detail information. In addition, U-Net supports end-to-end training, reduces intermediate processing steps, and improves processing efficiency; it excels in small sample learning and data enhancement, and can achieve good results with less training data, solving the problem of scarcity of high-quality labeled data.

[0093] On this basis, the embodiment of the present invention provides a specific implementation method for constructing a training data set, including: preprocessing the best medium-resolution satellite vegetation cover inversion product, the best high-resolution satellite vegetation cover inversion product, and related climate factor data, terrain factor data, and social factor data, and adjusting their value range to between [0,1] through normalization processing to generate a training data set suitable for a U-Net network. The best high-resolution satellite vegetation cover inversion product is used as label data (i.e., true value).

[0094] The embodiment of the present invention further provides a specific implementation method for training a U-Net network, including the following steps 2.1 to 2.2:

[0095] Step 2.1, through the U-Net network, based on the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution, climate factor data, terrain factor data and social factor data, generate the optimization result of the inversion product corresponding to the first resolution. Specifically:

[0096] (1) Through the encoder in the U-Net network, convolution and pooling operations are performed on the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution, climate factor data, terrain factor data, and social factor data to obtain a vegetation coverage feature map.

[0097] Encoder part of the U-Net network: The encoder extracts high-level features of the input data through a series of convolutional layers and maximum pooling layers.

[0098] The expression of the convolution operation is as follows:

[0099] Conv(X) = W*X+b;

[0100] Among them, W is the convolution kernel weight, b is the bias, X is the input feature map, and * is the convolution operation.

[0101] The expression of the pooling operation (max pooling) is as follows:

[0102]

[0103] Among them, window is the pooling window, and max means taking the maximum value in the window.

[0104] (2) Through the decoder in the U-Net network, deconvolution operations and jump connection operations are performed on the vegetation coverage feature map to obtain the optimized result of the inversion product corresponding to the first resolution.

[0105] The decoder part of the U-Net network: The decoder gradually restores the resolution of the image through a series of deconvolution (upsampling) operations, and finally outputs a vegetation coverage prediction map with the same resolution as the label data.

[0106] The expression of the deconvolution operation is as follows:

[0107] DeConv(X) = WT*X+b;

[0108] Among them, W T is the transposed convolution kernel.

[0109] Skip connection: One of the important features of U-Net, through the skip connection, the feature map of the encoder part is directly passed to the corresponding layer of the decoder to retain more detailed information. The expression of the skip connection is as follows:

[0110] SkipConnection(X encoder , X decoder )=Concat(X encoder , X decoder ).

[0111] Step 2.2, determine the loss value based on the optimal remote sensing vegetation coverage inversion product corresponding to the first resolution and the inversion product optimization result, and use the loss value to adjust the network parameters of the U-Net network to obtain the target neural network corresponding to the second resolution.

[0112] In one example, in order to evaluate the difference between the model output and the actual label, a loss function is defined. Commonly used loss functions include mean squared error loss (MSE) and cross entropy loss.

[0113] Exemplarily, the mean square error loss function is as follows:

[0114]

[0115] Among them, Y i True is the label value, is the model prediction value, and N is the number of samples.

[0116] In one example, during the training phase, the model parameters are continuously adjusted through the backpropagation algorithm and the optimization algorithm (such as the Adam optimizer). The expression of backpropagation is as follows:

[0117]

[0118] Among them, θ is the model parameter, η is the learning rate, L(θ is the loss function, is the gradient.

[0119] In one example, after training, the model is evaluated using a test data set, and metrics such as the root mean square error (RMSE) and correlation coefficient (R) are calculated to verify the performance of the model.

[0120]

[0121] Where X is the input data, This is the fusion result of the predicted vegetation coverage.

[0122] Furthermore, after the U-Net network training is completed, the trained U-Net network can be used to convert the medium-resolution satellite vegetation coverage inversion product to be processed into a high-resolution product optimization result, which is then fused with the obtained high-resolution satellite vegetation coverage inversion product to be processed to obtain the final vegetation coverage fusion product. In one example, the model is applied to daily data to generate daily high-resolution and high-precision vegetation coverage fusion products.

