Power transmission line three-dimensional reconstruction method based on unmanned aerial vehicle and laser radar

Abstract

The invention provides a power transmission line three-dimensional reconstruction method based on an unmanned aerial vehicle and a laser radar, and the method comprises the steps: dynamically adjusting the scanning field angle and scanning frequency of the laser radar according to the flight attitude of the unmanned aerial vehicle and the direction of a cable, obtaining the original point cloud data of the cable at different angles and different densities, carrying out the preprocessing of the obtained original point cloud data, and obtaining a pre-processed point cloud data; obvious noisy points and outliers are removed, and cable point cloud data after preliminary filtering are obtained; and in a local tangent coordinate system, point cloud simplification and curved surface reconstruction are carried out on each local cable point cloud segment to obtain a local cable three-dimensional model, and through coordinate system inverse transformation, the local cable three-dimensional models are spliced into a complete cable three-dimensional model to realize three-dimensional accurate reconstruction of the cable. According to the invention, the problem of accurate three-dimensional reconstruction of a power transmission line in a complex environment is effectively solved, the efficiency and precision of cable detection and maintenance are improved, and during later upgrading and reconstruction or new project construction, electric power personnel can conveniently and rapidly master cable information.

Description

Technical Field

[0001] The present invention relates to the field of power cable measurement, and in particular to a three-dimensional reconstruction method of power transmission lines based on unmanned aerial vehicles and laser radar. Background Art

[0002] The actual models of current distribution network lines, especially overhead line conductors, deviate from the models actually known by power engineers, which causes some obstacles to power engineers in distribution network operation and upgrade planning, and may also cause inaccurate power grid investment. Current technology requires power outage detection, which is not possible under the current power supply reliability requirements. Binocular cameras can be used in a small number of scenarios to implement image recognition-based methods to judge the wire diameter of cables, but for widely distributed overhead lines, the efficiency is low, and it is difficult to quickly form a three-dimensional reconstruction of the actual transmission line of the overhead line. Simple local image collection is not conducive to the rapid establishment of a three-dimensional model during power grid upgrades and renovations, and is difficult to apply continuously in data collection and later planning. The diameter of overhead cables is relatively thin, and there are difficulties in the data collection process and accurate matching, so in-depth research is needed. Summary of the invention

[0003] The present invention provides a three-dimensional reconstruction method for power transmission lines based on drones and laser radars, which is convenient for quickly realizing the judgment and modeling of the direction and diameter of overhead lines, and includes the following steps:

[0004] According to the UAV flight attitude and cable direction, the laser radar scanning field of view and scanning frequency are dynamically adjusted to obtain the original point cloud data of cables at different angles and densities. The original point cloud data is pre-processed to remove obvious noise points and outliers to obtain the cable point cloud data after preliminary filtering.

[0005] The adaptive point cloud density estimation algorithm is used to dynamically adjust the point cloud filtering parameters of the cable point cloud data after preliminary filtering. The neighborhood density of each point is calculated, and the filter window size and threshold are adaptively adjusted according to the density value to achieve adaptive filtering of different density areas. The cable point cloud data after preliminary filtering is filtered twice to further remove noise points, improve the quality of cable point cloud data, and obtain the rough registration result of cable point cloud data.

[0006] Based on the rough registration results, the curve fitting algorithm is used to achieve the precise registration of the cable point cloud data, and the complete cable point cloud data in a unified coordinate system is obtained;

[0007] According to the sag characteristics of the cable in space, a segmented processing strategy is adopted to divide the complete cable point cloud data into multiple local cable point cloud segments, and a coordinate system transformation is performed on each local cable point cloud segment to obtain a local tangent coordinate system, so as to eliminate the influence of the cable sag on the three-dimensional reconstruction;

[0008] In each local tangent coordinate system, each local cable point cloud segment is simplified and reconstructed to obtain a local cable 3D model; through inverse transformation of the coordinate system, all local cable 3D models are spliced ​​into a complete cable 3D model to achieve accurate 3D reconstruction of the cable.

