Obstacle framing method based on laser point cloud in converter station scene

Abstract

The invention relates to the technical field of obstacle detection, in particular to a laser point cloud-based obstacle framing method in a converter station scene, which comprises the following steps of: based on a converter station environment, performing three-dimensional scanning through a laser radar, collecting point cloud data, performing noise filtering on the data, performing spatial alignment of the data, and obtaining processed point cloud data. According to the invention, three-dimensional scanning is carried out through the laser radar, the scanning angle and the resolution are adjusted, the coverage rate and the precision of the point cloud data are optimized, the point cloud data are mapped to the adjacent grid center, the data structure is simplified, the point cloud data are horizontally cut, the linear features of the electric wire are identified, and the obstacle identification process is optimized. Through combination of spatial statistical analysis and clustering analysis, the accuracy and the meticulous degree of obstacle recognition are improved, and the obstacles are framed by constructing a bounding box of each obstacle, so that three-dimensional space positioning and form recognition of the obstacles are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of obstacle detection, and in particular to an obstacle framing method based on laser point cloud in a converter station scenario. Background Art

[0002] The field of obstacle detection technology focuses on identifying, locating, and classifying objects in the environment through various sensor systems. Obstacle detection technology is widely used in autonomous vehicles, drones, robot navigation, and security monitoring systems. Key technologies include image and signal processing, machine learning algorithms, and sensor fusion technology. Obstacle detection involves not only extracting features from sensor data, but also real-time analysis using methods such as deep learning to improve the accuracy and reliability of detection.

[0003] Among them, the obstacle framing method based on laser point cloud in the converter station scenario involves using laser point cloud technology to detect and calibrate the position of obstacles in the converter station environment. Laser point cloud is a large amount of accurate spatial point data collected by laser scanners, and the data can construct a three-dimensional digital model of the environment. In the converter station, this technology is used to identify obstacles that affect drone inspections, including equipment, pipelines, and wires in the converter station. By identifying and framing obstacles, it is ensured that drones can conduct inspections in the complex environment of the converter station, thereby improving the operational efficiency of the converter station.

[0004] Traditional methods rely on preset parameters and fixed algorithms, and are unable to cope with the challenges posed by the diversity of equipment, pipelines, and wire layouts inside converter stations. Traditional methods are limited in data processing efficiency by the complexity of the algorithms and computing resources, and suffer from low data processing efficiency and difficulty in accurately locating small-scale obstacles. Common problems include data overload, where a large amount of unoptimized point cloud data collected by sensors can easily lead to processing bottlenecks. Traditional methods are not sophisticated enough in the classification and specific attribute analysis of obstacles, and are unable to effectively distinguish different types of obstacles, which can easily lead to misjudgments or omissions, increasing the risk in the automated inspection process. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings existing in the prior art and propose an obstacle framing method based on laser point cloud in a converter station scenario.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a method for framing obstacles based on laser point cloud in a converter station scenario, comprising the following steps: S1: Based on the converter station environment, three-dimensional scanning is performed through the laser radar to collect point cloud data, and the data is filtered for noise and spatially aligned to obtain processed point cloud data; S2: Based on the processed point cloud data, each point in the data is mapped to the center of an adjacent grid, the coordinate average of each grid point is calculated, and the resulting point is set as a grid representative to obtain simplified point cloud data; S3: Based on the simplified point cloud data, filter the point cloud data that touches the ground, create a simulated cloth model, invert the point cloud data, adjust the position of the simulated cloth to fit the ground represented by the point cloud, segment the ground point cloud and the non-ground point cloud, and obtain ground segmentation data; S4: Based on the ground segmentation data, the point cloud data is horizontally cut, each layer of the point cloud is projected onto a two-dimensional plane, and the linear features of the wires on each layer are identified to obtain wire identification point cloud data; S5: Based on the wire recognition point cloud data, measure the density between the point clouds, mark densely clustered areas as potential obstacles, classify the obstacles, and obtain obstacle classification labels; S6: Based on the obstacle classification label, calculate the geometric center point and the circumscribed rectangular bounding box of each marked obstacle, construct a bounding box of each obstacle, and obtain an obstacle framing result.

[0007] As a further solution of the present invention, the processed point cloud data includes the position coordinates, reflection intensity and color information of the point cloud, the simplified point cloud data includes the voxelized point cloud, the average position point inside each voxel and the point cloud model with reduced spatial resolution after reconstruction, the ground segmentation data includes the ground point cloud marked after simulating the bonding of cloth, the non-ground point cloud part and the threshold value between the ground and the non-ground, the wire identification point cloud data includes the identified wire feature points and the point cloud segments of the wire, the obstacle classification label includes the divided obstacle category and the corresponding obstacle clustering mark, and the obstacle framing result includes the three-dimensional coordinates, length, width and height dimensions and obstacle volume size of each obstacle bounding box.

[0008] As a further solution of the present invention, based on the converter station environment, three-dimensional scanning is performed by laser radar to collect point cloud data, and noise is filtered out of the data, and spatial alignment of the data is performed to obtain the processed point cloud data. Specifically, the steps are as follows: S101: Based on the converter station environment, the scanning angle and resolution of the laser radar are adjusted to perform a panoramic three-dimensional scan of the converter station to obtain a basic point cloud data set; S102: Based on the basic point cloud data set, by analyzing the density and reflection characteristics of the point cloud, and filtering the data for noise, identifying and eliminating data points with abnormal reflection intensity and positions deviating from the normal, to obtain point cloud data with noise filtered out; S103: Based on the point cloud data from which noise has been filtered out, coordinates of data points are adjusted to eliminate offsets caused by scanning, and continuity of the point cloud data is verified through continuity detection to obtain processed point cloud data.

