Multi-source heterogeneous remote sensing data decision-level information fusion method and device

Abstract

The application discloses a kind of multi-source heterogeneous remote sensing data decision level information fusion method and device, belong to remote sensing technical field.The method includes: obtaining satellite remote sensing image and unmanned aerial vehicle remote sensing image and carrying out water body extraction, based on selected water quality parameter, satellite remote sensing water body area and unmanned aerial vehicle remote sensing water body area are carried out water quality parameter inversion;Based on selected precision evaluation index, the precision of satellite inversion grid data and unmanned aerial vehicle inversion grid data is evaluated;If the precision of satellite inversion grid data is better than the precision of unmanned aerial vehicle inversion grid data, then with satellite inversion grid data as benchmark, unmanned aerial vehicle inversion grid data is corrected, and the final remote sensing inversion result is obtained, otherwise, directly with unmanned aerial vehicle inversion grid data as final remote sensing inversion result.The application realizes precision and resolution collaborative optimization, significantly improves the reliability and practicality of water quality remote sensing monitoring.

Description

Technical Field

[0001] This invention relates to the field of remote sensing technology, and in particular to a method and apparatus for decision-level information fusion of multi-source heterogeneous remote sensing data. Background Technology

[0002] With the development of remote sensing technology, water quality remote sensing monitoring has gradually become an important means of aquatic ecosystem monitoring. Differences in the performance of different monitoring platforms and sensors lead to significant variations in spatial resolution, monitoring accuracy, and timeliness of multi-source remote sensing data, resulting in difficulties in data collaboration and the ineffective utilization of redundant data in water quality remote sensing monitoring.

[0003] Therefore, with the increasing demand for water environment monitoring, it is urgent to introduce new multi-source data fusion algorithms and establish a multi-source data fusion analysis model with high accuracy and wide application to make up for the shortcomings of single remote sensing data. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method and apparatus for decision-level information fusion of multi-source heterogeneous remote sensing data, achieving synergistic optimization of accuracy and resolution, and significantly improving the reliability and practicality of water quality remote sensing monitoring.

[0005] The technical solution provided by this invention is as follows:

[0006] A method for decision-level information fusion of multi-source heterogeneous remote sensing data, the method comprising:

[0007] S1: Acquire satellite and UAV remote sensing images covering the water body under study at the same time.

[0008] S2: Extract water bodies from the satellite remote sensing images and UAV remote sensing images to obtain the satellite remote sensing water body area and the UAV remote sensing water body area.

[0009] S3: Based on the selected water quality parameters, perform water quality parameter inversion on the satellite remote sensing water area and the UAV remote sensing water area to obtain satellite inversion raster data and UAV inversion raster data;

[0010] S4: Evaluate the accuracy of the satellite-inverted raster data and the UAV-inverted raster data based on the selected accuracy evaluation index;

[0011] S5: If the accuracy of the satellite-inverted raster data is better than that of the UAV-inverted raster data, then the UAV-inverted raster data is corrected based on the satellite-inverted raster data to obtain the final remote sensing inversion result; otherwise, the UAV-inverted raster data is directly used as the final remote sensing inversion result.

[0012] Furthermore, the step of correcting the UAV-inverted raster data based on the satellite-inverted raster data to obtain the final remote sensing inversion result includes:

[0013] For any satellite pixel included in the satellite inversion raster data, calculate the average water quality parameters of all UAV pixels within the coverage area of ​​the satellite pixel on the UAV inversion raster data;

[0014] The correction value is obtained by subtracting the average value from the water quality parameters of the satellite pixels;

[0015] The water quality parameters of each UAV pixel within the coverage area of ​​the satellite pixel are added to the correction value to correct the UAV inverted raster data, thus obtaining the final remote sensing inversion result.

[0016] Furthermore, S4 includes:

[0017] S41: Construct a hierarchical analysis structure model;

[0018] The hierarchical analysis structure model includes a scheme layer, a criterion layer, and a target layer. The scheme layer includes satellite inversion raster data and UAV inversion raster data. The criterion layer includes multiple accuracy evaluation indicators. The target layer is the data with the best accuracy evaluation among the satellite inversion raster data and UAV inversion raster data.

[0019] S42: Compare each accuracy evaluation index of the criterion layer pairwise to determine the importance of each accuracy evaluation index to the target layer, and construct the judgment matrix of the hierarchical analysis.