[0123] In summary, the multi-source remote sensing vegetation coverage inversion product fusion method provided by the embodiment of the present invention has at least the following characteristics:

[0124] (1) Improving the spatial resolution and accuracy of vegetation coverage products: By using the U-Net network model, the optimal medium-resolution satellite vegetation coverage product is integrated with the optimal high-resolution satellite product, effectively improving the spatial resolution and accuracy of the final vegetation coverage product. This enables the generated vegetation coverage product to more finely reflect the actual status of surface vegetation and is suitable for ecological and environmental monitoring and research with higher precision requirements.

[0125] (2) Improve the spatiotemporal consistency of data: The fusion method adopted by the present invention not only improves the spatial resolution, but also maintains the consistency of data in time, generating daily high-resolution vegetation coverage products. Compared with traditional single data source or low-frequency products, the results of the present invention can provide more frequent and more accurate time series data for monitoring vegetation dynamic changes.

[0126] (3) Enhance the comprehensive utilization efficiency of multi-source data: By integrating vegetation coverage products from different satellite platforms, the present invention maximizes the advantages of various types of satellite data. Medium-resolution satellites provide a wider coverage range, while high-resolution satellites provide fine spatial details. The combination of the two provides users with comprehensive vegetation coverage data with both macro and micro perspectives.

[0127] (4) Adapting to the diverse needs of different ecological environments: This paper takes into account multiple factors related to vegetation growth, such as climate, terrain, and social factors, and constructs a highly adaptable fusion model. The model can be adjusted according to the actual needs of different ecological environments to generate more targeted vegetation coverage products, providing strong data support for the management and protection of different ecosystems.

[0128] (5) Improve the versatility and scalability of the model: The application of the U-Net network model enables the present invention to have good versatility and scalability in applications in different regions and scales. By continuously updating training data and adjusting model parameters, the method can adapt to environmental conditions in different regions and ensure that the generated vegetation coverage products have high reliability in various application scenarios.

[0129] Based on the above embodiments, the embodiments of the present invention further provide Figure 2 The figure shows an overall process diagram of a multi-source remote sensing vegetation coverage inversion product fusion method, including: using the measured values ​​of vegetation coverage to verify the authenticity of medium-resolution remote sensing vegetation coverage inversion products (FY-3D, FY-3F, MODIS) and high-resolution remote sensing vegetation coverage inversion products (Landsat8 / 9, Sentinel-2A / B, GF-1 / 6), and obtain a comprehensive quality evaluation score, and then screen out the best medium-resolution satellite vegetation coverage inversion product and the best high-resolution satellite vegetation coverage inversion product; combining land use classification, GDP, population, temperature, precipitation, altitude, slope and other data to perform spatiotemporal matching; inputting the spatiotemporal matching data into the U-Net network to realize high spatiotemporal resolution vegetation coverage calculation, so as to output high spatiotemporal resolution vegetation coverage products.

[0130] In summary, the purpose of the embodiments of the present invention is to solve the limitations of spatial resolution and temporal resolution in current vegetation coverage monitoring, and to generate vegetation coverage data with high spatiotemporal resolution by proposing a multi-source remote sensing vegetation coverage product fusion technology based on the U-Net network, which aims to utilize the spatiotemporal advantages of Fengyun-3 series, MODIS, Landsat, Gaofen series and Sentinel series satellite data. The U-Net network has efficient image segmentation capabilities and the ability to handle complex scenes. By performing deep learning fusion on different satellite data, it not only maintains the accuracy of the original product data, but also improves the resolution of the monitoring results, thereby meeting the needs of large-scale and rapidly changing vegetation dynamic monitoring, providing more accurate vegetation coverage assessment, and providing strong data support for the refined monitoring of regional vegetation.

