Power transmission line tree height full-section automatic extraction method and system based on airborne laser radar, storage medium and computing device

Abstract

The present application belongs to the technical field of power transmission line survey and design, and particularly relates to a power transmission line tree height full-section automatic extraction method and system based on airborne laser radar, a storage medium and a computing device. The method comprises the following steps: S1, path parameterization and section generation; S2, point cloud block reading and pre-screening; S3, ground elevation extraction; S4, full-section highest vegetation point search; and S5, result output. The method solves the problems of low precision, poor efficiency and high missing detection rate of traditional methods. The method can also realize full-process automation and reduce labor intensity.

Description

Technical Field

[0001] This invention belongs to the field of power transmission line survey and design technology, specifically relating to a method, system, storage medium, and computing device for automatic extraction of tree height across the entire cross section of power transmission lines based on airborne lidar. Background Technology

[0002] During the design phase of overhead transmission line construction drawings, the height of trees along the route is a core foundational data point for tower positioning, conductor-to-ground distance verification, and delineation of logging areas. The accuracy of the tree height data directly affects the selection of tower height, the verification of conductor wind deflection safety clearance, and the rationality of the project investment estimate.

[0003] Currently, the following methods are mainly used by on-site surveyors to obtain tree height: First, the traditional endpoint sampling method, which only samples at both ends of the cross section, is prone to missing tall trees inside the buffer zone, with a miss rate as high as 12.7%; Second, the LiDAR manual interactive extraction method, which has not achieved full automation and cannot meet the high-efficiency requirements of construction drawing design.

[0004] The above methods have the following problems: First, visibility is limited, making it difficult to approach target trees in mountainous or dense forest areas; second, manual operation is inefficient and cannot achieve continuous measurement along the entire route; third, they are highly subjective, with estimation errors reaching several meters, which can easily lead to conservative designs resulting in unnecessary logging, or overly ambitious designs leaving operational safety hazards.

[0005] Airborne lidar can penetrate part of the vegetation canopy and quickly acquire high-density three-dimensional point cloud data of the ground surface and objects, providing a new direction for the digital extraction of vegetation information in the line corridor. At present, airborne lidar has been applied to the detection and operation and maintenance of tree obstacles in transmission lines, focusing on the identification of hidden dangers in existing lines. Regarding the automatic extraction of tree height in the construction drawing design stage of overhead transmission lines, some scholars have proposed an automatic tree height extraction method based on airborne LiDAR in the design stage of transmission lines. However, it has not designed a special optimization strategy for the problem of missed detection of trees in the buffer zone, and has not carried out multi-scenario verification, so its adaptability is limited (see: Chen Ming, Li Juan, Zhang Yu. Automatic extraction method of tree height in the design stage of transmission lines based on airborne LiDAR[J]. Power Survey and Design, 2023, 40(3): 58-63.). Other scholars have optimized the KD-Tree index for LiDAR point cloud vegetation height extraction, which improves the index query efficiency. However, it does not combine the parameterized features of the transmission line path and is difficult to apply directly to the construction drawing design scenario (see: Zhao Wei, Sun Qiang, Zhou Min. Improved KD-Tree index for LiDAR point cloud vegetation height extraction algorithm [J]. Surveying and Mapping Science, 2024, 49(2): 145-152.). Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a method, system, storage medium and computing device for automatic extraction of tree height across the entire cross section of power transmission lines based on airborne lidar, which solves the problems of low accuracy, poor efficiency and high false negative rate of traditional methods.

