A multi-scale vegetation coverage remote sensing calculation method based on UAV

Abstract

An embodiment of the present invention discloses a multi-scale vegetation coverage remote sensing calculation method based on drones, the method comprising: S1: acquiring satellite images and drone images; S2: spatially registering satellite images and drone images; S3: extracting vegetation information according to the drone images, and calculating vegetation index according to the satellite images; S4: aggregating the vegetation information of the drone images with the satellite images through a 2-D Gaussian sampling method and constructing a vegetation coverage model; S5: calculating vegetation index according to satellite images of the remaining periods, and correcting the vegetation index of the remaining periods, and obtaining the corrected vegetation coverage of the remaining periods according to the corrected vegetation index and the vegetation coverage model. The present invention solves the problem of spatial mismatch between the surface sample plot and the corresponding satellite pixel size, improves the calculation accuracy of vegetation coverage, and can achieve long-term and high-precision vegetation monitoring.

Description

Technical Field

[0001] The present invention relates to the application field of remote sensing technology, and in particular to a multi-scale vegetation coverage remote sensing calculation method based on unmanned aerial vehicle. Background Art

[0002] Vegetation coverage is an important indicator to measure the condition of surface vegetation, an important basic data to describe the ecosystem, and an important indicator of regional ecosystem environmental changes. It is of great significance to hydrology, ecology, regional changes, etc.

[0003] The traditional method of measuring vegetation coverage on the ground is time-consuming, laborious and not conducive to dynamic monitoring of vegetation over a large area. The development of remote sensing technology provides multi-scale remote sensing data for monitoring vegetation coverage. Satellite images, as representatives of low-resolution large-scale images, have the characteristics of long time series and large range. At the same time, satellite remote sensing is currently the mainstream way to calculate vegetation coverage, and many methods for estimating vegetation coverage have been developed, such as empirical model method, mixed pixel decomposition model method, etc. Among them, the most widely used method is to estimate vegetation coverage by establishing an empirical model between ground-measured vegetation coverage and vegetation index calculated from satellite images. Although these two methods can quickly obtain vegetation coverage, there is still the problem of mixed pixels, that is, a single pixel contains not only vegetation, but also other landforms such as soil, and the selection of vegetation index thresholds is often uncertain. Therefore, using the pixel binary model based on NDVI to estimate vegetation coverage will lead to inaccurate extraction of vegetation information, affecting the prediction accuracy of vegetation coverage. Manual ground sampling is somewhat subjective, and the sampling range is usually 1-10 4 m 2 This is quite different from the spatial resolution of satellite remote sensing product pixels. Therefore, there is a spatial mismatch between the surface sample and the corresponding satellite pixel size, which leads to inaccurate vegetation coverage estimation using the empirical model method based on ground measurements and satellite data.

[0004] In recent years, drones, as representatives of high-resolution small-scale images, can carry various micro-sensors and quickly generate digital orthophotos, and have been widely used in vegetation monitoring. Existing studies have shown that drone aerial photogrammetry technology provides images with spatial resolutions ranging from centimeters to meters, which can be used to improve the accuracy of satellite image vegetation coverage modeling. This is because compared with commercial satellites and traditional airborne sensors, drone remote sensing provides a fast and efficient way to obtain data with extremely high spatial resolution. These extremely high spatial resolution data can be upscaled to the required spatial resolution, and can also be matched with satellite data with global coverage, and vegetation coverage can be estimated at multiple scales. However, most current studies are based on a single scale (satellite / drone images) to study vegetation coverage, lacking the joint use of multi-scale remote sensing, resulting in the inability of drone-based remote sensing technology with extremely high spatial resolution to play its due role in spatiotemporal vegetation monitoring.

[0005] Previous studies generally used satellite remote sensing to monitor long-term vegetation changes. This is because some satellites were launched earlier (such as Landsat satellites) and can trace back to more ancient historical data, which facilitates the monitoring of global changes for decades. However, due to the coarse spatial resolution of satellite images and the limitations of the solar zenith angle and the sensor itself, the accuracy of vegetation change monitoring results cannot be guaranteed. In recent years, UAV remote sensing can also obtain high-spatial-resolution images for several years, which can be used for high-precision vegetation time series monitoring. However, due to the influence of different environmental factors (cloudy or sunny, windy or windless, etc.) during each flight mission, there are certain differences between multiple UAV images, which will have a certain impact on the results of long-term vegetation monitoring. In short, the current data sources for long-term monitoring are all low- and medium-resolution remote sensing images, which make it difficult to accurately control the changes in vegetation coverage, and lack high-precision vegetation coverage assessment methods for multi-scale remote sensing data. Summary of the invention

