A large-scale vegetation coverage evaluation method based on multi-source data fusion

Abstract

To enable rapid and reliable assessment of vegetation cover on a large scale, this invention provides a method for large-scale vegetation cover assessment based on multi-source data fusion. By combining large-scale satellite multispectral remote sensing images and high-resolution RGB images, this invention can acquire richer and more detailed vegetation information. Compared to traditional methods, it can accurately capture subtle changes in vegetation cover, improving the accuracy of the assessment. By acquiring high-resolution RGB images for vegetation cover calculation at sampling points, accurate vegetation information can be obtained over a wider area. Compared to traditional methods, this invention reduces the need for field verification, saves human and material resources, and improves the efficiency of the assessment. Furthermore, by analyzing the FVC (Frequency Value Categorization) of sample vegetation cover in typical areas... sam By calculating and weighting the sum, this invention can provide an assessment of vegetation cover for an entire region.

Description

Technical Field

[0001] This invention belongs to the field of vegetation cover assessment technology, and specifically relates to a large-scale vegetation cover assessment method based on multi-source data fusion. Background Technology

[0002] Current methods for assessing large-scale vegetation cover primarily rely on satellite remote sensing imagery. Satellite remote sensing images provide extensive and comprehensive surface spectral information, making them the primary data source for vegetation cover assessment. Researchers currently calculate vegetation cover from satellite images using band extraction and computation. However, relying solely on satellite remote sensing images for vegetation cover assessment has certain limitations: 1. Low resolution: While satellite remote sensing images provide extensive surface information, their resolution is limited by satellite technology and cost, potentially failing to capture richly detailed vegetation information, thus affecting the accuracy of the assessment results; 2. Insufficient data update frequency: Acquisition of satellite remote sensing images is often constrained by satellite orbital periods and weather conditions, which may result in insufficient data update frequency, making real-time or near-real-time vegetation cover assessment impossible; 3. Difficult field verification: The accuracy of vegetation cover calculations from satellite remote sensing images usually requires field verification, but this process consumes significant human and material resources and is often affected by terrain and weather conditions, making effective verification difficult.

[0003] In recent years, with the rapid popularization of drone technology, significant progress has been made in the technology of using drones' high-resolution RGB lenses to assess vegetation cover in regions. Compared with traditional remote sensing inversion methods, drones have several unique advantages and application scenarios. First, drones can be used in conjunction with other systems to achieve unmanned image acquisition. Through preset flight paths or autonomous flight, drones can efficiently acquire images in areas where vegetation cover needs to be assessed. This flexibility and controllability make drones an ideal choice for quickly acquiring high-resolution images. Second, the high-resolution RGB images provided by drones offer more detailed and accurate information. Compared to satellite remote sensing images, drone images have higher spatial resolution, capturing smaller-scale surface features and vegetation structures. This makes vegetation cover assessments based on drone images more accurate and detailed, capable of detecting smaller vegetation units and providing more refined vegetation cover information. Furthermore, the real-time or near-real-time nature of drone images is also an advantage. Because drones can rapidly acquire data at the required time and location, real-time or near-real-time vegetation cover assessments can be achieved. This is of great significance for applications that require rapid access to vegetation change information, such as agricultural monitoring and disaster response.

[0004] However, the limitations of using drones for vegetation cover assessment lie in the limited range and flight time of drones, and the limited image information acquired is confined to a small scale. Therefore, there is an urgent need for a new method that enables rapid and reliable assessment of vegetation cover in a large-scale spatial environment. Summary of the Invention

[0005] To enable rapid and reliable assessment of vegetation cover on a large scale, this invention provides a method for assessing large-scale vegetation cover based on multi-source data fusion.

