A method for obtaining key attributes of canopy trees based on UAV oblique photography

Abstract

A method for acquiring key attributes of canopy trees based on UAV oblique photography, relating to the field of image data processing, includes the following steps: S1, UAV oblique photography → S2, stitching DSM data → S3, acquiring DEM data → S4, data processing → S5, extracting key tree factors. This invention can cover large areas in a short time, rapidly collecting image data of trees for indoor data analysis to obtain key tree attribute elements. The cost is only a fraction of that of traditional manual surveys, with virtually no personnel safety risks. Furthermore, this invention achieves high accuracy by acquiring location signals from the air, using algorithms to identify the treetop position and obtain tree geographic coordinates and height; by acquiring image data from different angles such as oblique angles, and using algorithms to identify tree height and canopy width, it significantly reduces manual intervention and improves the efficiency and accuracy of the survey.

Description

Technical Field

[0001] This invention relates to the field of image data processing, specifically to a method for acquiring key attributes of canopy trees based on oblique photography by unmanned aerial vehicles. Background Technology

[0002] Tree geographic coordinates, height, crown width, and other forest attributes are essential elements for forestry production. Traditional methods for obtaining these attributes rely solely on manual, single-tree surveys. However, these methods have several drawbacks: 1. Traditional manual collection of tree geographic coordinates, height, and crown width requires venturing deep into the forest to collect data from each tree individually, resulting in high time and labor costs and significant safety risks. 2. Due to forest vegetation obstruction and poor GPS signal reception, manual single-tree geographic coordinate positioning is slow and inaccurate, and can only be performed around the tree's perimeter, leading to coordinate offsets and hindering the precision management required for forestry. 3. Due to forest canopy density, manual surveys often struggle to accurately and quickly identify the highest points of tree growth and the edges of the crown, resulting in slow and inaccurate collection of height and crown width data. Data collection efficiency and accuracy are significantly dependent on the experience of the survey personnel, requiring substantial human intervention and arduous work. Therefore, to significantly improve the accuracy and efficiency of forest attribute surveys and complete forest information collection while ensuring personnel safety, this applicant, after in-depth research, has developed this project. Summary of the Invention

[0003] This invention provides a method for obtaining key attributes of canopy trees based on UAV oblique photography. Its main purpose is to overcome the technical problems caused by the existing method of manually collecting tree attributes such as geographic coordinates, tree height, and canopy width.

[0004] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0005] A method for obtaining key attributes of canopy trees based on UAV oblique photography includes the following steps:

[0006] S1. UAV oblique photography: Plan the flight path for the measurement area, set the gimbal pitch angle to be within the range of 20-45°, set the flight path according to forward / backward or left / right view, and carry out UAV oblique photography of the measurement area.

[0007] S2. Stitching DSM data: The images captured by the drone are processed into DSM data using oblique photogrammetry processing software;

[0008] S3. Obtain DEM data: Interpolate the DSM data generated by UAV oblique photography to generate DEM data from areas with obvious no vegetation (such as forest gaps, forest roads, etc.);

[0009] S4. Data Processing: Adjust the resolution of DSM and DEM data to 1 / 10-3 / 10 of the average tree crown width to reduce the error rate of forest identification and improve computational efficiency. Subtract the DSM and DEM data to obtain the net vegetation height data.

[0010] S5. Extract key factors of trees: Generate contour lines according to the net height data of vegetation. The highest point of the contour line area is regarded as the tree top position, i.e. the tree's geographical coordinates, and the value is the tree height. The lowest point of the contour line area is regarded as the edge of the tree crown. The lowest point around the tree top is the tree crown width.

[0011] Furthermore, in step S2, the process of splicing DSM data is as follows:

[0012] S21. Check the data captured by the drone to ensure the accuracy of the captured geographical location information;

[0013] S22. Import the checked shooting data into a general 3D modeling software;

[0014] S23. Point cloud registration: Align the point cloud data in photos taken from different angles to form a complete point cloud model.

[0015] S24. Generate DSM data using point cloud data through the TIN triangulation model.

[0016] Furthermore, in step S3, the process of acquiring DEM data is as follows:

[0017] S31. Determine areas without vegetation: Subtract the vegetation height from the DSM data generated by UAV oblique photography to identify areas without vegetation (such as forest gaps, forest roads, etc.) or areas with clear vegetation height. Define a threshold h, that is, areas with height h are considered to be areas without vegetation.