[0007] The technical solution adopted by this invention to solve its technical problem is: a method for automatic extraction of the full cross-section of tree height along power transmission lines based on airborne lidar, comprising the following steps: S1, Path parameterization and cross-section generation: Construct the line centerline using the path turning point coordinate data, generate several first sampling points on the line centerline at fixed sampling intervals, and calculate the coordinates of the left and right endpoints of the cross-section perpendicular to the line direction at each first sampling point. S2, Point Cloud Block Reading and Pre-screening: Read the airborne lidar point cloud data, construct the overall outer rectangular expansion area of ​​the line, and remove the point cloud data outside the overall outer rectangular expansion area to obtain the target point cloud data; separate the ground point set from the target point cloud data. and high vegetation point set ; S3, Ground Elevation Extraction: Extraction of Ground Point Sets and high vegetation point set Two-dimensional KD-Tree spatial indexes are constructed separately, and the reference ground elevation of each section is obtained using the two-dimensional KD-Tree of ground points. For each section, the ground point closest to the center point of the line connecting the left and right endpoints of the section is found using the two-dimensional KD-Tree of ground points to obtain the reference ground elevation of the corresponding section. S4, Search for the highest vegetation point across the entire cross section: Generate multiple second sampling points at equal intervals along the line connecting the left and right endpoints of each cross section. Perform a neighborhood query on each second sampling point with a preset radius R. Take the maximum value of the neighborhood of all second sampling points as the tree vertex elevation of the cross section and calculate the tree height. S5, Output: Generates a data table containing section number, center point coordinates, reference ground elevation, tree top elevation, and tree height.

[0008] Furthermore, in step S1, the format of the path-to-point coordinate data is (x, y). Let the path transition sequence be Define the center line of the line L The total length of the polyline segment formed by connecting each path turning point is denoted as . S ; Let the sampling interval be Then at the center line of the line L above generated M The first sampling point mileage is ; in, ; The coordinates are obtained through linear interpolation; At each first sampling point At this point, calculate the coordinates of the left and right endpoints of the section perpendicular to the track direction; remember The unit tangent vector at is ,in , , This represents the distance between adjacent path turning points. Then the unit normal vector Given the widths of the left and right buffers W Then the left endpoint of the cross section With the right endpoint The mathematical expression is: ; ; Left end point of cross section With the right endpoint The length of the line connecting them is .

[0009] Furthermore, in step S2, a streaming readout strategy is used to read the airborne lidar point cloud data, reading in C points at a time, where C is 5 × 10⁻⁶. 6 .

[0010] Furthermore, in step S2, let the original point cloud set be... Separate the ground point set from it With vegetation point set ; ; ; in, p It represents a single point in a point cloud.

[0011] Furthermore, in step S3, if the distance between the center point of the line connecting the left and right endpoints of the cross section and the nearest ground point exceeds a preset threshold, then the linear interpolation result of the ground elevation of 1 to 5 adjacent cross sections is used as the substitute value of the reference ground elevation of the cross section.

[0012] Furthermore, in step S4, the interval between the second sampling points of each cross-section is K. ; The preset radius is R, where R = 1 to 4 m; Equivalently generated along the line connecting the left and right endpoints of each section The second sampling point , indicating the first j On the line connecting the left and right endpoints of the first cross section k The planar coordinates of the second sampling point ( (i.e., from the left end of the cross-section) To the right endpoint equidistant interpolation points; in: ; With the first sampling point The tree height at the corresponding cross-section is H. j , ; in, To the first sampling point The elevation of the tree apex at the corresponding cross section; To the first sampling point The ground elevation at the corresponding cross-section; If there are no effective vegetation sites, then =0.

[0013] The automatic extraction system for the full profile of tree height along power transmission lines based on airborne lidar adopts an automatic extraction method for the full profile of tree height along power transmission lines based on airborne lidar. It includes a path parameterization and profile generation module, a point cloud block reading and pre-screening module, a ground elevation extraction module, a full profile highest vegetation point search module, and a result output module. The path parameterization and cross-section generation module is used to construct the center line of the line through the coordinate data of the path turning points, generate a number of first sampling points on the center line of the line at a fixed sampling interval, and calculate the coordinates of the left and right endpoints of the cross-section perpendicular to the line direction at each first sampling point. The point cloud segmentation reading and pre-screening module is used to read airborne lidar point cloud data, construct a rectangular expansion area outside the overall line coverage, and remove point cloud data outside the rectangular expansion area to obtain target point cloud data; and separate the ground point set from the target point cloud data. and high vegetation point set ; The ground elevation extraction module is used to extract ground point sets. and high vegetation point set Two-dimensional KD-Tree spatial indexes are constructed separately, and the reference ground elevation of each section is obtained using the two-dimensional KD-Tree of ground points. For each section, the ground point closest to the center point of the line connecting the left and right endpoints of the section is found using the two-dimensional KD-Tree of ground points to obtain the reference ground elevation of the corresponding section. The full-section highest vegetation point search module is used to generate multiple second sampling points at equal intervals on the line connecting the left and right endpoints of each section, perform a neighborhood query for each second sampling point with a preset radius R, take the maximum value of the neighborhood of all second sampling points as the tree vertex elevation of the section, and calculate the tree height. The output module is used to generate a data table containing section number, center point coordinates, reference ground elevation, tree top elevation, and tree height.

