Classification method of three-dimensional point cloud of power transmission channel
By setting the pole and tower area in the transmission channel and determining the terrain, dividing ground and non-ground points, and determining the conductor segmentation lines in the non-roof points, the problem of point cloud classification in the transmission channel is solved, and efficient three-dimensional point cloud classification is achieved. Meets the requirements of real-time and accuracy.
Patent Information
- Application Number
- CN202210529346.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-05-16
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2042-05-16
AI Technical Summary
In the case where there are many ground points and vegetation points in the transmission channel, few pole tower points and conductor points and extremely uneven distribution of categories, the requirements of real-time and accuracy cannot be met at the same time, and it is difficult to effectively classify three-dimensional point clouds. data.
By obtaining the three-dimensional point cloud file of the transmission channel, setting the tower area, determining the terrain and dividing ground points and non-ground points, obtaining the plane coordinates of non-ground points, dividing them into tower points and non-roof points, and at non-roof points, The wire division line is determined in the middle, and further divided into wire points and other points.
It realizes efficient classification of ground points, tower points, wire points and other points in the transmission channel, improves classification efficiency and accuracy, and can meet the requirements of immediacy and accuracy.
Smart Images

Figure CN115019199B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of intelligent operation and inspection of electric power equipment, and in particular relates to a three-dimensional point cloud classification method for a power transmission channel. Background Art
[0002] With the popularization of 3D point cloud technology, the inspection method of drones equipped with laser radars has been widely introduced into the operation, maintenance and overhaul of power transmission channels. It is used to collect 3D point cloud information in the power transmission channel. Compared with the inspection of monitoring equipment in the power transmission channel or the inspection of drones alone, the collected 3D point cloud information can more accurately reflect the scene in the power transmission channel. However, in the inspection method of drones equipped with laser radars, the 3D point clouds collected by the laser radars are not classified, and it is impossible to effectively conduct a deeper analysis of a part of the point cloud in the channel under real-time conditions, such as the distribution of pole towers or the sag changes of conductors.
[0003] To solve the problem of 3D point cloud classification, the methods commonly used in the prior art include:
[0004] 1. Use traditional artificial point cloud feature classification; that is, classification is performed based on normal vector, echo intensity, local density, local curvature and other information, where echo intensity is related to the target's surface material, roughness, incident angle direction, instrument emission energy, and laser wavelength. However, the transmission channel scene is complex, with a variety of terrain, vegetation, buildings and other information. Traditional artificial point cloud features are difficult to accurately classify.
[0005] 2. Classification based on 3D CNN (3D Convolutional Neural Networks) voxel model; that is, first map the point cloud to voxel space, and then classify or segment through 3D CNN. The computational complexity of this method is limited by hardware capabilities, and in the process of mapping to voxels, the point cloud will lose a lot of surface features.
[0006] 3. Classification is performed based on the PointNet or PointNet++ point cloud classification / segmentation deep learning framework. This type of deep learning network can perform three-dimensional point cloud classification of small objects very well, but it is difficult to apply to large transmission channel scenarios with a span of 100 meters or even 300 meters.
[0007] Considering the characteristics of transmission channels with many ground points and vegetation points, few pole tower points and conductor points, and extremely uneven category distribution, the above three traditional classification methods cannot meet the requirements of real-time and accuracy at the same time. Summary of the invention
[0008] In view of the problem that traditional point cloud classification methods cannot meet the requirements of real-time and accuracy at the same time when faced with the characteristics of transmission channels with a large number of ground points and vegetation points, sparse pole tower points and conductor points, and extremely uneven category distribution, the present invention provides a three-dimensional point cloud classification method for transmission channels.