[0009] Preferably, preprocessing the acquired original point cloud data includes:

[0010] Performing statistical analysis on the original point cloud data to obtain density distribution and discreteness of the point cloud data; judging whether there are noise points and outliers in the original point cloud data according to the density distribution and a preset discreteness threshold; if there are noise points and outliers, using an outlier detection method based on Euclidean distance to calculate the distance between each point and surrounding points, determining outliers according to a preset distance threshold, and removing the outliers from the original point cloud data to obtain processed point cloud data.

[0011] Preferably, based on the filtered point cloud data, the spatial orientation of the cable is extracted by principal component analysis, the cable point cloud data is decomposed into a covariance matrix, the eigenvector corresponding to the maximum eigenvalue is taken as the direction vector of the cable, a point cloud subset is intercepted along the direction vector with a certain step size, and a three-dimensional straight line is fitted to obtain the skeleton curve of the cable in space.

[0012] Preferably, dividing the local cable point cloud segments includes: performing point cloud segmentation using a region growing algorithm according to the spatial distribution characteristics of the processed point cloud data, and dividing point clouds with similar spatial positions and similar geometric features into the same cable segment by setting a distance threshold and an angle threshold for region growth, thereby obtaining segmented cable point cloud data.

[0013] The present invention discloses a method for three-dimensional reconstruction of power transmission lines based on drones and laser radars. Aiming at the characteristics of slender transmission cables and weak reflection signals, the method obtains multi-angle point cloud data by dynamically adjusting the laser radar scanning parameters, and uses an adaptive density estimation algorithm to filter the point cloud. It combines the cable length constraint and shape prior knowledge to build a parameterized model to achieve precise point cloud registration. The innovative use of segmented processing strategies and local tangent coordinate system transformation eliminates the influence of cable sag, performs surface reconstruction in the local coordinate system, and finally obtains a complete three-dimensional cable model. The present invention effectively solves the problem of accurate three-dimensional reconstruction of power transmission lines in complex environments, improves the efficiency and accuracy of cable detection and maintenance, and can efficiently and quickly determine the wire diameter of overhead cables. In the later stage of upgrading and reconstruction or new project construction, it is convenient for power personnel to quickly grasp cable information, avoid repeated measurements, and avoid waste of investment caused by wrong or imperfect evaluation data. At the same time, it is of great significance to the safe operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The present invention is a flow chart of a method for three-dimensional reconstruction of power transmission lines based on UAV and laser radar. DETAILED DESCRIPTION

[0015] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0016] like Figure 1 In this embodiment, a 3D reconstruction method of a power transmission line based on a drone and a laser radar may specifically include:

[0017] S101. According to the UAV flight posture and cable direction, dynamically adjust the laser radar scanning field of view and scanning frequency to obtain the original point cloud data of cables at different angles and densities. Preprocess the obtained original point cloud data to remove obvious noise points and outliers to obtain the cable point cloud data after preliminary filtering.

[0018] The drone obtains flight attitude information, including pitch, roll and yaw angles, through the inertial measurement unit (IMU), and obtains the three-dimensional spatial coordinate information of the cable through the Velodyne VLP-16 laser radar. According to the flight attitude and cable spatial information, the scanning field of view and scanning frequency parameters of the laser radar are dynamically calculated. For example, when the drone's flight altitude is 50 meters and the angle between the cable and the horizontal plane is 30 degrees, the laser radar's scanning field of view is set to 60 degrees and the scanning frequency is set to 20Hz. The calculated scanning parameters are used to control the laser radar to scan and obtain the original point cloud data of different angles and densities of the cable. According to the flight attitude and spatial orientation information, the scanning field of view angle and scanning frequency parameters of the laser radar are dynamically calculated; the scanning field of view angle and scanning frequency parameters are used to control the laser radar to obtain original point cloud data of different angles and different densities of the cable; the original point cloud data is statistically analyzed to obtain the density distribution and discreteness of the point cloud data; according to the density distribution and the preset discreteness threshold, it is judged whether there are noise points and outliers in the original point cloud data; if there are noise points and outliers, an outlier detection method based on Euclidean distance is used to calculate the distance between each point and the surrounding points, and the outliers are determined according to the preset distance threshold, and the outliers are removed from the original point cloud data, such as calculating the average distance between each point and the 10 points around it. If the distance is greater than the threshold of 1 meter, it is determined to be an outlier and is removed from the original point cloud data, thereby obtaining the processed point cloud data.

