Tree obstacle analysis method based on unmanned aerial vehicle image and line parametric modeling

Abstract

The invention discloses a tree obstacle analysis method based on unmanned aerial vehicle images and line parametric modeling, and the method comprises the following steps: S1, unmanned aerial vehicle image point cloud reconstruction: carrying out the three-dimensional reconstruction through a multi-angle aerial image, and generating geographically registered dense point cloud data; s2, parameterized line modeling: constructing a line model comprising towers, wires and ancillary facilities based on the collected line parameters; and S3, tree obstacle risk point analysis: through fusion of the point cloud and the line model, calculating the height difference between the lead and the vegetation point cloud, and identifying a risk point set with insufficient safety distance. According to the method, a rapid ground three-dimensional reconstruction and line modeling method is found to meet the timeliness requirement of tree obstacle analysis, three-dimensional reconstruction is conducted through images shot by an unmanned aerial vehicle, analysis of obstacles below a line is rapidly achieved by constructing a line model through parameterization, and the tree obstacle analysis processing efficiency and accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) inspection systems, and particularly to a tree obstacle analysis method based on UAV images and line parametric modeling. Background Art

[0002] With the expansion of the scale of power lines and the increase in vegetation coverage, tree obstacle hazards have become one of the core risks threatening the safety of the power grid. The current technology takes UAV three-dimensional modeling as the core, combines lidar point cloud processing and deep learning algorithms, and gradually replaces traditional manual inspections. However, there are still significant deficiencies in aspects such as monitoring coverage, dynamic assessment, and line modeling, which restrict the accuracy and practicality of tree obstacle analysis.

[0003] (1) Insufficient monitoring coverage

[0004] The current tree obstacle monitoring technology is limited by the low density of hardware deployment and poor adaptability to complex scenarios. UAVs have coverage blind spots when collecting point clouds in densely populated areas of distribution lines. Relying on manual supplementary measurements leads to low efficiency. The cost of high-precision lidar equipment is high, making it difficult to meet the full-coverage requirements of long-distance transmission lines.

[0005] (2) Lagging dynamic assessment ability

[0006] The existing tree obstacle assessment methods have a low update frequency and lack health monitoring. The three-dimensional modeling cycle of the transmission corridor is long, far lower than the annual growth of fast-growing tree species, resulting in a lag in risk assessment compared to actual changes. At the same time, the existing methods only focus on the geometric distance between the conductor and the tree, lacking dynamic monitoring of the health status such as pests, diseases, and loose roots, and unable to warn of sudden toppling risks.

[0007] (3) Low quality of line modeling

[0008] The extraction of line modeling depends on fixed-scale parameters for point cloud denoising, making it difficult to adapt to point cloud data with different densities and noise distributions, resulting in a risk of misjudgment in tree obstacle hazard assessment. For example, the point cloud characteristics in sparse and dense vegetation areas near transmission lines are significantly different, and fixed parameters are likely to cause loss of key information or residual noise. Summary of the Invention

[0009] The present invention aims to solve at least one of the technical problems existing in the prior art. For this reason, an object of the present invention is to propose a tree obstacle analysis method based on UAV images and line parametric modeling. This method aims to find a fast ground three-dimensional reconstruction and line modeling method to meet the timeliness requirements of tree obstacle analysis. It uses the images taken by UAVs for three-dimensional reconstruction and quickly realizes the analysis of obstacles under the line by parametrically constructing a line model, improving the efficiency and accuracy of tree obstacle analysis and processing.

[0010] To solve the above problems, the present invention provides a tree obstacle analysis method based on UAV images and line parametric modeling, including the following steps:

[0011] S1. UAV image point cloud reconstruction: Perform three-dimensional reconstruction through multi-angle aerial images to generate georegistered dense point cloud data;

[0012] S2. Parametric line modeling: Based on the collected line parameters, construct a line model including poles, conductors, and auxiliary facilities;

[0013] S3. Tree obstacle risk point analysis: By fusing the point cloud and the line model, calculate the height difference between the conductor and the vegetation point cloud, and identify the set of risk points with insufficient safety distance.

