Pixel-level anisotropy correction method for multispectral image of unmanned aerial vehicle

Abstract

The invention discloses a pixel-level anisotropy correction method for a multispectral image of an unmanned aerial vehicle, and belongs to the technical field of radiation correction of the multispectral image of the unmanned aerial vehicle. The invention aims to solve the problem of poor image radiation precision caused by limitation of an existing anisotropy correction method in complex earth surface coverage and high-resolution image processing. Comprising the following steps: performing three-dimensional reconstruction based on unmanned aerial vehicle multispectral sub-images, geographic position data and attitude data to obtain a spliced image, a digital elevation model and sub-image optimization camera parameters; pixel area coordinates in the spliced image are converted into an image coordinate system from an earth-centered earth-fixed coordinate system through a camera coordinate system; extracting a projection area of the pixel area in the sub-image to obtain multi-view-angle reflectivity, and calculating an observation imaging view angle to obtain multi-view-angle data; a universal BRDF model is used for fitting multi-view data to obtain a fitting result, and then downward-view reflectivity is extracted to replace original reflectivity of a spliced image to achieve pixel-level anisotropy correction. The method is used for anisotropic correction of the multi-spectral image of the unmanned aerial vehicle.

Description

Technical Field

[0001] The invention relates to a pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images, and belongs to the technical field of radiation correction of unmanned aerial vehicle multispectral images. Background Art

[0002] Unmanned aerial vehicle (UAV) multispectral images collect reflectance spectral information of the surface of objects by carrying multispectral sensors to obtain spatial and spectral characteristics of the objects. Unlike traditional remote sensing images, UAV multispectral images have higher spatial resolution and flexible shooting angles, and can provide more detailed information about objects. These images contain the spectral characteristics of objects in various bands, reflecting the subtle differences in the spatial and spectral dimensions of objects. Due to the anisotropy of the reflectance characteristics of the surface of objects, changes in observation angles and lighting conditions will cause obvious radiation errors in the image, affecting the accuracy and consistency of the image. Therefore, anisotropy correction plays a vital role in improving the quantitative accuracy of remote sensing images.

[0003] Anisotropy correction of UAV multispectral images can effectively eliminate the radiation error caused by changes in observation angles, making the image data consistent and comparable under different conditions. Research on pixel-level anisotropy correction methods has important theoretical significance and application value. Existing anisotropy correction methods mainly focus on "regional level" or "coarse level" correction methods, which have certain limitations in processing complex surface coverage and high-resolution images. Summary of the invention

[0004] Aiming at the limitation of existing anisotropy correction methods in processing complex surface coverage and high-resolution images, resulting in poor image radiation accuracy, the present invention provides a pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images.

[0005] A method for pixel-level anisotropy correction of multispectral images of unmanned aerial vehicles of the present invention comprises:

[0006] 3D reconstruction is performed based on UAV multispectral sub-images, geographic location data, and attitude data to obtain stitched images, digital elevation models, and sub-image optimized camera parameters;

[0007] For each pixel area in the stitched image, the three-dimensional coordinates of the pixel area in the geocentric earth-fixed coordinate system are converted into the three-dimensional coordinates in the camera coordinate system based on the digital elevation model and the sub-image optimization camera parameters, and then the three-dimensional coordinates in the camera coordinate system are converted into the coordinates in the image coordinate system;

[0008] Extract the projection area of ​​the pixel area in the corresponding multispectral sub-image of the drone according to the coordinates of the pixel area in the image coordinate system, and obtain the multi-view reflectivity of the pixel area; at the same time, calculate the observation imaging viewing angle of the pixel area according to the three-dimensional coordinates of the pixel area in the geocentric earth-fixed coordinate system and the three-dimensional coordinates of the corresponding camera position in the geocentric earth-fixed coordinate system; the multi-view reflectivity and the observation imaging viewing angle constitute multi-view data;

[0009] The multi-view data is fitted using a universal BRDF model based on basis functions to obtain the multi-view data fitting results for each pixel area. The downward-looking reflectivity is then extracted from the multi-view data fitting results, and the downward-looking reflectivity is used to replace the original reflectivity of the stitched image to achieve pixel-level anisotropy correction of the UAV multispectral image.