[0019] S102. Adopting an adaptive point cloud density estimation algorithm to dynamically adjust the point cloud filtering parameters of the cable point cloud data after preliminary filtering, the neighborhood density of each point is calculated, and the filter window size and threshold are adaptively adjusted according to the density value to achieve adaptive filtering of different density areas. The cable point cloud data after preliminary filtering is subjected to secondary filtering to further remove noise points, improve the quality of the cable point cloud data, and obtain a coarse registration result of the cable point cloud data.

[0020] In view of the characteristics of thin cable diameter and weak laser reflection signal, an adaptive point cloud density estimation algorithm is used for filtering. By calculating the local density of the point cloud, the point cloud density distribution in different areas is judged, and the filtering parameters are dynamically adjusted according to the density distribution. For areas where the point cloud density is higher than the preset threshold, a filtering radius smaller than the preset threshold and a density threshold higher than the preset threshold are used. For areas where the point cloud density is lower than the preset threshold, a filtering radius larger than the preset threshold and a density threshold lower than the preset threshold are used to achieve adaptive filtering. With each point as the center and a radius of 2 meters, a local spherical neighborhood is constructed, and the number of points in the neighborhood is counted to obtain the local density distribution histogram of the point cloud. According to the peak and mean of the histogram, the point cloud density threshold is set to 100 / cubic meter. For areas where the local point cloud density is higher than the threshold, the filtering radius is set to 1 meter and the density threshold is set to 150 / cubic meter; for areas where the local point cloud density is lower than the threshold, the filtering radius is set to 3 meters and the density threshold is set to 50 / cubic meter. After adaptive filtering, the uniformity and integrity of the point cloud data are improved.

[0021] S103, based on the rough registration result, a curve fitting algorithm is used to achieve a precise registration of the cable point cloud data, and obtain complete cable point cloud data in a unified coordinate system.

[0022] According to the filtered point cloud data, the spatial direction of the cable is extracted by principal component analysis, and the covariance matrix of the point cloud data is decomposed. The eigenvector corresponding to the maximum eigenvalue is taken as the direction vector of the cable. The point cloud subset is intercepted along the direction vector with a certain step length, and a three-dimensional straight line is fitted to obtain the complete skeleton curve data of the cable in space; based on the coarse alignment results, the cable length constraint and shape prior knowledge are used to construct a cable parameterized model; the cable length constraint means that the total length of the cable should be consistent with the actual measured value, and the shape prior knowledge includes the sag characteristics and tension distribution of the cable. The cable is measured on the spot using a total station to measure the actual length of the cable. Through curve fitting algorithms, etc., the precise alignment of the cable point cloud data is achieved to obtain the cable point cloud data in a unified coordinate system. The least squares method is used to construct a curve fitting optimization problem to minimize the sum of the squares of the distances between the fitted cable curve and the point cloud data; the optimal parameters of the spline curve are obtained by solving the least squares problem, so that the fitted cable curve model is consistent with the point cloud data; the length of the optimized cable model is calculated and compared with the measured length; if the length deviation exceeds a preset threshold, the position of the spline curve control point is adjusted according to the deviation size and direction, and the curve fitting optimization process is repeated until the length constraint is met; based on the optimal parameters of the spline curve obtained by solving, a precisely aligned three-dimensional cable model is generated; the aligned cable point cloud data is fused with the generated spline curve model to obtain a final high-precision three-dimensional cable reconstruction result.