[0014] Preferably, S1 includes the following sub-steps:

[0015] S1.1. The UAV obtains multi-angle aerial images, records POS information, and imports the data into a computer for processing;

[0016] S1.2. Data preprocessing, creating a processing data set containing images and POS data;

[0017] S1.3. Feature extraction, using the DSPSIFT feature point extraction algorithm to extract feature points in the images and establish feature correspondence relationships;

[0018] S1.4. Sparse reconstruction, based on the OpenSFM algorithm to recover camera parameters and generate a sparse point cloud;

[0019] S1.5. Dense reconstruction, using the OpenMVS algorithm for dense point cloud reconstruction and performing voxel filtering for denoising;

[0020] S1.6. Georegistration, converting the model to a geographic coordinate system;

[0021] S1.7. Result output, outputting the point cloud data results.

[0022] Preferably, in S1.1, when the UAV obtains multi-angle aerial images, the shooting overlap rate requirements are as follows: the forward overlap rate requirement is ≥70% to prevent matching faults; the lateral overlap rate requirement is ≥60% to ensure multi-view coverage; multi-angle shooting needs to include orthophoto and oblique photography with an inclination angle of ≥45° to avoid model holes or elevation missing caused by a single angle.

[0023] Preferably, S2 includes the following sub-steps:

[0024] S2.1. Parameter collection, collecting the parameter information required for poles, auxiliary facilities, on-pole equipment, conductors, and line topology equipment;

[0025] S2.2. Construction of poles and towers and their ancillary facilities: Construct a model of poles and towers and their ancillary facilities according to the parameter information;

[0026] S2.3. Construction of on-pole equipment: Construct a model of on-pole equipment such as transformers, disconnect switches, and fuses according to the parameter information;

[0027] S2.4. Construction of conductors according to the line topology: Calculate the sag of the conductors according to the line topology connection relationship information and generate a line model;

[0028] S2.5. Model output: Output the line model data.

[0029] Preferably, S3 includes the following sub-steps:

[0030] S3.1. Data fusion: Use a point cloud display tool to fuse and load the point cloud data and the line model;

[0031] S3.2. Calculate the conductor fitting points: Calculate the set m (L, B, H) of conductor fitting points according to the requirements for collecting conductor monitoring points, where L represents longitude, B represents latitude, and H represents height;

[0032] S3.3. Calculate the corresponding points in the point cloud: According to the longitude and latitude information of the set m of conductor fitting points, use a spatial query method to obtain all points P (L, B, H) in the point cloud with a tolerance within 0.5 meters, and calculate the height H of the highest point using a single-round bubble sort method max , forming the set n (L, B, H) of the highest points in the corresponding point cloud, where L represents longitude, B represents latitude, and H represents height;

[0033] S3.4. Calculate the height difference: Subtract the set n from the set m to obtain the height difference set f;

[0034] The calculation formula is: Lm - Ln = Lf; Bm - Bn = Lf; Hm - Hn = Hf;

[0035] Where: Lm, Ln, and Lf are the longitudes of the elements in the sets m, n, and f respectively; Bm, Bn, and Bf are the latitudes of the elements in the sets m, n, and f respectively; Hm, Hn, and Hf are the heights of the elements in the sets m, n, and f respectively;

[0036] S3.5. Analyze the tree obstacle risk points: According to the safety distance D of the conductors at different voltage levels, calculate the set e in the set f that is less than D to obtain the tree obstacle risk point set;

[0037] The calculation formula is: Hf - D = He; where Hf and He are the heights of the elements in the sets f and e respectively.

[0038] Preferably, the calculation method of the conductor fitting points in S3.2 is:

[0039] Sag calculation formula: ;

[0040] Where: F is the sag, unit m; Fo is the sag of the representative span, unit m; Lo is the representative span, unit m; l is the observed span, unit m; α is the height difference angle, α = tan-1(h / l); h is the height difference between the suspension points of the conductor in the observed span, unit m;

[0041] The relationship between the sag F and the parameter a is: ;

[0042] The formula for solving a using the Newton-Raphson method is: ;

[0043] After determining a, a coordinate system is established with the midpoint of the span as the origin, and the catenary equation is:

[0044] ;

[0045] If there is a height difference, the height difference angle α is introduced, and the catenary equation is adjusted to:

[0046] .