[0009] The invention discloses a three-dimensional point cloud classification method for a power transmission channel, comprising the following steps: obtaining a three-dimensional point cloud file of the power transmission channel, wherein the power transmission channel includes at least two pole towers; obtaining the coordinates of the pole towers; setting the pole tower area; determining the topography of the power transmission channel, and dividing the ground point cloud and the non-ground point cloud according to the determined topography of the power transmission channel; obtaining the plane coordinates of each non-ground point in the non-ground point cloud; dividing the non-ground point cloud into the pole tower point cloud and the non-pole tower point cloud; if the plane coordinates of the non-ground point belong to any pole tower area, the non-ground point is divided into the pole tower point, and if the plane coordinates of the non-ground point do not belong to any pole tower area, the non-ground point is divided into the non-pole tower point; determining the conductor segmentation line in the non-pole tower point cloud; and dividing the non-pole tower point cloud into the conductor point cloud and other point cloud according to the conductor segmentation line.
[0010] Compared with the prior art, the advantages and positive effects of the present invention are:
[0011] The present invention is designed based on the operation, maintenance and inspection characteristics of the transmission channel. Compared with methods such as the three-dimensional voxel model and the PointNet deep learning model, it greatly improves the efficiency and can meet the requirements of immediacy and accuracy at the same time.
[0012] After reading the specific embodiments of the present invention in conjunction with the accompanying drawings, other features and advantages of the present invention will become more clear. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0014] Figure 1 A flowchart of an embodiment of a three-dimensional point cloud classification method for a power transmission channel disclosed in the present invention;
[0015] Figure 2 For Figure 1 The first flow chart of the method for classifying the three-dimensional point cloud of the power transmission channel shown in the figure is for determining the topography of the power transmission channel and dividing the ground point cloud and the non-ground point cloud according to the determined topography of the power transmission channel;
[0016] Figure 3 For Figure 1A second flow chart for determining the topography of a transmission channel and dividing a ground point cloud and a non-ground point cloud according to the determined topography of the transmission channel in the three-dimensional point cloud classification method of the transmission channel shown;
[0017] Figure 4 A flowchart for dividing ground points and non-ground points based on terrain values in a two-dimensional array of terrain values;
[0018] Figure 5 A flowchart for determining the conductor segmentation line in a non-tower point cloud;
[0019] Figure 6 It is a schematic diagram of the three-dimensional point cloud file of the power transmission channel;
[0020] Figure 7 This is the overall effect diagram after classification;
[0021] Figure 8 It is a schematic diagram of the ground point cloud;
[0022] Fig. 9 This is a schematic diagram of the point cloud of the tower;
[0023] Fig.10 It is a schematic diagram of the point cloud of the wire;
[0024] Fig.11 Schematic diagram of other point clouds. DETAILED DESCRIPTION
[0025] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.
[0026] The terms "first", "second", "third", etc. in the specification and claims of the present invention and the drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof represent non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or devices.
[0027] In the present invention, "embodiment" means that a specific feature, structure or characteristic described in conjunction with the embodiment can be included in at least one embodiment of the present application. In the specification, the phrase does not necessarily refer to the same embodiment at various locations, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It will be understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.
[0028] In view of the problem that the traditional point cloud classification method cannot meet the requirements of real-time and accuracy at the same time when there are many ground points and vegetation points in the transmission channel, but few pole tower points and conductor points, and the category distribution is extremely uneven, an embodiment of the present invention provides a three-dimensional point cloud classification method for the transmission channel. Figure 1 As shown, the three-dimensional point cloud classification method of the transmission channel includes multiple steps.
[0029] Obtain a three-dimensional point cloud file of the scanned power transmission channel; illustratively, the three-dimensional point cloud file may be as follows: Figure 6 The three-dimensional point cloud image shown in the figure. The transmission channel includes at least two towers.
[0030] Get the coordinates of the tower;
[0031] Set the tower area.