[0023] According to the sag characteristics and tension distribution law of the transmission line, a parameterized cable model is constructed using a cubic B-spline curve. According to the suspension state and force analysis of the cable, 8 control points are initially set, located at the two ends of the cable and 6 evenly distributed points in the middle. The least squares method is used to construct a curve fitting optimization model, and the objective function is to minimize the sum of squares of the Euclidean distances between the sampling points on the cable model and the corresponding point cloud data points. The Levenberg-Marquardt algorithm is used to optimize and solve the objective function to obtain the optimal control point coordinates of the spline curve, so that the average distance between the cable model and the point cloud data is reduced to an acceptable range. For example, the length of the optimized cable spline curve is 95 meters, which is 5 meters different from the measured length of 100 meters, exceeding the preset threshold of 2 meters. According to the direction of the length deviation, the control points at both ends of the cable are expanded by 3 meters along the normal vector direction, and 2 control points are added to 10, and the above curve fitting optimization process is repeated. After 3 iterative optimizations, the length of the cable spline curve finally fitted is 95 meters, which meets the length constraint. The coordinates of the optimal control points of the solved spline curve are used as parameters to generate the precisely aligned 3D geometric model of the cable. The weighted ICP algorithm is used to align and fuse the cable point cloud data with the spline curve model. The weight coefficient is adaptively adjusted according to the distance from the point to the curve to obtain the final high-precision 3D reconstruction model of the cable.

[0024] S104. According to the sag characteristics of the cable in space, a segmented processing strategy is adopted to divide the complete cable point cloud data into multiple local cable point cloud segments, and a coordinate system transformation is performed on each local cable point cloud segment to obtain a local tangent coordinate system, so as to eliminate the influence of the cable sag on the three-dimensional reconstruction.

[0025] One of the methods is to use a region growing algorithm to segment the point cloud according to the spatial distribution characteristics of the processed point cloud data. By setting the distance threshold and angle threshold of the region growing, point clouds with similar spatial positions and similar geometric features are divided into the same cable segment to obtain segmented cable point cloud data.

[0026] Another method is to use the DBSCAN clustering algorithm to segment the cable point cloud. According to the density distribution characteristics of the cable point cloud, the radius parameter ε and the minimum point number threshold minPts of the DBSCAN algorithm are set to cluster the cable point cloud into different cable segments.

[0027] Perform cylinder fitting on each subset of the cable segment point cloud, take the cable skeleton curve as the cylinder axis, set the initial cylinder radius along the normal direction, optimize the cylinder parameters by the least squares method, minimize the distance error between the cylinder surface and the cable point cloud, and obtain the optimal cylinder parameters through multiple iterative optimizations to obtain the average diameter of the cable segment;

[0028] By calculating the radius of each fitted cylinder, the diameter distribution of different sections of the cable is obtained. The diameter distribution is statistically analyzed, and outliers are removed to obtain the average diameter and diameter variation range of the cable.

[0029] The fitting results were compared with the measured diameter to evaluate the accuracy of the cylinder fitting.

