A power transmission line intrusion object positioning and ranging method fusing images and point clouds

Abstract

This invention relates to a method for locating and ranging intrusions into power lines by fusing images and point clouds. The method includes: aligning LiDAR point clouds to the viewpoint of an online monitoring device through coordinate transformation, and reconstructing power lines from these points, replacing direct power line reconstruction from the online monitoring point clouds; accurately detecting and locating intrusions online, extracting point clouds that can be mapped to 2D warning boxes, and extracting the intrusion point cloud using Euclidean clustering after removing ground data; and measuring the distance between the intrusion and the power line: calculating the shortest distance from the highest point of the intrusion point cloud to the power line using a numerical optimization method. This invention reduces the computational burden on edge devices for real-time processing and effectively reduces resource consumption. The invention uses a catenary model to fit the power line model on the XOY and XOZ planes respectively, constructing a system of linear equations and solving it using the SVD method, improving the accuracy of power line reconstruction and accurately measuring the distance between the highest point of the intrusion and the nearest point on the nearest power line.

Description

Technical Field

[0001] This invention relates to the field of dynamic monitoring technology for safety hazards of ultra-high voltage transmission lines, and in particular to a method for locating and ranging intruders on transmission lines by fusing images and point clouds. Background Technology

[0002] In modern power systems, the stable operation of ultra-high voltage (UHV) transmission lines is crucial for ensuring power grid security. Dynamic monitoring of safety hazards along transmission line corridors is a complex and critical task, primarily aimed at preventing intrusions by external objects such as cranes. Such intrusions can not only damage transmission equipment but also potentially trigger safety accidents and even pose a serious threat to public safety. Currently, although image monitoring devices are widely used in transmission line monitoring, these systems often struggle to accurately identify the actual distance between the intruder and the transmission line, leading to misjudgments and numerous false alarms. This problem severely reduces the accuracy of intelligent identification of UHV transmission line hazards based on visual information.

[0003] Current 2D / 3D fusion sensing technologies still face significant challenges in achieving accuracy and efficiency in monitoring the safe distance of power transmission lines. Firstly, because the distance between ultra-high-voltage transmission line towers far exceeds the target positioning distance in autonomous driving scenarios, accurately locating distant dynamic targets such as engineering vehicles in sparse point cloud environments presents a huge challenge. This challenge ultimately stems from two aspects: firstly, the near-dense, far-sparse characteristic of lidar point cloud data limits the quality and accuracy of point cloud data captured from a distance; secondly, the changing speed and positional uncertainty of dynamic targets increase the complexity of real-time monitoring and positioning. Secondly, due to the low reflectivity of transmission lines and the long distances between transmission line towers, the viewing angle and ranging range of monitoring lidar installed on the towers are limited, allowing only partial acquisition of transmission line point clouds, which also exhibit near-dense, far-sparse characteristics. Under these circumstances, how to effectively utilize the point cloud data acquired by airborne lidar to assist in monitoring the safe distance of potential hazards using parameterized transmission line models has become an unresearched issue. Summary of the Invention

[0004] To address the problem in existing technologies that cannot accurately measure the distance between an intruder and a power transmission line, the present invention aims to provide a method for locating and ranging power transmission line intruders by fusing images and point clouds that can accurately measure the distance between the highest point of an intruder and the nearest point on the nearest power transmission line.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for locating and ranging intrusions on power transmission lines by fusing images and point clouds, the method comprising the following sequential steps:

[0006] (1) Power line reconstruction in UAV point cloud: The complete, high-quality LiDAR point cloud obtained by UAV scanning is aligned to the view of the online monitoring device through coordinate transformation, and power line reconstruction is performed from it, instead of directly reconstructing power lines from the online monitoring point cloud.

[0007] (2) Online precise detection and positioning of intruders: Using the camera and lidar of the online monitoring device, 3D detection of distant targets based on image guidance is carried out; through camera-radar extrinsic parameter calibration, a precise mapping from point cloud to image is established, and then the point cloud that can be mapped to the 2D warning box is extracted. After removing the ground, the intruder point cloud is extracted using the Euclidean clustering algorithm.

[0008] (3) Distance between intruder and power line: The shortest distance from the highest point of the intruder point cloud to the power line is calculated using numerical optimization methods.