[0014] A storage medium storing a computer program, which, when executed by a processor, implements the steps of a method for automatically extracting the full cross-section height of trees along power transmission lines based on airborne lidar.

[0015] A computing device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a method for automatic full-section extraction of tree height along power transmission lines based on airborne lidar.

[0016] Compared with existing technologies, the beneficial effects of this invention are: This invention provides a method, system, storage medium, and computing device for automatic extraction of tree height across the entire cross-section of power transmission lines based on airborne lidar. It solves the problems of low accuracy, poor efficiency, and high false negative rate of traditional methods. It also achieves full-process automation, reducing labor intensity. This invention has been practically applied, and its specific advantages include the following: First, by employing a full-section dense sampling search strategy, the problem of missed detection of trees within the buffer zone by the traditional endpoint search method is effectively solved. Compared with existing technologies, the root mean square error (RMSE) of tree height extraction is reduced from 0.24m to 0.065m, a reduction of 72.9%; the missed detection rate of trees within the buffer zone is reduced from 12.7% to 0.48%, a reduction of 96.2%, fully meeting the centimeter-level accuracy requirements of construction drawing design.

[0017] Secondly, by employing point cloud segmented streaming reading and two-dimensional KD-Tree spatial indexing technology, a tree height survey of an entire 50-kilometer route can be completed within one minute. Compared with traditional manual operations (3-5 days), efficiency is improved by 4320 times; compared with traditional LiDAR manual interaction methods (2-3 hours), efficiency is improved by more than 120 times, and manual workload is reduced by more than 95%.

[0018] Third, by directly reading the pole or path turning point coordinate data and lidar point cloud data provided by the design professionals, a formatted tree height data table that meets the design requirements is automatically generated without additional data preprocessing or manual interaction, realizing full automation from data input to output.

[0019] Fourth, through actual testing, this invention demonstrates good adaptability to various terrain types, including high mountains, plains, and hills, as well as different vegetation types such as cypress, birch, camphor, poplar, pine, and oak. It is also applicable to the construction drawing design of overhead transmission lines at all voltage levels, including 220kV, 500kV, 750kV, 1000kV, and above.

[0020] Fifth, accurate tree height data allows for more reasonable delineation of logging areas, reducing unnecessary logging by an average of about 12%, and lowering project investment and compensation costs for crop damage. Attached Figure Description

[0021] Figure 1 This is a flowchart of the present invention; Figure 2 This is a schematic diagram showing the superposition of the extraction results and point cloud data in the main experimental area using the present invention; Figure 3 This is a schematic diagram showing the superposition of the extraction results in auxiliary experimental area one with point cloud data using the present invention; Figure 4 This is a schematic diagram showing the superposition of the extraction results and point cloud data in auxiliary experimental area two using the present invention. Detailed Implementation

[0022] The following is in conjunction with the appendix Figure 1 , 2 The invention is further illustrated in sections 3 and 4, as well as in the embodiments.

[0023] A method for automatic full-section extraction of tree height along power transmission lines based on airborne lidar, such as Figure 1 As shown, the steps include: S1, Path parameterization and cross-section generation: Construct the line centerline using the path turning point coordinate data, generate several first sampling points on the line centerline at fixed sampling intervals, and calculate the coordinates of the left and right endpoints of the cross-section perpendicular to the line direction at each first sampling point.

[0024] The path turning point coordinate data was provided by the surveying and mapping professionals.

[0025] Specifically, in step S1, the format of the path-to-point coordinate data is (x, y). Let the path transition sequence be Define the center line of the line L The total length of the polyline segment formed by connecting each path turning point is denoted as . S ; Let the sampling interval be Then at the center line of the line L above generated M The first sampling point mileage is ; in, ; The coordinates are obtained through linear interpolation; Sampling interval The optimal value is 5m.