[0009] As a further solution of the present invention, based on the processed point cloud data, each point in the data is mapped to the adjacent grid center, the coordinate average of each grid point is calculated, and the resulting point is set as a grid representative. The steps of obtaining simplified point cloud data are specifically as follows: S201: performing spatial compression based on the processed point cloud data, and mapping the point cloud data to a preset spatial grid according to the spatial position of each point to obtain grid-mapped point cloud data; S202: Based on the grid mapping point cloud data, calculating the coordinate average value in the point cloud set of each grid to obtain a grid coordinate average value calculation result; S203: Based on the calculation result of the grid coordinate average value, the coordinate average value is set as the representative point of each grid to obtain simplified point cloud data.

[0010] As a further solution of the present invention, based on the simplified point cloud data, the point cloud data contacting the ground is screened, a simulated cloth model is created, the point cloud data is inverted, and the position of the simulated cloth is adjusted to fit the ground represented by the point cloud, and the ground and non-ground point clouds are segmented to obtain the ground segmentation data. Specifically, the steps are as follows: S301: Based on the simplified point cloud data, by analyzing the difference between the Z coordinate of each point and a predetermined ground height, filtering the data points in contact with the ground, identifying a point set close to or equal to the ground height, and obtaining ground point cloud information; S302: Based on the ground point cloud information, simulate the physical properties of the cloth, simulate the covering behavior of the cloth in the virtual environment by setting the elastic coefficient and grid size of the cloth model, adjust the cloth to adapt to the terrain, and obtain a simulated cloth configuration; S303: Based on the simulated cloth configuration, the cloth is matched with the inverted point cloud, the position of the cloth model is adjusted, the cloth model is fitted to the inverted terrain features, and the ground and non-ground areas are segmented through the contact points between the cloth and the terrain to obtain ground segmentation data.

[0011] As a further solution of the present invention, based on the ground segmentation data, the point cloud data is horizontally cut, each layer of the point cloud is projected onto a two-dimensional plane, and the linear features of the wires on each layer are identified. The steps of obtaining the wire identification point cloud data are specifically as follows: S401: Based on the ground segmentation data, the entire point cloud data set is subjected to height interval layering processing, each layer has a fixed height, and the point cloud is divided into multiple horizontal layers according to a preset interval to obtain a layer segmentation point cloud; S402: Based on the layer segmentation point cloud, perform two-dimensional plane projection on the point cloud of each horizontal layer, draw the XY coordinates of the point cloud on the two-dimensional plane, and obtain two-dimensional projection data; S403: Based on the two-dimensional projection data, linear feature detection technology is used to analyze the linear arrangement characteristics of the projection graphics, screen the projection point cloud with linear characteristics, mark the point set whose linear arrangement meets the characteristics of the wire, and obtain the wire identification point cloud data.

[0012] As a further solution of the present invention, the linear feature detection technology is according to the formula: ; Calculate the linear feature score of the point cloud data, where represents the length of the longest linear feature in the projected data, represents the average distance between adjacent features, represents the standard deviation in the direction of the linear feature, , and is the adjustment coefficient, Represents the linear feature scores of point cloud data.

[0013] As a further solution of the present invention, based on the wire recognition point cloud data, the density between the point clouds is measured, densely clustered areas are marked as potential obstacles, and the obstacles are classified to obtain obstacle classification labels. Specifically, the steps are as follows: S501: Based on the wire identification point cloud data, using the kernel density estimation method, perform spatial statistical analysis on each point cloud block, calculate the number and distribution of points inside the point cloud block, identify the density of each point cloud area, and generate point cloud density information; S502: Based on the point cloud density information, by comparing with a preset density reference, a point cloud area exceeding the preset density reference is identified as a potential obstacle, and obstacle marked area data is obtained; S503: Based on the marked obstacle area data, analyze the color information of the point cloud area, and classify the obstacles into corresponding types, including metal, wood and plastic, according to the color characteristics of the point cloud, to obtain obstacle classification labels.

[0014] As a further solution of the present invention, the kernel density estimation method is according to the formula: ; Calculate point cloud density ,in, represents the density of the point cloud, Represents the number of points in the point cloud block, Representative The location of the point, is the average position of the point cloud block, is the standard deviation of the point positions, is the base of natural logarithms, is the ratio of pi.