[0030] Specifically, in steps S103-S104, the filtered point cloud data is subjected to principal component analysis to extract the spatial orientation characteristics of the cable. For example, the 3×3 covariance matrix of the point cloud data is calculated, and the matrix is ​​subjected to eigenvalue decomposition to obtain three eigenvalues ​​of 85, 12, and 3, respectively, and the corresponding eigenvectors are (5, 8, 99), (10, 99, -8), and (99, -10, 5), respectively. The eigenvector (5, 8, 99) corresponding to the maximum eigenvalue 85 is taken as the direction vector of the cable, and a point cloud subset is intercepted along this direction with a step length of 1 meter. Each subset contains 1000 points, and a three-dimensional straight line is fitted to the subset to obtain the skeleton curve of the cable in space, with a curve length of 98 meters. The DBSCAN clustering algorithm is used to segment the cable point cloud, and the radius parameter ε is set to 5 meters, and the minimum point number threshold minPts is set to 100, to obtain 8 cable segment point cloud clusters. A cylinder is fitted for each cable segment point cloud, with the skeleton curve as the axis, and the initial cylinder radius is set to 3 cm. The Levenberg-Marquardt least squares optimization algorithm is used to iteratively optimize the cylinder parameters to minimize the average distance from the cylinder surface to the cable point. After 10 iterations, the diameters of the 8 cable segment fitting cylinders converged to 1.95 cm, 2.03 cm, 1.98 cm, 2.06 cm, 1.93 cm, 2.01 cm, 1.97 cm and 2.04 cm, respectively, with an average diameter of 2 cm. The cable diameter obtained by cylinder fitting is compared with the prior measured diameter of 2.5 cm, with an error of 0.5 cm and a relative error of 25%. This shows that the cable diameter estimate obtained by the cylinder fitting method is too large due to the quality and density of the cable point cloud data. The point cloud processing algorithm needs to be further optimized to improve the accuracy of cable diameter estimation.

[0031] S105. In each local tangent coordinate system, perform point cloud simplification and surface reconstruction on each local cable point cloud segment to obtain a local cable three-dimensional model; through inverse transformation of the coordinate system, splice all local cable three-dimensional models into a complete cable three-dimensional model to achieve accurate three-dimensional reconstruction of the cable. Perform curve fitting on the segmented cable point cloud segment data, and use the least squares method to fit the spatial curve equation of each cable point cloud segment; calculate the length, curvature and sag geometric parameters of each cable point cloud segment based on the fitted spatial curve equation; construct a cable status assessment model based on the density, discreteness, segmentation characteristics and geometric parameters of the segmented cable point cloud segment data.

[0032] For each local cable point cloud segment, down-sampling is performed through the voxelization method to obtain a simplified local cable point cloud segment; according to the simplified local cable point cloud segment, the surface of the local cable point cloud segment is fitted by the Poisson surface reconstruction algorithm to obtain a local cable three-dimensional model; the coordinate transformation relationship between the local tangent coordinate system and the global coordinate system is obtained, and the coordinates of the local cable three-dimensional model are transformed to the global coordinate system through inverse coordinate transformation; the transformed local cable three-dimensional models are spliced ​​according to their positional relationship on the cable to obtain a complete cable three-dimensional model; the complete cable three-dimensional model is meshed and Laplace smoothed to obtain a processed cable three-dimensional model; the processed cable three-dimensional model is stored in the specified path in OBJ format.

[0033] Specifically, for the local point cloud data of each span segment, the least squares method is used to fit the quadratic polynomial equation y=ax that reflects the sag characteristics of the conductor. 2 +bx+c. According to the projection of the fitting curve on the XOZ plane, the segmentation position of the curve is adaptively determined, and the wire point cloud of each span segment is divided into 80 local point cloud units. For each local point cloud unit, the line connecting its head and tail endpoints is used as the X-axis to establish a local rectangular coordinate system. The local point cloud unit is downsampled by the voxelization method, and the voxel size is set to 5cm to obtain a simplified local cable point cloud segment. In the local tangent coordinate system, the Poisson surface reconstruction algorithm is used to perform surface fitting on the simplified local cable point cloud segment. The depth value of the Poisson equation is taken as 8 to reconstruct the local cable three-dimensional model. According to the transformation matrix between the local tangent coordinate system and the global geographic coordinate system, the vertex coordinates of the local cable three-dimensional model are converted to the global coordinate system through coordinate inverse transformation. According to the spatial position relationship of the local cable three-dimensional model on the line, the closest point iteration algorithm (ICP) is used to finely align the overlapping parts of the two adjacent local models and splice them into a complete cable three-dimensional model. The spliced ​​cable 3D model was meshed and smoothed using the Laplace smoothing algorithm. The number of iterations was set to 20 to maintain the overall shape characteristics of the cable model. Finally, the smoothed cable 3D model was stored in OBJ format and texture coordinates were set for subsequent visualization.

