Three-dimensional laser point cloud redundancy removal method, system and equipment for power transmission line inspection and medium

Abstract

The invention discloses a three-dimensional laser point cloud redundancy removal method, system and device for power transmission line inspection and a medium, and belongs to the technical field of three-dimensional point cloud redundancy removal, and the method comprises the steps: determining three-dimensional laser point cloud data, carrying out the normal vector estimation, and constructing an enhancement matrix for each point cloud data point; obtaining a normal vector coordinate for the enhanced matrix decomposition, determining a normal vector direction through a unit vector, and constructing a data structure; and carrying out normal vector analysis on the feature extraction in the normal vector direction, carrying out segmentation processing on an analysis result, and judging an area in the point cloud set based on the segmentation result to carry out point cloud elimination. According to the method, various redundant plane structures such as a horizontal plane, a vertical plane and an inclined plane can be effectively identified, the application range of a point cloud redundancy removal technology is effectively expanded, and the limitation that a traditional redundancy judgment mode based on a geometric height or density threshold value is sensitive to topographic change and poor in stability is overcome.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of three-dimensional point cloud de-redundancy, in particular to a three-dimensional laser point cloud de-redundancy method, system, device and medium for power line inspection. BACKGROUND

[0002] At present, in the process of power line inspection, three-dimensional laser radar (LiDAR) technology is widely used in tasks such as environment modeling, equipment identification and obstacle detection. However, due to the large amount of point cloud data and complex structure, especially in the field scene, there are a large number of plane structures (such as ground, building facade, terrain steps, etc.) irrelevant to the inspection target. These redundant points not only occupy storage space and increase computational burden, but also interfere with the accuracy of subsequent target identification and defect diagnosis.

[0003] In the prior art, height threshold-based or simple ground segmentation methods are often used to remove some inefficient information, but these methods are mainly aimed at the ground or horizontal plane, lack the ability to identify and process vertical structures (such as wall surface, slope surface, etc.), and cannot effectively deal with the complex three-dimensional space environment in the power transmission channel. In addition, some methods use voxel filtering or statistical analysis for downsampling, which can reduce the amount of data, but often loses key information, affecting the subsequent modeling and identification effect.

[0004] Therefore, how to effectively remove the redundant plane data in the point cloud irrelevant to the inspection while retaining the key structure (such as conductor, tower, etc.) information is one of the technical bottlenecks in the current three-dimensional point cloud processing of power transmission lines. SUMMARY

[0005] In view of the above problems, the present application is proposed.

[0006] Therefore, the present application aims to solve the problem that redundant plane structure point clouds in multiple directions (horizontal and vertical) cannot be effectively identified and removed, resulting in low analysis efficiency and poor recognition accuracy in the subsequent analysis; the existing methods lack sufficient utilization of the directional features of point cloud normal vectors, and cannot accurately distinguish the target area from the background area from the geometric structure level; the de-redundancy process has the risk of mistakenly deleting key structures or retaining a large amount of invalid data, affecting the system performance and actual inspection effect.

[0007] To solve the above technical problems, the present application provides the following technical solutions: a three-dimensional laser point cloud de-redundancy method for power line inspection, comprising, determining the three-dimensional laser point cloud data for normal vector estimation, constructing an enhanced matrix for each point cloud data point; obtaining the normal vector coordinates by decomposing the enhanced matrix, determining the normal vector direction through the unit vector, and constructing the data structure; performing normal vector analysis on the feature extraction of the normal vector direction, performing segmentation processing on the analysis result, and judging the point cloud concentrated area based on the segmentation result to remove the point cloud.

[0008] As a preferred scheme of the three-dimensional laser point cloud de-redundancy method for power transmission line inspection, the three-dimensional laser point cloud data is determined by radar reconstruction point cloud. After obtaining the point cloud data, surface normal vector estimation and principal component analysis are performed.

[0009] As a preferred scheme of the three-dimensional laser point cloud de-redundancy method for power transmission line inspection, the enhanced matrix is constructed by constructing a principal component analysis plane composed of each point cloud and the corresponding nearest neighbor point cloud. Nearest neighbor search is performed to construct a data structure.