[0026] At each first sampling point At this point, calculate the coordinates of the left and right endpoints of the section perpendicular to the track direction; remember The unit tangent vector at is ,in , , This represents the distance between adjacent path turning points. Then the unit normal vector Given the widths of the left and right buffers W Then the left endpoint of the cross section With the right endpoint The mathematical expression is: ; ; Left end point of cross section With the right endpoint The length of the line connecting them is .

[0027] S2, Point Cloud Segmentation and Pre-screening: Read the airborne LiDAR point cloud data, construct a rectangular extension area encompassing the entire line, and remove point cloud data outside the rectangular extension area to obtain the target point cloud data; according to the ASPRS standard (the full name of the LAS file format specification developed by the American Society for Photogrammetry and Remote Sensing, which defines the land cover type by assigning a uniform classification code and classification label to each point in the point cloud) classification system, separate the ground point set from the target point cloud data. (Classification code for ground points is 2) and high vegetation point set (The classification code for high vegetation points is 5).

[0028] Specifically, in step S2, a streaming readout strategy is used to read the airborne lidar point cloud data, reading in C points at a time. The optimal value of C is 5 × 10⁻⁶. 6 .

[0029] Specifically, in step S2, let the original point cloud set be... Separate the ground point set from it With vegetation point set ; ; ; in, p It represents a single point in a point cloud.

[0030] S3, Ground Elevation Extraction: Extraction of Ground Point Sets and high vegetation point set Two-dimensional KD-Tree spatial indexes are constructed separately, and the reference ground elevation of each section is obtained using the two-dimensional KD-Tree of ground points. For each section, the ground point closest to the center point of the line connecting the left and right endpoints of the section is found using the two-dimensional KD-Tree of ground points to obtain the reference ground elevation of the corresponding section.

[0031] The query requirement in this embodiment is to find the highest vegetation point and the nearest ground point in the neighborhood of a given point on a horizontal plane. The elevation dimension is only used for result calculation and not for spatial partitioning. The time complexity of nearest neighbor lookup in a two-dimensional KD-Tree in low-dimensional space (k≤5) is O(n). O ( It reduces the computational cost of one dimension compared to the three-dimensional KD-Tree, and improves query efficiency by about 30%.

[0032] Specifically, in step S3, if the distance between the center point of the line connecting the left and right endpoints of the cross section and the nearest ground point exceeds a preset threshold, then the linear interpolation result of the ground elevation of 1 to 5 adjacent cross sections is used as the substitute value of the reference ground elevation of the cross section.

[0033] The preset threshold ranges from 0.1 to 1m, with the optimal value being 0.5m.

[0034] Preferably, the linear interpolation result of the ground elevation of three adjacent cross sections is used as the substitute value of the reference ground elevation of the cross section.

[0035] S4, Search for the highest vegetation point across the entire cross section: Generate multiple second sampling points at equal intervals along the line connecting the left and right endpoints of each cross section. Perform a neighborhood query on each second sampling point with a preset radius R. Take the maximum value of the neighborhood of all second sampling points as the tree vertex elevation of the cross section and calculate the tree height.

[0036] Specifically, in step S4, the number of second sampling point intervals for each cross-section is K. The optimal value of K is ; The preset radius is R, where R = 1 to 4 m; the optimal value of R is 2 m.

[0037] Equivalently generated along the line connecting the left and right endpoints of each section The second sampling point , indicating the first j On the line connecting the left and right endpoints of the first cross section kThe planar coordinates of the second sampling point ( (i.e., from the left end of the cross-section) To the right endpoint equidistant interpolation points; in: ; With the first sampling point The tree height at the corresponding cross-section is , ; in, To the first sampling point The elevation of the tree apex at the corresponding cross section; To the first sampling point The ground elevation at the corresponding cross-section; If there are no effective vegetation sites, then =0.

[0038] S5, Output: Generates a formatted data table containing section number, center point coordinates, reference ground elevation, tree top elevation, and tree height.

[0039] Specifically, the formatted data table is an Excel or CSV table, which can be directly imported into the power transmission line design software.