[0018] A binary mask M represents the area without vegetation:

[0019] ;

[0020] Where (x,y) represents the coordinate position in the DSM data;

[0021] S32. Treat the height of the unvegetated area as the ground height: For the area with mask M(x,y) = 1, directly treat the height value of the DSM as the height value of the DEM;

[0022] Generate initial DEM data: DEMinit(x, y):

[0023] DEMinit(x,y)=M(x,y)·DSM(x,y);

[0024] S33. Interpolation to generate complete DEM data: For the region M(x, y) = 0, the DEM height value is estimated using an interpolation algorithm:

[0025] ;

[0026] in, This means interpolating and estimating the values ​​of the nearest DEM values. Specifically, it involves finding the nearest points (x', y') with M(x', y') = 1 and interpolating based on the height values ​​of these points.

[0027] Furthermore, in step S4, the data processing procedure is as follows:

[0028] S41. Define input and output grids:

[0029] Input raster data: Raster input (i,j), where (i,j) represents the pixel position of the input raster data.

[0030] Output raster data: Raster output (m,n), where (m,n) represents the pixel position of the output raster data.

[0031] Input raster resolution: R input ,

[0032] Output raster resolution: R output = M•R input ,

[0033] M represents the target resolution to be adjusted;

[0034] S42. Interpolate to target resolution:

[0035] The output raster position (m,n) corresponds to the input raster position (i,j):

[0036] ;

[0037] For the data at position (m,n), perform nearest neighbor interpolation based on the position value at (i,j):

[0038] ;

[0039] S43. Obtain vegetation net height data DSM0: Subtract the DSM and DEM raster data adjusted to the same resolution to obtain vegetation net height data DSM0.

[0040] DSM0 = DSM-DEM.

[0041] Furthermore, in step S5, the data processing procedure is as follows: the DSM0 data generated in S43 is used as a two-dimensional matrix DSM(i,j), where (i,j) represents the coordinates in the grid, h(i,j) represents the elevation value of the point, the neighborhood is defined as a 3×3 window, and the tree top position (tree geographic coordinates), tree height elevation value (tree height), crown edge position and crown area (crown width) are marked.

[0042] S51. Locate the tree height:

[0043] 1) Initialize the set of highest point positions:

[0044] Peaks = Ø;

[0045] 2) Traverse each point in the DSM data:

[0046] For each (i,j) in DSM;

[0047] 3) Obtain the neighborhood of point (i,j):

[0048]

[0049] 4) Determine if it is the treetop location:

[0050] ;

[0051] S52, Locate the edge of the tree canopy:

[0052] 1) Initialize the set of lowest point positions:

[0053] Vallerys=Ø;

[0054] 2) Traverse each point in the DSM data:

[0055] For each (i,j) in DSM;

[0056] 3) Obtain the neighborhood of point (i,j):

[0057] ;

[0058] 4) Determine if it is the treetop location:

[0059] ;

[0060] S53, Mark tree height, crown edge, and crown width:

[0061] 1) Initialize the tag set:

[0062] Markedpeaks = Ø;

[0063] 2) Iterate through each point in the highest position:

[0064] For each (i,j) in Peaks;

[0065] 3) Find the nearest lowest point in the X direction:

[0066] For each (i,j) in Peaks;

[0067] 3) Find the nearest lowest point in the X direction:

[0068] ;

[0069] ;

[0070] 4) Find the nearest minimum point in the Y direction:

[0071] ;

[0072] ;

[0073] 5) Mark the location of the highest point and the edge of its canopy:

[0074] ;

[0075] 6) Calculate the canopy area:

[0076] Long axis:

[0077] ;

[0078] Short axis:

[0079] ;

[0080] Furthermore, in step S4, the resolution of the DSM and DEM data is adjusted to 1 / 5 of the average tree crown width.

[0081] As can be seen from the above description of the present invention, compared with the prior art, the present invention has the following advantages:

[0082] 1. This invention can cover a large area in a short time, quickly collect image data of trees, conduct indoor data analysis, and obtain key attribute elements of trees. The cost is only a fraction of that of traditional manual surveys, and there is almost no risk to personnel safety.