[0010] As a preferred scheme of the three-dimensional laser point cloud de-redundancy method for power transmission line inspection, the enhanced matrix further includes singular value decomposition of the principal component analysis plane generated and represented as the enhanced matrix to obtain normal vector coordinates. The data matrix empirical mean removal of principal component analysis is used for variance reduction, and singular value decomposition is performed on the corrected data matrix.

[0011] As a preferred scheme of the three-dimensional laser point cloud de-redundancy method for power transmission line inspection, the data structure is constructed by, Surface normal vector estimation is performed by determining a unit vector as a reference for the normal direction. The unit vector is determined by normalizing the coordinates of each normal vector and using to perform normalization processing, where , is the normal vector, and the coordinates obtained by singular value decomposition of the principal component analysis plane of the enhanced matrix ; After the surface normal vector estimation process, a new data structure matrix is generated, representing the coordinates of each point and the corresponding unit normal vector, and the data structure matrix is defined as: where, is the point of each point cloud data on the principal component analysis plane, is the coordinate of the point on the principal component analysis plane.

[0012] As a preferred scheme of the three-dimensional laser point cloud de-redundancy method for power transmission line inspection, the feature extraction comprises feature extraction on the normal vector direction, and a plurality of types of normal vectors are used as the distinguishing features to remove the points located on the plane. In the normal vector, if the component of the unit normal vector of any point on any coordinate axis is 1 or 0, it indicates that the current normal vector is perpendicular to the plane, plane or plane.

[0013] The preferred technical scheme in the embodiment of the application has the beneficial effect that the efficient organization and management of the point cloud data are realized by generating the data structure R matrix containing the point coordinates and the unit normal vector, and the efficiency of the subsequent normal vector analysis and processing is improved.

[0014] As a preferred scheme of the three-dimensional laser point cloud de-redundancy method for power transmission line inspection, the segmentation processing comprises histogram segmentation processing after the analysis of the normal vector direction. For each , coordinate axis direction, the component value interval [0, 1] of the unit normal vector is divided into 20 intervals with an interval of 0.05, and the number of normal vectors in each interval is counted to verify whether the normal vectors with similar directions, i.e., the vectors representing the plane structure, are concentrated in the interval of 0 or 1, and the point cloud removal processing is performed in the interval with the concentrated points in the histogram.

[0015] The preferred technical scheme in the embodiment of the application has the beneficial effect that the histogram segmentation method is used to analyze the normal vector direction, the plane structure area in the point cloud can be accurately identified, the redundant point cloud can be effectively removed, and the compactness and quality of the point cloud data are improved.

[0016] Another object of the application is to provide a three-dimensional laser point cloud de-redundancy system for power transmission line inspection.

[0017] To solve the above technical problems, the application provides the following technical scheme: a three-dimensional laser point cloud de-redundancy system for power transmission line inspection, comprising a normal vector estimation module, a normal vector processing module and a point cloud de-redundancy module. The normal vector estimation module determines the normal vector estimation of the three-dimensional laser point cloud data, and constructs an enhanced matrix for each point cloud data point. The normal vector processing module obtains the normal vector coordinates by decomposing the enhanced matrix, determines the normal vector direction through the unit vector, and constructs a data structure. The point cloud redundancy removing module performs normal vector analysis on feature extraction in the direction of the normal vector, performs segmentation processing on the analysis result, and judges the point cloud removal based on the segmentation result.

[0018] The application provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the three-dimensional laser point cloud redundancy removing method for power transmission line inspection when executing the computer program.

[0019] The application provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the three-dimensional laser point cloud redundancy removing method for power transmission line inspection when executed by a processor.

[0020] The application has the beneficial effect that the normal vector direction feature is introduced to realize multi-directional plane structure recognition and removal, compared with the traditional point cloud segmentation processing mode for the ground (horizontal plane) only, the normal vector direction information of the point cloud is extracted through principal component analysis (PCA) and singular value decomposition (SVD), and the component distribution of the unit normal vector in the three axes (x, y and z) directions, so that the horizontal plane, the vertical plane and the inclined plane and other redundant plane structures can be effectively recognized. This method is suitable for the complex environment with a large number of vertical poles, tower bodies and building backgrounds in the power transmission line, and effectively expands the application range of the point cloud redundancy removing technology.