[0032] Exemplarily, any pole tower is set as the origin pole tower, which is recorded as the origin pole tower O;
[0033] Set the origin tower area D around the origin tower O , origin tower area D O Denoted as [o x1 , o x2 , o y1 , o y2 ], that is, the origin tower area D O The four corner points D OC1 , D OC2 , D OC3 and D OC4 The coordinates are [o x1 , o y1 ],[o x1 , o y2 ],[o x1 , o y2 ],[o x2 , o y2 ]; origin tower O center point D O 'The coordinates are [o x , o y ], where o x =(o x1 +o x2 ) / 2,o y =(o y1 +o y2 ) / 2;
[0034] The tower adjacent to the origin tower O along the positive direction of the y-axis is set as the secondary tower, denoted as secondary tower T;
[0035] Set the secondary tower area T around the secondary tower T , secondary tower area TT Denoted as [t x1 , t x2 , t y1 , t y2 ], that is, the secondary tower area T T The four corner points T TC1 , T TC2 , T TC3 and T TC4 The coordinates are [t x1 , t y1 ],[t x1 , t y2 ],[t x1 , t y2 ],[t x2 , t y2 ]; The center point T of the secondary tower T T 'The coordinates are [t x , t y ], where t x =(t x1 +t x2 ) / 2,t y =(t y1 +t y2 ) / 2.
[0036] In an optional implementation, the transmission channel may also include three or more towers; that is, based on the above steps, it also includes:
[0037] The tower adjacent to the secondary tower T along the positive direction of the y-axis is set as the secondary tower, denoted as the secondary tower Q;
[0038] Set the secondary tower area Q around the secondary tower Q , the secondary tower area Q Q Denoted as [Q x1 , Q x2 , Q y1 , Q y2 ], that is, the secondary tower area Q Q The four corner points Q QC1 , Q QC2 , Q QC3 and Q QC4 The coordinates are [Q x1 , Q y1 ],[Q x1 , Q y2 ],[Q x1 , Q y2 ],[Q x2 , Q y2 ]; The center point Q of the secondary tower Q Q 'The coordinates are [q x ,qy ], where q x =(Q x1 +Q x2 ) / 2,q y =(Q y1 +Q y2 ) / 2;
[0039] When a plurality of towers are arranged in a transmission channel, the tower areas of the towers in the transmission channel are set sequentially.
[0040] Determine the topography of the transmission channel, and divide the ground point cloud and the non-ground point cloud according to the determined topography of the transmission channel;
[0041] Get the plane coordinates of each non-ground point in the non-ground point cloud;
[0042] If a non-ground point belongs to any tower area, the point is classified as a tower point; for example, if the x coordinate of one of the non-ground points is in [o x1 , o x2 ] interval and the Y coordinate is between [o y1 , o y2 ] interval, then the point is a tower point; or the X coordinate of one of the non-ground points is between [t x1 , t x2 ] interval and the Y coordinate is between [t y1 , t y2 ] interval, then this point is also the tower point;
[0043] If a non-ground point does not belong to any tower area, the point is classified as a non-tower point.
[0044] The non-ground point cloud is divided into tower point cloud and non-tower point cloud.
[0045] A conductor dividing line is determined in the non-tower point, and the non-tower point cloud is divided into a conductor point cloud and other point clouds according to the conductor dividing line.
[0046] The following is a detailed introduction to determining the topography of the transmission channel and dividing the ground point cloud and non-ground point cloud according to the topography of the determined transmission channel:
[0047] like Figure 2 As shown, a plane composed of the x-axis and the y-axis is extracted from the three-dimensional point cloud file and defined as a reference plane;
[0048] The reference plane is divided into several terrain reference intervals; in an optional implementation, the terrain reference interval is an interval of s meters × s meters, where s is a unit length, which can be set according to the accuracy. The larger the value of s, the better the noise reduction capability; the smaller the value of s, the higher the accuracy. For example, s can be set to 1 meter. Assuming that m intervals can be divided on the x-axis of the terrain reference plane and n intervals can be divided on the y-axis, there are a total of m × n terrain reference intervals, and the m × n terrain reference intervals constitute a terrain reference interval set M. The unique representation of each terrain reference interval can be obtained through the x-axis coordinates and the y-axis coordinates.