[0040] The automatic extraction system for the full profile of tree height along power transmission lines based on airborne lidar adopts an automatic extraction method for the full profile of tree height along power transmission lines based on airborne lidar. It includes a path parameterization and profile generation module, a point cloud block reading and pre-screening module, a ground elevation extraction module, a full profile highest vegetation point search module, and a result output module. The path parameterization and cross-section generation module is used to construct the center line of the line through the coordinate data of the path turning points, generate a number of first sampling points on the center line of the line at a fixed sampling interval, and calculate the coordinates of the left and right endpoints of the cross-section perpendicular to the line direction at each first sampling point. The point cloud segmentation reading and pre-screening module is used to read airborne lidar point cloud data, construct a rectangular expansion area outside the overall line coverage, and remove point cloud data outside the rectangular expansion area to obtain target point cloud data; and separate the ground point set from the target point cloud data. and high vegetation point set ; The ground elevation extraction module is used to extract ground point sets. and high vegetation point set Two-dimensional KD-Tree spatial indexes are constructed separately, and the reference ground elevation of each section is obtained using the two-dimensional KD-Tree of ground points. For each section, the ground point closest to the center point of the line connecting the left and right endpoints of the section is found using the two-dimensional KD-Tree of ground points to obtain the reference ground elevation of the corresponding section. The full-section highest vegetation point search module is used to generate multiple second sampling points at equal intervals on the line connecting the left and right endpoints of each section, perform a neighborhood query for each second sampling point with a preset radius R, take the maximum value of the neighborhood of all second sampling points as the tree vertex elevation of the section, and calculate the tree height. The output module is used to generate a data table containing section number, center point coordinates, reference ground elevation, tree top elevation, and tree height.

[0041] A storage medium storing a computer program, which, when executed by a processor, implements the steps of a method for automatically extracting the full cross-section height of trees along power transmission lines based on airborne lidar.

[0042] A computing device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a method for automatic full-section extraction of tree height along power transmission lines based on airborne lidar.

[0043] Example

[0044] A 50km section of a proposed 1000kV transmission line in Sichuan Province was selected as the main experimental area. Two auxiliary experimental areas were also selected: the Chengdu Plain (mainly covered with camphor and poplar trees, with tree heights ranging from 3 to 20 m) and the Chengdu Hills (mainly covered with pine and oak trees, with tree heights ranging from 8 to 35 m) to verify the method's universality. The main experimental area consisted of high mountain terrain, with altitudes ranging from 1500 to 3500 m and slopes from 0° to 75°. The vegetation consisted mainly of cypress, birch, and oak trees, with tree heights generally ranging from 6 to 48 m. The coordinates of 56 path turning points were provided by the surveying professionals after the construction drawings were positioned. Point cloud data was collected by a Pegasus V10 UAV equipped with a LiDAR-22 laser scanner, with a point density of approximately 8 pts / m². Ground points and high-vegetation points have been classified, and the data summary is shown in Table 1.

[0045] Table 1: Summary of Experimental Data

[0046] In the auxiliary experimental area 1, the Chengdu Plain 500kV line experimental section has a path length of 50km, a point cloud file size of 6.82 GB, and a total of 230 million points, of which ground points account for 15.2% and high vegetation points account for 84.8%.

[0047] The auxiliary experimental area 2, the Chengdu hilly 220kV line experimental section, has a path length of 50km, a point cloud file size of 7.15 GB, and a total of 240 million points, of which ground points account for 13.7% and high vegetation points account for 86.3%.

[0048] The algorithm is implemented in Python 3.9 and runs on Windows 11 with an Intel Core i7-10700 CPU and 32GB of memory.

[0049] Based on the specific design requirements of the main experimental area and the two auxiliary experimental areas, the buffer widths are respectively set to W. 主 =30m, W 辅1 =18m, W 辅2 =12m. Set the sampling interval Δs=5m, and generate multiple second sampling points at equal intervals along the line connecting the left and right endpoints of each section of the main experimental area and the two auxiliary experimental areas. The number of these points is K. 主 =20、K 辅1 =12、K 辅2 =8, the neighborhood search radius is R=2m, and a total of 8998 tree vertices are collected. For example... Figure 2 , Figure 3 and Figure 4 As shown, red dots represent tree heights extracted using this invention, green areas represent buffer zones, clearly demonstrating the matching relationship between tree height extraction results and point cloud data, and orange dots represent ground points.