[0021] The application constructs a histogram model based on the unit normal vector direction distribution for redundancy area judgment, designs a three-axis histogram analysis mechanism based on the unit normal vector direction distribution, divides the unit normal vector component interval of each direction into several fixed intervals, and takes the point density as the judgment basis to automatically identify the point cloud area with consistent normal direction and plane structure characteristics. This method overcomes the limitations of the traditional redundancy judgment mode based on geometric height or density threshold, which is sensitive to terrain changes and has poor stability. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0023] Figure 1 A three-dimensional laser point cloud redundancy removing method for power transmission line inspection is provided for an embodiment of the application. DETAILED DESCRIPTION

[0024] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present application.

[0025] Embodiment 1, refer to Figure 1 For an embodiment of the present application, the embodiment provides a three-dimensional laser point cloud de-redundancy method for power transmission line inspection, comprising: S100, determining three-dimensional laser point cloud data for normal vector estimation, constructing an enhanced matrix for each point cloud data point; S200, obtaining normal vector coordinates by decomposing the enhanced matrix, determining the normal vector direction through the unit vector, and constructing a data structure; S300, performing normal vector analysis on the feature extraction of the normal vector direction, performing segmentation processing on the analysis result, and judging the point cloud elimination based on the segmentation result; It should be noted that the kdTree-NN is used to find the neighboring points of each reference point, so that the principal component analysis plane composed of the reference points is used as the axis, and the singular value decomposition is used to estimate the normal vector perpendicular to the principal component analysis plane. The previous research only eliminated the points representing the land blocks, and the planes of these land blocks were horizontally copied. The data points of the previous research were also relatively small, only human objects and land blocks, without other non-planar objects. Based on these reasons, the present application proposes to eliminate horizontal and vertical planes in a wider point cloud data representation than previous research.

[0026] Therefore, in view of the above problems, through the steps of S100-S300, the present application uses a normal vector estimation method based on principal component analysis (PCA) and singular value decomposition (SVD) to extract the unit normal vector of each point in the point cloud, and combines the distribution histogram analysis of the three-axis direction to accurately identify the point set corresponding to the typical plane structure. By eliminating the regions with high density and concentrated distribution, the automatic and high-fidelity removal of the point cloud redundancy surface can be realized under the premise of ensuring the integrity of the target, thereby improving the inspection data processing efficiency and intelligent recognition accuracy.

[0027] Embodiment 2, refer to Figure 1 For an embodiment of the present application, the embodiment provides a three-dimensional laser point cloud de-redundancy method for power transmission line inspection, comprising: To support the data elimination process, previous studies refer to the difference in the direction of the normal vector perpendicular to the plane. If the support plane is similar, the normal vector is in the same direction as another normal vector. Therefore, the direction of the normal vector of each point on the plane has a similar direction. The present application uses a histogram to analyze the similarity of the vector direction of the entire point cloud.

[0028] In an embodiment of the present application, the normal vector estimation of the three-dimensional laser point cloud data is determined in S100, and an enhanced matrix is constructed for each point cloud data point, including the following steps S101-S102: S101, since the data is a point cloud reproduced by a 3D laser radar, it is necessary to first determine the point cloud data.

[0029] The data set of n points in matrix P is represented as: wherein, : wherein, is the point of each point cloud data on the principal component analysis plane, is the coordinate of the point on the principal component analysis plane, and R is a real number set.

[0030] S102, after obtaining the three-dimensional LiDAR point cloud data, surface normal vector estimation is performed.

[0031] At this stage, the principal component analysis is implemented by kdTree-NN and SVD.

[0032] Thus, the normal vector coordinates of each point are obtained.

[0033] In an embodiment of the present application, the normal vector coordinates are obtained by decomposing the enhanced matrix in S200, the normal vector direction is determined by a unit vector, and a data structure is constructed, including the following steps S201-S203: S201, nearest neighbor search is performed, and a data structure is constructed, including the following steps A1-A2: A1, the starting step of the normal vector estimation is to construct a principal component analysis (PCA) plane composed of each point and its nearest neighbor points , wherein .

[0034] A2, the nearest neighbor search is implemented by kdTree-NN, and the data structure and is constructed.

[0035] wherein, represents a matrix composed of adjacent points in the PCA plane, and representing an enhanced matrix comprising a reference point and its neighboring points.