[0049] Sample the z-axis coordinates of all points in each terrain reference interval in the terrain reference interval set M, and sort the z-axis coordinates of all points in each terrain reference interval to obtain the point with the smallest z-axis coordinate in each terrain reference interval. The z-axis coordinate value of the point with the smallest z-axis coordinate is used as the terrain value of the terrain reference interval. For example, if the x-axis sequence number of a terrain reference interval is 1 and the y-axis sequence number is also 1, that is, m=1 and n=1, the z-axis coordinate of the point with the smallest z-axis coordinate in the terrain reference interval is the terrain value of the terrain reference interval, that is, g 11 The terrain values of all terrain reference intervals form a two-dimensional array, which is defined as the terrain value two-dimensional array G. The size of the terrain value two-dimensional array G is m*n. If there is no point in one of the terrain reference intervals, the terrain value of the terrain reference interval is recorded as an invalid value, that is, None.
[0050] Since the 3D point cloud file scanned by the laser radar only records the surface information of the object, it is not difficult to understand that vegetation, buildings and other objects will block the ground, so the ground points obtained from the 3D point cloud file are likely to be incomplete. Figure 3 As shown, the two-dimensional array G of terrain values is further corrected, and some erroneous terrain values therein are corrected.
[0051] The calibration process specifically includes the following steps:
[0052] A correction threshold e1 is set, and the correction threshold e1 is set according to the terrain. The higher the terrain, the larger the correction threshold e1. In an optional embodiment, the correction threshold e1 set for the power transmission channel in the mountainous terrain is greater than the correction threshold e1 set for the power transmission channel in the plain terrain.
[0053] Take any two adjacent terrain values in the two-dimensional array G of terrain values, for example, g 11 and g 12 ;
[0054] Calculate the difference between two adjacent terrain values g', g' = |g 11 -g 12 |;
[0055] If the difference g' between the terrain values is less than or equal to the correction threshold e1, the terrain value is kept unchanged;
[0056] If g 11 and / or 12 If the invalid value is None, the terrain value remains unchanged;
[0057] If the difference g' between the terrain values is greater than the correction threshold e1, then g 11 and g 12 The larger one is used for correction.
[0058] The following steps are performed during calibration:
[0059] Calculate the larger terrain value and correction value h e The difference, correction value h e Satisfy h e =|g 11 -g 12 |-e1;
[0060] The above steps are repeated until all terrain values in the terrain value two-dimensional array G no longer change, and the corrected terrain value two-dimensional array G' is obtained. G' is also the preliminary terrain of the point cloud.
[0061] In an optional implementation, ground points and non-ground points may be divided based on a two-dimensional array G of terrain values.
[0062] But more preferably, the ground points and non-ground points are divided based on the two-dimensional array G' of corrected terrain values, such as Figure 4 As shown, the specific steps include:
[0063] Set the height threshold e2;
[0064] The points in each terrain reference interval in the terrain reference interval set M are divided in turn according to the height threshold e2; the division specifically includes the following steps:
[0065] Calculate the sum sum1 of the terrain value and the height threshold e2 in each terrain reference interval, and record it as the terrain determination reference;
[0066] If the z-axis coordinate of a point in the terrain reference interval is less than the terrain determination reference sum1, then the point is a ground point;
[0067] If the z-axis coordinate of a point in the terrain reference interval is greater than or equal to the terrain determination reference sum1, the point is a non-ground point.
[0068] For example, it is assumed that in the two-dimensional array G' of corrected terrain values, the terrain value of the terrain reference interval (1, 1) is g 11, then in the terrain reference interval (1, 1), all z-axis coordinates are less than g 11 The points with +e2 are all ground points, and the rest are non-ground points.
[0069] Thereby obtaining the ground point cloud and non-ground point cloud respectively.
[0070] The non-ground point cloud is further divided into tower points and non-tower points.
[0071] The conductor segmentation line is further determined in the non-tower point cloud, and the non-tower points are divided into conductor points and other points, thereby realizing the division of ground points, tower points, conductor points and other points.
[0072] The following introduces the detailed process of determining the conductor segmentation line in the non-tower point and dividing the non-tower point cloud into conductor points and other points.