[0010] According to the pixel-level anisotropy correction method for multispectral images of unmanned aerial vehicles of the present invention, the method for converting the three-dimensional coordinates of the pixel area in the Earth-centered Earth-fixed coordinate system into the three-dimensional coordinates in the camera coordinate system is:

[0011]

[0012] In the formula [x c ,y c ,z c ] is the three-dimensional coordinate of the center point of the pixel area in the camera coordinate system, [x e ,y e ,z e ] is the three-dimensional coordinate of the center point of the pixel area in the Earth-centered Earth-fixed coordinate system, is the camera extrinsic matrix from the Earth-centered Earth-fixed coordinate system to the camera coordinate system;

[0013]

[0014] In the formula is the rotation matrix from the Earth-centered Earth-fixed coordinate system to the camera coordinate system, is the translation vector from the Earth-centered Earth-fixed coordinate system to the camera coordinate system;

[0015]

[0016] In the formula [t x ,t y ,t z ] is the three-dimensional displacement from the Earth-centered Earth-fixed coordinate system to the camera coordinate system, [R x ,R y ,R z ] is the three-dimensional rotation matrix from the Earth-centered Earth-fixed coordinate system to the camera coordinate system, is the three-dimensional rotation angle from the Earth-centered Earth-fixed coordinate system to the camera coordinate system.

[0017] According to the pixel-level anisotropy correction method for multispectral images of unmanned aerial vehicles of the present invention, the method for converting the three-dimensional coordinates in the camera coordinate system into the coordinates in the image coordinate system is:

[0018]

[0019] In the formula [u i ,v i ] is the coordinate of the pixel area in the image coordinate system, α is the pixel resolution of the image coordinate system X axis, β is the pixel resolution of the image coordinate system Y axis, f is the focal length of the camera, [c x ,c y ] is the coordinate of the camera center point in the image coordinate system, The camera intrinsic parameter matrix from the camera coordinate system to the image coordinate system:

[0020]

[0021] According to the pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images of the present invention, based on the conversion relationship between the earth-centered earth-fixed coordinate system and the camera coordinate system, and the conversion relationship between the camera coordinate system and the image coordinate system, the conversion relationship between the earth-centered earth-fixed coordinate system and the image coordinate system is obtained:

[0022]

[0023] According to the pixel-level anisotropy correction method for multispectral images of unmanned aerial vehicles of the present invention, the method for obtaining the multi-viewing angle reflectivity of the pixel area includes:

[0024] Select a 3×3 pixel area and record it as S={P(p,q)p∈[x-1,x+1],q∈[y-1,y+1]}, where P(p,q) represents the pixel point of the pixel area in the stitched image, (p,q) represents the pixel point coordinates of the pixel area, and (x,y) represents the pixel point coordinates of the stitched image in the image coordinate system;

[0025] According to the expression of the pixel area S, the projection coordinates of the boundary points of the pixel area on the corresponding drone multispectral sub-image are calculated, so as to extract the projection area of ​​the pixel area on the corresponding drone multispectral sub-image and obtain the multi-view reflectivity of the pixel area:

[0026]

[0027] Where R represents the multi-view reflectivity of the pixel area in each UAV multispectral sub-image, represents the projection area of ​​the pixel area S in the w-th UAV multispectral sub-image, Indicates the projection area The average reflectivity of , mean represents the arithmetic mean.

[0028] According to the pixel-level anisotropy correction method for multispectral images of unmanned aerial vehicles of the present invention, the observation imaging viewing angle of the pixel area includes the solar zenith angle θ s 、Sun azimuth φ s , observation zenith angle θ v and the observation azimuth φ v ; Solar zenith angle θ s and the solar azimuth φ s are the horizontal angles of the sun and the camera relative to the ground, measured clockwise from true north;

[0029] Observation zenith angle θ v and the observation azimuth φ v are the vertical angles between the sun and the camera and the observation point:

[0030]

[0031] In the formula Represents the three-dimensional coordinates of the camera position in the Earth-centered Earth-fixed coordinate system.

[0032] According to the pixel-level anisotropy correction method for multispectral images of unmanned aerial vehicles of the present invention, an infinite series basis function is used to approximate the theoretical BRDF model, and a hemispherical harmonic basis function is used to represent the BRDF model:

[0033]

[0034] In the formula is a hemispherical harmonic basis function with order l and component m, are the coefficients corresponding to the hemispherical harmonic basis functions with order l and components m;

[0035]

[0036] In the formula is the normalization coefficient, is cosθ v The associated Legendre polynomials of .