[0083] 2. This invention obtains location signals from the air with high accuracy, identifies the treetop position using algorithms, and obtains the tree's geographical coordinates and height with high data precision. By collecting image data from different angles such as tilt, and identifying tree height and crown width using algorithms, it greatly reduces manual intervention and improves the efficiency and accuracy of the survey. Detailed Implementation

[0084] A method for obtaining key attributes of canopy trees based on UAV oblique photography includes the following steps:

[0085] S1. UAV oblique photography: Plan the flight path for the measurement area, set the gimbal pitch angle to be within the range of 20-45°, set the flight path according to forward / backward or left / right view, and carry out UAV oblique photography of the measurement area.

[0086] S2. Stitching DSM (Digital Surface Model) data: The images captured by the drone are processed into DSM data using oblique photogrammetry processing software;

[0087] S3. Obtain DEM (Digital Elevation Model) data: Interpolate the DSM data generated by UAV oblique photography to generate DEM data from areas with obvious lack of vegetation (such as forest gaps, forest roads, etc.);

[0088] S4. Data Processing: Adjust the resolution of DSM and DEM data to 1 / 10-3 / 10 of the average tree crown width to reduce the error rate of forest identification and improve computational efficiency. Subtract the DSM and DEM data to obtain the net vegetation height data.

[0089] S5. Extract key factors of trees: Generate contour lines according to the net height data of vegetation. The highest point of the contour line area is regarded as the tree top position, i.e. the tree's geographical coordinates, and the value is the tree height. The lowest point of the contour line area is regarded as the edge of the tree crown. The lowest point around the tree top is the tree crown width.

[0090] In step S1, when the UAV plans the flight path for oblique photography, it only sets the flight path according to forward / backward or left / right view, and determines the gimbal pitch angle to be within the range of 20~45°. This allows for the acquisition of image data from different angles, and the workload of data acquisition and post-processing is only 40% of that of traditional oblique photography. While meeting the data acquisition accuracy requirements, the work efficiency is greatly improved.

[0091] Specifically, in step S2, the process of splicing DSM data is as follows:

[0092] S21. Check the data captured by the drone to ensure the accuracy of the captured geographical location information;

[0093] S22. Import the checked shooting data into common 3D modeling software (such as Pix4D, ContextCapture, Agisoft Metashape, DJI Terra, etc.).

[0094] S23. Point cloud registration: Align the point cloud data in photos taken from different angles to form a complete point cloud model.

[0095] S24. Generate DSM data using point cloud data through the TIN triangulation model.

[0096] Specifically, in step S3, the process of obtaining DEM data is as follows:

[0097] S31. Determine areas without vegetation: Subtract the vegetation height from the DSM data generated by UAV oblique photography to identify areas without vegetation (such as forest gaps, forest roads, etc.) or areas with clear vegetation height. Define a threshold h, that is, areas with height h are considered to be areas without vegetation.

[0098] A binary mask M represents the area without vegetation:

[0099] ;

[0100] Where (x,y) represents the coordinate position in the DSM data;

[0101] S32. Treat the height of the unvegetated area as the ground height: For the area with mask M(x,y) = 1, directly treat the height value of the DSM as the height value of the DEM;

[0102] Generate initial DEM data: DEMinit(x, y):

[0103] DEMinit(x,y)=M(x,y)·DSM(x,y);

[0104] S33. Interpolation to generate complete DEM data: For the region M(x, y) = 0, the DEM height value is estimated using an interpolation algorithm:

[0105] ;

[0106] in, This means interpolating and estimating the values ​​of the nearest DEM values. Specifically, it involves finding the nearest points (x', y') with M(x', y') = 1 and interpolating based on the height values ​​of these points.

[0107] Specifically, in step S4, the data processing procedure is as follows:

[0108] S41. Define input and output grids:

[0109] Input raster data: Raster input (i,j), where (i,j) represents the pixel position of the input raster data.

[0110] Output raster data: Raster output (m,n), where (m,n) represents the pixel position of the output raster data.