[0073] In some embodiments of the present invention, determining the conductor segmentation line at the non-tower point includes: Figure 5 The following steps are shown:
[0074] Perform preliminary filtering based on the x-axis coordinate of the tower area.
[0075] The preliminary filtering based on the x-axis coordinate of the tower area includes:
[0076] Set Q x1 , Q x2 , t x1 , t x2 The minimum value in x left ;
[0077] Set Q x1 , Q x2 , t x1 , t x2 The maximum value in is x right ;
[0078] Compare the x-axis coordinates of the non-tower points in the non-tower point cloud with the minimum value x left Compare;
[0079] Compare the x-axis coordinates of the non-tower points in the non-tower point cloud with the maximum value x right Compare;
[0080] If the x-axis coordinate of the non-tower point is greater than x right or less than x left , then the non-tower points are divided into other points and removed from the non-tower point cloud.
[0081] In the y-axis direction, the non-tower point cloud after step filtering is divided into multiple non-tower point intervals with a step length of s meters; assuming that a total of k non-tower point intervals are divided, the k non-tower point intervals constitute a non-tower point interval set K.
[0082] Further classification is carried out based on the two characteristics that the conductor is higher than other objects in the transmission channel and the conductor changes continuously, which specifically includes the following steps:
[0083] Sort the z-axis coordinates of the non-tower points in each non-tower point interval in the non-tower point interval set K, determine the point with the largest z-axis coordinate, and record it as the maximum height point;
[0084] The z-axis coordinate of the maximum height point in each non-tower point interval is set as the conductor value of the non-tower point interval; if there is no point in the non-tower point interval, the conductor value of the non-tower point interval is recorded as None.
[0085] The conductor values of all non-tower point intervals form an array, which is recorded as the conductor value array C; the size of the conductor value array C is k.
[0086] Determine whether the wire value array C changes continuously, specifically including:
[0087] Calculate the proportion of invalid values (i.e. None) in the wire value array C;
[0088] If the proportion of invalid values is greater than or equal to the set ratio, it is considered that the wire value array C changes discontinuously; the set ratio can be 30%;
[0089] If the invalid value ratio is less than the set ratio, two consecutive wire values are randomly selected from the wire value array C, exemplarily denoted as c i and c i+1 ;
[0090] Calculate the absolute value of the difference between two consecutive wire values;
[0091] The absolute value of the difference between two consecutive wire values is compared with the boundary threshold value e3;
[0092] If the difference between two consecutive wire values is greater than the boundary threshold, |c i -c i+1 |>e3, the wire value array C is considered to change discontinuously;
[0093] Traverse the wire value array C. If there is no case where the difference between two consecutive wire values is greater than the boundary threshold, there is no |c i -c i+1 |>e3, the wire value array C is considered to change continuously;
[0094] In the process of traversing the wire value array C, if ci and / or c i+1 If it is None, skip this group of wire values and execute the judgment of the next group of wire values;
[0095] If the wire value array C changes continuously, then the wire value array C can represent the direction of the wire;
[0096] Since there are multiple groups of conductors on the tower, the above steps are further executed in a loop starting from dividing the non-tower point interval until the conductor value array C is determined to be discontinuously changed, and the last group of continuously changed conductor value array C is the conductor segmentation line, that is, the conductor segmentation line array D.
[0097] The linear interpolation method is used to assign invalid values (None) in the wire segmentation line array D to obtain the corrected wire segmentation line array D'.
[0098] The wire points and non-wire points are divided based on the corrected wire segmentation line array D', specifically including the following steps:
[0099] Set the wire threshold e4;
[0100] The points in each non-tower point interval in the non-tower point interval set K are divided in turn according to the conductor threshold value e4; the division specifically includes the following steps:
[0101] Calculate the sum of the conductor value and the conductor threshold e4 in each non-tower point interval, and record it as the conductor determination benchmark;
[0102] If the z-axis coordinate of a point in the non-tower point interval is less than or equal to the conductor determination criterion sum2, then the point is another point;
[0103] If the z-axis coordinate of a point in the non-tower point interval is greater than the conductor determination benchmark sum2, then the point is a conductor point.