[0015] As a further solution of the present invention, based on the obstacle classification label, the geometric center point and the circumscribed rectangular bounding box of each marked obstacle are calculated, and the bounding box of each obstacle is constructed to obtain the obstacle framing result. Specifically, the steps are as follows: S601: Based on the obstacle classification label, the geometric center of the cluster is identified by calculating the average value of the coordinates of each cluster point cloud, and the center point is used as a representative of the obstacle position to obtain the coordinates of the obstacle center point; S602: Based on the coordinates of the obstacle center point, measure the distance from the farthest point of each cluster to the center point, identify the size and position of the circumscribed rectangular bounding box of each obstacle, and obtain obstacle bounding box parameters; S603: Based on the obstacle bounding box parameters, a three-dimensional bounding box of each obstacle is constructed through three-dimensional modeling, each obstacle is covered by the bounding box, and the position and shape of the bounding box in space are recorded to obtain an obstacle framing result.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, three-dimensional scanning is performed by laser radar, and the scanning angle and resolution are adjusted to optimize the coverage and accuracy of point cloud data. The data availability is enhanced by optimizing point cloud data processing, such as noise filtering and spatial alignment. The point cloud data is mapped to adjacent grid centers, and the coordinate average is calculated to simplify the data structure, making subsequent data processing more efficient. The point cloud data in contact with the ground is screened, and a simulated cloth model is used to segment ground and non-ground point clouds, horizontally cut point cloud data and identify linear features of wires, and optimize the obstacle recognition process. The accuracy and meticulousness of obstacle recognition are improved by combining spatial statistical analysis with cluster analysis, and the obstacle is framed by constructing a bounding box for each obstacle, thereby realizing three-dimensional spatial positioning and morphological recognition of the obstacle, ensuring that the UAV can safely and effectively avoid obstacles when performing tasks, and enhancing the safety and efficiency of converter station inspections. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the main steps of the present invention; Figure 2 It is a schematic diagram of the refinement of S1 of the present invention; Figure 3 It is a schematic diagram of the refinement of S2 of the present invention; Figure 4 It is a schematic diagram of the refinement of S3 of the present invention; Figure 5 It is a schematic diagram of the refinement of S4 of the present invention; Figure 6 It is a detailed schematic diagram of S5 of the present invention; Figure 7 It is a detailed schematic diagram of S6 of the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0019] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0020] See also Figure 1 The present invention provides a technical solution: a method for framing obstacles based on laser point cloud in a converter station scenario, comprising the following steps: S1: Based on the converter station environment, three-dimensional scanning is performed through the laser radar, and the scanning angle and resolution are adjusted to optimize the coverage and accuracy of the point cloud data. The point cloud data is collected and noise is filtered out of the data to remove abnormal points caused by scanning errors, and the coordinate offset of the scanned data is corrected. The data is spatially aligned to optimize the continuity between point clouds to obtain the processed point cloud data. S2: Based on the processed point cloud data, the point cloud data is spatially compressed, each point in the data is mapped to the adjacent grid center, the coordinate average of each grid point is calculated, and the resulting point is set as the grid representative to obtain simplified point cloud data; S3: Based on the simplified point cloud data, the point cloud data touching the ground is screened, a simulated cloth model is created, the point cloud data is inverted, the position of the simulated cloth is adjusted to fit the ground represented by the point cloud, the ground point cloud and the non-ground point cloud are segmented, and the ground segmentation data is obtained; S4: Based on the ground segmentation data, the point cloud data is horizontally cut, and the point cloud is divided into multiple horizontal layers according to the preset height interval. Each layer of the point cloud is projected onto a two-dimensional plane, and the linear features of the wires on each layer are identified. The point cloud of the wire part is marked to obtain the wire identification point cloud data; S5: Based on the wire recognition point cloud data, the density between point clouds is measured through spatial statistical analysis, and cluster analysis is performed using the density benchmark to mark densely clustered areas as potential obstacles. The obstacles are classified according to the color characteristics of the point cloud to obtain obstacle classification labels; S6: Based on the obstacle classification label, the geometric center point and circumscribed rectangular bounding box of each marked obstacle are calculated. Through three-dimensional space modeling, the bounding box of each obstacle is constructed to locate and characterize the spatial position and shape of the obstacle to obtain the obstacle framing result.

[0021] The processed point cloud data includes the position coordinates, reflection intensity and color information of the point cloud. The simplified point cloud data includes the voxelized point cloud, the average position point inside each voxel and the point cloud model with reduced spatial resolution after reconstruction. The ground segmentation data includes the ground point cloud marked after simulating cloth fitting, the non-ground point cloud part and the threshold between ground and non-ground. The wire recognition point cloud data includes the identified wire feature points and the point cloud segments of the wire. The obstacle classification label includes the divided obstacle category and the corresponding obstacle clustering mark. The obstacle framing result includes the three-dimensional coordinates, length, width and height dimensions and obstacle volume size of each obstacle bounding box.

[0022] See also Figure 2 Based on the converter station environment, three-dimensional scanning is performed by laser radar, and the scanning angle and resolution are adjusted to optimize the coverage and accuracy of the point cloud data. The point cloud data is collected and noise is filtered out to remove abnormal points caused by scanning errors. The coordinate offset of the scanned data is corrected, and the spatial alignment of the data is performed to optimize the continuity between the point clouds. The specific steps to obtain the processed point cloud data are as follows: S101: Based on the converter station environment, the scanning angle and resolution of the laser radar are adjusted to perform a panoramic three-dimensional scan of the converter station to obtain a basic point cloud data set. The specific process is as follows; Based on the converter station environment, the laser radar scanning parameters are set, including adjusting the laser radar scanning angle to a suitable range to cover the full range of the converter station. At the same time, the scanning resolution is set to ensure high-precision data capture. Use high-precision laser radar for three-dimensional scanning to cover every corner of the converter station and obtain a basic point cloud data set. During this process, the laser radar repeatedly scans along a preset path to ensure the integrity and accuracy of the data, and records the environmental conditions of each scan, such as temperature and humidity, for correction in subsequent data analysis.