[0047] The advantages of the present invention compared with the prior art are:

[0048] 1. The point cloud reconstruction is based on UAV images. The image acquisition method is convenient, without relying on special equipment. The point cloud update period is short and the frequency is high. The reconstruction effect can meet the requirements of quasi-real-time point cloud application scenarios.

[0049] 2. The line model is based on parametric modeling, with high model accuracy and small line sag deviation, solving the problem of poor extraction quality in traditional line information extraction based on point clouds.

[0050] 3. Based on the fact that the model parameters are obtained through on-site collection and the conductor is generated by data algorithms, it is simple to obtain the coordinates of any point on the conductor, and the efficiency and accuracy of tree obstacle analysis and processing are high. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0052] Figure 1 It is the flowchart of point cloud reconstruction processing based on UAV images in the present invention;

[0053] Figure 2It is the process flow chart of parametric line modeling for the present invention;

[0054] Figure 3 It is the process flow chart of tree obstacle risk point analysis in the present invention. Specific embodiments

[0055] The embodiments of the present application will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary only for explaining the present application and should not be construed as a limitation to the present application.

[0056] The present invention will be further described in detail below with reference to the accompanying drawings.

[0057] Combined with Figures 1 to 3 , the present invention provides a tree obstacle analysis method based on UAV images and line parametric modeling, which performs three-dimensional reconstruction based on the image data captured by UAV operators during daily inspection tasks, and realizes the process of rapid tree obstacle risk point analysis by parametrically constructing a line model.

[0058] To achieve the above object, a tree obstacle analysis method based on UAV images and line parametric modeling includes the following steps:

[0059] S1. UAV image point cloud reconstruction

[0060] S1.1. The UAV obtains multi-angle aerial images, records POS information (GPS coordinates, altitude, attitude angle), and imports the data into a computer for processing. The requirements for the shooting overlap rate are as follows: the forward overlap rate requirement is ≥70% to prevent matching faults; the side overlap rate requirement is ≥60% to ensure multi-view coverage. Multi-angle shooting needs to include orthophoto and oblique photography with an inclination angle ≥45° to avoid model holes or elevation missing caused by a single angle.

[0061] S1.2. Data preprocessing, creating a processing data set containing images and POS data.

[0062] S1.3. Feature extraction, using the DSPSIFT feature point extraction algorithm to extract feature points in the images and establish feature correspondence relationships.

[0063] S1.4. Sparse reconstruction, restoring camera parameters based on the OpenSFM algorithm to generate a sparse point cloud.

[0064] S1.5. Dense reconstruction, using the OpenMVS algorithm for dense point cloud reconstruction and performing voxel filtering for denoising.

[0065] S1.6. Geometric registration, converting the model to a geographic coordinate system.

[0066] S1.7, Output of results, output the point cloud data results.

[0067] S2, Parametric line modeling

[0068] S2.1, Parameter collection, collect the parameter information required for poles and towers, auxiliary facilities, on-pole equipment, conductors, and line topology equipment.

[0069] S2.2, Construction of poles and towers and auxiliary facilities, construct the models of poles and towers and auxiliary facilities according to the parameter information.

[0070] S2.3, Construction of on-pole equipment, construct the on-pole equipment models of transformers, disconnect switches, and fuses according to the parameter information.

[0071] S2.4, Construct conductors according to the line topology, calculate the sag of the conductors based on the line topology connection relationship information, and generate the line model.

[0072] S2.5, Model output, output the line model data.

[0073] S3, Analysis of tree obstacle risk points

[0074] S3.1, Data fusion, use point cloud display tools (such as Three.JS, Cesium.JS, etc.) to fuse and load the point cloud data and the line model.

[0075] S3.2, Calculate the conductor fitting points, calculate the set M (L, B, H) of conductor fitting points according to the requirements for collecting conductor monitoring points (for example: 2 collection points per meter), where L represents longitude, B represents latitude, and H represents height.