[0111] Input raster resolution: R input ,

[0112] Output raster resolution: R output = M•R input ,

[0113] M represents the target resolution to be adjusted;

[0114] S42. Interpolate to target resolution:

[0115] The output raster position (m,n) corresponds to the input raster position (i,j):

[0116] ;

[0117] For the data at position (m,n), perform nearest neighbor interpolation based on the position value at (i,j):

[0118] ;

[0119] S43. Obtain vegetation net height data DSM0: Subtract the DSM and DEM raster data adjusted to the same resolution to obtain vegetation net height data DSM0.

[0120] DSM0 = DSM-DEM.

[0121] During data processing, DEM and DSM data were preprocessed, and the raster resolution was adjusted to 1 / 5 of the average crown width of trees in the survey area. Adjusting the DEM and DSM data to the same resolution can significantly reduce the workload of data computation and improve data processing efficiency. Adjusting the resolution to 1 / 5 of the average crown width can significantly reduce the misidentification rate of evergreen broad-leaved forests.

[0122] Specifically, in step S5, the data processing process is as follows: the DSM0 data generated in S43 is used as a two-dimensional matrix DSM(i,j), where (i,j) represents the coordinates in the grid, h(i,j) represents the elevation value of the point, the neighborhood is defined as a 3×3 window, and the tree top position (tree geographic coordinates), tree height elevation value (tree height), crown edge position and crown area (crown width) are marked.

[0123] S51. Locate the tree height:

[0124] 1) Initialize the set of highest point positions:

[0125] Peaks = Ø;

[0126] 2) Traverse each point in the DSM data:

[0127] For each (i,j) in DSM;

[0128] 3) Obtain the neighborhood of point (i,j):

[0129] ;

[0130] 4) Determine if it is the treetop location:

[0131] ;

[0132] S52, Locate the edge of the tree canopy:

[0133] 1) Initialize the set of lowest point positions:

[0134] Vallerys=Ø;

[0135] 2) Traverse each point in the DSM data:

[0136] For each (i,j) in DSM;

[0137] 3) Obtain the neighborhood of point (i,j):

[0138] ;

[0139] 4) Determine if it is the treetop location:

[0140] ;

[0141] S53, Mark tree height, crown edge, and crown width:

[0142] 1) Initialize the tag set:

[0143] Markedpeaks = Ø;

[0144] 2) Iterate through each point in the highest position:

[0145] For each (i,j) in Peaks;

[0146] 3) Find the nearest lowest point in the X direction:

[0147] ;

[0148] ;

[0149] 4) Find the nearest minimum point in the Y direction:

[0150] ;

[0151] ;

[0152] 5) Mark the location of the highest point and the edge of its canopy:

[0153] ;

[0154] 6) Calculate the canopy area:

[0155] Long axis:

[0156] ;

[0157] Short axis:

[0158] ;

[0159] The algorithm described above can obtain axial attributes of trees, such as tree height, by generating a 3D model through oblique photography. However, due to the complexity of the forest canopy, it is difficult to perform individual tree segmentation and obtain the radial attributes of trees. The algorithm described above can obtain the key radial attribute of trees—canopy width.

[0160] This invention can cover large areas in a short time, rapidly collect forest image data, conduct indoor data analysis, and obtain key tree attribute elements at a cost only a fraction of that of traditional manual surveys, with virtually no personnel safety risks. Furthermore, this invention achieves high accuracy by acquiring location signals from the air, using algorithms to identify treetop positions and obtain tree geographic coordinates and height; by collecting image data from different angles such as tilt, and using algorithms to identify tree height and crown width, it significantly reduces manual intervention and improves the efficiency and accuracy of surveys.

[0161] The above are merely specific embodiments of the present invention, but the design concept of the present invention is not limited thereto. Any non-substantial modifications made to the present invention using this concept shall be considered as infringing upon the protection scope of the present invention.