[0023] S102: Based on the basic point cloud data set, by analyzing the density and reflection characteristics of the point cloud, and filtering the data for noise, identifying and removing data points with abnormal reflection intensity and positions deviating from the normal, the process of obtaining the point cloud data after noise is specifically as follows; Based on the obtained basic point cloud data set, the point cloud density distribution map is used to analyze the data point density of each area. By setting the density threshold, the low-density area is identified and excluded. The low-density area usually has incomplete data due to the scanning angle or occlusion. The reflection characteristics of the point cloud are analyzed, and the reflection intensity threshold is set. For example, if it is greater than 800 units, it is a high reflection intensity, and the point is marked as abnormal. The entire point cloud is processed using a Gaussian blur filter to remove environmental noise and errors caused by the equipment. For data points whose positions deviate from the norm, their spatial relationship with the surrounding points is compared. Data points that exceed the preset spatial distance threshold (for example, 5 cm) are eliminated to obtain clean point cloud data after noise filtering.

[0024] S103: Based on the point cloud data after filtering out noise, the coordinates of the data points are adjusted to eliminate the offset caused by scanning, and the continuity of the point cloud data is verified through continuity detection to obtain the processed point cloud data. Specifically, the process is as follows; Based on the point cloud data after noise filtering, the data coordinates are adjusted. According to the position and scanning angle information of the laser radar, the exact position of each data point in the global coordinate system is calculated, and the coordinates of the data points are adjusted using the transformation matrix to eliminate the coordinate deviation caused by the movement or rotation of the scanning device. The continuity test is performed by establishing a distance matrix between points and analyzing the distance between each point and its nearest neighbor. For example, if the distance is less than 3 cm, the data is considered continuous, and the continuity of the point cloud data is verified. Ensure that there are no obvious jumps or breaks between data points, ensure the accurate reconstruction of the converter station model, and form processed point cloud data.

[0025] See also Figure 3 Based on the processed point cloud data, the point cloud data is spatially compressed, each point in the data is mapped to the adjacent grid center, the coordinate average of each grid point is calculated, and the resulting point is set as the grid representative. The specific steps to obtain simplified point cloud data are: S201: Based on the processed point cloud data, spatial compression is performed, and the point cloud data is mapped to a preset spatial grid according to the spatial position of each point, and the specific process of obtaining the grid-mapped point cloud data is as follows; Based on the processed point cloud data, a three-dimensional spatial grid is designed according to the actual structure and spatial characteristics of the converter station. The size and shape of the grid are preset to adapt to the spatial distribution of each part of the converter station. For example, the volume of each grid can be set to 3 cubic centimeters to ensure sufficient resolution and processing efficiency. The point cloud data is mapped to these preset spatial grids, and each data point is assigned to the corresponding grid according to its coordinates. Through this mapping, the original point cloud data is effectively compressed, and the point cloud set within each grid reflects the geometric and spatial characteristics of the area, obtaining grid-mapped point cloud data, while recording the original reflection characteristics and color information of each point to ensure the integrity of the data after compression.

[0026] S202: Based on the grid-mapped point cloud data, the coordinate average value in the point cloud set of each grid is calculated to obtain the grid coordinate average value calculation result.

[0027] Based on the grid-mapped point cloud data, the coordinate average of each grid's point cloud set is calculated according to the formula:

[0028] and

[0029] and ; The average value of the grid coordinates is calculated. , , represents the average value of x, y, and z coordinates. Represents the total number of points in the point cloud set, , , Representative The x, y, z coordinates of a point.

[0030] Detailed explanation of the formula and the process of formula calculation and derivation: In point cloud data processing, calculating the average value of coordinates is a basic operation used to determine the center position or mean position of all points in each grid. For each grid, set its content points, each with three-dimensional coordinates , , By summing up the coordinates of all points in the same dimension and dividing by the total number of points , the average coordinate of this dimension can be calculated. The process is performed separately for the three coordinate dimensions x, y, and z.

[0031] To demonstrate the calculation, set up a grid containing the following points: Point 1: (2,3,5).

[0032] Point 2: (4,1,6).

[0033] Point 3: (3,4,4).

[0034] Compute the mean of the x-coordinates: ; Similarly, the average value of the y coordinate is: ; The average value of the z-coordinate: ; The procedure shows how to use mathematical operations to determine the center or mean location of the point cloud data in each grid.

[0035] S203: Based on the calculation result of the average value of the grid coordinates, the coordinate average value is set as the representative point of each grid, and the process of obtaining simplified point cloud data is specifically as follows; Based on the coordinate average calculation results of each grid, the point cloud data is simplified. The coordinate average of each grid is set as the representative point of the grid, so that each grid only retains the data of one point and removes other points, reducing the size of the data set while retaining key structural information. Simplified point cloud data facilitates more advanced analysis and processing, such as modeling and visualization. In the simplified point cloud data, the selection of each grid representative point is based on the geometric centrality and the statistical importance of the data point to ensure the representativeness and accuracy of the data.

[0036] See also Figure 4 Based on the simplified point cloud data, the point cloud data touching the ground is screened, a simulated cloth model is created, the point cloud data is inverted, and the position of the simulated cloth is adjusted to fit the ground represented by the point cloud, and the ground and non-ground point clouds are segmented to obtain the ground segmentation data. The specific steps are as follows: S301: Based on the simplified point cloud data, by analyzing the difference between the Z coordinate of each point and the predetermined ground height, filtering the data points in contact with the ground, identifying the point set close to and equal to the ground height, and obtaining the ground point cloud information in detail; Based on the simplified point cloud data, the ground point cloud is identified and screened. For example, the predetermined ground height is defined as 0 cm. Based on this, the Z coordinate of each data point is checked and the height difference with the ground is calculated. A height difference threshold, such as ±4 cm, is set to identify all points close to the ground height. All data points that are in contact with or almost in contact with the ground are screened out to form a new point cloud dataset, namely the ground point cloud information. The dataset includes all points on the ground or near the ground. The selection of points is based on the comparative analysis of the Z coordinate and the predetermined ground height to ensure that only points that are really close to the ground are selected.