[0034] The establishment of a three-dimensional model helps to realize the electronicization of cable information. When there is a deviation between the traditional distribution network line and the model mastered by power engineers, it will cause certain obstacles to the operation and planning of the power distribution network, and may also cause inaccurate power grid investment. With the help of cable three-dimensional model information, such problems can be effectively solved.

[0035] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A three-dimensional reconstruction method of power transmission lines based on drones and laser radar, characterized in that: The method includes the following steps: dynamically adjusting the laser radar scanning field of view and scanning frequency according to the UAV flight attitude and cable direction, obtaining the original point cloud data of the cable at different angles and densities, pre-processing the obtained original point cloud data, removing obvious noise points and outliers, and obtaining the cable point cloud data after preliminary filtering; The adaptive point cloud density estimation algorithm is used to dynamically adjust the point cloud filtering parameters of the cable point cloud data after preliminary filtering. The neighborhood density of each point is calculated, and the filter window size and threshold are adaptively adjusted according to the density value to achieve adaptive filtering of different density areas. The cable point cloud data after preliminary filtering is filtered twice to further remove noise points, improve the quality of cable point cloud data, and obtain the rough registration result of cable point cloud data. Based on the rough registration results, the curve fitting algorithm is used to achieve the precise registration of the cable point cloud data, and the complete cable point cloud data in a unified coordinate system is obtained; According to the sag characteristics of the cable in space, a segmented processing strategy is adopted to divide the complete cable point cloud data into multiple local cable point cloud segments, and a coordinate system transformation is performed on each local cable point cloud segment to obtain a local tangent coordinate system, so as to eliminate the influence of the cable sag on the three-dimensional reconstruction; In each local tangent coordinate system, each local cable point cloud segment is simplified and reconstructed to obtain a local cable 3D model; through inverse transformation of the coordinate system, all local cable 3D models are spliced ​​into a complete cable 3D model to achieve accurate 3D reconstruction of the cable.

2. The method for three-dimensional reconstruction of power transmission lines based on drones and laser radar according to claim 1, characterized in that: Preprocessing the acquired raw point cloud data includes: Perform statistical analysis on the original point cloud data to obtain the density distribution and discreteness of the point cloud data; determine whether there are noise points and outliers in the original point cloud data based on the density distribution and a preset discreteness threshold; if there are noise points and outliers, use an outlier detection method based on Euclidean distance to calculate the distance between each point and surrounding points, determine the outliers based on a preset distance threshold, and remove the outliers from the original point cloud data to obtain processed point cloud data.

3. The method for three-dimensional reconstruction of power transmission lines based on drones and laser radar according to claim 1, characterized in that: According to the filtered point cloud data, the spatial direction of the cable is extracted by principal component analysis. The covariance matrix of the cable point cloud data is decomposed, and the eigenvector corresponding to the maximum eigenvalue is taken as the direction vector of the cable. The point cloud subset is intercepted along the direction vector with a certain step size, and a three-dimensional straight line is fitted to obtain the skeleton curve of the cable in space.

4. The method for three-dimensional reconstruction of power transmission lines based on drones and laser radar according to claim 3 is characterized in that: Dividing the local cable point cloud segments includes: performing point cloud segmentation using a region growing algorithm according to the spatial distribution characteristics of the processed point cloud data, and dividing point clouds with similar spatial positions and similar geometric features into the same cable segment by setting a distance threshold and an angle threshold for region growing, thereby obtaining segmented cable point cloud data.