[0036] In an optional embodiment, the nearest neighbor search of S201 can be a radius neighbor search, for each point cloud data point, a fixed search radius is defined, all neighbor points within the spherical radius range are collected to form a principal component analysis (PCA) plane. Neighbor point set Q; the nearest neighbor search is realized by radius neighbor search, and the data structure is constructed; but the radius neighbor search causes large fluctuation of the number of neighbor points when the point cloud density is uneven, which affects the consistency of the normal vector estimation.

[0037] In another optional embodiment, the nearest neighbor search of S201 can also be a grid division search, which divides the entire three-dimensional point cloud space into uniform cubic grid units, for each point, determines the grid where it is located, and searches all points in the adjacent grid units as neighbor points to form a principal component analysis (PCA) plane. Neighbor point set; the nearest neighbor search is realized by grid division search, and the data structure is constructed; but the grid division search is inefficient and the neighbor point search is not comprehensive when the point cloud is sparse or the grid size is not properly selected.

[0038] S202, using principal component analysis data matrix empirical mean removal for variance reduction, including the following steps B1-B2: B1, to the generated and represented as PCA plane of matrix singular value decomposition, get the normal vector with coordinates .

[0039] B2, using PCA data matrix empirical mean removal for variance reduction, singular value decomposition is performed on the modified data matrix, as shown in the following formula: Construct , where is the mean vector of the enhanced matrix , and the calculation method is .

[0040] Since the normal vector is perpendicular to the plane represented by the matrix , singular value decomposition is applied to the modified matrix , and the unique normal vector solution can be solved.

[0041] In an optional embodiment, the variance reduction of S202 can be direct singular value decomposition, for each point and its nearest neighbor point, directly applying singular value decomposition (SVD) without any centering processing. The vector corresponding to the minimum singular value is extracted from the SVD result as the normal vector, without calculating the mean or modifying the data matrix; but since the data is not centered, the normal vector estimation is easily affected by the local translation bias of the point cloud, resulting in a decrease in direction accuracy.

[0042] In another optional embodiment, the variance reduction of S202 can also be median centering and singular value decomposition, for the augmented matrix, calculating the median vector (not the mean vector) of each coordinate dimension as the center point, then subtracting the median vector from the augmented matrix to obtain a centered matrix, and then applying singular value decomposition (SVD) to the centered matrix to solve the normal vector; but although median centering is more robust to outliers, it may introduce unnecessary bias in the case of uniform point cloud distribution, reducing the consistency of the normal vector direction.

[0043] S203, unit vector The determination method is to normalize the coordinates of each normal vector to , wherein , is the normal vector, is the principal component analysis plane of the principal component analysis plane singular value decomposition of the augmented matrix; After the surface normal estimation process, a new data structure matrix is generated, which represents the coordinates of each point and the corresponding unit normal vector, and the data structure matrix is defined as: wherein is the point of each point cloud data on the principal component analysis plane, is the coordinate of the point on the principal component analysis plane.

[0044] In an embodiment of the present application, the normal vector analysis of the normal vector direction feature extraction in S300 is segmented, and the segmented result is used to judge the point cloud cluster area and remove the point cloud, including the following steps S301-S302: In an embodiment of the present application, S301, the normal vector analysis of the normal vector direction feature extraction includes the following steps C1-C2: C1, since the normal vector is perpendicular to the surface it is on, a variety of types of normal vectors can be used as discriminant features for removing points located on the plane.

[0045] C2, in the normal vector, if the component of the unit normal vector of any point on any coordinate axis is 1 or 0, it indicates that the current normal vector is perpendicular to the plane, , , or .

[0046] In an optional embodiment, the feature extraction of S301 can be a range judgment based on the normal vector component threshold. For the unit normal vector of each point, the component value on the x, y, z coordinate axis is calculated, and a component threshold range (such as the absolute value of the component being greater than 0.9 or less than 0.1) is set. If the component of the normal vector of a certain point falls within the threshold range on any coordinate axis, it is determined that the normal vector is approximately perpendicular to the corresponding coordinate plane, and the point is marked as a point on the plane. Based on the marking result, the regions in the point cloud set are segmented and removed. However, the selection of the threshold range depends on the noise level and density of the point cloud data, leading to excessive removal or omission, and the effect is not good in complex terrain.