[0037] S302: Based on the ground point cloud information, simulate the physical properties of the cloth, simulate the covering behavior of the cloth in the virtual environment by setting the elastic coefficient and grid size of the cloth model, adjust the cloth to adapt to the terrain, and obtain the simulated cloth configuration process. Based on the obtained ground point cloud information, modeling software is used to simulate the physical properties of cloth to adapt to the undulations of the terrain. The elastic coefficient of the cloth model is set, for example, to 0.05 Newtons per square meter to simulate the elastic behavior of real cloth. The size of the cloth grid is set to 5 cm × 5 cm to ensure that the model can accurately cover the complex terrain. Simulate in a virtual environment and adjust the cloth to make it naturally cover the undulations of the terrain. By gradually adjusting the position and shape of the cloth grid, the cloth model is made to fit the terrain perfectly. The simulated behavior of the cloth reflects the real physical properties, such as stretching and compression, to ensure the practicality and accuracy of the simulation results.

[0038] S303: Based on the simulated cloth configuration, the cloth is matched with the inverted point cloud, the position of the cloth model is adjusted, the cloth model is fitted to the inverted terrain features, and the ground and non-ground areas are segmented through the contact points between the cloth and the terrain to obtain the ground segmentation data. The specific process is as follows; S302: simulation and configuration process of the cloth model.

[0039] Based on the simulated cloth configuration, the cloth model is matched to the inverted terrain features. The terrain point cloud data is inverted, that is, the Z coordinate value of each point is reversed to simulate the natural behavior of cloth drooping. The position of the cloth model is adjusted to ensure that the cloth can accurately cover the inverted terrain features. Through this matching, the points where the cloth touches the terrain are used to segment the ground and non-ground areas. At each contact point, the contact between the cloth and the terrain is accurately recorded, and these contact points are used as boundaries to clearly separate the ground area from the non-ground area. The accuracy and practicality of the ground segmentation data are ensured through geometric and physical matching.

[0040] See also Figure 5 Based on the ground segmentation data, the point cloud data is horizontally cut, and the point cloud is divided into multiple horizontal layers according to the preset height interval. Each layer of point cloud is projected onto a two-dimensional plane, and the linear features of the wires on each layer are identified. The point cloud of the wire part is identified, and the specific steps for obtaining the wire recognition point cloud data are as follows: S401: Based on the ground segmentation data, the entire point cloud data set is subjected to height interval layering processing, each layer has a fixed height, and the point cloud is divided into multiple horizontal layers according to a preset interval, and the specific process of obtaining the layer segmentation point cloud is as follows; Based on the ground segmentation data, the point cloud dataset is layered at height intervals. For example, the fixed height of each layer is set to 10 cm, and the entire point cloud dataset is layered according to the standard. Use a preset height interval, for example, starting from a ground height of 0 cm, and setting each increase of 10 cm as a layer until the highest point is reached. In this way, the entire point cloud dataset is divided into multiple horizontal layers, each layer contains all the point cloud data within the height range, and the point cloud data is organized into multiple independent layers, each layer evenly covers a specific height area, and a layer-segmented point cloud is obtained. Such layering helps with subsequent detailed analysis and data processing.

[0041] S402: Based on the layer segmentation point cloud, the point cloud of each horizontal layer is projected on a two-dimensional plane, and the XY coordinates of the point cloud are plotted on the two-dimensional plane to obtain the two-dimensional projection data. Specifically, the process is as follows; Based on the obtained layer segmentation point cloud, the point cloud of each horizontal layer is projected into a two-dimensional plane. The XY coordinates of each point are selected and these coordinates are plotted on a two-dimensional plane. In the process, the Z coordinate of each point is ignored and only the horizontal distribution is considered. In this way, each layer of point cloud is converted into a two-dimensional graph to form two-dimensional projection data. The two-dimensional graph reflects the distribution characteristics of each layer of point cloud on the horizontal plane, which is convenient for further graphic analysis and feature extraction.

[0042] S403: Based on the two-dimensional projection data, linear feature detection technology is used to analyze the linear arrangement characteristics of the projection graphics, screen the projection point cloud with linear characteristics, mark the point set whose linear arrangement meets the characteristics of the wire, and obtain the wire recognition point cloud data.

[0043] Linear feature detection technology, according to the formula: ; Calculate the linear feature score of the point cloud data, where represents the length of the longest linear feature in the projected data, represents the average distance between adjacent features, represents the standard deviation in the direction of the linear feature, , and is an adjustment factor used to balance the effects of length, spacing, and direction deviation on the overall recognition result. The comprehensive consideration of parameters helps to accurately define which point cloud data belongs to wires, thereby improving selectivity and accuracy. Adjust the weight of the longest linear feature length in recognition, The effect of adjusting the spacing between adjacent features on the results, It is used to balance the impact of direction deviation on recognition accuracy. A linear feature score representing the point cloud data, used to determine which point cloud collections have significant linear arrangements that match the characteristics of wires. By calculating this score, wires can be accurately distinguished from other nonlinear structures.