[0006] In view of this, the first embodiment of the present invention provides a multi-scale vegetation coverage remote sensing calculation method based on an unmanned aerial vehicle, comprising:

[0007] S1: Acquire satellite images and drone images, wherein the coverage area of ​​the satellite images includes the coverage area of ​​the drone images;

[0008] S2: spatial registration of satellite images and drone images of the same period and the same coverage area;

[0009] S3: extracting vegetation information according to the drone image, and calculating vegetation index according to the satellite image;

[0010] S4: Aggregating the vegetation information of the UAV image with the satellite image through a 2-D Gaussian sampling method to obtain the vegetation coverage of each satellite pixel, and constructing a vegetation coverage model with the vegetation index calculated from the corresponding satellite pixel;

[0011] S5: Calculate the vegetation index according to the satellite images of the remaining period, and correct the vegetation index of the remaining period, and obtain the corrected vegetation coverage of the remaining period according to the corrected vegetation index and the vegetation coverage model.

[0012] In a specific embodiment, the S3 includes:

[0013] S31: classifying the same coverage area as the satellite image according to the drone image into a vegetation area and a non-vegetation area;

[0014] S32: Binarization is performed according to the classification result of the drone image to obtain a binarization result.

[0015] In a specific embodiment, the S4 includes:

[0016] S41: Determine the window size of a satellite image after aggregation as α×α, where

[0017]

[0018] Among them, Rsatellite is the spatial resolution of each pixel of satellite images in the same period, and RUAV is the spatial resolution of each pixel of drone images in the same period;

[0019] S42: Calculate the pixel weight of each drone image in the aggregated satellite image window using 2-D Gaussian sampling:

[0020]

[0021] Among them, W (i,j) represents the weight value of the drone image pixel in the i-th row and j-th column in a satellite pixel,

[0022] σ represents the standard deviation, and σ=α / 6;

[0023] S43: Calculate the vegetation coverage of each pixel of the aggregated satellite image according to the binarization result and the pixel weight of the drone image.

[0024]

[0025] Among them, ρi ,j is the binarization result of the pixels in the i-th row and j-th column of the drone image.

[0026] In a specific embodiment, the S4 further includes:

[0027] S44: establishing a first linear regression equation by using the corresponding relationship between the vegetation coverage of the aggregated satellite image and the vegetation index of the corresponding area calculated according to the satellite image;

[0028] S45: The vegetation coverage of the remaining coverage areas may be obtained according to the first linear regression equation and the vegetation index of the remaining coverage areas calculated according to the satellite image.

[0029] In a specific embodiment, an ID number is assigned to each pixel of the aggregated satellite image, and the vegetation coverage of each pixel of the aggregated satellite image and the vegetation index calculated based on each pixel are associated with each other through the ID number.

[0030] In a specific embodiment, the S5 includes:

[0031] S51: Obtain satellite images of the same period as the drone image and satellite images of the same area at other periods, display them in true color, and select multiple types of ground coverage in the target area;

[0032] S52: Calculating the vegetation index of the ground object coverage according to the satellite image of the same period as the UAV image and the satellite image of the same area at other periods, and using the corresponding relationship between the vegetation index of the ground object coverage calculated according to the satellite image of the same period as the UAV image and the vegetation index of the ground object coverage calculated according to the satellite image of the same area at other periods, establishing a second linear regression equation to complete the correction of the vegetation index of any period;

[0033] S53: According to the first linear regression equation, the vegetation index of any period after correction is substituted into the first linear regression equation to obtain the vegetation coverage of the remaining periods.

[0034] In a specific embodiment, the S1 includes:

[0035] The satellite image and the drone image are preprocessed.