[0037] According to the pixel-level anisotropy correction method for UAV multispectral images of the present invention, given the order n, the BRDF model is obtained as follows:

[0038]

[0039] set up:

[0040]

[0041] where k∈[1,2n+1];

[0042] M represents the total number of multi-view data, ignoring the spectral correlation, and establishing a linear equation system for each band:

[0043]

[0044] Where R M is the multi-view reflectivity of the Mth viewing angle; and Indicates the Mth value corresponding to the variable;

[0045] Solve the linear equations to obtain the multi-view data fitting results for each pixel area.

[0046] According to the pixel-level anisotropy correction method for multispectral images of unmanned aerial vehicles of the present invention, the method for solving the linear equations is:

[0047] Represent the linear system in matrix form:

[0048] HA=R,where:

[0049]

[0050] and

[0051] Convert the problem of solving a system of linear equations into a convex optimization problem; construct the objective function:

[0052]

[0053] The least squares method is used to obtain the multi-view data fitting results

[0054]

[0055] In the formula is the optimal coefficient.

[0056] According to the pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images of the present invention, the method for extracting downward reflectivity is as follows:

[0057] Let θ v =0,φ v = 0, extract the downward reflectivity R nadir :

[0058]

[0059] Beneficial effects of the present invention: The pixel-level anisotropy correction of the method of the present invention can more accurately process the local features of the ground object by extracting multi-view reflection data at the pixel level, overcoming the shortcomings of traditional methods in capturing details. In the application of high-resolution remote sensing images, the method of the present invention can effectively alleviate the radiation error introduced by the change of imaging perspective during the acquisition process of drone multispectral images, thereby more effectively improving the radiation accuracy of the image. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic diagram of the process of the pixel-level anisotropy correction method for multispectral images of unmanned aerial vehicles according to the present invention;

[0061] Figure 2 is a schematic diagram of the stitched image without anisotropy correction;

[0062] Figure 3 It is a schematic diagram of the spliced ​​image after correction by the method of the present invention. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0065] The present invention will be further described below in conjunction with the accompanying drawings, but is not intended to be a limitation of the present invention.

[0066] Specific implementation method 1. Combination Figure 1 As shown, the present invention provides a method for pixel-level anisotropy correction of multispectral images of unmanned aerial vehicles, comprising:

[0067] 3D reconstruction is performed based on UAV multispectral sub-images, geographic location data, and attitude data to obtain stitched images, digital elevation models, and sub-image optimized camera parameters;

[0068] For each pixel area in the stitched image, the three-dimensional coordinates of the pixel area in the geocentric earth-fixed coordinate system are converted into the three-dimensional coordinates in the camera coordinate system based on the digital elevation model and the sub-image optimization camera parameters, and then the three-dimensional coordinates in the camera coordinate system are converted into the coordinates in the image coordinate system;

[0069] Extract the projection area of ​​the pixel area in the corresponding multispectral sub-image of the drone according to the coordinates of the pixel area in the image coordinate system, and obtain the multi-view reflectivity of the pixel area; at the same time, calculate the observation imaging viewing angle of the pixel area according to the three-dimensional coordinates of the pixel area in the geocentric earth-fixed coordinate system and the three-dimensional coordinates of the corresponding camera position in the geocentric earth-fixed coordinate system; the multi-view reflectivity and the observation imaging viewing angle constitute multi-view data;

[0070] The multi-view data is fitted using a universal BRDF model based on basis functions to obtain the multi-view data fitting results for each pixel area. The downward-looking reflectivity is then extracted from the multi-view data fitting results, and the downward-looking reflectivity is used to replace the original reflectivity of the stitched image to achieve pixel-level anisotropy correction of the UAV multispectral image.

[0071] This implementation first processes the input data, then extracts pixel-level multi-view data for each pixel area based on the spliced ​​image, digital elevation model and optimized camera parameters; then performs BRDF model fitting and correction, and uses a general bidirectional reflectance distribution function (BRDF) model based on basis functions to fit the extracted multi-view data, and predicts the downward reflectivity to achieve pixel-level anisotropy correction. This implementation can effectively alleviate the radiation error introduced by the change of imaging perspective during the acquisition process of drone multispectral images.