[0044] formula: ; Detailed explanation of the formula and the process of formula calculation and derivation: In the formula, A linear feature score representing the point cloud data, used to determine which point cloud collections have significant linear arrangements that match the characteristics of wires. By calculating this score, wires can be accurately distinguished from other nonlinear structures.

[0045] Represents the longest linear feature length observed in the projected data. For example, through point cloud data analysis, the longest straight length of a potential wire feature is determined to be 10 meters.

[0046] is the average spacing between adjacent features, which characterizes the spatial distribution between linear features in wire detection. For example, the average spacing between wires calculated from point cloud data is 0.5 meters.

[0047] Represents the standard deviation of the linear feature direction, which measures the consistency of the linear feature in direction. For example, through directional analysis, the standard deviation of the wire direction is 5 degrees.

[0048] , and are adjustment coefficients, for example, they are set to 0.7, 0.2, and 0.1 respectively. The selection of the coefficients is based on the relative importance of linear features in wire detection. Higher, because the longest feature length is the most critical in recognition; and is low, indicating that spacing and orientation biases, while important, have little direct impact on the results.

[0049] Formula calculation derivation process: , , , rice, rice, Spend.

[0050] Substitute into the formula to calculate: ; Calculated linear feature score The score is 9.07, indicating that the strength of the identified linear features is high and has typical characteristics of wires. Based on this score, this point cloud data can be marked as wires.

[0051] See also Figure 6 Based on the wire recognition point cloud data, the density between point clouds is measured through spatial statistical analysis, and cluster analysis is performed using density benchmarks. The densely clustered areas are marked as potential obstacles, and obstacles are classified according to the color characteristics of the point clouds. The specific steps to obtain obstacle classification labels are as follows: S501: Based on the wire recognition point cloud data, use the kernel density estimation method to perform spatial statistical analysis on each point cloud block, calculate the number and distribution of points inside the point cloud block, identify the density of each point cloud area, and generate point cloud density information.

[0052] Kernel density estimation method, according to the formula: ; Calculate point cloud density ,in, represents the density of the point cloud, Represents the number of points in the point cloud block, Representative The location of the point, is the average position of the point cloud block, is the standard deviation of the point positions, is the base of natural logarithms, is the ratio of pi.

[0053] formula: ; Detailed explanation of the formula and the process of formula calculation and derivation: : is the number of points in the point cloud block, directly counted from the dataset. For example, 200 points are counted in a particular point cloud block.

[0054] : The position of each point in the point cloud is usually defined by three-dimensional space coordinates, for example, It may be the 3D coordinate values ​​directly obtained from the laser scanning device.

[0055] : The average position of the point cloud block, calculated by taking the arithmetic mean of the coordinates of all points. For example, if all points The coordinate values ​​are summed and divided by the number of points , get the average Coordinate value.

[0056] : The standard deviation of the point position, calculated by taking the square root of the average of the squares of the deviations of each point's coordinates from the mean position. This indicates how dispersed the points in the point cloud are relative to the mean position.

[0057] Specific calculation example: , average position m, standard deviation Meters, point location Ranging from 12 meters to 18 meters.

[0058] Calculation process: ; To simplify the calculation process, use and , calculate the two extreme cases : ;

[0059] ;

[0060] Approximately calculated: ; The results show that the point cloud density yes , which is a relatively low density value, indicating that the points inside this point cloud area are relatively scattered and may not belong to wires or other densely arranged linear structures.

[0061] S502: Based on the point cloud density information, by comparing with a preset density reference, a point cloud area exceeding the preset density reference is identified as a potential obstacle, and the process of obtaining the marked obstacle area data is specifically as follows; Based on the generated point cloud density information, by comparing it with the preset density benchmark, the point cloud areas that exceed this density benchmark are identified as potential obstacles. The process includes comparing the density of each point cloud block with the benchmark value, identifying all areas that exceed the standard, and marking the area as a potential obstacle area. This marking helps to quickly identify and respond to possible obstacles in subsequent maintenance and safety operations, and obtain the marked obstacle area data.

[0062] S503: Based on the marked obstacle area data, the color information of the point cloud area is analyzed, and according to the color characteristics of the point cloud, the obstacles are classified into corresponding types, including metal, wood and plastic. The specific process of obtaining the obstacle classification label is as follows; Based on the marked obstacle area data, the color information of the point cloud area is analyzed. Obstacles are classified according to the RGB color value of each point in the point cloud data. The point cloud color features in each marked area are analyzed, and obstacles are classified into metal (usually gray or silver), wood (brown or yellow), and plastic (may appear in a variety of bright colors) based on the typical color range. The color features of each material type are set in advance to ensure the accuracy of the classification, and specific material type labels are assigned to each obstacle area to generate obstacle classification labels.

[0063] See also Figure 7 Based on the obstacle classification label, the geometric center point and the circumscribed rectangular bounding box of each marked obstacle are calculated. Through three-dimensional space modeling, the bounding box of each obstacle is constructed to locate and characterize the spatial position and shape of the obstacle. The specific steps to obtain the obstacle framing result are as follows: S601: Based on the obstacle classification label, the geometric center of the cluster is identified by calculating the average value of the coordinates of each cluster point cloud, and the center point is used as a representative of the obstacle position to obtain the coordinates of the obstacle center point.

[0064] Calculate the average value of each cluster point cloud coordinate based on the obstacle classification label according to the formula:

[0065] and

[0066] and

[0067] Calculate the geometric center of the cluster, where , , represents the average value of x, y, and z coordinates. represents the total number of points in the cluster, , , Representative The x, y, z coordinates of a point.