[0020] S43: Perform a consistency check on the judgment matrix. If the consistency check passes, proceed to the next step; otherwise, return to S42.

[0021] S44: Derive the weights of each accuracy evaluation index from the judgment matrix;

[0022] S45: Based on the satellite inversion raster data and UAV inversion raster data, as well as the measured values ​​of the water quality parameters, calculate the various accuracy evaluation indicators of the satellite inversion raster data and UAV inversion raster data.

[0023] S46: Based on the various accuracy evaluation indicators and their weights of the satellite inverted raster data and the UAV inverted raster data, a weighted calculation is performed to obtain the comprehensive accuracy score of the satellite inverted raster data and the UAV inverted raster data;

[0024] S47: Compare the comprehensive accuracy scores of the satellite-inverted raster data and the UAV-inverted raster data, and take the data with the best accuracy evaluation among the satellite-inverted raster data and UAV-inverted raster data as the output of the target layer.

[0025] Furthermore, the accuracy evaluation indicators include correlation coefficient, mean absolute error, root mean square error, and mean relative error.

[0026] A multi-source heterogeneous remote sensing data decision-level information fusion device, the device comprising:

[0027] The data acquisition module is used to acquire satellite remote sensing images and UAV remote sensing images covering the water body under study at the same time.

[0028] The water extraction module is used to extract water from the satellite remote sensing images and UAV remote sensing images to obtain the satellite remote sensing water area and the UAV remote sensing water area.

[0029] The remote sensing inversion module is used to invert water quality parameters of the satellite remote sensing water area and the UAV remote sensing water area based on the selected water quality parameters, and obtain satellite inversion raster data and UAV inversion raster data.

[0030] The accuracy assessment module is used to assess the accuracy of the satellite-inverted raster data and the UAV-inverted raster data based on selected accuracy assessment indicators.

[0031] The information fusion module is used to correct the UAV-inverted raster data based on the satellite-inverted raster data if the accuracy of the satellite-inverted raster data is better than that of the UAV-inverted raster data, so as to obtain the final remote sensing inversion result; otherwise, the UAV-inverted raster data is directly used as the final remote sensing inversion result.

[0032] Furthermore, in the information fusion module, the UAV-inverted raster data is corrected based on the satellite-inverted raster data to obtain the final remote sensing inversion result, including the following process:

[0033] For any satellite pixel included in the satellite inversion raster data, calculate the average water quality parameters of all UAV pixels within the coverage area of ​​the satellite pixel on the UAV inversion raster data;

[0034] The correction value is obtained by subtracting the average value from the water quality parameters of the satellite pixels;

[0035] The water quality parameters of each UAV pixel within the coverage area of ​​the satellite pixel are added to the correction value to correct the UAV inverted raster data, thus obtaining the final remote sensing inversion result.

[0036] Furthermore, the accuracy evaluation module includes:

[0037] Model building unit, used to build hierarchical analysis structure models;

[0038] The hierarchical analysis structure model includes a scheme layer, a criterion layer, and a target layer. The scheme layer includes satellite inversion raster data and UAV inversion raster data. The criterion layer includes multiple accuracy evaluation indicators. The target layer is the data with the best accuracy evaluation among the satellite inversion raster data and UAV inversion raster data.

[0039] The judgment matrix construction unit is used to compare each accuracy evaluation index of the criterion layer pairwise to determine the importance of each accuracy evaluation index to the target layer and construct the judgment matrix of the hierarchical analysis.

[0040] The verification unit is used to perform consistency verification on the judgment matrix. If the consistency verification passes, the next step is executed; otherwise, the process returns to the judgment matrix construction module.

[0041] The weight calculation unit is used to derive the weights of each accuracy evaluation index from the judgment matrix;

[0042] The accuracy evaluation index calculation unit is used to calculate various accuracy evaluation indices of the satellite inversion raster data and the UAV inversion raster data based on the satellite inversion raster data, the UAV inversion raster data, and the measured values ​​of the water quality parameters.

[0043] The accuracy score calculation unit is used to perform weighted calculations based on the various accuracy evaluation indicators and their weights of the satellite inverted raster data and the UAV inverted raster data to obtain the comprehensive accuracy score of the satellite inverted raster data and the UAV inverted raster data.