5. The method for three-dimensional reconstruction of power transmission lines based on drones and laser radar according to claim 3, characterized in that: Dividing the local cable point cloud segments includes: using the DBSCAN clustering algorithm to segment the cable point cloud, setting the radius parameter ε and the minimum point count threshold minPts of the DBSCAN algorithm according to the density distribution characteristics of the cable point cloud, and clustering the cable point cloud into different cable segments.

6. The method for three-dimensional reconstruction of power transmission lines based on drones and laser radar according to claim 1, characterized in that: Obtaining a local cable three-dimensional model includes: performing curve fitting on the segmented cable point cloud segment data, and using the least square method to fit the spatial curve equation of each cable point cloud segment; According to the fitted spatial curve equation, the length, curvature and sag geometric parameters of each cable point cloud segment are calculated; cylinder fitting is performed on each cable segment point cloud subset, with the cable skeleton curve as the cylinder axis, the initial cylinder radius is set along the normal direction, and the cylinder parameters are optimized by the least squares method to minimize the distance error between the cylinder surface and the cable point cloud. The optimal cylinder parameters are obtained through multiple iterative optimizations to obtain the average diameter of the cable segment; By calculating the radius of each fitted cylinder, the diameter distribution of different sections of the cable is obtained. The diameter distribution is statistically analyzed, and outliers are removed to obtain the average diameter and diameter variation range of the cable. The cable status assessment model is constructed by comprehensively considering the density, discreteness, segmentation characteristics and geometric parameters of the segmented cable point cloud data.

7. The method for three-dimensional reconstruction of power transmission lines based on drones and laser radar according to claim 1, characterized in that: Based on the rough registration results, the cable parameterized model is constructed using the length constraint and shape prior knowledge of the cable. The cable length constraint means that the total length of the cable should be consistent with the actual measured value, and the shape prior knowledge includes the sag characteristics and tension distribution of the cable.

8. The method for 3D reconstruction of power transmission lines based on drones and laser radar according to claim 1, characterized in that: A precise registration of the cable point cloud data is achieved by a curve fitting algorithm, including: constructing a curve fitting optimization problem using a least squares method to minimize the sum of squares of distances between the fitted cable curve and the point cloud data; obtaining the optimal parameters of the spline curve by solving the least squares problem to make the fitted cable curve model match the point cloud data; The length of the optimized cable model is calculated and compared with the measured length; if the length deviation exceeds a preset threshold, the position of the spline curve control point is adjusted according to the deviation size and direction, and the curve fitting optimization process is repeated until the length constraint is met; based on the optimal parameters of the spline curve obtained by solving, a precisely aligned three-dimensional cable model is generated; the aligned cable point cloud data is fused with the generated spline curve model to obtain a final high-precision three-dimensional cable reconstruction result.

9. The method for three-dimensional reconstruction of power transmission lines based on drones and laser radar according to claim 1, characterized in that: In the local tangent coordinate system, the cable point cloud data is obtained, and the cable point cloud is divided into multiple local cable point cloud segments by using the local tangent coordinate system method; for each of the local cable point cloud segments, down-sampling is performed by using the voxelization method to obtain a simplified local cable point cloud segment; according to the simplified local cable point cloud segment, the surface of the local cable point cloud segment is fitted by using the Poisson surface reconstruction algorithm to obtain a local cable three-dimensional model; the coordinate transformation relationship between the local tangent coordinate system and the global coordinate system is obtained, and the coordinates of the local cable three-dimensional model are transformed to the global coordinate system by using the coordinate inverse transformation; the transformed local cable three-dimensional models are spliced ​​according to their positional relationship on the cable to obtain a complete cable three-dimensional model; the complete cable three-dimensional model is meshed and Laplace smoothed to obtain a processed cable three-dimensional model; the processed cable three-dimensional model is stored in the specified path in the OBJ format.

Images