[0050] Comparative Example 1

[0051] 521 cross-sections were manually measured in the main experimental area. Sampling ensured coverage of all tree height ranges (6–48 m, with each 6 m layer containing approximately 87 samples), all terrain slopes (0°–15°, 15°–30°, 30°–45°, approximately 174 samples each), and all vegetation types (cypress, birch, and oak, approximately 174 samples each). Experienced surveyors manually measured each cross-section using LiDAR 360 software, and these measurements were used as reference values ​​for cross-comparison with the results obtained using the full-section automatic extraction method. A portion of the comparison results is shown in Table 2.

[0052] Table 2: Comparison of Automatic Extraction Results and Manual Measurement Results for the Entire Cross-Section

[0053] This indicates that the error between the full-section automatic extraction method and the manual measurement results is very small, and the extraction results are accurate.

[0054] Comparative Example 2

[0055] The error statistics of the full-section automatic extraction method and the traditional endpoint search method are shown in Table 3. The left and right buffer widths W and sampling interval Δs are kept consistent between the full-section automatic extraction method and the traditional endpoint search method.

[0056] Table 3: Error Statistics Results

[0057] The full-section automatic extraction method effectively covered the entire 30 m buffer zone through dense sampling along the line. The extraction results were highly consistent with the manual measurement values ​​(R²=0.9991), and RMSE=0.065 m. Compared with the traditional endpoint search method, the RMSE was reduced by 0.175 m, and the missed detection rate of tall trees inside the buffer zone was reduced from 12.7% to 0.48%, effectively solving the problem of missed detection of trees inside the buffer zone. RMSE (Root Mean Square Error) and MAE (Mean Absolute Error) are commonly used accuracy indicators to measure the deviation between the extracted values ​​and the true values ​​of manual measurement. The smaller the value, the higher the extraction accuracy.

[0058] The accuracy statistics of the main experimental area and the two auxiliary experimental areas are shown in Table 4, indicating that the full-section automatic extraction method has good adaptability under different terrain and vegetation types.

[0059] Table 4: Accuracy Statistics

[0060] The specific embodiments described are preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. All equivalent changes made to the structure, shape, and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for automatic extraction of tree height across the entire cross-section of power transmission lines based on airborne lidar, characterized in that, Including the following steps: S1, Path parameterization and cross-section generation: Construct the line centerline using the path turning point coordinate data, generate several first sampling points on the line centerline at fixed sampling intervals, and calculate the coordinates of the left and right endpoints of the cross-section perpendicular to the line direction at each first sampling point. S2, Point Cloud Block Reading and Pre-screening: Read the point cloud data of the airborne lidar, construct the overall outer rectangular expansion area of ​​the line, and remove the point cloud data outside the overall outer rectangular expansion area of ​​the line to obtain the target point cloud data; Separate the ground point set from the target point cloud data. and high vegetation point set ; S3, Ground Elevation Extraction: Extraction of Ground Point Sets and high vegetation point set Two-dimensional KD-Tree spatial indexes are constructed separately, and the reference ground elevation of each section is obtained using the two-dimensional KD-Tree of ground points. For each section, the ground point closest to the center point of the line connecting the left and right endpoints of the section is found using the two-dimensional KD-Tree of ground points to obtain the reference ground elevation of the corresponding section. S4, Search for the highest vegetation point across the entire cross section: Generate multiple second sampling points at equal intervals along the line connecting the left and right endpoints of each cross section. Perform a neighborhood query on each second sampling point with a preset radius R. Take the maximum value of the neighborhood of all second sampling points as the tree vertex elevation of the cross section and calculate the tree height. S5, Output: Generates a data table containing section number, center point coordinates, reference ground elevation, tree top elevation, and tree height.