[0068] Detailed explanation of the formula and the process of formula calculation and derivation: The key step in identifying obstacle locations is to determine the geometric center of the clustered point cloud, which is usually achieved by calculating the average of the coordinates. For example, a cluster contains points, each with three-dimensional coordinates , , The average value for each coordinate dimension can be calculated by summing the coordinate values ​​of each point and then dividing by the total number of points. Come get.

[0069] Set a cluster to contain the following points: Point 1: (3,2,7).

[0070] Point 2: (5,3,8).

[0071] Point 3: (4,5,6).

[0072] Compute the mean of the x-coordinates: ; Similarly, the average value of the y coordinate is: ; The average value of the z-coordinate: ; The calculation method accurately determines the geometric center of the obstacle, which is the basis for locating and handling obstacles. The process relies on actual point cloud data.

[0073] S602: Based on the coordinates of the obstacle center point, measure the distance from the farthest point of each cluster to the center point, identify the size and position of the circumscribed rectangular bounding box of each obstacle, and obtain the obstacle bounding box parameters. Based on the coordinates of the obstacle center point, measure the distance from the farthest point to the center point in each cluster to identify the size and position of each obstacle's circumscribed rectangular bounding box. Find the farthest point by calculating the Euclidean distance from each point to the center point, which is defined as the radius of the bounding box. Considering the shape of the obstacle, determine the length, width, and height of the bounding box by calculating the maximum and minimum coordinates independently for each dimension. Obtain the bounding box parameters of each obstacle, including its position and size in three-dimensional space.

[0074] S603: Based on the obstacle bounding box parameters, a three-dimensional bounding box of each obstacle is constructed through three-dimensional modeling, each obstacle is covered by the bounding box, and the position and shape of the bounding box in space are recorded to obtain the obstacle frame result. The specific process is as follows; Based on the obtained obstacle bounding box parameters, 3D modeling is performed to construct a 3D bounding box for each obstacle. Using a standard cube or cuboid model, the size and position of the bounding box are adjusted according to the bounding box parameters to ensure that the bounding box completely covers the corresponding obstacle cluster. In this way, each obstacle is surrounded by an accurate 3D bounding box, which records the specific position and shape of the obstacle in space, providing the necessary data for visualization and further processing of the obstacle. Such a framed result is crucial information for obstacle management and space planning.

[0075] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for identifying obstacles based on laser point cloud in a converter station scenario, characterized in that: The following steps are involved: Based on the converter station environment, three-dimensional scanning is performed through LiDAR to collect point cloud data, and the data is filtered for noise and spatially aligned to obtain processed point cloud data. Based on the processed point cloud data, each point in the data is mapped to the center of an adjacent grid, the coordinate average of each grid point is calculated, and the resulting point is set as a grid representative to obtain simplified point cloud data; Based on the simplified point cloud data, the point cloud data touching the ground is screened, a simulated cloth model is created, the point cloud data is inverted, the position of the simulated cloth is adjusted to fit the ground represented by the point cloud, and the ground and non-ground point clouds are segmented to obtain ground segmentation data; Based on the ground segmentation data, the point cloud data is horizontally cut, each layer of the point cloud is projected onto a two-dimensional plane, and the linear features of the wires on each layer are identified to obtain wire identification point cloud data; Based on the wire recognition point cloud data, the density between the point clouds is measured, densely clustered areas are marked as potential obstacles, the obstacles are classified, and obstacle classification labels are obtained; Based on the obstacle classification label, the geometric center point and the circumscribed rectangular bounding box of each marked obstacle are calculated, and the bounding box of each obstacle is constructed to obtain the obstacle framing result.

2. The obstacle framing method based on laser point cloud in the converter station scenario according to claim 1 is characterized in that: The processed point cloud data includes the position coordinates, reflection intensity and color information of the point cloud; the simplified point cloud data includes the voxelized point cloud, the average position point inside each voxel and the point cloud model with reduced spatial resolution after reconstruction; the ground segmentation data includes the ground point cloud marked after simulating the bonding of cloth, the non-ground point cloud part and the threshold value between the ground and the non-ground; the wire identification point cloud data includes the identified wire feature points and the point cloud line segments of the wire; the obstacle classification label includes the divided obstacle category and the corresponding obstacle clustering mark; the obstacle framing result includes the three-dimensional coordinates, length, width and height dimensions and obstacle volume size of each obstacle bounding box.

3. The obstacle framing method based on laser point cloud in the converter station scenario according to claim 1 is characterized in that: Based on the converter station environment, three-dimensional scanning is performed by LiDAR to collect point cloud data, filter out noise from the data, and perform spatial alignment of the data to obtain the processed point cloud data. The specific steps are as follows: Based on the converter station environment, the scanning angle and resolution of the LiDAR are adjusted to perform a panoramic 3D scan of the converter station to obtain a basic point cloud data set. Based on the basic point cloud data set, by analyzing the density and reflection characteristics of the point cloud and filtering the data for noise, data points with abnormal reflection intensity and positions deviating from the normal are identified and eliminated, thereby obtaining point cloud data with noise filtered out; Based on the point cloud data from which noise has been filtered out, the coordinates of the data points are adjusted to eliminate the offset caused by scanning, and the continuity of the point cloud data is verified through continuity detection to obtain processed point cloud data.