[0006] The technical solution of the large-scale vegetation cover assessment method based on multi-source data fusion of the present invention is as follows:

[0007] A method for assessing large-scale vegetation cover based on multi-source data fusion includes the following steps:

[0008] S1, acquire large-scale satellite multispectral remote sensing images of the target area, and preprocess the large-scale satellite multispectral remote sensing images;

[0009] S2, calculate the vegetation index of each pixel in the preprocessed large-scale satellite multispectral remote sensing image;

[0010] S3, using the vegetation index, divide each pixel in the large-scale satellite multispectral remote sensing image into various desertification sub-regions, and determine the number of pixels Pi in each desertification sub-region. n Where n is the number of the desertification sub-region; the classification method and classification basis (the required vegetation index) of the desertification sub-region can be determined according to the researcher's requirements;

[0011] S4. Select several sampling points in each desertification sub-region;

[0012] S5. Acquire high-resolution RGB images of each sampling point, and calculate the sample vegetation coverage (FVC) of each sampling point based on the high-resolution RGB images. sam ;

[0013] S6, The sample vegetation cover (FVC) of each sampling point in each desertification sub-region. sam The average value of vegetation cover f for each desertification sub-region is determined by averaging. n ;

[0014] S7 represents the vegetation cover value f of each desertification sub-region. n Perform a weighted summation to calculate the total vegetation cover (FVC) of the target area. tol Among them, the vegetation cover representative value f of each desertification sub-region nThe weight is the area percentage of each desertification sub-region, specifically calculated using the formula: FVC. tol = Pi1 / Pi*f1 + Pi2 / Pi*f2 + ... + Pi n / Pi*f n Where Pi is the total number of pixels in a large-scale satellite multispectral remote sensing image.

[0015] The technical advantages of the large-scale vegetation cover assessment method based on multi-source data fusion of the present invention are as follows:

[0016] I. Improving the Accuracy of Vegetation Cover Assessment: This invention combines large-scale satellite multispectral remote sensing images and high-resolution RGB images to obtain richer and more detailed vegetation information. Compared with traditional methods, it can accurately capture subtle changes in vegetation cover, thus improving the accuracy of the assessment.

[0017] Second, it reduces the need for field verification: This invention calculates vegetation cover at sampling points by acquiring high-resolution RGB images, enabling the acquisition of accurate vegetation information over a wider area. Compared to traditional methods, this invention reduces the need for field verification, saves human and material resources, and improves the efficiency of the assessment.

[0018] III. Providing comprehensive vegetation cover assessment results: through the FVC of sample vegetation cover in typical areas. sam By calculating and weighted summing, this invention can provide an overall assessment of vegetation cover in a region. This enables fields such as agricultural monitoring and ecological environment assessment to obtain comprehensive and integrated vegetation information, providing a scientific basis for decision-making.

[0019] Furthermore, in the large-scale vegetation cover assessment method based on multi-source data fusion, S1, the preprocessing of the large-scale satellite multispectral remote sensing image includes geometric correction, atmospheric correction, noise removal, and image stitching. This operation can be implemented in ENVI software or through libraries such as GDAL and Rasterio in Python.

[0020] Furthermore, in the aforementioned large-scale vegetation cover assessment method based on multi-source data fusion, specifically in S2, the vegetation index is the vegetation cover FVC based on satellite remote sensing images. sat Improved Adjusted Vegetation Index (MSAVI) sat Enhanced Vegetation Index (EVI) sat The calculation formula is as follows:

[0021] FVC sat =(NDVI) sat -NDVI soil ) / (NDVI veg -NDVIsoil ); among which, NDVI sat NDVI is the pixel normalized vegetation index. soil NDVI is the NDVI value of a bare soil pixel. veg NDVI value of vegetation pixels;

[0022] NDVI sat = (NIR-R) / (NIR+R), where NIR is the near-infrared reflectance of the pixel and R is the red reflectance of the pixel;

[0023] MSAVI sat =((2NIR+1) 2 -8(NIR-R)) 0.5 ) / 2;

[0024] EVI sat =2.5*(NIR-R) / (NIR+0.6R+7.6B+1); where B is the reflectivity of the blue band of the pixel;

[0025] In S3, the desertification sub-regions are divided into non-desertification regions, slightly desertified regions, moderately desertified regions, and severely desertified regions.

[0026] The pixel division method is as follows:

[0027] S31, respectively set FVC sat MSAVI sat EVI sat The corresponding threshold TH fvc TH msavi TH evi The three thresholds were determined by relevant experts based on the actual situation. FVC was selected as the indicator to distinguish between mild and moderate desertification, MSAVI as the indicator to distinguish between moderate and severe desertification, and EVI as the indicator to distinguish between non-desertification and mild desertification. The specific methods are as follows:

[0028] S32, Comparison of FVC sat With TH fvc When FVC sat >TH fvc If the condition is met, proceed to S33; otherwise, proceed to S34.