[0047] In another optional embodiment, the feature extraction of S301 can also be to construct a normal vector direction histogram of the entire point cloud, divide the normal vector direction into multiple angle intervals, analyze the peak values of the histogram, identify the normal vector directions with the highest frequency (i.e. the dominant direction), which usually correspond to the plane regions (such as the ground or device surface) in the point cloud. For each point, the similarity (such as the dot product or the included angle) between its normal vector direction and the dominant direction is calculated. If the similarity is higher than a set threshold, it is determined that the point is on the plane. Based on the similarity judgment, the regions in the point cloud set are segmented and removed. The resolution and peak detection of the histogram are sensitive to the density changes of the point cloud, and may produce deviations in non-uniform point clouds, leading to inaccurate segmentation.

[0048] S302, after analyzing the normal vector direction, histogram segmentation processing is adopted, including the following steps D1-D2: D1, for each , coordinate axis direction, the component value interval [0, 1] of the unit normal vector is divided into 20 intervals with an interval of 0.05, and the number of normal vectors in each interval is counted; D2, verify the normal vectors with similar directions, i.e. whether the vectors representing the plane structure are concentrated in the interval of 0 or 1. The interval with concentrated points in the histogram is subjected to point cloud removal processing, because such concentrated distribution usually represents the existence of plane structure.

[0049] In an optional embodiment, the segmentation processing of S302 can divide the value range [0, 1] of each component (x, y, z) of the unit normal vector into larger intervals (for example, divided into 10 intervals with an interval of 0.1), and use a sliding window method to traverse these intervals: for each interval, the total number of normal vectors in the window is counted to smooth the distribution and identify the concentrated area, and when the number of normal vectors in a certain window exceeds a preset threshold (for example, 10% of the total number of points), it is determined that the normal vector direction corresponding to the window represents a planar structure, and all point clouds corresponding to the normal vectors in the window are removed; but the size of the sliding window and the threshold need to be manually adjusted, and the stability is poor in different point cloud data, resulting in excessive removal or omission.

[0050] In another optional embodiment, the segmentation processing of S302 can also sort the values of each component (x, y, z) of the unit normal vector to generate an ordered list; on the ordered list, the local density of each point normal vector component is calculated (for example, using the difference between each point and its adjacent points to estimate the density), and the high-density area (for example, a continuous point group with a difference less than 0.05) is marked; identify the area where the normal vector component is close to 0 or 1 in these high-density areas (for example, the component value is in the range of [0, 0.1] or [0.9, 1]), as an indication of planar structure; perform batch removal processing on the point cloud corresponding to these areas; but the sorting and density calculation has large computational overhead, especially for large-scale point cloud data, the processing efficiency is low, and it is not suitable for real-time applications.

[0051] Embodiment 3 is an embodiment of the present application, and the above is a schematic scheme of a three-dimensional laser point cloud de-redundancy method for power line inspection. It should be noted that the technical scheme of a three-dimensional laser point cloud de-redundancy system for power line inspection and the technical scheme of the above three-dimensional laser point cloud de-redundancy method for power line inspection belong to the same concept. The technical scheme of the three-dimensional laser point cloud de-redundancy system for power line inspection in this embodiment is not described in detail, and can be referred to the description of the technical scheme of the three-dimensional laser point cloud de-redundancy method for power line inspection.

[0052] The embodiment provides a three-dimensional laser point cloud de-redundancy system for power line inspection, which comprises a normal vector estimation module, a normal vector processing module and a point cloud de-redundancy module. The normal vector estimation module determines the normal vector estimation of the three-dimensional laser point cloud data, constructs an enhanced matrix for each point cloud data point; The normal vector processing module obtains the normal vector coordinates by decomposing the enhanced matrix, determines the normal vector direction through the unit vector, and constructs a data structure; The point cloud redundancy removal module performs normal vector analysis on the feature extraction of the normal vector direction, segments the analysis results, and removes point clouds from concentrated areas based on the segmentation results.

[0053] This embodiment also provides an electronic device applicable to a three-dimensional laser point cloud redundancy removal method for power transmission line inspection, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the three-dimensional laser point cloud redundancy removal method for power transmission line inspection as proposed in the above embodiment.

[0054] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a three-dimensional laser point cloud redundancy removal method for power transmission line inspection as proposed in the above embodiment.