[0009] Step (1) specifically includes the following sequential steps:

[0010] (1a) Using a point cloud registration method based on homonymous point pairs, static and unchanging ground features are selected, and the airborne LiDAR point cloud is aligned to the monitoring view to obtain the attitude transformation matrix between the two views; the two views are the view of the lidar on the UAV and the view of the lidar of the monitoring equipment installed on the tower.

[0011] (1b) Preprocess the airborne LiDAR point cloud to obtain the preprocessed airborne LiDAR point cloud. The preprocessing refers to performing ground filtering and uniform downsampling of the point cloud.

[0012] (1c) The tower point cloud and the transmission line point cloud are extracted from the preprocessed airborne LiDAR point cloud using the grid height difference method and the octree connected component clustering algorithm, respectively.

[0013] (1d) Align the extracted transmission line point cloud to the monitoring lidar coordinate system, replace the monitoring lidar point cloud with limited range, and reconstruct the catenary model of the transmission line.

[0014] Step (2) specifically refers to: firstly, establishing an accurate mapping from point cloud to image; based on the pinhole camera model, using the camera's intrinsic parameter matrix K, distortion coefficients, and extrinsic parameters to realize the projection of point cloud from three-dimensional space to two-dimensional image; the extrinsic parameters include the rotation matrix R and the translation vector t, accurately mapping the fixed-viewpoint LiDAR point cloud under the monitoring lidar perspective onto the two-dimensional image of the monitoring camera; for point P in the monitoring lidar coordinate system l Its position P in the coordinate system of the monitoring camera c Calculated using the following formula:

[0015] P c =R·Pl +t

[0016] Then, these points are projected onto the imaging plane using the intrinsic parameter matrix K, and their positions are fine-tuned using the distortion coefficients. The points P projected onto the image plane are... i Calculated using the following formula:

[0017] P i =K·P c

[0018] After completing the mapping from point cloud to image, for all warning boxes in the two-dimensional image, the subset corresponding to the intruding object is identified and extracted from the fixed-view radar point cloud through the mapping relationship from point cloud to image.

[0019] Ground points were removed using a plane fitting method based on random sampling consistency, and then Euclidean clustering algorithm was used to determine the intruder point cloud by identifying the cluster with the most points.

[0020] Step (3) specifically refers to: assuming the coordinates of the highest point of the intruder are (x h ,y h ,z h First, it is necessary to determine the point on the transmission line closest to that point;

[0021] Define the objective function f(x,y) as the distance from the highest point of the intruder to the power line:

[0022]

[0023] The minimum value of f(x,y) is solved by numerical optimization to find the point where the highest point of the intrusion is closest to the transmission line; the numerical optimization method can be any one of gradient descent, Newton's method, conjugate gradient method, or genetic algorithm.

[0024] Assume the coordinates of the nearest point are (x nearest ,y nearest ,z nearest Then, the three-dimensional Euclidean distance D between these two points is calculated using the following formula:

[0025]

[0026] Calculate the distance D for the two lowest transmission lines respectively, find the transmission line closest to the highest point of the intruder, and this distance D is the three-dimensional distance measurement result of the intruder, which is also the shortest distance from the highest point of the intruder's point cloud to the transmission line.

[0027] Step (1a) specifically refers to: employing a point cloud registration method based on corresponding point pairs, selecting corresponding point pairs on statically invariant ground features from airborne LiDAR point clouds and fixed-view LiDAR point clouds, and using singular value decomposition to estimate the optimal rigid body transformation between the corresponding point pairs on the statically invariant ground features. The rigid body transformation formula is:

[0028] Q = RP + t

[0029] Where R is the rotation matrix, t is the translation vector, P is the airborne LiDAR point cloud, and Q is the fixed-view LiDAR point cloud.

[0030] Step (1c) specifically refers to: extracting tower point clouds from preprocessed airborne LiDAR point cloud data using a grid height difference method, and extracting power line point clouds from preprocessed airborne LiDAR point cloud data using an octree connected component clustering algorithm; the tower is the tallest object in the point cloud data, and its location is determined by analyzing the number of grids along the Z-axis; the octree connected component clustering algorithm effectively distinguishes different power lines that are close in space by recursively dividing the three-dimensional space into voxel grids with specific resolutions, adjusting the resolution of the voxel grids, and achieving accurate segmentation for power line point clouds with different densities and spacings.

[0031] Specifically, step (1d) refers to the following: when reconstructing the catenary model, after projecting the transmission line point cloud onto the XOY and XOZ planes respectively, the catenary model is represented by a straight line equation and a quadratic equation in two variables:

[0032] y = ax + b

[0033]

[0034] Where a, b, A, B, and C are all parameters of the equation.