Claims

1. A method for acquiring key attributes of canopy trees based on UAV oblique photography, characterized in that: The following steps are involved: S1. UAV oblique photography: Plan the flight path for the measurement area, set the gimbal pitch angle to be within the range of 20-45°, set the flight path according to forward / backward or left / right view, and carry out UAV oblique photography of the measurement area. S2. Stitching DSM data: The images captured by the drone are processed into DSM data using oblique photogrammetry processing software; S3. Obtain DEM data: Interpolate the DSM data generated by UAV oblique photography to generate DEM data from areas with obvious no vegetation. S4. Data Processing: Adjust the resolution of DSM and DEM data to 1 / 10-3 / 10 of the average tree crown width to reduce the error rate of forest identification and improve computational efficiency. Subtract the DSM and DEM data to obtain the net vegetation height data. S5. Extract key factors of trees: Generate contour lines according to the net height data of vegetation. The highest point of the contour line area is regarded as the tree top position, i.e. the tree's geographical coordinates, and the value is the tree height. The lowest point of the contour line area is regarded as the edge of the tree crown. The lowest point around the tree top is the tree crown width. In step S2, the process of splicing DSM data is as follows: S21. Check the data captured by the drone to ensure the accuracy of the captured geographical location information; S22. Import the checked shooting data into a general 3D modeling software; S23. Point cloud registration: Align the point cloud data in photos taken from different angles to form a complete point cloud model. S24. Generate DSM data from point cloud data using the TIN triangulation model; In step S3, the process of obtaining DEM data is as follows: S31. Determine areas without vegetation: Subtract the vegetation height from the DSM data generated by UAV oblique photography to determine areas without vegetation in the DSM. Define a threshold h, that is, areas with height h are considered areas without vegetation. A binary mask M represents the area without vegetation: ； Where (x,y) represents the coordinate position in the DSM data; S32. Treat the height of the unvegetated area as the ground height: For the area with mask M(x,y) = 1, directly treat the height value of the DSM as the height value of the DEM; Generate initial DEM data: DEMinit(x, y): DEMinit(x,y)=M(x,y)·DSM(x,y); S33. Interpolation to generate complete DEM data: For the region M(x, y) = 0, the DEM height value is estimated using an interpolation algorithm: ； in, This means interpolating and estimating the values ​​of the nearest DEM values. Specifically, it involves finding the nearest points (x', y') with M(x', y') = 1 and interpolating based on the height values ​​of these points.

2. The method for obtaining key attributes of canopy trees based on UAV oblique photography as described in claim 1, characterized in that: In step S4, the data processing procedure is as follows: S41. Define input and output grids: Input raster data: Raster input (i,j), where (i,j) represents the pixel position of the input raster data. Output raster data: Raster output (m,n), where (m,n) represents the pixel position of the output raster data. Input raster resolution: R input , Output raster resolution: R output = M•R input , M represents the target resolution to be adjusted; S42. Interpolate to target resolution: The output raster position (m,n) corresponds to the input raster position (i,j): ； For the data at position (m,n), perform nearest neighbor interpolation based on the position value at (i,j): ； S43. Obtain vegetation net height data DSM0: Subtract the DSM and DEM raster data adjusted to the same resolution to obtain vegetation net height data DSM0. DSM0 = DSM-DEM.

3. The method for obtaining key attributes of canopy trees based on UAV oblique photography as described in claim 2, characterized in that: In step S5, the data processing process is as follows: the DSM0 data generated in S43 is used as a two-dimensional matrix DSM(i,j), where (i,j) represents the coordinates in the grid, h(i,j) represents the elevation value of the point, the neighborhood is defined as a 3×3 window, and the tree top position, tree height elevation value, tree crown edge position and tree crown area are marked. S51. Locate the tree height: 1) Initialize the set of highest point positions: Peaks = Ø; 2) Traverse each point in the DSM data: For each (i,j) in DSM; 3) Obtain the neighborhood of point (i,j): ； 4) Determine if it is the treetop location: ； S52, Locate the edge of the tree canopy: 1) Initialize the set of lowest point positions: Vallerys=Ø; 2) Traverse each point in the DSM data: For each (i,j) in DSM; 3) Obtain the neighborhood of point (i,j): ； 4) Determine if it is the treetop location: ； S53, Mark tree height, crown edge, and crown width: 1) Initialize the tag set: Markedpeaks = Ø; 2) Iterate through each point in the highest position: For each (i,j) in Peaks; 3) Find the nearest lowest point in the X direction: ； ； 4) Find the nearest minimum point in the Y direction: ； ； 5) Mark the location of the highest point and the edge of its canopy: ； 6) Calculate the canopy area: Long axis: ； Short axis: 。 4. The method for obtaining key attributes of canopy trees based on UAV oblique photography as described in claim 1, characterized in that: In step S4, the resolution of DSM and DEM data is adjusted to 1 / 5 of the average tree crown width.

Images