[0131] Based on the above-mentioned embodiment, the present invention provides a multi-source remote sensing vegetation coverage inversion product fusion device, see Figure 3 The schematic diagram of the structure of a multi-source remote sensing vegetation coverage inversion product fusion device is shown in FIG. The device mainly includes the following parts:

[0132] A product acquisition module 302 is used to acquire multi-source remote sensing vegetation coverage inversion products, where the multi-source remote sensing vegetation coverage inversion products include multiple remote sensing vegetation coverage inversion products corresponding to different resolutions;

[0133] The product screening module 304 is used to verify the authenticity of the remote sensing vegetation coverage inversion product, so as to screen out the best remote sensing vegetation coverage inversion product corresponding to each resolution from multiple remote sensing vegetation coverage inversion products corresponding to each resolution;

[0134] The training set construction module 306 is used to construct a training data set using the optimal remote sensing vegetation coverage inversion product corresponding to the first resolution as the true value and the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution other than the first resolution as input data, wherein the first resolution is higher than the second resolution;

[0135] A model training module 308 is used to train the neural network using the training data set to obtain a target neural network corresponding to the second resolution;

[0136] Among them, the target neural network is used to generate an optimization result of the inversion product corresponding to the first resolution based on the remote sensing vegetation cover inversion product to be processed corresponding to the second resolution, and the optimization result of the inversion product is used to fuse with the remote sensing vegetation cover inversion product to be processed corresponding to the first resolution to obtain a vegetation coverage fusion product.

[0137] The multi-source remote sensing vegetation cover inversion product fusion device provided in the embodiment of the present invention selects the optimal remote sensing vegetation cover inversion product from remote sensing vegetation cover inversion products of different resolutions through authenticity verification, and uses it to construct a training data set so that the trained target neural network can optimize the lower resolution remote sensing vegetation cover inversion product to be processed, and fuse it with the higher resolution remote sensing vegetation cover inversion product to be processed. Compared with the prior art, the observed spectral data is fundamentally changed. The embodiment of the present invention realizes remote sensing vegetation cover monitoring from the product level, and significantly improves the spatiotemporal resolution and accuracy of the remote sensing vegetation cover inversion product.

[0138] In one embodiment, the product screening module 304 is specifically used to:

[0139] Obtain the latitude and longitude information and measured values ​​of vegetation coverage corresponding to the ground station;

[0140] Use the latitude and longitude information to match the remote sensing vegetation coverage inversion product with the measured vegetation coverage value in time and space, and use the measured vegetation coverage value matched in time and space to verify the authenticity of the remote sensing vegetation coverage inversion product, and obtain the comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product;

[0141] For each resolution, based on the comprehensive quality evaluation score, the optimal remote sensing vegetation cover inversion product corresponding to the resolution is screened out from multiple remote sensing vegetation cover inversion products corresponding to the resolution.

[0142] In one embodiment, the product screening module 304 is specifically used to:

[0143] Based on the measured values ​​of vegetation coverage matched in time and space, multiple evaluation indicators corresponding to the remote sensing vegetation coverage inversion products are determined;

[0144] The evaluation indicators are standardized to obtain multiple standardized evaluation indicators corresponding to the remote sensing vegetation coverage inversion products;

[0145] Determine the weight value corresponding to each standardized evaluation index, and use the weight value to fuse the standardized evaluation indicators to obtain the comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product.

[0146] In one implementation, the evaluation index includes a root mean square error index, a correlation coefficient index, and a deviation index, and the weight values ​​corresponding to the root mean square error index and the deviation index are higher than the weight value corresponding to the correlation coefficient index.

[0147] In one embodiment, the comprehensive quality evaluation score is negatively correlated with the quality of the remote sensing vegetation coverage inversion product; the product screening module 304 is specifically used to:

[0148] According to the order of comprehensive quality evaluation scores from low to high, the optimal remote sensing vegetation cover inversion product corresponding to the resolution is screened out from multiple remote sensing vegetation cover inversion products corresponding to the resolution.

[0149] In one embodiment, the neural network adopts a U-Net network; the model training module 308 is specifically used for:

[0150] Through the U-Net network, based on the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution, climate element data, terrain element data and social element data, the optimization result of the inversion product corresponding to the first resolution is generated;

[0151] The loss value is determined based on the optimal remote sensing vegetation cover inversion product corresponding to the first resolution and the inversion product optimization result, and the network parameters of the U-Net network are adjusted using the loss value to obtain the target neural network corresponding to the second resolution.