[0035] As can be seen from the above technical solution, the beneficial effects of the present invention are as follows: First, the present invention accurately extracts the point cloud of the transmission line using UAV mapping technology and aligns it to the coordinate system of the monitoring LiDAR, replacing the monitoring LiDAR point cloud with limited range measurement to reconstruct the catenary model of the transmission line, reducing the computational pressure of real-time processing of edge devices and effectively reducing resource consumption; Second, the present invention adopts a point cloud registration method based on corresponding point pairs to align the airborne LiDAR point cloud to the monitoring LiDAR coordinate system, solving the problem of registration algorithm failure when "the scanning time of the airborne LiDAR point cloud and the fixed-view LiDAR point cloud are different, and the ground features in the transmission line scene may change significantly"; Third, the present invention uses the catenary model to fit the transmission line model on the XOY and XOZ planes respectively, constructs a linear equation system and solves it using the SVD method, improving the accuracy of transmission line reconstruction and providing a solid foundation for subsequent three-dimensional ranging; Fourth, the present invention uses the catenary model of the transmission line to accurately measure the distance between the highest point of the intrusion and the nearest point of the nearest transmission line. Attached Figure Description

[0036] Figure 1 This is a flowchart of the method for image-guided 3D detection of distant targets in this invention.

[0037] Figure 2 This is a schematic diagram of the three-dimensional distance measurement results of the intruder according to the present invention;

[0038] Figure 3 This is a schematic diagram of the power transmission line extraction and reconstruction results of the present invention;

[0039] Figure 4 This is a schematic diagram illustrating the effect of aligning the reconstruction result of the two-split transmission line of the present invention to the monitoring view.

[0040] Figure 5 This is a schematic diagram showing the three-dimensional distance measurement results of intruders at different distances according to the present invention. Detailed Implementation

[0041] A method for locating and ranging intrusions into power transmission lines by fusing images and point clouds, the method comprising the following sequential steps:

[0042] (1) Power line reconstruction in UAV point cloud: The complete, high-quality LiDAR point cloud obtained by UAV scanning is aligned to the view of the online monitoring device through coordinate transformation, and power line reconstruction is performed from it, instead of directly reconstructing power lines from the online monitoring point cloud.

[0043] (2) Online precise detection and positioning of intruders: Using the camera and lidar of the online monitoring device, 3D detection of distant targets based on image guidance is carried out; through camera-radar extrinsic parameter calibration, a precise mapping from point cloud to image is established, and then the point cloud that can be mapped to the 2D warning box is extracted. After removing the ground, the intruder point cloud is extracted using the Euclidean clustering algorithm.

[0044] (3) Distance between intruder and power line: The shortest distance from the highest point of the intruder point cloud to the power line is calculated using numerical optimization methods.

[0045] Step (1) specifically includes the following sequential steps:

[0046] (1a) Using a point cloud registration method based on corresponding point pairs, static and unchanging ground features are selected, and the airborne LiDAR point cloud is aligned to the monitoring view to obtain the attitude transformation matrix between the two views; the two views are the view of the lidar on the UAV and the view of the lidar of the monitoring device installed on the tower; the static and unchanging ground features include road signs, street lights and buildings, etc.

[0047] (1b) Preprocess the airborne LiDAR point cloud to obtain the preprocessed airborne LiDAR point cloud. The preprocessing refers to performing ground filtering and uniform downsampling of the point cloud.

[0048] (1c) The tower point cloud and the transmission line point cloud are extracted from the preprocessed airborne LiDAR point cloud using the grid height difference method and the octree connected component clustering algorithm, respectively.

[0049] (1d) Align the extracted transmission line point cloud to the monitoring lidar coordinate system, replace the monitoring lidar point cloud with limited range, and reconstruct the catenary model of the transmission line.