[0055] The storage medium proposed in this embodiment and the method for redundancy removal of three-dimensional laser point clouds for power transmission line inspection proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0056] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0057] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for redundancy removal from three-dimensional laser point clouds for power transmission line inspection, characterized in that: include, Normal vector estimation is performed on the 3D laser point cloud data, and an enhancement matrix is ​​constructed for each point cloud data point; The normal vector coordinates are obtained from the enhanced matrix decomposition, the normal vector direction is determined by the unit vector, and a data structure is constructed. The features of the normal vector direction are extracted and normal vector analysis is performed. The analysis results are segmented, and point cloud concentration areas are identified based on the segmentation results for point cloud removal.

2. The three-dimensional laser point cloud redundancy removal method for power transmission line inspection as described in claim 1, characterized in that: The determination of the three-dimensional laser point cloud data includes determining the point cloud data by reconstructing the point cloud using radar. After obtaining the point cloud data, surface normal vector estimation and principal component analysis are performed.

3. The three-dimensional laser point cloud redundancy removal method for power transmission line inspection as described in claim 2, characterized in that: The construction of the enhancement matrix includes constructing a principal component analysis plane composed of each point cloud and its corresponding nearest neighbor point cloud. Perform nearest neighbor search and construct the data structure.

4. The three-dimensional laser point cloud redundancy removal method for power transmission line inspection as described in claim 3, characterized in that: The construction of the enhancement matrix also includes performing singular value decomposition on the principal component analysis plane generated and represented as the enhancement matrix to obtain the normal vector coordinates; Variance reduction is performed by removing the empirical mean from the data matrix using principal component analysis, and singular value decomposition is then performed on the corrected data matrix.

5. The three-dimensional laser point cloud redundancy removal method for power transmission line inspection as described in claim 4, characterized in that: The constructed data structure includes, To estimate the surface normal vector, a unit vector is determined as a reference for the normal direction. Unit vector The method for determining it is to use each normal vector Normalize the coordinates using Normalization is performed, where , It is the normal vector. To enhance the matrix The coordinates obtained by singular value decomposition of the principal component analysis plane; After the surface normal vector estimation process, a new data structure is generated. A matrix representing the coordinates of each point and its corresponding unit normal vector; data structure. A matrix is ​​defined as: in, For each point cloud data point on the principal component analysis plane Let be the coordinates of the point on the principal component analysis plane.

6. The three-dimensional laser point cloud redundancy removal method for power transmission line inspection as described in claim 5, characterized in that: The feature extraction includes feature extraction of the normal vector direction, using multiple types of normal vectors as discriminative features to remove points located on the plane; In the normal vector, if any point If the component of the unit normal vector on any coordinate axis is 1 or 0, it indicates that the current normal vector is perpendicular to the coordinate axis. flat, plane or flat.

7. A three-dimensional laser point cloud redundancy removal method for power transmission line inspection as described in claim 6, characterized in that: The segmentation process includes segmenting the vectors using histograms after analyzing the normal vector directions. For each , In the coordinate axis direction, the component value interval [0,1] of the unit normal vector is divided into 20 intervals with an interval of 0.05, and the number of normal vectors in each interval is counted to verify normal vectors with similar directions, that is, whether the vectors representing the planar structure are concentrated in the interval of 0 or 1. Point cloud removal is performed on the intervals with concentrated points in the histogram.

8. A three-dimensional laser point cloud redundancy removal system for transmission line inspection, employing the three-dimensional laser point cloud redundancy removal method for transmission line inspection as described in any one of claims 1 to 7, characterized in that, include: Normal vector estimation module, normal vector processing module, point cloud redundancy removal module; The normal vector estimation module determines the normal vector of the 3D laser point cloud data and constructs an enhancement matrix for each point cloud data point. The normal vector processing module obtains the normal vector coordinates from the enhanced matrix decomposition, determines the normal vector direction through the unit vector, and constructs the data structure. The point cloud redundancy removal module performs normal vector analysis on the feature extraction of the normal vector direction, segments the analysis results, and removes point clouds from concentrated areas based on the segmentation results.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the three-dimensional laser point cloud redundancy removal method for power transmission line inspection as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the three-dimensional laser point cloud redundancy removal method for power transmission line inspection as described in any one of claims 1 to 7.

Images