[0036] The beneficial effects of the present invention are as follows:

[0037] The invention provides a multi-scale vegetation coverage remote sensing calculation method based on unmanned aerial vehicle. The vegetation information of the unmanned aerial vehicle image and the satellite image are aggregated by a 2-D Gaussian sampling method to obtain the vegetation coverage of each satellite pixel in the corresponding area, thereby solving the problem of inaccurate vegetation coverage calculation and the problem of spatial mismatch between the surface sample plot and the corresponding satellite pixel size in the prior art. The vegetation coverage model is constructed and the vegetation index calculated by the satellite image of the remaining period is corrected to obtain the vegetation coverage of the satellite image of the arbitrary period, thereby realizing long-time series, high-precision and large-range vegetation coverage inversion, that is, not only the long-time series and large-range advantages of the satellite image are brought into play, but also the high spatial resolution advantage of the unmanned aerial vehicle image is brought into play, so that the recovery status of the vegetation in the region in recent years can be more accurately understood, and a reference is provided for the evaluation of the regional ecological environment restoration effect. The invention has the advantages of strong adaptability, high precision and easy operation, can greatly reduce the field artificial field observation, and has a wide application prospect. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0039] Figure 1 A flowchart of a multi-scale vegetation coverage remote sensing calculation method based on an unmanned aerial vehicle according to an embodiment of the present invention is shown;

[0040] Figure 2 A schematic diagram of vegetation coverage calculated by aggregating satellite images and drone images of different resolutions according to an embodiment of the present invention is shown;

[0041] Figure 3 A schematic diagram showing a comparison between a multi-scale vegetation coverage remote sensing calculation method based on a drone according to an embodiment of the present invention and the prior art is shown. DETAILED DESCRIPTION

[0042] To make the technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0043] like Figure 1 As shown, an embodiment of the present invention provides a multi-scale vegetation coverage remote sensing calculation method based on an unmanned aerial vehicle, comprising:

[0044] S1: Acquire satellite images and drone images, wherein the coverage area of ​​the satellite images includes the coverage area of ​​the drone images;

[0045] S2: spatial registration of satellite images and drone images of the same period and the same coverage area;

[0046] S3: extracting vegetation information according to the drone image, and calculating vegetation index according to the satellite image;

[0047] S4: Aggregating the vegetation information of the UAV image with the satellite image through a 2-D (two-dimensional) Gaussian sampling method to obtain the vegetation coverage of each satellite pixel, and constructing a vegetation coverage model with the vegetation index calculated from the corresponding satellite pixel;

[0048] S5: Calculate the vegetation index according to the satellite images of the remaining period, and correct the vegetation index of the remaining period, and obtain the corrected vegetation coverage of the remaining period according to the corrected vegetation index and the vegetation coverage model.

[0049] It is understandable that in this embodiment, the satellite image and the drone image are not specifically limited. Figure 2 As shown, the satellite image can be a Gaofen-2 (GF-2) satellite remote sensing image, a Sentinel-2 satellite remote sensing image or a Landsat5 / Landsat7 / Landsat8 satellite remote sensing image, wherein the spatial resolution of the Gaofen-2 satellite remote sensing image is 4 meters, the spatial resolution of the Sentinel-2 satellite remote sensing image is 10 meters, and the spatial resolution of the Landsat series satellite remote sensing image is 30 meters; the drone image can be an image obtained by a drone, which can be a fixed-wing drone, a multi-rotor drone or other forms. The specific setting forms are well known to those skilled in the art, and this application does not impose too many restrictions, as long as the image acquisition requirements can be met.

[0050] In a specific embodiment, Table 1 shows a comparison between a multi-scale vegetation coverage remote sensing calculation method based on a drone in the prior art and a vegetation coverage remote sensing calculation method in this embodiment.

[0051] From the content in Table 1, it can be seen that the pixel dichotomy method based on satellite images used in the prior art has a determination coefficient R when calculating vegetation coverage. 2 is 0.301, and the root mean square error RMSE is 0.349. It can be understood by those skilled in the art that the coefficient of determination R 2The larger the value, the higher the correlation between the method and the actual situation, and the more significant the effect; the smaller the root mean square error RMSE value, the smaller the error between the method and the actual situation, and the more significant the effect. According to the content of Table 1, in this embodiment, when any satellite remote sensing image is selected and the vegetation coverage is calculated, the error with the actual situation is smaller than the error caused by the calculation using the pixel dichotomy method in the prior art. For example, when the Landsat series satellite remote sensing images are selected, the vegetation coverage is calculated by the method described in this embodiment, and the calculated determination coefficient R 2 The root mean square error (RMSE) is 0.738 and 0.149, which is significantly improved compared with the existing technology. Figure 3 As shown, it can be seen that this embodiment aggregates the vegetation information of the drone image and the satellite image through the 2-D Gaussian sampling method to obtain the vegetation coverage of each satellite pixel in the corresponding area, and establishes a vegetation coverage estimation model with the vegetation coverage of the corresponding satellite pixels. This process solves the problem of inaccurate vegetation coverage calculation in the prior art and the problem of spatial mismatch between the surface sample and the corresponding satellite pixel size, and improves the calculation accuracy of vegetation coverage.