[0050] like Figure 1 As shown, step (2) specifically refers to: firstly, establishing an accurate mapping from point cloud to image; based on the pinhole camera model, using the camera's intrinsic parameter matrix K, distortion coefficients, and extrinsic parameters to realize the projection of point cloud from three-dimensional space to two-dimensional image; the extrinsic parameters include the rotation matrix R and translation vector t, accurately mapping the fixed-view LiDAR point cloud under the monitoring LiDAR perspective onto the two-dimensional image of the monitoring camera; this process accurately maps the fixed-view LiDAR point cloud under the monitoring LiDAR perspective onto the two-dimensional image of the monitoring camera, providing a foundation for subsequent 3D target detection. For point P in the monitoring LiDAR coordinate system... l Its position P in the coordinate system of the monitoring camera c Calculated using the following formula:

[0051] P c =R·P l +t

[0052] Then, these points are projected onto the imaging plane using the intrinsic parameter matrix K, and their positions are fine-tuned using the distortion coefficients. The points P projected onto the image plane are... i Calculated using the following formula:

[0053] P i =K·P c

[0054] After completing the mapping from point cloud to image, for all warning boxes in the two-dimensional image, the subset corresponding to the intruding object is identified and extracted from the fixed-view radar point cloud through the mapping relationship from point cloud to image. The key to this step is to use efficient two-dimensional image target detection technology and two-dimensional to three-dimensional mapping relationship to achieve three-dimensional target detection, thereby accurately determining the location and range of the intruder.

[0055] After successfully extracting the point cloud corresponding to the image target, the next step is to accurately segment the intruder from these point clouds. Since the extracted point cloud may contain some ground and background point clouds, a plane fitting method based on random sampling consistency is first used to remove ground points. Then, a Euclidean clustering algorithm is used to identify the intruder point cloud by recognizing the cluster with the most points. This process not only effectively improves the accuracy of point cloud segmentation but also increases processing speed.

[0056] By leveraging the coordinate transformation relationship between the monitoring camera and the monitoring lidar, three-dimensional target detection is achieved on the basis of two-dimensional target detection. While ensuring the response speed of the monitoring system, the accuracy and robustness of 3D detection of distant targets are improved, providing a precise location of the highest point of the intruder for subsequent three-dimensional distance measurement.

[0057] Step (3) specifically refers to: assuming the coordinates of the highest point of the intruder are (x h ,y h ,z h First, it is necessary to determine the point on the transmission line closest to that point;

[0058] Define the objective function f(x,y) as the distance from the highest point of the intruder to the power line:

[0059]

[0060] The minimum value of f(x,y) is solved by numerical optimization to find the point where the highest point of the intrusion is closest to the transmission line; the numerical optimization method can be any one of gradient descent, Newton's method, conjugate gradient method, or genetic algorithm.

[0061] Assume the coordinates of the nearest point are (x nearest ,y nearest ,z nearestThen, the three-dimensional Euclidean distance D between these two points is calculated using the following formula:

[0062]

[0063] Calculate the distance D for the two lowest transmission lines respectively, and find the transmission line closest to the highest point of the intruder. This distance D is the three-dimensional distance measurement result of the intruder, and also the shortest distance from the highest point of the intruder's point cloud to the transmission line. Figure 2 As shown. The three-dimensional distance measurement results of the intruder at different distances are as follows. Figure 5 As shown, from left to right are close-range, medium-range, and long-range, and from top to bottom are crane, bulldozer, and excavator.

[0064] Step (1a) specifically refers to: employing a point cloud registration method based on corresponding point pairs, selecting corresponding point pairs on statically invariant ground features from airborne LiDAR point clouds and fixed-view LiDAR point clouds, and using the SVD singular value decomposition method to estimate the optimal rigid body transformation between the corresponding point pairs on the statically invariant ground features. The rigid body transformation formula is:

[0065] Q = RP + t

[0066] Where R is the rotation matrix, t is the translation vector, P is the airborne LiDAR point cloud, and Q is the fixed-view LiDAR point cloud.

[0067] The point cloud registration method based on corresponding point pairs is adopted mainly because the scanning time of airborne LiDAR point clouds and fixed-view LiDAR point clouds is different, and the ground objects (such as vegetation and building materials) in the power transmission line scene may change significantly, making their registration a huge challenge. In the power transmission line scene, the distance between power towers is usually between 200 and 500 meters, making it difficult for the monitoring LiDAR to obtain complete and clear point clouds of the opposite towers.

[0068] Step (1c) specifically refers to: extracting tower point clouds from preprocessed airborne LiDAR point cloud data using a grid height difference method, and extracting power line point clouds from preprocessed airborne LiDAR point cloud data using an octree connected component clustering algorithm; the tower is the tallest object in the point cloud data, and its location is determined by analyzing the number of grids along the Z-axis; the octree connected component clustering algorithm effectively distinguishes spatially close different power lines by recursively dividing the three-dimensional space into voxel grids with specific resolution, such as... Figure 3 As shown, in Figure 3In the figures, (a) shows the segmentation result of a two-split transmission line, (b) shows the reconstruction result of a two-split transmission line, (c) shows the segmentation result of a four-split transmission line, and (d) shows the reconstruction result of a four-split transmission line. By adjusting the resolution of the voxel mesh, accurate segmentation is achieved for transmission line point clouds with different densities and spacings. This method not only improves the accuracy of point cloud extraction for towers and transmission lines but also facilitates subsequent analysis and processing.