[0152] In one embodiment, the model training module 308 is specifically used to:

[0153] Through the encoder in the U-Net network, convolution and pooling operations are performed on the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution, climate factor data, terrain factor data, and social factor data to obtain a vegetation coverage feature map;

[0154] Through the decoder in the U-Net network, deconvolution operations and jump connection operations are performed on the vegetation coverage feature map to obtain the optimized result of the inversion product corresponding to the first resolution.

[0155] The device provided in the embodiment of the present invention has the same implementation principle and technical effects as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0156] An embodiment of the present invention provides an electronic device. Specifically, the electronic device includes a processor and a storage device. The storage device stores a computer program, and when the computer program is executed by the processor, it executes the method described in any one of the above-mentioned implementation methods.

[0157] Figure 4 A structural diagram of an electronic device provided in an embodiment of the present invention, the electronic device 100 includes: a processor 40, a memory 41, a bus 42 and a communication interface 43, wherein the processor 40, the communication interface 43 and the memory 41 are connected via the bus 42; the processor 40 is used to execute an executable module stored in the memory 41, such as a computer program.

[0158] The memory 41 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 43 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0159] The bus 42 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0160] Among them, the memory 41 is used to store programs, and the processor 40 executes the program after receiving the execution instruction. The method executed by the device for flow process definition disclosed in any embodiment of the above-mentioned embodiment of the present invention can be applied to the processor 40 or implemented by the processor 40.

[0161] The processor 40 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 40. The above processor 40 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiments of the present invention can be directly embodied as a hardware decoding processor to execute, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 41, and the processor 40 reads the information in the memory 41 and completes the steps of the above method in combination with its hardware.

[0162] The computer program product of the readable storage medium provided in the embodiment of the present invention includes a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the methods described in the previous method embodiments. The specific implementation can be referred to the previous method embodiments, which will not be repeated here.

[0163] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0164] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for fusing multi-source remote sensing vegetation coverage inversion products, characterized in that: include: Acquire a multi-source remote sensing vegetation coverage inversion product, wherein the multi-source remote sensing vegetation coverage inversion product includes a plurality of remote sensing vegetation coverage inversion products corresponding to different resolutions; Verifying the authenticity of the remote sensing vegetation coverage inversion product, so as to screen out the best remote sensing vegetation coverage inversion product corresponding to each resolution from a plurality of remote sensing vegetation coverage inversion products corresponding to each resolution; Taking the optimal remote sensing vegetation coverage inversion product corresponding to a first resolution as a true value and taking the optimal remote sensing vegetation coverage inversion product corresponding to a second resolution other than the first resolution as input data, a training data set is constructed, wherein the first resolution is higher than the second resolution; Using the training data set to train a neural network to obtain a target neural network corresponding to the second resolution; Among them, the target neural network is used to generate an optimization result of the inversion product corresponding to the first resolution based on the remote sensing vegetation cover inversion product to be processed corresponding to the second resolution, and the optimization result of the inversion product is used to fuse with the remote sensing vegetation cover inversion product to be processed corresponding to the first resolution to obtain a vegetation coverage fusion product.

2. The multi-source remote sensing vegetation coverage inversion product fusion method according to claim 1 is characterized in that: The remote sensing vegetation coverage inversion product is verified for authenticity, so as to select the optimal remote sensing vegetation coverage inversion product corresponding to each resolution from the multiple remote sensing vegetation coverage inversion products corresponding to each resolution, including: Obtain the latitude and longitude information and measured values ​​of vegetation coverage corresponding to the ground station; Using the latitude and longitude information, the remote sensing vegetation coverage inversion product and the vegetation coverage measured value are temporally and spatially matched, so as to verify the authenticity of the remote sensing vegetation coverage inversion product using the temporally and spatially matched vegetation coverage measured value, and obtain a comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product; For each of the resolutions, based on the comprehensive quality evaluation score, the optimal remote sensing vegetation coverage inversion product corresponding to the resolution is screened out from the multiple remote sensing vegetation coverage inversion products corresponding to the resolution.