4. The obstacle framing method based on laser point cloud in the converter station scenario according to claim 1 is characterized in that: Based on the processed point cloud data, each point in the data is mapped to the adjacent grid center, the coordinate average of each grid point is calculated, and the result point is set as a grid representative. The specific steps of obtaining simplified point cloud data are as follows: Based on the processed point cloud data, spatial compression is performed, and according to the spatial position of each point, the point cloud data is mapped to a preset spatial grid to obtain grid-mapped point cloud data; Based on the grid-mapped point cloud data, calculating the coordinate average value in the point cloud set of each grid to obtain a grid coordinate average value calculation result; Based on the grid coordinate average calculation result, the coordinate average is set as the representative point of each grid to obtain simplified point cloud data.

5. The obstacle framing method based on laser point cloud in the converter station scenario according to claim 1 is characterized in that: Based on the simplified point cloud data, the point cloud data touching the ground is screened, a simulated cloth model is created, the point cloud data is inverted, the position of the simulated cloth is adjusted to fit the ground represented by the point cloud, and the ground and non-ground point clouds are segmented to obtain the ground segmentation data. Specifically, the steps are as follows: Based on the simplified point cloud data, by analyzing the difference between the Z coordinate of each point and a predetermined ground height, filtering the data points in contact with the ground, identifying a point set close to or equal to the ground height, and obtaining ground point cloud information; Based on the ground point cloud information, the physical properties of the cloth are simulated, the elastic coefficient and the grid size of the cloth model are set, the covering behavior of the cloth in the virtual environment is simulated, the cloth is adjusted to adapt to the terrain, and a simulated cloth configuration is obtained; Based on the simulated cloth configuration, the cloth is matched with the inverted point cloud, the position of the cloth model is adjusted, the cloth model is fitted to the inverted terrain features, and the ground and non-ground areas are segmented through the contact points between the cloth and the terrain to obtain ground segmentation data.

6. The obstacle framing method based on laser point cloud in the converter station scenario according to claim 1 is characterized in that: Based on the ground segmentation data, the point cloud data is horizontally cut, each layer of the point cloud is projected onto a two-dimensional plane, and the linear features of the wires on each layer are identified. The specific steps of obtaining the wire identification point cloud data are as follows: Based on the ground segmentation data, the entire point cloud data set is subjected to height interval layering processing, each layer has a fixed height, and the point cloud is divided into multiple horizontal layers according to a preset interval to obtain a layer segmentation point cloud; Based on the layer segmentation point cloud, the point cloud of each horizontal layer is projected into a two-dimensional plane, and the XY coordinates of the point cloud are plotted on the two-dimensional plane to obtain two-dimensional projection data; Based on the two-dimensional projection data, linear feature detection technology is used to analyze the linear arrangement characteristics of the projection graphics, screen the projection point cloud with linear characteristics, mark the point set whose linear arrangement meets the characteristics of the wire, and obtain the wire identification point cloud data.

7. The obstacle framing method based on laser point cloud in the converter station scenario according to claim 6 is characterized in that: The linear feature detection technology is based on the formula: ； Calculate the linear feature score of the point cloud data, where represents the length of the longest linear feature in the projected data, represents the average distance between adjacent features, represents the standard deviation in the direction of the linear feature, , and is the adjustment coefficient, Represents the linear feature scores of point cloud data.

8. The obstacle framing method based on laser point cloud in the converter station scenario according to claim 1 is characterized in that: Based on the wire recognition point cloud data, the density between the point clouds is measured, densely clustered areas are marked as potential obstacles, and the obstacles are classified to obtain obstacle classification labels. Specifically, the steps are as follows: Based on the wire identification point cloud data, a kernel density estimation method is used to perform spatial statistical analysis on each point cloud block, calculate the number and distribution of points inside the point cloud block, identify the density of each point cloud area, and generate point cloud density information; Based on the point cloud density information, by comparing with a preset density benchmark, a point cloud area exceeding the preset density benchmark is identified as a potential obstacle, and marked obstacle area data is obtained; Based on the marked obstacle area data, the color information of the point cloud area is analyzed, and according to the color characteristics of the point cloud, the obstacles are classified into corresponding types, including metal, wood and plastic, to obtain obstacle classification labels.

9. The obstacle framing method based on laser point cloud in the converter station scenario according to claim 8 is characterized in that: The kernel density estimation method is based on the formula: ； Calculate point cloud density ,in, represents the density of point cloud, Represents the number of points in the point cloud block, Representative The location of the point, is the average position of the point cloud block, is the standard deviation of the point positions, is the base of natural logarithms, is the ratio of pi.

10. The obstacle framing method based on laser point cloud in the converter station scenario according to claim 1, characterized in that: Based on the obstacle classification label, the geometric center point and the circumscribed rectangular bounding box of each marked obstacle are calculated, and the bounding box of each obstacle is constructed to obtain the obstacle framing result. Specifically, the steps are as follows: Based on the obstacle classification label, the geometric center of the cluster is identified by calculating the average value of the coordinates of each cluster point cloud, and the center point is used as a representative of the obstacle position to obtain the coordinates of the obstacle center point; Based on the coordinates of the obstacle center point, measure the distance from the farthest point of each cluster to the center point, identify the size and position of the circumscribed rectangular bounding box of each obstacle, and obtain the obstacle bounding box parameters; Based on the obstacle bounding box parameters, a three-dimensional bounding box of each obstacle is constructed through three-dimensional modeling, each obstacle is covered by the bounding box, and the position and shape of the bounding box in space are recorded to obtain an obstacle framing result.

Images