[0069] Specifically, step (1d) refers to the following: when reconstructing the catenary model, after projecting the transmission line point cloud onto the XOY and XOZ planes respectively, the catenary model is represented by a straight line equation and a quadratic equation in two variables:

[0070] y = ax + b

[0071]

[0072] Where a, b, A, B, and C are all parameters of the equation. By constructing a system of linear equations and solving it using the SVD method, polynomial coefficients can be obtained. This method improves the accuracy of transmission line reconstruction and provides a solid foundation for subsequent three-dimensional ranging. Figure 4 As shown, in Figure 4 In (a) the green power transmission line is the result of mapping the reconstructed catenary model onto the image; in (b) the red power transmission line is the power transmission line point cloud that has been reconstructed and aligned to the radar coordinate system.

[0073] In summary, this invention accurately extracts power transmission line point clouds using UAV mapping technology and aligns them to the coordinate system of a monitoring LiDAR system, replacing the range-limited monitoring LiDAR point clouds for reconstructing power transmission line models. This reduces the computational burden on edge devices for real-time processing and effectively reduces resource consumption. Second, this invention employs a point cloud registration method based on corresponding point pairs to align the airborne LiDAR point cloud to the monitoring LiDAR coordinate system, solving the problem of registration algorithm failure when "the scanning time of the airborne LiDAR point cloud and the fixed-view LiDAR point cloud are different, and the ground features in the power transmission line scene may change significantly." Third, this invention uses a catenary model to fit the power transmission line model on the XOY and XOZ planes respectively, constructs a system of linear equations, and solves them using the SVD method, improving the accuracy of power transmission line reconstruction and providing a solid foundation for subsequent 3D ranging. Fourth, this invention utilizes the catenary model of the power transmission line to accurately measure the distance between the highest point of the intrusion and the nearest point of the nearest power transmission line.

[0074] In summary, this invention accurately extracts the point cloud of power transmission lines using UAV mapping technology and aligns it to the coordinate system of a monitoring LiDAR system. This replaces the monitoring LiDAR point cloud, which has limited range measurement, to reconstruct the catenary model of the power transmission line, reducing the computational burden on edge devices for real-time processing and effectively reducing resource consumption. This invention employs a point cloud registration method based on corresponding point pairs to align the airborne LiDAR point cloud to the monitoring LiDAR coordinate system, solving the problem of registration algorithm failure when the scanning time of the airborne LiDAR point cloud and the fixed-view LiDAR point cloud differs, and ground features in the power transmission line scene may change significantly. This invention uses a catenary model to fit the power transmission line model on the XOY and XOZ planes respectively, constructs a system of linear equations, and solves it using the SVD method, improving the accuracy of power transmission line reconstruction and providing a solid foundation for subsequent 3D ranging. This invention utilizes the catenary model of the power transmission line to accurately measure the distance between the highest point of the intrusion and the nearest point of the nearest power transmission line.

Claims

1. A method for power line intrusion object positioning and ranging by fusing images and point clouds, characterized in that: The method includes the following steps in sequence: (1) Power line reconstruction in UAV point cloud: The complete LiDAR point cloud obtained by UAV scanning is aligned to the view of the online monitoring device through coordinate transformation, and power line reconstruction is performed from it, instead of directly reconstructing power lines from the online monitoring point cloud; (2) Online precise detection and positioning of intruders: Using the camera and lidar of the online monitoring device, 3D detection of distant targets based on image guidance is carried out; through camera-radar extrinsic parameter calibration, a precise mapping from point cloud to image is established, and then the point cloud that can be mapped to the 2D warning box is extracted. After removing the ground, the Euclidean clustering algorithm is used to extract the intruder point cloud. (3) Distance between intruder and power line: The shortest distance from the highest point of the intruder point cloud to the power line is calculated using numerical optimization methods; Step (1) specifically includes the following sequential steps: (1a) Using a point cloud registration method based on corresponding point pairs, static and unchanging ground features are selected, and the airborne LiDAR point cloud is aligned to the monitoring view to obtain the attitude transformation matrix between the two views; the two views are the view of the lidar on the UAV and the view of the lidar of the monitoring equipment installed on the tower. (1b) Preprocess the airborne LiDAR point cloud to obtain the preprocessed airborne LiDAR point cloud. The preprocessing refers to ground filtering and uniform downsampling of the point cloud. (1c) The tower point cloud and the transmission line point cloud are extracted from the preprocessed airborne LiDAR point cloud using the grid height difference method and the octree connected component clustering algorithm, respectively. (1d) Align the extracted transmission line point cloud to the monitoring lidar coordinate system, replace the monitoring lidar point cloud with limited range, and reconstruct the catenary model of the transmission line.