[0044] The accuracy evaluation unit is used to compare the comprehensive accuracy scores of the satellite inverted raster data and the UAV inverted raster data, and to take the data with the best accuracy evaluation among the satellite inverted raster data and the UAV inverted raster data as the output of the target layer.

[0045] Furthermore, the accuracy evaluation indicators include correlation coefficient, mean absolute error, root mean square error, and mean relative error.

[0046] The present invention has the following beneficial effects:

[0047] This invention breaks through the technical bottleneck of a single remote sensing platform by integrating AHP objective assessment with adaptive decision-level assessment, achieving synergistic optimization of accuracy and resolution, and significantly improving the reliability and practicality of water quality remote sensing monitoring. Attached Figure Description

[0048] Figure 1 This is a flowchart of the multi-source heterogeneous remote sensing data decision-level information fusion method of the present invention;

[0049] Figure 2 This is a structural diagram of the analytic hierarchy process (AHP) model;

[0050] Figure 3 This is a schematic diagram illustrating information fusion in this invention;

[0051] Figure 4 This is a comparison diagram showing the effects of the fusion process in this invention;

[0052] Figure 5 This is a schematic diagram of the multi-source heterogeneous remote sensing data decision-level information fusion device of the present invention. Detailed Implementation

[0053] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0054] This invention provides a method for decision-level information fusion of multi-source heterogeneous remote sensing data, such as... Figure 1 As shown, the method includes:

[0055] S1: Acquire satellite and UAV remote sensing images covering the water body under study at the same time.

[0056] S2: Extract water bodies from satellite remote sensing images and UAV remote sensing images to obtain the water body areas from satellite remote sensing images and UAV remote sensing images.

[0057] Water body information can be extracted using the Normalized Difference Water Index (NDWI, see formula) to eliminate interference from non-water targets such as land, floating objects, and ships.

[0058]

[0059] Green indicates the green light band, and NIR indicates the near-infrared band.

[0060] S3: Based on the selected water quality parameters, perform water quality parameter inversion on the satellite remote sensing water area and the UAV remote sensing water area to obtain satellite inversion raster data and UAV inversion raster data.

[0061] Different inversion methods are selected for different water quality parameters. For example, water quality parameters can be selected based on chlorophyll or suspended solids concentration, and a suitable inversion method can be chosen. Each pixel in the satellite inversion raster data obtained after inversion is called a satellite pixel, and its value represents the value of that water quality parameter (i.e., chlorophyll or suspended solids concentration). Each pixel in the UAV inversion raster data obtained after inversion is called a UAV pixel, and its value represents the value of that water quality parameter (i.e., chlorophyll or suspended solids concentration).

[0062] S4: Evaluate the accuracy of satellite-inverted raster data and UAV-inverted raster data based on the selected accuracy evaluation metrics.

[0063] In one example, the Analytic Hierarchy Process (AHP) can be chosen for evaluation. The AHP decomposes the decision problem into multiple levels, such as the objective level, alternative level, and criterion level. It then performs quantitative analysis and comparison at each level to determine the relative weight of each factor, ultimately providing a reference for selecting the optimal solution.

[0064] Specifically, the process includes:

[0065] S41: Construct a hierarchical analysis structure model;

[0066] The hierarchical analysis (AHP) model decomposes the decision problem into three levels: the scheme level, the criterion level, and the target level. The target level (O) selects the best inversion data for the water quality index across multiple study areas. The scheme level contains alternative schemes (C1, C2, ..., Cn) for the best inversion data. The criterion level uses the accuracy assessment results of each water quality index as the indicators for this hierarchical analysis (B1, B2, ..., Bn). In this technique, the accuracy assessment indicators selected for the criterion level are: correlation coefficient, mean absolute error, root mean square error, and mean relative error. The scheme level includes satellite inversion raster data and UAV inversion raster data. The target level is the data with the best accuracy assessment among the satellite inversion raster data and UAV inversion raster data. The structure is as follows: Figure 2 As shown.

[0067] S42: Compare each accuracy evaluation index of the criterion layer pairwise to determine the importance of each accuracy evaluation index to the target layer. Determine the scaling value between each element through scoring and construct the judgment matrix between the target layer and the criterion layer of the hierarchical analysis.