2. The method for automatic full-section extraction of tree height along power transmission lines based on airborne lidar as described in claim 1, characterized in that, In step S1, the format of the path turning point coordinate data is (x, y). Let the path transition sequence be Define the center line of the line L The total length of the polyline segment formed by connecting each path turning point is denoted as . S ; Let the sampling interval be Then at the center line of the line L above generated M The first sampling point , mileage is ; in, ; The coordinates are obtained through linear interpolation; At each first sampling point At this point, calculate the coordinates of the left and right endpoints of the section perpendicular to the track direction; remember The unit tangent vector at is ,in , , The distance between adjacent path turning points; Then the unit normal vector Given the widths of the left and right buffers W Then the left end point of the cross section With the right endpoint The mathematical expression is: ； ； Left end point of cross section With the right endpoint The length of the line connecting them is .

3. The method for automatic full-section extraction of tree height along power transmission lines based on airborne lidar as described in claim 1, characterized in that, In step S2, a streaming readout strategy is used to read the airborne lidar point cloud data, reading in C points at a time, where C is 5 × 10⁻⁶. 6 .

4. The method for automatic full-section extraction of tree height along power transmission lines based on airborne lidar as described in claim 1, characterized in that, In step S2, let the original point cloud set be... Separate the ground point set from it With vegetation point set ; ； ； in, p It represents a single point in a point cloud.

5. The method for automatic extraction of tree height across the entire cross-section of power transmission lines based on airborne lidar as described in claim 1, characterized in that, In step S3, if the distance between the center point of the line connecting the left and right endpoints of the cross section and the nearest ground point exceeds a preset threshold, then the linear interpolation result of the ground elevation of 1 to 5 adjacent cross sections is used as the substitute value of the reference ground elevation of the cross section.

6. The method for automatic full-section extraction of tree height along power transmission lines based on airborne lidar as described in claim 2, characterized in that, In step S4, the number of intervals between the second sampling points of each cross section is K. ; The preset radius is R, where R = 1 to 4 m; Equivalently generated along the line connecting the left and right endpoints of each section The second sampling point ; in: ; The tree height at the cross section corresponding to the first sampling point j is H. j , ; in, The elevation of the tree vertex at the cross section corresponding to the first sampling point j; This refers to the ground elevation at the cross-section corresponding to the first sampling point j. If there are no effective vegetation sites, then =0.

7. An automatic full-section extraction system for tree height along power transmission lines based on airborne lidar, characterized in that, The automatic extraction method for the full cross-section of tree height of power transmission lines based on airborne lidar as described in any one of claims 1-6 includes a path parameterization and cross-section generation module, a point cloud block reading and pre-screening module, a ground elevation extraction module, a full cross-section highest vegetation point search module, and a result output module. The path parameterization and cross-section generation module is used to construct the center line of the line through the coordinate data of the path turning points, generate a number of first sampling points on the center line of the line at a fixed sampling interval, and calculate the coordinates of the left and right endpoints of the cross-section perpendicular to the line direction at each first sampling point. The point cloud segmentation reading and pre-screening module is used to read airborne lidar point cloud data, construct a rectangular expansion area outside the overall line coverage, and remove point cloud data outside the rectangular expansion area to obtain target point cloud data; and separate the ground point set from the target point cloud data. and high vegetation point set ; The ground elevation extraction module is used to extract ground point sets. and high vegetation point set Two-dimensional KD-Tree spatial indexes are constructed separately, and the reference ground elevation of each section is obtained using the two-dimensional KD-Tree of ground points. For each section, the ground point closest to the center point of the line connecting the left and right endpoints of the section is found using the two-dimensional KD-Tree of ground points to obtain the reference ground elevation of the corresponding section. The full-section highest vegetation point search module is used to generate multiple second sampling points at equal intervals on the line connecting the left and right endpoints of each section, perform a neighborhood query for each second sampling point with a preset radius R, take the maximum value of the neighborhood of all second sampling points as the tree vertex elevation of the section, and calculate the tree height. The output module is used to generate a data table containing section number, center point coordinates, reference ground elevation, tree top elevation, and tree height.

8. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the automatic extraction method for full-section height of trees along power transmission lines based on airborne lidar as described in any one of claims 1-6.

9. A computing device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the automatic full-section extraction method for tree height of power transmission lines based on airborne lidar as described in any one of claims 1-6.

Images