[0076] Sag calculation formula: F = Fo / cosα * (l / Lo) 2

[0077] Where: F is the sag (m), Fo is the sag of the representative span (m), Lo is the representative span (m), l is the observed span (m), α is the elevation angle, α = tan-1(h / l), and h is the elevation difference between the suspension points of the observed span of the conductor (m).

[0078] The relationship between the sag F and the parameter a is: ;

[0079] The formula for solving a using the Newton iteration method is: ;

[0080] After determining a, establish a coordinate system with the midpoint of the span as the origin, and the catenary equation is:

[0081] ;

[0082] If there is a height difference (unequal suspension points), introduce the height difference angle α and adjust the catenary equation as follows:

[0083] ;

[0084] S3.3. Calculate the corresponding points in the point cloud. According to the longitude and latitude information of the set m of the fitted points of the conductor, use the spatial query method to obtain all the points P (L, B, H) in the point cloud with a tolerance within 0.5 meters, and calculate the height H of the highest point (maximum value) by using a round of bubble sorting method (subtracting two by two, that is, Hp1 - Hp2, where Hp1 and Hp2 are the heights of the elements in the set P). max , forming the set N (L, B, H) of the highest points of the corresponding point cloud, where L represents longitude, B represents latitude, and H represents height.

[0085] S3.4. Calculate the height difference. Subtract the set n from the set m to obtain the height difference set f.

[0086] The calculation formula is: Lm - Ln = Lf; Bm - Bn = Lf; Hm - Hn = Hf;

[0087] Where: Lm, Ln, and Lf are the longitudes of the elements in the sets m, n, and f respectively; Bm, Bn, and Bf are the latitudes of the elements in the sets m, n, and f respectively; Hm, Hn, and Hf are the heights of the elements in the sets m, n, and f respectively.

[0088] S3.5. Analyze the tree obstacle risk points. According to the safety distance D of the conductors with different voltage levels, calculate the set e in the set f that is less than D, that is, the set of tree obstacle risk points.

[0089] The calculation formula is: Hf - D = He;

[0090] Where Hf and He are the heights of the elements in the sets f and e respectively.

[0091] In addition, the implementation parameters of the preferred embodiment of the present invention are as follows:

[0092] 1. The drone uses DJI M300RTK, is equipped with a P1 camera (35mm lens), and the flight height is 80 meters;

[0093] 2. OpenMVS 1.0 is used for point cloud reconstruction, and the voxel filtering parameter is set to 0.03 meters;

[0094] 3. The improved catenary equation is used for calculating the sag of the conductor: ; where H is the horizontal tension and w is the weight per unit length;

[0095] 4. The safety distance standard refers to "DL / T 741-2019", and the minimum vertical distance of 4.0 meters is set for the 110kV line.

[0096] Verified by actual measurement in a certain provincial power grid: By adopting the method of the present invention, the modeling period is shortened from 14 days to 72 hours; the accuracy rate of tree obstacle recognition reaches 98.7% (the traditional method is 82.3%); the false alarm rate is reduced to 1.2% (the lidar solution is 5.8%).

[0097] Finally, for the parts not described in the present invention, mature products and mature technical means in the prior art are adopted.

[0098] The present invention and its implementation manners have been described above. Such description is not restrictive. What is shown in the drawings is only one of the implementation manners of the present invention, and the actual structure is not limited thereto. Generally speaking, if those of ordinary skill in the art are inspired by it and design similar structural manners and embodiments to this technical solution without creative work without departing from the purpose of the present invention, they shall fall within the protection scope of the present invention.

Claims

1. A tree obstacle analysis method based on UAV images and line parametric modeling, characterized in that, It includes the following steps: S1. UAV image point cloud reconstruction: Perform three-dimensional reconstruction through multi-angle aerial images to generate georegistered dense point cloud data; S2. Parametric line modeling: Based on the collected line parameters, construct a line model including poles, conductors, and ancillary facilities; S3. Tree obstacle risk point analysis: By fusing the point cloud and the line model, calculate the height difference between the conductor and the vegetation point cloud, and identify the set of risk points with insufficient safety distance.

2. The tree obstacle analysis method based on UAV images and line parameterized modeling according to claim 1, characterized in that: The S1 includes the following sub-steps: S1.