[0068] Specifically, the 1-9 basic scaling method can be used to determine the scaling values ​​between various accuracy evaluation indicators.

[0069] S43: Perform a consistency check on the judgment matrix. If the consistency check passes, proceed to the next step; otherwise, return to S42.

[0070] Before using the judgment matrix to calculate the weights, a consistency check is required. First, the consistency index (CI) is calculated using the following formula:

[0071]

[0072] in Let be the largest eigenvalue, and n be the number of rows and columns of the matrix. Then, find the corresponding RI in the table and substitute it into the formula:

[0073]

[0074] If CR < 0.1, the consistency of the judgment matrix can be considered acceptable; otherwise, return to S42 to modify the matrix.

[0075] S44: Derive the weights of each accuracy evaluation index from the judgment matrix.

[0076] Specifically, the weight vector W of the judgment matrix A is its largest eigenvalue. The corresponding eigenvectors. This is determined by solving the characteristic equation. We can obtain the feature vector W. Then, we normalize W (making the sum of all its components equal to 1) to obtain the weights of each accuracy evaluation index.

[0077] S45: Based on satellite inversion raster data and UAV inversion raster data, as well as the measured values ​​of water quality parameters, calculate various accuracy evaluation indicators for satellite inversion raster data and UAV inversion raster data.

[0078] Sampling points were set up in the water body, with sampling cross-sections spaced approximately 100 meters apart, and about 8-10 cross-sections were set up in each study area. The data were based on measured values ​​of water quality parameters (i.e., chlorophyll or suspended solids concentration data) obtained from laboratory measurements.

[0079] S46: Based on the various accuracy evaluation indicators and their weights of the satellite-inverted raster data and the UAV-inverted raster data, a weighted calculation is performed to obtain the comprehensive accuracy score of the satellite-inverted raster data and the UAV-inverted raster data.

[0080] S47: Compare the comprehensive accuracy scores of satellite-inverted raster data and UAV-inverted raster data, and use the data with the best accuracy evaluation among the satellite-inverted raster data and UAV-inverted raster data as the output of the target layer.

[0081] S5: If the accuracy of the satellite-inverted raster data is better than that of the UAV-inverted raster data, then the satellite-inverted raster data will be used as the benchmark to correct the UAV-inverted raster data and obtain the final remote sensing inversion result. Otherwise, the UAV-inverted raster data will be used directly as the final remote sensing inversion result.

[0082] In one example, the process of correcting UAV-retrieved raster data based on satellite-retrieved raster data includes:

[0083] 1. For any satellite pixel contained in the satellite inversion raster data, calculate the average water quality parameters of all UAV pixels within the coverage area of ​​the satellite pixel on the UAV inversion raster data.

[0084] Because of the significant resolution difference between the two heterogeneous data sets, the coverage area of ​​one satellite pixel in the satellite inversion raster data is used as the correction unit. Assuming the water quality parameters of the satellite pixel in the m-th row and n-th column of the satellite inversion raster data are... The average water quality parameter of all UAV pixels included in the satellite pixel coverage area of ​​the m-th row and n-th column is [value missing]. .

[0085]

[0086] x and y represent the x and y coordinates of the top left corner of the UAV pixel corresponding to the top left corner of the satellite pixel coverage area of ​​m rows and n columns of satellite inversion raster data, p represents the number of rows of UAV pixels in this area after division, and p represents the number of columns of UAV pixels in this area after division.

[0087] 2. Subtract the average value from the water quality parameters of the satellite pixels to obtain the correction value.

[0088] in , This represents the correction value for the UAV-inverted raster data in row i and column j.

[0089] 3. Add the water quality parameters of each UAV pixel within the coverage area of ​​the satellite pixels to the correction value, correct the UAV inverted raster data, and obtain the final remote sensing inversion result.

[0090] use The correction value is applied to each UAV pixel in the UAV-inverted raster data for that region, and the corrected UAV-inverted result is taken as the final inversion result. The specific correction process is as follows: Figure 3 As shown.