[0029] S33, compared to MSAVI sat With TH msavi When MSAVI sat >TH msavi If the condition is met, the pixel is classified as a severely desertified area; otherwise, the pixel is classified as a moderately desertified area.

[0030] S34, compared to EVIsat With TH evi When EVI sat >TH evi If the condition is met, the pixel is assigned to a slightly desertified area; otherwise, the pixel is assigned to a non-desertified area.

[0031] Furthermore, the aforementioned large-scale vegetation cover assessment method based on multi-source data fusion specifically includes NDVI. soil The value is the NDVI of all pixels in the satellite remote sensing image. sat The value that ranks in the 5th percentile after being sorted from smallest to largest; NDVI veg The value is the NDVI of all pixels in the satellite remote sensing image. sat The value that ranks at the 95th percentile after being arranged from smallest to largest.

[0032] Furthermore, in the large-scale vegetation cover assessment method based on multi-source data fusion, to ensure the authenticity of the data, in S4, the sampling points are randomly selected, with at least 3 sampling points selected in each desertification sub-region.

[0033] Furthermore, in the aforementioned large-scale vegetation cover assessment method based on multi-source data fusion, in S5, the high-resolution RGB image is obtained through drone photography. Traditional satellite remote sensing images have limited data update frequency, making real-time or near-real-time vegetation cover assessment impossible. Utilizing the flexibility and rapid response capabilities of drones, vegetation cover calculations can be performed quickly when needed, enabling more timely vegetation monitoring and change detection. Additionally, large-scale satellite multispectral remote sensing images are used for typical area division and area determination, followed by drone-based calculation of vegetation cover at sampling points. Compared to traditional field survey methods, this approach can acquire data from a large number of sampling points in a shorter time, improving the efficiency of large-scale vegetation monitoring.

[0034] Furthermore, in the large-scale vegetation cover assessment method based on multi-source data fusion, in S5, the sample vegetation cover (FVC) of the sampling points... sam The calculation method is as follows:

[0035] S51, use Python to read high-definition RGB images and use the OpenCV library in Python to perform color space conversion, converting the RGB color space of the high-definition RGB image to the LAB color space;

[0036] S52, using a traversal loop to extract the LAB chromaticity of each pixel block in the high-definition RGB image;

[0037] S53, determine whether a pixel block belongs to greenery by judging the extracted LAB chromaticity; the judgment method is that if all three channels of the LAB chromaticity of a pixel block are within the threshold, it is judged as a pixel belonging to greenery; otherwise, it is not a pixel belonging to greenery; where the minimum value of the threshold brightness L in the chromaticity judgment is... min Maximum brightness L max Minimum red-green hue A min The maximum value of red-green color A max Minimum value of yellow-blue saturation B min The maximum value of yellow-blue saturation B max Determined by experts based on the actual situation;

[0038] S54, count the number of pixels belonging to green plants P gre And the total number of pixel blocks P of the high-definition RGB image;

[0039] S55, Calculate the sample vegetation cover (FVC) sam =P gre / P. Attached Figure Description

[0040] Figure 1 This is a flowchart of a large-scale vegetation cover assessment method based on multi-source data fusion according to the present invention.

[0041] Figure 2 This is a flowchart of a pixel division method for a large-scale vegetation cover assessment method based on multi-source data fusion according to the present invention.

[0042] Figure 3 This invention relates to a method for assessing large-scale vegetation cover based on multi-source data fusion, specifically the sample vegetation cover (FVC) at sampling points. sam The flowchart of the calculation method. Detailed Implementation

[0043] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become clearer from the following description and claims. It should be noted that the drawings are all in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.