[0052] Table 1

[0053] Pixel dichotomy Gaofen-2 satellite remote sensing image Sentinel-2 satellite remote sensing image Landsat series satellite remote sensing images Spatial resolution 30 meters 4 meters 10 meters 30 meters <![CDATA[Coefficient of determination R 2 > 0.301 0.546 0.579 0.738 Root mean square error RMSE 0.349 0.277 0.247 0.149

[0054] In a specific embodiment, the multi-scale vegetation coverage remote sensing calculation method based on drone specifically includes:

[0055] S1: Acquire satellite images and drone images, wherein the coverage area of ​​the satellite images includes the coverage area of ​​the drone images;

[0056] In this embodiment, when acquiring the satellite images, satellite images of all previous years covering all study areas should be acquired, such as satellite images of a mining area from 2010 to 2021. When acquiring drone images, several typical areas should be selected from all study areas, and drone images of these several typical areas in specific periods should be acquired. For example, several areas with different vegetation coverage should be selected in a mining area, and several drone images of these areas with different vegetation coverage in 2021 should be acquired, so that the drone images can be subsequently aggregated with the satellite images through the 2-D Gaussian sampling method, thereby obtaining high-precision vegetation coverage in the study area and greatly reducing manual field observations in the field.

[0057] In a specific embodiment, the S1 further comprises:

[0058] The satellite image and the drone image are preprocessed, for example, the satellite image is subjected to radiation calibration and atmospheric correction, and the drone image is spliced ​​to obtain an orthophoto of the corresponding area.

[0059] In this embodiment, by preprocessing the satellite image, such as performing radiation calibration, the difference caused by the uneven radiation response of the original image is corrected, and the real information of the ground object reflected by the image is restored. By preprocessing the satellite image, such as performing atmospheric correction, the influence of factors such as atmosphere and light on the reflection of the ground object is eliminated, and the calculation accuracy of vegetation coverage is improved. The orthophoto of the study area is obtained by splicing the acquired drone images using professional software.

[0060] S2: Spatial registration of satellite and drone images of the same period and the same coverage area.

[0061] In this embodiment, satellite images and drone images of the same period and the same coverage area are spatially aligned. For example, the satellite image of a mining area in 2021 is cropped to obtain a satellite image of part of the mining area, and at the same time, several drone images of 2021 corresponding to the part of the area are obtained and spatially aligned so that the drone images and satellite images can be aggregated subsequently.

[0062] S3: Extracting vegetation information according to the drone image, and calculating vegetation index according to the satellite image.

[0063] In a specific embodiment, the S3 includes:

[0064] S31: Classifying the drone image into vegetation area and non-vegetation area;

[0065] S32: Binarization is performed according to the classification result of the drone image to obtain a binarization result.

[0066] In this embodiment, the drone image covering the same area as the satellite image can be classified into vegetation areas and non-vegetation areas through a supervised classification method, and the classification results can be binarized, that is, the vegetation area is marked as 1 and the non-vegetation area is marked as 0, so that the vegetation information can be subsequently aggregated with the satellite image through a 2-D Gaussian sampling method.

[0067] It should be noted that the classification method may be any classification method, and those skilled in the art may select one according to actual conditions, such as selecting an unsupervised classification method.

[0068] S4: Aggregating the vegetation information of the UAV image with the satellite image through a 2-D Gaussian sampling method to obtain the vegetation coverage of each satellite pixel, and constructing a vegetation coverage model with the vegetation index calculated from the corresponding satellite pixels.

[0069] In a specific embodiment, the S4 further comprises:

[0070] S41: Determine the window size of a satellite image after aggregation as α×α, where

[0071]

[0072] Among them, R satellite is the spatial resolution of each pixel of the satellite image in the same period, R UAV is the spatial resolution per pixel of drone images from the same period;

[0073] S42: Calculate the pixel weight of each drone image in the window of the aggregated satellite image using 2-D Gaussian sampling:

[0074]

[0075] In this embodiment, W (i,j) represents the weight value of the pixel in the i-th row and j-th column of the drone image. i and j are positive integers, which are the row and column numbers of each pixel on the image. The pixel in the upper left corner is defined as (1,1), and based on this, i gradually increases to the right and j gradually increases downward. Since the window size defined by the general Gaussian sampling is (2k+1)(2k+1), let the window size be equal to α, then σ represents the standard deviation. Since the Gaussian curve occupies 99.7% of the range in the interval (μ-3σ, μ+3σ), the values ​​outside 3σ are generally close to 0 and can be ignored, that is, the window size is 6σ×6σ. Usually, the nearest odd number is taken, then σ=α / 6. For example, when Landsat series satellite remote sensing images are selected as satellite images in order to obtain more accurate calculation results, that is, R satellite For 30m, k=375,σ=125 should be used.