[0091] Remote sensing technology has continuously enhanced its Earth observation capabilities across spatial, temporal, and spectral dimensions, playing a significant role in water resource monitoring. However, relying solely on remote sensing methods presents numerous challenges. For instance, limitations in satellite remote sensing technology, data noise, the high image resolution requirements for water body information retrieval, and restrictions on water quality index retrieval methods can all lead to inaccurate water quality monitoring. Furthermore, water quality monitoring based on satellite remote sensing data is significantly affected by factors such as clouds and atmospheric conditions. Unmanned aerial vehicle (UAV) remote sensing technology also suffers from drawbacks such as relatively small monitoring range, limited parameter retrieval accuracy, and susceptibility to water surface ripples, solar flares, and water reflections.

[0092] Monitoring data from different sensors exhibit heterogeneous characteristics such as varying resolution, monitoring accuracy, and timeliness, leading to discrepancies in the analysis of the results. This invention proposes a decision-level information fusion analysis method that integrates multi-source heterogeneous remote sensing data. Based on the analytic hierarchy process (AHP), a decision-level information fusion analysis algorithm is constructed to comprehensively evaluate the water quality monitoring accuracy of satellites and UAVs. Based on the evaluation results, the optimal data is selected, achieving information fusion analysis of different inversion data.

[0093] While UAVs offer higher resolution than satellites, the accuracy of UAV imagery remains compromised by factors such as water ripples, solar flares, and water reflections. If the evaluation indicates satellite data is superior, the UAV retrieval results should be corrected using satellite data as a benchmark. Spatial matching correction should be performed on the high-resolution UAV data within the satellite's pixel coverage area, balancing the stability of satellites with the resolution advantages of UAVs to significantly improve retrieval accuracy. If the evaluation indicates UAVs are superior, their high-resolution results should be directly adopted to avoid introducing unnecessary errors.

[0094] This invention utilizes the Analytic Hierarchy Process (AHP) to construct an evaluation system comprising an objective layer, a scheme layer, and a criterion layer. Consistency checks ensure the objectivity of weight allocation, scientifically selecting the optimal single data source for specific water quality indicators. Based on the AHP evaluation results, a dynamic fusion strategy is selected to achieve an adaptive, high-precision fusion strategy. Through quantitative evaluation and data fusion, the complementary value of multi-source heterogeneous data is fully explored, significantly improving data utilization efficiency and the overall accuracy of monitoring results, providing a more reliable basis for water environment management.

[0095] In summary, this invention breaks through the technical bottleneck of a single remote sensing platform by integrating AHP objective assessment with adaptive decision-level assessment, achieving synergistic optimization of accuracy and resolution, and significantly improving the reliability and practicality of water quality remote sensing monitoring.

[0096] The invention will be illustrated below with a specific experimental example:

[0097] This study selected a section of the Parlung Tsangpo River as the experimental area, and conducted on-site water sampling, UAV and satellite remote sensing water quality monitoring. Sampling sections were spaced approximately 100 meters apart, with about 8-10 sections in each study area. Based on laboratory-measured Chl-a concentration data, the accuracy of the UAV and satellite remote sensing inversion results was evaluated. Specifically, the water sample test results were considered as approximate true values ​​of water quality parameters, and four accuracy evaluation indicators—correlation coefficient (R), mean absolute error (MAE), root mean square error (RMSE), and mean relative error (MRE)—were calculated to assess the quality of the satellite and UAV inversion results. Furthermore, the analytic hierarchy process (AHP) was applied for comprehensive evaluation, and based on this evaluation result, information fusion analysis of multi-source heterogeneous data was achieved. Experimental data and quality evaluation indicators are shown in Table 1.

[0098] surface Experimental data and evaluation indicators used in this invention

[0099]

[0100] The satellite remote sensing data used in this study came from the European Space Agency's (ESA) Sentinel-2 satellite, with a spatial resolution of 10 meters and a revisit period of 10 days. UAV data was acquired by a platform equipped with a five-band (Blue, Green, Red, Red Edge, NIR) multispectral camera. Before analyzing the spectral characteristics of the water bodies, the Normalized Difference Water Index (NDWI) was used to extract water body information to eliminate interference from non-water targets such as land, floating objects, and vessels.

[0101] The UAV data preprocessing process includes steps such as image denoising, radiometric calibration, flare removal, and solar elevation angle correction.

[0102] This study selected Chl-a as a water quality monitoring indicator and used the method of this invention to evaluate the accuracy of the Chl-a inversion results in the Parlung Tsangpo River study area. Based on this, the weight coefficients of each accuracy indicator were determined. The results are detailed in Table 2.