[0072] Furthermore, the method of converting the three-dimensional coordinates of the pixel area in the Earth-Centered Earth-Fixed (ECEF) coordinate system into the three-dimensional coordinates in the camera coordinate system is as follows:

[0073]

[0074] In the formula [x c ,y c ,z c ] is the three-dimensional coordinate of the center point of the pixel area in the camera coordinate system, [x e ,y e ,z e ] is the three-dimensional coordinate of the center point of the pixel area in the Earth-centered Earth-fixed coordinate system, is the camera extrinsic matrix from the Earth-centered Earth-fixed coordinate system to the camera coordinate system;

[0075]

[0076] In the formula is the rotation matrix from the Earth-centered Earth-fixed coordinate system to the camera coordinate system, is the translation vector from the Earth-centered Earth-fixed coordinate system to the camera coordinate system;

[0077]

[0078] In the formula [t x ,t y ,t z ] is the three-dimensional displacement from the Earth-centered Earth-fixed coordinate system to the camera coordinate system, [R x ,R y ,R z ] is the three-dimensional rotation matrix from the Earth-centered Earth-fixed coordinate system to the camera coordinate system, is the three-dimensional rotation angle from the Earth-centered Earth-fixed coordinate system to the camera coordinate system. x ,t y ,t z ]and are the optimized camera parameters.

[0079] The method of converting the three-dimensional coordinates in the camera coordinate system into the coordinates in the image coordinate system is:

[0080]

[0081] In the formula [u i ,v i ] is the coordinate of the pixel area in the image coordinate system, α is the pixel resolution of the image coordinate system X axis, β is the pixel resolution of the image coordinate system Y axis, f is the focal length of the camera, [c x ,c y ] is the coordinate of the camera center point in the image coordinate system, The camera intrinsic parameter matrix from the camera coordinate system to the image coordinate system:

[0082]

[0083] In summary, according to the conversion relationship between the Earth-centered Earth-fixed coordinate system and the camera coordinate system, and the conversion relationship between the camera coordinate system and the image coordinate system, the conversion relationship between the Earth-centered Earth-fixed coordinate system and the image coordinate system is obtained:

[0084]

[0085] According to the above formula, the projection area of ​​a given pixel area in the corresponding sub-image can be extracted to realize the extraction of pixel-level reflectivity data.

[0086] Furthermore, the method for obtaining the multi-viewing angle reflectivity of the pixel area includes:

[0087] Select a 3×3 pixel area and record it as S={P(p,q)p∈[x-1,x+1],q∈[y-1,y+1]}, where P(p,q) represents the pixel point of the pixel area in the stitched image, (p,q) represents the pixel point coordinates of the pixel area, and (x,y) represents the pixel point coordinates of the stitched image in the image coordinate system;

[0088] According to the expression of the pixel area S, the projection coordinates of the boundary points of the pixel area on the corresponding drone multispectral sub-image are calculated, so as to extract the projection area of ​​the pixel area on the corresponding drone multispectral sub-image and obtain the multi-view reflectivity of the pixel area:

[0089]

[0090] Where R represents the multi-view reflectivity of the pixel area in each UAV multispectral sub-image, represents the projection area of ​​the pixel area S in the w-th UAV multispectral sub-image, Indicates the projection area The average reflectivity of , mean represents the arithmetic mean.

[0091] In this embodiment, the imaging angle is a set of angle parameters that define the spatial relationship between the sun, the camera, and the ground target; the observation imaging angle of the pixel area includes the solar zenith angle θ s 、Sun azimuth φ s , observation zenith angle θ v and the observation azimuth φ v ; Solar zenith angle θ s and the solar azimuth φ s are the horizontal angles of the sun and the camera relative to the ground, measured clockwise from true north;

[0092] Observation zenith angle θ v and the observation azimuth φ v are the vertical angles between the sun and the camera and the observation point:

[0093]

[0094] In the formula Represents the three-dimensional coordinates of the camera position in the Earth-centered Earth-fixed coordinate system.

[0095] The infinite series basis function is used to approximate the theoretical BRDF model. In order to ensure the alignment of the definition domain, the hemispherical harmonic basis function is used to represent the BRDF model:

[0096]

[0097] In the formula is a hemispherical harmonic basis function with order l and component m, are the coefficients corresponding to the hemispherical harmonic basis functions with order l and components m;

[0098]

[0099]

[0100] In the formula is the normalization coefficient, is cosθ v The associated Legendre polynomials of .