[0044] Example 1:

[0045] refer to Figures 1 to 3 This embodiment provides a method for assessing large-scale vegetation cover based on multi-source data fusion, including the following steps:

[0046] S1, acquire large-scale satellite multispectral remote sensing images of the target area, and preprocess the large-scale satellite multispectral remote sensing images;

[0047] S2, calculate the vegetation index of each pixel in the preprocessed large-scale satellite multispectral remote sensing image;

[0048] S3, using the vegetation index, divide each pixel in the large-scale satellite multispectral remote sensing image into various desertification sub-regions, and determine the number of pixels Pi in each desertification sub-region. n Where n is the number of the desertification sub-region; the classification method and classification basis (the required vegetation index) of the desertification sub-region can be determined according to the researcher's requirements;

[0049] S4. Select several sampling points in each desertification sub-region;

[0050] S5. Acquire high-resolution RGB images of each sampling point, and calculate the sample vegetation coverage (FVC) of each sampling point based on the high-resolution RGB images. sam ;

[0051] S6, The sample vegetation cover (FVC) of each sampling point in each desertification sub-region. sam The average value of vegetation cover f for each desertification sub-region is determined by averaging. n ;

[0052] S7 represents the vegetation cover value f of each desertification sub-region. n Perform a weighted summation to calculate the total vegetation cover (FVC) of the target area. tol Among them, the vegetation cover representative value f of each desertification sub-region n The weight is the area percentage of each desertification sub-region, specifically calculated using the formula: FVC. tol = Pi1 / Pi*f1 + Pi2 / Pi*f2 + ... + Pi n / Pi*f n Where Pi is the total number of pixels in a large-scale satellite multispectral remote sensing image.

[0053] The technical advantages of the large-scale vegetation cover assessment method based on multi-source data fusion in this embodiment are as follows:

[0054] I. Improving the Accuracy of Vegetation Cover Assessment: This invention combines large-scale satellite multispectral remote sensing images and high-resolution RGB images to obtain richer and more detailed vegetation information. Compared with traditional methods, it can accurately capture subtle changes in vegetation cover, thus improving the accuracy of the assessment.

[0055] Second, it reduces the need for field verification: This invention calculates vegetation cover at sampling points by acquiring high-resolution RGB images, enabling the acquisition of accurate vegetation information over a wider area. Compared to traditional methods, this invention reduces the need for field verification, saves human and material resources, and improves the efficiency of the assessment.

[0056] III. Providing comprehensive vegetation cover assessment results: through the FVC of sample vegetation cover in typical areas. sam By calculating and weighted summing, this invention can provide an overall assessment of vegetation cover in a region. This enables fields such as agricultural monitoring and ecological environment assessment to obtain comprehensive and integrated vegetation information, providing a scientific basis for decision-making.

[0057] As a preferred implementation, the large-scale vegetation cover assessment method based on multi-source data fusion, in step S1, includes the preprocessing of the large-scale satellite multispectral remote sensing image, which includes geometric correction, atmospheric correction, noise removal, and image stitching. This operation can be implemented in ENVI software or through libraries such as GDAL and Raster io in Python.

[0058] As a preferred implementation, the large-scale vegetation cover assessment method based on multi-source data fusion specifically includes, in S2, the vegetation index being the vegetation cover FVC based on satellite remote sensing images. sat Improved Adjusted Vegetation Index (MSAVI) sat Enhanced Vegetation Index (EVI) sat The calculation formula is as follows:

[0059] FVC sat =(NDVI) sat -NDVI soil ) / (NDVI veg -NDVI soil ); among which, NDVI sat NDVI is the pixel normalized vegetation index. soil NDVI is the NDVI value of a bare soil pixel. veg NDVI value of vegetation pixels;

[0060] NDVI sat = (NIR-R) / (NIR+R), where NIR is the near-infrared reflectance of the pixel and R is the red reflectance of the pixel;

[0061] MSAVI sat =((2NIR+1) 2 -8(NIR-R)) 0.5 ) / 2;

[0062] EVI sat=2.5*(NIR-R) / (NIR+0.6R+7.6B+1); where B is the reflectivity of the blue band of the pixel;

[0063] In S3, the desertification sub-regions are divided into non-desertification regions, slightly desertified regions, moderately desertified regions, and severely desertified regions.

[0064] The pixel division method is as follows:

[0065] S31, respectively set FVC sat MSAVI sat EVI sat The corresponding threshold TH fvc TH msavi TH evi The three thresholds were determined by relevant experts based on the actual situation. FVC was selected as the indicator to distinguish between mild and moderate desertification, MSAVI as the indicator to distinguish between moderate and severe desertification, and EVI as the indicator to distinguish between non-desertification and mild desertification. The specific methods are as follows:

[0066] S32, Comparison of FVC sat With TH fvc When FVC sat >TH fvc If the condition is met, proceed to S33; otherwise, proceed to S34.