[0103] Table 2: Scoring of Chl-a multi-source monitoring data in the Parlung Tsangpo River study area

[0104]

[0105] Table 2 shows that, in the accuracy assessment process, four accuracy indicators were comprehensively evaluated using an equal-weighted approach to assess the Chl-a inversion quality of the study area, and the comprehensive scores for UAV data and satellite data were calculated separately. The results show that in the Parlung Tsangpo River study area, the Chl-a inversion score for satellite data was 0.636, significantly higher than the 0.364 for UAV data.

[0106] This indicates that, in the plateau river environment covered in this study, the Chl-a concentration retrieved from satellite data is closer to the true value. The main reason for this is that the study area is a typical plateau river with a large elevation difference and rapid currents. The numerous waves generated on the river surface significantly interfere with the signal reception of the UAV's spectral sensors, reducing their monitoring accuracy. On the other hand, satellite remote sensing data, due to its relatively low spatial resolution, primarily reflects the average Chl-a concentration level of the water body in the study area, and is relatively less affected by local water turbulence and waves.

[0107] Based on the above scoring results, the Chl-a inversion results for the study area need to undergo information fusion processing to improve accuracy. After information fusion analysis based on the method of this invention, a comparison is made between the fusion results and individual UAV and satellite monitoring results. Figure 4 As shown, Figure 4 In the diagram, (a) represents the fusion result of the present invention, (b) represents the UAV inversion result, and (c) represents the satellite inversion result.

[0108] Depend on Figure 4 As can be seen, the spatial distribution results of Parlung Tsangpo Chl-a analyzed using the method of this invention show a significant reduction in high-concentration areas (red in the figure) compared to the results obtained from a single UAV inversion. The overall spatial distribution characteristics are closer to the satellite inversion results, while the spatial resolution is significantly improved. This indicates that the fusion results are superior to UAV inversion in terms of accuracy and superior to satellite inversion in terms of spatial detail. Therefore, the fusion analysis method of this invention effectively integrates the advantages of two monitoring data sources, significantly optimizing the final monitoring results.

[0109] This invention provides a multi-source heterogeneous remote sensing data decision-level information fusion device, such as... Figure 5 As shown, the device includes:

[0110] Data acquisition module 1 is used to acquire satellite remote sensing images and UAV remote sensing images covering the water body under study at the same time.

[0111] Water extraction module 2 is used to extract water from satellite remote sensing images and UAV remote sensing images to obtain satellite remote sensing water areas and UAV remote sensing water areas.

[0112] Remote sensing inversion module 3 is used to invert water quality parameters for satellite-sensed water areas and UAV-sensed water areas based on selected water quality parameters, and obtain satellite inversion raster data and UAV inversion raster data.

[0113] Accuracy assessment module 4 is used to assess the accuracy of satellite-inverted raster data and UAV-inverted raster data based on selected accuracy assessment indicators.

[0114] The information fusion module 5 is used to correct the UAV-inverted raster data based on the satellite-inverted raster data if the accuracy of the satellite-inverted raster data is better than that of the UAV-inverted raster data, so as to obtain the final remote sensing inversion result; otherwise, the UAV-inverted raster data is directly used as the final remote sensing inversion result.

[0115] In one example, the information fusion module uses satellite-inverted raster data as a benchmark to correct UAV-inverted raster data to obtain the final remote sensing inversion result, including the following process:

[0116] For any satellite pixel contained in the satellite inversion raster data, calculate the average water quality parameters of all UAV pixels within the coverage area of ​​the satellite pixel on the UAV inversion raster data.

[0117] The correction value is obtained by subtracting the average value from the water quality parameters of the satellite pixels.

[0118] The water quality parameters of each UAV pixel within the coverage area of ​​the satellite pixels are added to the correction value to correct the UAV-inverted raster data, thus obtaining the final remote sensing inversion result.

[0119] As an improvement to this embodiment of the invention, the accuracy evaluation module includes:

[0120] Model building unit, used to build hierarchical analysis structure models.

[0121] The hierarchical analysis structure model includes a scheme layer, a criterion layer, and a target layer. The scheme layer includes satellite-inverted raster data and UAV-inverted raster data. The criterion layer includes multiple accuracy evaluation indicators. The target layer is the data with the best accuracy evaluation among the satellite-inverted raster data and UAV-inverted raster data.