3. The multi-source remote sensing vegetation coverage inversion product fusion method according to claim 2 is characterized in that: The remote sensing vegetation coverage inversion product is verified for authenticity using the temporally and spatially matched measured values ​​of vegetation coverage to obtain a comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product, including: Determine a plurality of evaluation indicators corresponding to the remote sensing vegetation coverage inversion product based on the measured value of the vegetation coverage matched in time and space; Standardizing the evaluation index to obtain multiple standardized evaluation indexes corresponding to the remote sensing vegetation coverage inversion product; Determine the weight value corresponding to each of the standardized evaluation indicators, and use the weight value to fuse the standardized evaluation indicators to obtain the comprehensive quality evaluation score corresponding to the remote sensing vegetation coverage inversion product.

4. The multi-source remote sensing vegetation coverage inversion product fusion method according to claim 3 is characterized in that: The evaluation index includes a root mean square error index, a correlation coefficient index and a deviation index. The weight values ​​corresponding to the root mean square error index and the deviation index are higher than the weight value corresponding to the correlation coefficient index.

5. The multi-source remote sensing vegetation coverage inversion product fusion method according to claim 2 is characterized in that: The comprehensive quality evaluation score is negatively correlated with the quality of the remote sensing vegetation coverage inversion product; based on the comprehensive quality evaluation score, the optimal remote sensing vegetation coverage inversion product corresponding to the resolution is screened out from the multiple remote sensing vegetation coverage inversion products corresponding to the resolution, including: According to the order of the comprehensive quality evaluation scores from low to high, the optimal remote sensing vegetation coverage inversion product corresponding to the resolution is screened out from the multiple remote sensing vegetation coverage inversion products corresponding to the resolution.

6. The multi-source remote sensing vegetation coverage inversion product fusion method according to claim 1 is characterized in that: The neural network adopts U-Net network; Using the training data set to train a neural network to obtain a target neural network corresponding to the second resolution includes: Generate an optimization result of the inversion product corresponding to the first resolution through the U-Net network based on the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution, climate element data, terrain element data, and social element data; A loss value is determined based on the optimal remote sensing vegetation coverage inversion product corresponding to the first resolution and the optimization result of the inversion product, so as to use the loss value to adjust the network parameters of the U-Net network to obtain a target neural network corresponding to the second resolution.

7. The multi-source remote sensing vegetation coverage inversion product fusion method according to claim 6 is characterized in that: The optimization result of the inversion product corresponding to the first resolution is generated by the U-Net network based on the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution, the climate element data, the terrain element data and the social element data, including: Performing convolution and pooling operations on the optimal remote sensing vegetation coverage inversion product corresponding to the second resolution, the climate element data, the terrain element data, and the social element data through an encoder in the U-Net network to obtain a vegetation coverage feature map; Through the decoder in the U-Net network, deconvolution operations and jump connection operations are performed on the vegetation coverage feature map to obtain an optimized inversion product result corresponding to the first resolution.

8. A multi-source remote sensing vegetation coverage inversion product fusion device, characterized in that: include: A product acquisition module, used to acquire multi-source remote sensing vegetation coverage inversion products, wherein the multi-source remote sensing vegetation coverage inversion products include multiple remote sensing vegetation coverage inversion products corresponding to different resolutions; A product screening module, used for verifying the authenticity of the remote sensing vegetation coverage inversion product, so as to screen out the best remote sensing vegetation coverage inversion product corresponding to each resolution from a plurality of remote sensing vegetation coverage inversion products corresponding to each resolution; A training set construction module, used to construct a training data set using the optimal remote sensing vegetation coverage inversion product corresponding to a first resolution as a true value and the optimal remote sensing vegetation coverage inversion product corresponding to a second resolution other than the first resolution as input data, wherein the first resolution is higher than the second resolution; A model training module, used to train a neural network using the training data set to obtain a target neural network corresponding to the second resolution; Among them, the target neural network is used to generate an optimization result of the inversion product corresponding to the first resolution based on the remote sensing vegetation cover inversion product to be processed corresponding to the second resolution, and the optimization result of the inversion product is used to fuse with the remote sensing vegetation cover inversion product to be processed corresponding to the first resolution to obtain a vegetation coverage fusion product.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the method according to any one of claims 1 to 7.

Images