[0101] The parameter fitting and downward reflectivity extraction process of the BRDF model are as follows:

[0102] Ideally, the BRDF is represented with an infinite order, but in practice, it is often approximated with a given order.

[0103] Given the order n, the BRDF model is:

[0104]

[0105] set up:

[0106]

[0107] where k∈[1,2n+1];

[0108] M represents the total number of multi-view data, n represents the highest order of HSH approximation, ignoring the spectral correlation, and establishing a linear equation system for each band:

[0109]

[0110] Where R M is the multi-view reflectivity of the Mth viewing angle; and Indicates the Mth value corresponding to the variable;

[0111] Solve the linear equations to obtain the multi-view data fitting results for each pixel area.

[0112] Going further, the method for solving the linear equations is:

[0113] Represent the linear system in matrix form:

[0114] HA=R,where:

[0115]

[0116] and

[0117] Considering that the system of equations is overdetermined and does not have a unique solution, the problem of solving the linear system of equations is transformed into a convex optimization problem; by constructing the objective function and minimizing it, the optimal solution is obtained:

[0118]

[0119] The least squares method is used to obtain the multi-view data fitting results

[0120]

[0121] In the formula is the optimal coefficient.

[0122] The method for extracting downward reflectivity is:

[0123] Let θ v =0,φ v = 0, extract the downward reflectivity R nadir :

[0124]

[0125] Anisotropy correction can be achieved by replacing the original reflectivity with the downward-looking reflectivity.

[0126] Verification experiment:

[0127] The multispectral image data collected by the drone is obtained. The data contains 10 bands with a wavelength range of 400-840nm and a ground resolution of 3.5cm. The data has been pre-processed with radiation correction and converted into reflectance data. Figure 2 is the stitched image without anisotropy correction, Figure 3 Table 1 is a comparison of the reflectivity errors of ground samples before and after correction.

[0128] Table 1

[0129]

[0130] It can be seen that the image corrected by the method of the present invention can effectively alleviate the radiation error introduced by the change of imaging viewing angle, which proves the effectiveness of the method of the present invention.

[0131] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that the features described in conjunction with a single embodiment may be used in other described embodiments.

Claims

1. A pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images, characterized in that: include, 3D reconstruction is performed based on UAV multispectral sub-images, geographic location data and attitude data to obtain stitched images, digital elevation models and sub-image optimized camera parameters; For each pixel area in the stitched image, the three-dimensional coordinates of the pixel area in the geocentric earth-fixed coordinate system are converted into the three-dimensional coordinates in the camera coordinate system based on the digital elevation model and the sub-image optimization camera parameters, and then the three-dimensional coordinates in the camera coordinate system are converted into the coordinates in the image coordinate system; Extracting the projection area of ​​the pixel area in the corresponding multispectral sub-image of the drone according to the coordinates of the pixel area in the image coordinate system, and obtaining the multi-viewing angle reflectivity of the pixel area; At the same time, the observation imaging angle of the pixel area is calculated according to the three-dimensional coordinates of the pixel area in the geocentric earth-fixed coordinate system and the three-dimensional coordinates of the corresponding camera position in the geocentric earth-fixed coordinate system; the multi-angle reflectivity and the observation imaging angle constitute multi-angle data; The multi-view data is fitted using a universal BRDF model based on basis functions to obtain the multi-view data fitting results for each pixel area. The downward-looking reflectivity is then extracted from the multi-view data fitting results, and the downward-looking reflectivity is used to replace the original reflectivity of the stitched image to achieve pixel-level anisotropy correction of the UAV multispectral image.

2. The pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images according to claim 1 is characterized in that: The method for converting the three-dimensional coordinates of the pixel area in the geocentric earth-fixed coordinate system into the three-dimensional coordinates in the camera coordinate system is: In the formula [x c ,y c ,z c ] is the three-dimensional coordinate of the center point of the pixel area in the camera coordinate system, [x e ,y e ,z e ] is the three-dimensional coordinate of the center point of the pixel area in the Earth-centered Earth-fixed coordinate system, is the camera extrinsic matrix from the Earth-centered Earth-fixed coordinate system to the camera coordinate system; In the formula is the rotation matrix from the Earth-centered Earth-fixed coordinate system to the camera coordinate system, is the translation vector from the Earth-centered Earth-fixed coordinate system to the camera coordinate system; In the formula [t x ,t y ,t z ] is the three-dimensional displacement from the Earth-centered Earth-fixed coordinate system to the camera coordinate system, [R x ,R y ,R z ] is the three-dimensional rotation matrix from the Earth-centered Earth-fixed coordinate system to the camera coordinate system, is the three-dimensional rotation angle from the Earth-centered Earth-fixed coordinate system to the camera coordinate system.