[0122] The judgment matrix construction unit is used to compare each accuracy evaluation index of the criterion layer pairwise to determine the importance of each accuracy evaluation index to the target layer and construct the judgment matrix of the hierarchical analysis.

[0123] The verification unit is used to perform consistency verification on the judgment matrix. If the consistency verification passes, the next step is executed; otherwise, the judgment matrix construction module is returned.

[0124] The weight calculation unit is used to derive the weights of each accuracy evaluation index from the judgment matrix.

[0125] The accuracy assessment index calculation unit is used to calculate various accuracy assessment indices for satellite inversion raster data and UAV inversion raster data based on satellite inversion raster data, UAV inversion raster data, and measured values ​​of water quality parameters.

[0126] The accuracy score calculation unit is used to perform weighted calculations based on the various accuracy evaluation indicators and their weights of the satellite-inverted raster data and the UAV-inverted raster data to obtain the comprehensive accuracy score of the satellite-inverted raster data and the UAV-inverted raster data.

[0127] The accuracy evaluation unit is used to compare the comprehensive accuracy scores of satellite-inverted raster data and UAV-inverted raster data, and to take the data with the best accuracy evaluation from the satellite-inverted raster data and UAV-inverted raster data as the output of the target layer.

[0128] Specifically, the accuracy evaluation indicators include correlation coefficient, mean absolute error, root mean square error, and mean relative error.

[0129] The apparatus provided in this embodiment of the invention operates on the same principle and produces the same technical effects as the aforementioned method embodiments. For the sake of brevity, any parts not mentioned in the apparatus embodiments can be referred to the corresponding content in the aforementioned method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the apparatus and units described above can all be referred to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

[0130] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the scope of the technology disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention.

Claims

1. A method for decision-level information fusion of multi-source heterogeneous remote sensing data, characterized in that, The method includes: S1: Acquire satellite and UAV remote sensing images covering the water body under study at the same time. S2: Extract water bodies from the satellite remote sensing images and UAV remote sensing images to obtain the satellite remote sensing water body area and the UAV remote sensing water body area. S3: Based on the selected water quality parameters, perform water quality parameter inversion on the satellite remote sensing water area and the UAV remote sensing water area to obtain satellite inversion raster data and UAV inversion raster data; S4: Evaluate the accuracy of the satellite-inverted raster data and the UAV-inverted raster data based on the selected accuracy evaluation index; S5: If the accuracy of the satellite inverted raster data is better than that of the UAV inverted raster data, then the UAV inverted raster data is corrected based on the satellite inverted raster data to obtain the final remote sensing inversion result; otherwise, the UAV inverted raster data is directly used as the final remote sensing inversion result. The step of correcting the UAV-inverted raster data based on the satellite-inverted raster data to obtain the final remote sensing inversion result includes: For any satellite pixel included in the satellite inversion raster data, calculate the average water quality parameters of all UAV pixels within the coverage area of ​​the satellite pixel on the UAV inversion raster data; The correction value is obtained by subtracting the average value from the water quality parameters of the satellite pixels; The water quality parameters of each UAV pixel within the coverage area of ​​the satellite pixel are added to the correction value to correct the UAV inverted raster data, thus obtaining the final remote sensing inversion result.

2. The method for decision-level information fusion of multi-source heterogeneous remote sensing data according to claim 1, characterized in that, S4 includes: S41: Construct a hierarchical analysis structure model; The hierarchical analysis structure model includes a scheme layer, a criterion layer, and a target layer. The scheme layer includes satellite inversion raster data and UAV inversion raster data. The criterion layer includes multiple accuracy evaluation indicators. The target layer is the data with the best accuracy evaluation among the satellite inversion raster data and UAV inversion raster data. S42: Compare each accuracy evaluation index of the criterion layer pairwise to determine the importance of each accuracy evaluation index to the target layer, and construct the judgment matrix of the hierarchical analysis. S43: Perform a consistency check on the judgment matrix. If the consistency check passes, proceed to the next step; otherwise, return to S42. S44: Derive the weights of each accuracy evaluation index from the judgment matrix; S45: Based on the satellite inversion raster data and UAV inversion raster data, as well as the measured values ​​of the water quality parameters, calculate the various accuracy evaluation indicators of the satellite inversion raster data and UAV inversion raster data. S46: Based on the various accuracy evaluation indicators and their weights of the satellite inverted raster data and the UAV inverted raster data, a weighted calculation is performed to obtain the comprehensive accuracy score of the satellite inverted raster data and the UAV inverted raster data; S47: Compare the comprehensive accuracy scores of the satellite-inverted raster data and the UAV-inverted raster data, and take the data with the best accuracy evaluation among the satellite-inverted raster data and UAV-inverted raster data as the output of the target layer.