[0067] S33, compared to MSAVI sat With TH msavi When MSAVI sat >TH msavi If the condition is met, the pixel is classified as a severely desertified area; otherwise, the pixel is classified as a moderately desertified area.

[0068] S34, compared to EVI sat With TH evi When EVI sat >TH evi If the condition is met, the pixel is assigned to a slightly desertified area; otherwise, the pixel is assigned to a non-desertified area.

[0069] As a preferred implementation, the large-scale vegetation cover assessment method based on multi-source data fusion specifically includes NDVI. soil The value is the NDVI of all pixels in the satellite remote sensing image. sat The value that ranks in the 5th percentile after being sorted from smallest to largest; NDVI veg The value is the NDVI of all pixels in the satellite remote sensing image. sat The value that ranks at the 95th percentile after being arranged from smallest to largest.

[0070] As a preferred implementation method, in order to ensure the authenticity of the data, the sampling points in S4 of the large-scale vegetation cover assessment method based on multi-source data fusion are randomly selected, with at least 3 sampling points selected in each desertification sub-region.

[0071] As a preferred implementation, in the large-scale vegetation cover assessment method based on multi-source data fusion, in S5, the high-resolution RGB image is obtained by taking pictures with a drone. Traditional satellite remote sensing images have limited data update frequency, making real-time or near-real-time vegetation cover assessment impossible. Utilizing the flexibility and rapid response capabilities of drones, vegetation cover calculations can be performed quickly when needed, enabling more timely vegetation monitoring and change detection. Furthermore, large-scale satellite multispectral remote sensing images are used to divide typical areas and determine their areas, and then the drone is used to calculate the vegetation cover at sampling points. Compared to traditional field survey methods, this method can acquire data from a large number of sampling points in a shorter time, improving the efficiency of monitoring large-scale vegetation.

[0072] As a preferred implementation, in the large-scale vegetation cover assessment method based on multi-source data fusion, in S5, the sample vegetation cover value (FVC) of the sampling points is... sam The calculation method is as follows:

[0073] S51, use Python to read high-definition RGB images and use the OpenCV library in Python to perform color space conversion, converting the RGB color space of the high-definition RGB image to the LAB color space;

[0074] S52, using a traversal loop to extract the LAB chromaticity of each pixel block in the high-definition RGB image;

[0075] S53, determine whether a pixel block belongs to greenery by judging the extracted LAB chromaticity; the judgment method is that if all three channels of the LAB chromaticity of a pixel block are within the threshold, it is judged as a pixel belonging to greenery; otherwise, it is not a pixel belonging to greenery; where the minimum value of the threshold brightness L in the chromaticity judgment is... min Maximum brightness L max Minimum red-green hue A min The maximum value of red-green color A max Minimum value of yellow-blue saturation B min The maximum value of yellow-blue saturation B max Determined by experts based on the actual situation;

[0076] S54, count the number of pixels belonging to green plants P gre And the total number of pixel blocks P of the high-definition RGB image;

[0077] S55, Calculate the sample vegetation cover (FVC)sam =P gre / P.

[0078] The above description is merely a description of preferred embodiments of the present invention and is not intended to limit the scope of the present invention in any way. Any changes or modifications made by those skilled in the art based on the above disclosure shall fall within the protection scope of the claims.

Claims

1. A method for assessing large-scale vegetation cover based on multi-source data fusion, characterized in that, Includes the following steps: S1, acquire large-scale satellite multispectral remote sensing images of the target area, and preprocess the large-scale satellite multispectral remote sensing images; S2, calculate the vegetation index of each pixel in the preprocessed large-scale satellite multispectral remote sensing image; S3, using the vegetation index, divide each pixel in the large-scale satellite multispectral remote sensing image into various desertification sub-regions, and determine the number of pixels Pi in each desertification sub-region. n Where n is the number of the desertification sub-region; S4. Select several sampling points in each desertification sub-region; S5. Acquire high-resolution RGB images of each sampling point, and calculate the sample vegetation coverage (FVC) of each sampling point based on the high-resolution RGB images. sam ; S6, The sample vegetation cover (FVC) of each sampling point in each desertification sub-region. sam The average value of vegetation cover f for each desertification sub-region is determined by averaging. n ; S7, Calculate the total vegetation cover (FVC) of the target area. tol FVC tol = Pi1 / Pi*f1 + Pi2 / Pi*f2 + ... + Pi n / Pi*f n Where Pi is the total number of pixels in a large-scale satellite multispectral remote sensing image.