3. The pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images according to claim 2 is characterized in that: The method of converting the three-dimensional coordinates in the camera coordinate system into the coordinates in the image coordinate system is: In the formula [u i ,v i ] is the coordinate of the pixel area in the image coordinate system, α is the pixel resolution of the image coordinate system X axis, β is the pixel resolution of the image coordinate system Y axis, f is the focal length of the camera, [c x ,c y ] is the coordinate of the camera center point in the image coordinate system, The camera intrinsic parameter matrix from the camera coordinate system to the image coordinate system:

4. The pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images according to claim 3 is characterized in that: According to the conversion relationship between the Earth-centered Earth-fixed coordinate system and the camera coordinate system, and the conversion relationship between the camera coordinate system and the image coordinate system, the conversion relationship between the Earth-centered Earth-fixed coordinate system and the image coordinate system is obtained:

5. The pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images according to claim 4 is characterized in that: The method for obtaining the multi-viewing angle reflectivity of the pixel area includes: Select a 3×3 pixel area and record it as S={P(p,q)|p∈[x-1,x+1],q∈[y-1,y+1]}, where P(p,q) represents the pixel point of the pixel area in the stitched image, (p,q) represents the pixel point coordinates of the pixel area, and (x,y) represents the pixel point coordinates of the stitched image in the image coordinate system; According to the expression of the pixel area S, the projection coordinates of the boundary points of the pixel area on the corresponding drone multispectral sub-image are calculated, so as to extract the projection area of ​​the pixel area on the corresponding drone multispectral sub-image and obtain the multi-view reflectivity of the pixel area: Where R represents the multi-view reflectivity of the pixel area in each UAV multispectral sub-image, represents the projection area of ​​the pixel area S in the w-th UAV multispectral sub-image, Indicates the projection area The average reflectivity of , mean represents the arithmetic mean.

6. The pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images according to claim 5 is characterized in that: The observation imaging angle of the pixel area includes the solar zenith angle θ s 、Sun azimuth φ s , observation zenith angle θ v and the observation azimuth φ v ; Solar zenith angle θ s and the solar azimuth φ s are the horizontal angles of the sun and the camera relative to the ground, measured clockwise from true north; Observation zenith angle θ v and the observation azimuth φ v are the vertical angles between the sun and the camera and the observation point: In the formula Represents the three-dimensional coordinates of the camera position in the Earth-centered Earth-fixed coordinate system.

7. The pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images according to claim 6 is characterized in that: The infinite series basis function is used to approximate the theoretical BRDF model, and the hemispherical harmonic basis function is used to represent the BRDF model: In the formula is a hemispherical harmonic basis function with order l and component m, are the coefficients corresponding to the hemispherical harmonic basis functions with order l and component m; In the formula is the normalization coefficient, is cosθ v The associated Legendre polynomials of .

8. The pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images according to claim 7 is characterized in that: Given the order n, the BRDF model is: set up: where k∈[1,2n+1]; M represents the total number of multi-view data, ignoring the spectral correlation, and establishing a linear equation system for each band: Where R M is the multi-view reflectivity of the Mth viewing angle; θ v M and φ v M Indicates the Mth value corresponding to the variable; Solve the linear equations to obtain the multi-view data fitting results for each pixel area.

9. The pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images according to claim 8 is characterized in that: The method to solve the linear system of equations is: Represent the linear system in matrix form: HA=R,where: and Convert the problem of solving a system of linear equations into a convex optimization problem; construct the objective function: The least squares method is used to obtain the multi-view data fitting results In the formula is the optimal coefficient.

10. The pixel-level anisotropy correction method for unmanned aerial vehicle multispectral images according to claim 9 is characterized in that: The method for extracting downward reflectivity is: Let θ v =0,φ v = 0, extract the downward reflectivity R nadir :

Images