3. The method for decision-level information fusion of multi-source heterogeneous remote sensing data according to claim 2, characterized in that, The accuracy evaluation indicators include correlation coefficient, mean absolute error, root mean square error, and mean relative error.

4. A multi-source heterogeneous remote sensing data decision-level information fusion device, characterized in that, The device includes: The data acquisition module is used to acquire satellite remote sensing images and UAV remote sensing images covering the water body under study at the same time. The water extraction module is used to extract water from the satellite remote sensing images and UAV remote sensing images to obtain the satellite remote sensing water area and the UAV remote sensing water area. The remote sensing inversion module is used to invert water quality parameters of the satellite remote sensing water area and the UAV remote sensing water area based on the selected water quality parameters, and obtain satellite inversion raster data and UAV inversion raster data. The accuracy assessment module is used to assess the accuracy of the satellite-inverted raster data and the UAV-inverted raster data based on selected accuracy assessment indicators. The information fusion module is used to correct the UAV inverted raster data based on the satellite inverted raster data if the accuracy of the satellite inverted raster data is better than that of the UAV inverted raster data, so as to obtain the final remote sensing inversion result; otherwise, the UAV inverted raster data is directly used as the final remote sensing inversion result. In the information fusion module, the satellite-inverted raster data is used as a reference to correct the UAV-inverted raster data to obtain the final remote sensing inversion result, including the following process: For any satellite pixel included in the satellite inversion raster data, calculate the average water quality parameters of all UAV pixels within the coverage area of ​​the satellite pixel on the UAV inversion raster data; The correction value is obtained by subtracting the average value from the water quality parameters of the satellite pixels; The water quality parameters of each UAV pixel within the coverage area of ​​the satellite pixel are added to the correction value to correct the UAV inverted raster data, thus obtaining the final remote sensing inversion result.

5. The multi-source heterogeneous remote sensing data decision-level information fusion device according to claim 4, characterized in that, The accuracy assessment module includes: Model building unit, used to build hierarchical analysis structure models; The hierarchical analysis structure model includes a scheme layer, a criterion layer, and a target layer. The scheme layer includes satellite inversion raster data and UAV inversion raster data. The criterion layer includes multiple accuracy evaluation indicators. The target layer is the data with the best accuracy evaluation among the satellite inversion raster data and UAV inversion raster data. The judgment matrix construction unit is used to compare each accuracy evaluation index of the criterion layer pairwise to determine the importance of each accuracy evaluation index to the target layer and construct the judgment matrix of the hierarchical analysis. The verification unit is used to perform consistency verification on the judgment matrix. If the consistency verification passes, the next step is executed; otherwise, the process returns to the judgment matrix construction module. The weight calculation unit is used to derive the weights of each accuracy evaluation index from the judgment matrix. The accuracy evaluation index calculation unit is used to calculate various accuracy evaluation indices of the satellite inversion raster data and the UAV inversion raster data based on the satellite inversion raster data, the UAV inversion raster data, and the measured values ​​of the water quality parameters. The accuracy score calculation unit is used to perform weighted calculations based on the various accuracy evaluation indicators and their weights of the satellite inverted raster data and the UAV inverted raster data to obtain the comprehensive accuracy score of the satellite inverted raster data and the UAV inverted raster data. The accuracy evaluation unit is used to compare the comprehensive accuracy scores of the satellite-inverted raster data and the UAV-inverted raster data, and to take the data with the best accuracy evaluation among the satellite-inverted raster data and the UAV-inverted raster data as the output of the target layer.

6. The multi-source heterogeneous remote sensing data decision-level information fusion device according to claim 5, characterized in that, The accuracy evaluation indicators include correlation coefficient, mean absolute error, root mean square error, and mean relative error.

Images