[0076] S43: Calculate the vegetation coverage of each pixel of the aggregated satellite image according to the binarization result and the pixel weight of the drone image,

[0077]

[0078] Among them, ρ i,j is the binarization result of the pixels in the i-th row and j-th column of the drone image.

[0079] In this embodiment, the vegetation coverage of each pixel of the aggregated satellite image is calculated according to the binarization result and the pixel weight of the drone image, thereby solving the problem of inaccurate vegetation coverage calculation in the prior art and the spatial mismatch between the surface sample and the corresponding satellite pixel size, weakening the boundary weight of the satellite image, and further improving the accuracy of vegetation coverage.

[0080] In a specific embodiment, the S4 further includes:

[0081] S44: establishing a first linear regression equation by using the corresponding relationship between the vegetation coverage of the aggregated satellite image and the vegetation index of the corresponding area calculated according to the satellite image;

[0082] S45: The vegetation coverage of the remaining coverage areas may be obtained according to the first linear regression equation and the vegetation index of the remaining coverage areas calculated according to the satellite image.

[0083] In this embodiment, the surface reflectance of the red band and the near-infrared band of each pixel in the satellite image can be used to calculate the Normalized Difference Vegetation Index (NDVI) to establish a vegetation coverage model, that is, to establish a first linear regression equation, wherein NDVI is an important indicator of vegetation growth conditions and is often used to monitor vegetation growth status and calculate vegetation coverage, etc.

[0084]

[0085] Among them, NIR represents the near-infrared band of the satellite image, and R represents the red band of the satellite image.

[0086] For example, the vegetation coverage y of a part of a mining area in 2021 calculated based on the aggregated satellite image FVC and the corresponding relationship between the vegetation index NDVI of a certain area of ​​a mining area in 2021 calculated according to the satellite image, and establishing a first linear regression equation;

[0087] y FVC =a×NDVI+b

[0088] Finally, the vegetation index can be calculated based on the satellite images of other covered areas and substituted into the first linear regression equation to obtain the vegetation coverage of other satellite covered areas. This process improves the accuracy of vegetation coverage calculation, and not only takes advantage of the large range of satellite images, but also takes advantage of the high spatial resolution of drone images. At the same time, it provides a reference for the evaluation of regional ecological environment restoration effects, has the advantages of strong adaptability, high accuracy, and easy operation, can greatly reduce field manual observations, and has broad application prospects.

[0089] It should be noted that other vegetation indices can also be used to calculate the vegetation index of the target area based on the satellite image. Those skilled in the art can choose according to actual conditions, such as the ratio vegetation index calculated by the ratio of the near-infrared and red bands, the difference vegetation index calculated by the difference between the near-infrared and red bands, etc.

[0090] S5: Calculate the vegetation index according to the satellite images of the remaining period, and correct the vegetation index of the remaining period, and obtain the corrected vegetation coverage of the remaining period according to the corrected vegetation index and the vegetation coverage model.

[0091] In a specific embodiment, the S5 further comprises:

[0092] S51: Obtain satellite images of the same period as the drone image and satellite images of the same area at other periods, display them in true color, and select multiple types of ground coverage in the target area;

[0093] S52: Calculate the vegetation index of the ground object coverage based on the satellite image of the same period as the drone image and the satellite image of the same area at other times, and use the vegetation index VI of the ground object coverage calculated based on the satellite image of the same period as the drone image reference The difference between the vegetation index of land cover calculated from satellite images of the same area at other times is VI i The corresponding relationship between the two is used to establish the second linear regression equation and complete the correction of vegetation index in any period.

[0094] VI i =c+d·VI reference

[0095] Among them, VI i is the vegetation index of the i-th year, VI reference Vegetation index calculated for satellite imagery contemporaneous with drone imagery.

[0096] S53: According to the first linear regression equation, the corrected vegetation index of any other period is substituted into the first linear regression equation to obtain the vegetation coverage of the remaining periods.