2. The method of claim 1, wherein the method further comprises: Step (2) specifically refers to: firstly establishing an accurate mapping from point cloud to image, based on the pinhole camera model, using the camera's intrinsic parameter matrix. The distortion coefficients and their extrinsic parameters are used to project a point cloud from a three-dimensional space to a two-dimensional image. The extrinsic parameters include a rotation matrix. Translation vector This accurately maps the fixed-viewpoint LiDAR point cloud from the perspective of the monitoring LiDAR onto the two-dimensional image of the monitoring camera; for points in the monitoring LiDAR coordinate system... Its position in the coordinate system of the surveillance camera Calculated using the following formula: ； Then, using the intrinsic parameter matrix These points are projected onto the imaging plane, and their positions are fine-tuned using distortion coefficients. The points projected onto the image plane... Calculated using the following formula: ； After completing the mapping from point cloud to image, for all warning boxes in the two-dimensional image, the subset corresponding to the intruding object is identified and extracted from the fixed-view radar point cloud through the mapping relationship from point cloud to image. Ground points were removed using a plane fitting method based on random sampling consistency, and then Euclidean clustering algorithm was used to determine the intruder point cloud by identifying the cluster with the most points.

3. The method of claim 1, wherein: The step (3) is specifically as follows: assuming that the coordinates of the highest point of the intruder are , firstly, the point on the power transmission line closest to the highest point needs to be determined; Define the objective function Distance from the highest point of the intruder to the power line: ； in, The minimum value is solved by numerical optimization to find the point where the highest point of the intrusion is closest to the transmission line; the numerical optimization method can be any one of gradient descent, Newton's method, conjugate gradient method, or genetic algorithm. Assuming the coordinates of the nearest point are the three-dimensional Euclidean distance between the two points is calculated by : ； Calculate the distance for the two lowest transmission lines separately. Find the nearest power line to the highest point of the intruder. This is the result of the three-dimensional distance measurement of the intruder, and also the shortest distance from the highest point of the intruder's point cloud to the power transmission line.

4. The method for locating and ranging intrusions into power transmission lines by fusing images and point clouds according to claim 1, characterized in that: Step (1a) specifically refers to: employing a point cloud registration method based on corresponding point pairs, selecting corresponding point pairs on statically invariant ground features from airborne LiDAR point clouds and fixed-view LiDAR point clouds, and using singular value decomposition to estimate the optimal rigid body transformation between the corresponding point pairs on the statically invariant ground features. The rigid body transformation formula is: ； wherein, is a rotation matrix, is a translation vector, is an airborne LiDAR point cloud, is a fixed-view LiDAR point cloud.

5. The method for locating and ranging intrusions into power transmission lines by fusing images and point clouds according to claim 1, characterized in that: The step (1c) specifically refers to: extracting the tower point cloud from the preprocessed airborne LiDAR point cloud data using the grid height difference method, and extracting the power transmission line point cloud from the preprocessed airborne LiDAR point cloud data using the octree connected component clustering algorithm. The tower is the tallest object in the point cloud data. The tower's location is determined by analyzing the number of grids along the Z-axis. The octree connected domain clustering algorithm effectively distinguishes different power transmission lines that are close in space by recursively dividing the three-dimensional space into voxel grids. By adjusting the resolution of the voxel grids, accurate segmentation is achieved for power transmission line point clouds with different densities and spacings.

6. The method of claim 1, wherein: Specifically, step (1d) refers to the following: when reconstructing the catenary model, after projecting the transmission line point cloud onto the XOY and XOZ planes respectively, the catenary model is represented by a straight line equation and a quadratic equation in two variables: ； wherein , , , , are parameters of the equation.

Images