1. The UAV acquires multi-angle aerial images, records POS information, and imports the data into a computer for processing; S1.

2. Data preprocessing, creating a processing data set containing images and POS data; S1.

3. Feature extraction, using the DSPSIFT feature point extraction algorithm to extract feature points in the images and establish feature correspondence relationships; S1.

4. Sparse reconstruction, based on the OpenSFM algorithm, recover the camera parameters and generate a sparse point cloud; S1.

5. Dense reconstruction, using the OpenMVS algorithm for dense point cloud reconstruction and performing voxel filtering for denoising; S1.

6. Georegistration, converting the model to a geographic coordinate system; S1.

7. Result output, outputting the point cloud data result.

3. The tree obstacle analysis method based on UAV images and line parametric modeling according to claim 2, characterized in that: In the S1.1, when the UAV acquires multi-angle aerial images, the shooting overlap rate requirements are as follows: the forward overlap rate requirement is ≥70% to prevent matching faults; the side overlap rate requirement is ≥60% to ensure multi-view coverage; multi-angle shooting needs to include orthophoto and oblique photography with an inclination angle ≥45° to avoid model holes or elevation missing caused by a single angle.

4. The tree obstacle analysis method based on UAV images and line parametric modeling according to claim 1, wherein: The S2 includes the following sub-steps: S2.

1. Parameter collection, collecting the parameter information required for poles, ancillary facilities, on-pole equipment, conductors, and line topology equipment; S2.

2. Pole and ancillary facility construction, constructing pole and ancillary facility models according to the parameter information; S2.

3. On-pole equipment construction, constructing on-pole equipment models of transformers, disconnectors, and fuses according to the parameter information; S2.

4. Construct conductors according to the line topology, calculate the conductor sag according to the line topology connection relationship information, and generate a line model; S2.

5. Model output, outputting the line model data.

5. A tree obstacle analysis method based on UAV images and line parameterized modeling according to claim 1, characterized in that: The S3 includes the following sub-steps: S3.

1. Data fusion, using a point cloud display tool to fuse and load the point cloud data and the line model; S3.

2. Calculate the conductor fitting points, calculate the set m (L, B, H) of conductor fitting points according to the conductor monitoring point collection requirements, where L represents longitude, B represents latitude, and H represents height; S3.

3. Calculate the corresponding points in the point cloud. According to the longitude and latitude information of the set m of wire-fitting points, use the spatial query method to obtain all points P(L, B, H) in the point cloud with a tolerance within 0.5 meters, and calculate the height H of the highest point by means of a round of bubble sort. max , to form a set n(L, B, H) of the highest points in the corresponding point cloud, where L represents longitude, B represents latitude, and H represents height. S3.

4. Calculate the height difference, subtract the set n from the set m to obtain the height difference set f; The calculation formula is: Lm - Ln = Lf; Bm - Bn = Lf; Hm - Hn = Hf; Where: Lm, Ln, Lf are the longitudes of the elements in the sets m, n, f respectively; Bm, Bn, Bf are the latitudes of the elements in the sets m, n, f respectively; Hm, Hn, Hf are the heights of the elements in the sets m, n, f respectively; S3.

5. Analyze the tree obstacle risk points, calculate the set e in the set f that is less than the safety distance D of the conductors of different voltage levels to obtain the set of tree obstacle risk points; The calculation formula is: Hf - D = He; where Hf and He are the heights of the elements in sets f and e respectively.

6. The tree obstacle analysis method based on UAV images and line parametric modeling according to claim 5, characterized in that: The calculation method of the wire fitting points in S3.2 is as follows: Sag calculation formula: ; Where: F is the sag, unit m, Fo is the sag of the representative span, unit m, Lo is the representative span, unit m, l is the observed span, unit m, α is the elevation angle, α = tan-1(h / l), h is the elevation difference between the suspension points of the observed span wire, unit m; The relationship between the sag F and the parameter a is as follows: ; The formula for solving a using the Newton-Raphson method is: ; After determining a, a coordinate system is established with the midpoint of the span as the origin, and the catenary equation is: ； If there is an elevation difference, introduce the elevation angle α and adjust the catenary equation to: 。

Images