[0097] In this embodiment, considering that satellite images of different periods come from different sensors and are obtained in different weather conditions, the calculated vegetation index is different. Therefore, the vegetation index of the ground object coverage calculated according to the satellite image of the same period of the drone image and the vegetation index of the ground object coverage calculated according to the satellite image of other periods are used to establish a second linear regression equation. For example, the vegetation index of the ground object coverage is calculated according to the satellite image of a mining area in 2021 and the satellite image of a mining area in 2020 in the same period of the drone image to establish a second linear regression equation, and the vegetation index of other periods after correction is calculated according to the second linear regression equation, and then substituted into the first linear equation to obtain the vegetation coverage of other periods. This process realizes the inversion of vegetation coverage with long time series, high precision and large range, which not only plays the advantages of long time series and large range of satellite images, but also plays the advantages of high spatial resolution of drone images. At the same time, the recovery of vegetation in the region in recent years can be more accurately understood, and a reference can be provided for the evaluation of environmental restoration effects in typical ecological areas, with the advantages of strong adaptability, high precision and easy operation.

[0098] In an optional embodiment, the method further includes: assigning an ID number to each aggregated satellite image pixel, and associating the vegetation coverage of each pixel of the aggregated satellite image with the vegetation index of each pixel calculated based on the satellite image through the ID number.

[0099] Obviously, the above embodiments of the present invention are only examples for clearly illustrating the present invention, and are not limitations on the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made on the basis of the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the protection scope of the present invention.

Claims

1. A multi-scale vegetation coverage remote sensing calculation method based on drone, characterized in that: include: S1: Acquire satellite images and drone images, wherein the coverage area of ​​the satellite images includes the coverage area of ​​the drone images; S2: spatial registration of satellite images and drone images of the same period and the same coverage area; S3: extracting vegetation information according to the drone image, and calculating vegetation index according to the satellite image; S4: Aggregating the vegetation information of the UAV image with the satellite image through a 2-D Gaussian sampling method to obtain the vegetation coverage of each satellite pixel, and constructing a vegetation coverage model with the vegetation index calculated from the corresponding satellite pixel; S5: Calculating vegetation index according to satellite images of the remaining period, and correcting the vegetation index of the remaining period, and obtaining the corrected vegetation coverage of the remaining period according to the corrected vegetation index and the vegetation coverage model; The S3 includes: S31: Classifying the drone image into vegetation area and non-vegetation area; S32: Binarization is performed according to the classification result of the drone image to obtain a binarization result; The S4 includes: S41: Determine the window size of a satellite image after aggregation as α×α, where Among them, R satellite is the spatial resolution of each pixel of satellite images in the same period, R UAV is the spatial resolution of each pixel of drone images during the same period; S42: Calculate the pixel weight of each drone image in the aggregated satellite image window using 2-D Gaussian sampling: Among them, W (i,j) represents the weight value of the drone image pixel in the i-th row and j-th column in a satellite pixel, σ represents the standard deviation, and σ=α / 6; S43: Calculate the vegetation coverage of each pixel of the aggregated satellite image according to the binarization result and the pixel weight of the drone image. Among them, ρ i,j is the binarization result of the pixels in the i-th row and j-th column of the drone image.

2. The method according to claim 1, characterized in that The S4 further comprises: S44: establishing a first linear regression equation by using the corresponding relationship between the vegetation coverage of the aggregated satellite image and the vegetation index of the corresponding area calculated according to the satellite image; S45: The vegetation coverage of the remaining coverage areas may be obtained according to the first linear regression equation and the vegetation index of the remaining coverage areas calculated according to the satellite image.

3. The method according to claim 1, characterized in that An ID number is assigned to each pixel of the aggregated satellite image, and the vegetation coverage of each pixel of the aggregated satellite image and the vegetation index calculated based on each pixel are associated with each other through the ID number.

4. The method according to claim 2, characterized in that: The S5 includes: S51: Obtain satellite images of the same period as the drone image and satellite images of the same area at other periods, display them in true color, and select multiple types of ground coverage in the target area; S52: Calculating the vegetation index of the ground object coverage according to the satellite image of the same period as the UAV image and the satellite image of the same area at other periods, and using the corresponding relationship between the vegetation index of the ground object coverage calculated according to the satellite image of the same period as the UAV image and the vegetation index of the ground object coverage calculated according to the satellite image of the same area at other periods, establishing a second linear regression equation to complete the correction of the vegetation index of any period; S53: According to the first linear regression equation, the vegetation index of any period after correction is substituted into the first linear regression equation to obtain the vegetation coverage of the remaining periods.

5. The method according to claim 1, characterized in that The S1 includes: The satellite image and the drone image are preprocessed.

Images