2. The method for assessing large-scale vegetation cover based on multi-source data fusion as described in claim 1, characterized in that, In S1, the preprocessing of the large-scale satellite multispectral remote sensing image includes geometric correction, atmospheric correction, noise removal, and image stitching.

3. The method for assessing large-scale vegetation cover based on multi-source data fusion as described in claim 1, characterized in that, In S2, the vegetation index is the vegetation cover (FVC) based on satellite remote sensing imagery. sat Improved Adjusted Vegetation Index (MSAVI) sat Enhanced Vegetation Index (EVI) sat The calculation formula is as follows: FVC sat =(NDVI) sat -NDVI soil ) / (NDVI veg -NDVI soil ); among which, NDVI sat NDVI is the pixel normalized vegetation index. soil NDVI is the NDVI value of a bare soil pixel. veg NDVI value of vegetation pixels; NDVI sat = (NIR-R) / (NIR+R), where NIR is the near-infrared reflectance of the pixel and R is the red reflectance of the pixel; MSAVI sat =(((2NIR+1) 2 -8(NIR-R)) 0.5 ) / 2; EVI sat =2.5*(NIR-R) / (NIR+0.6R+7.6B+1); where B is the reflectivity of the blue band of the pixel; In S3, the desertification sub-regions are divided into non-desertification regions, slightly desertified regions, moderately desertified regions, and severely desertified regions. The pixel division method is as follows: S31, respectively set FVC sat MSAVI sat EVI sat The corresponding threshold TH fvc TH msavi TH evi ; S32, Comparison of FVC sat With TH fvc When FVC sat >TH fvc If the condition is met, proceed to S33; otherwise, proceed to S34. S33, compared to MSAVI sat With TH msavi When MSAVI sat >TH msavi If the condition is met, the pixel is classified as a severely desertified area; otherwise, the pixel is classified as a moderately desertified area. S34, compared to EVI sat With TH evi When EVI sat >TH evi If the condition is met, the pixel is assigned to a slightly desertified area; otherwise, the pixel is assigned to a non-desertified area.

4. The method for assessing large-scale vegetation cover based on multi-source data fusion as described in claim 3, characterized in that: NDVI soil The value is the NDVI of all pixels in the satellite remote sensing image. sat The value that ranks in the 5th percentile after being sorted from smallest to largest; NDVI veg The value is the NDVI of all pixels in the satellite remote sensing image. sat The value that ranks at the 95th percentile after being arranged from smallest to largest.

5. The method for assessing large-scale vegetation cover based on multi-source data fusion as described in claim 1, characterized in that, In S4, the sampling points are randomly selected, with at least 3 sampling points selected in each desertification sub-region.

6. The method for assessing large-scale vegetation cover based on multi-source data fusion as described in claim 1, characterized in that, In S5, the high-definition RGB image is obtained by taking pictures with a drone.

7. The method for assessing large-scale vegetation cover based on multi-source data fusion as described in claim 1, characterized in that, In S5, the sample vegetation cover (FVC) at the sampling points sam The calculation method is as follows: S51, use Python to read high-definition RGB images and use the OpenCV library in Python to perform color space conversion, converting the RGB color space of the high-definition RGB image to the LAB color space; S52, using a traversal loop to extract the LAB chromaticity of each pixel block in the high-definition RGB image; S53, determine whether a pixel block is a green plant by judging the extracted LAB chromaticity; the judgment method is that if all three channels of the LAB chromaticity of a pixel block are within the threshold, it is judged as a pixel belonging to green plants, otherwise it is not a pixel belonging to green plants. S54, count the number of pixels belonging to green plants P gre And the total number of pixel blocks P of the high-definition RGB image; S55, Calculate the sample vegetation cover (FVC) sam =P gre / P.

Images