[0104] Further, we can obtain the wire point cloud and other point clouds.
[0105] This enables the four-category division of ground points, tower points, conductor points and other points.
[0106] It is not difficult to understand that there may be other auxiliary components such as auxiliary conductors and spacers on the conductor. These components also belong to the conductor but often extend in a different direction from the conductor. Under the premise of considering the auxiliary components, the conductor dividing line D is determined through the following steps.
[0107] If the wire value array C changes continuously, set the auxiliary component threshold e5;
[0108] Calculate the wire value array C, C = {c1, c2, ..., c k} and the threshold value e5 of the accessory component, that is, each point in the wire value C is reduced by e5 meters as a whole, and the modified wire value array C1 is obtained, C1 = {c1-e5, c2-e5, ..., c k -e5};
[0109] Determine whether there is a non-tower point in the corresponding non-tower interval whose z-axis coordinate is between the value of the modified conductor value array C1 and the corresponding value of the conductor value array C, that is, between [(c k -e5), c k ]between.
[0110] If there is, replace the corresponding value in the wire value array C with the value corresponding to the modified wire value array C1;
[0111] If None, leave the array of wire values unchanged.
[0112] Repeat the above process until there is no non-tower point whose z-axis coordinate is between the value of the modified conductor value array C and the corresponding value of the conductor value array C. The conductor value array C after the assignment is recorded as the conductor value array, and the step of dividing the non-tower point interval is performed until the conductor value array C is discontinuous, and the conductor segmentation line array D is obtained.
[0113] Figure 6 The figure is an example of a three-dimensional point cloud file of an exemplary acquired scanned transmission channel, wherein the total number of point cloud points is 2,452,826, including two pole towers.
[0114] Set the origin tower according to the positive direction of the y-axis and set the origin tower area D O Denoted as [o x1 , o x2 , o y1 , o y2 ], that is [-13.18,12.36,-2.41,4.05];
[0115] Set another tower as the secondary tower, record it as secondary tower T, set the secondary tower area T T Denoted as [t x1 , t x2 , t y1 , t y2 ], that is [-13.57,12.17,262.92,268.9];
[0116] The point cloud in the reference plane is divided into several terrain reference intervals, each of which is an interval of s meters × s meters, and s can be optionally set to 1 meter.
[0117] The minimum value of the X coordinate in the three-dimensional point cloud is -45.72, the maximum value is 45.0, the minimum value of the Y coordinate is -5.29, and the maximum value is 299.84; therefore, 91 intervals can be divided in the x-axis direction, and 306 intervals can be divided in the y-axis direction, that is, m = 91, n = 306, and the total number of terrain reference intervals is 27846.
[0118] The terrain reference interval set M can be expressed as:
[0119] M={[-45.72,-5.29),[-45.72,-4.29),[-45.72,-3.29),[-45.72,-2.29),[-45.72,-1.2 9),[-45.72,-0.29),[-45.72,0.71),[-45.72,1.71),[-45.72,2.71),[-45.72,3.71),[ -45.72,4.71),[-45.72,5.71),[-45.72,6.71),[-45.72,7.71),[-45.72,8.71),[-45.72,9.71),[-45.72,10.71),[-45.72,11.71),[-45.72,12.71),[-45.72,13.71),......}
[0120] Find the point with the smallest z-axis coordinate in each terrain reference interval and establish a two-dimensional array G of terrain values.
[0121] The two-dimensional array G of terrain values can be expressed as G={[-2.52,-3.5,-3.51,-5.49,-8.72,-8.61,-14.24,-14.49,5.23,5.25,5.27,5.26,5.34,5.31,5.36,5.35,5.34,5.33,5.33,5.39,......],......};
[0122] Set the correction threshold e1 to 3, cyclically adjust the terrain values in G, and obtain the two-dimensional array G' of corrected terrain values;
[0123] The two-dimensional array G' of corrected terrain values can be expressed as G'={[-2.52,-3.5,-3.51,-5.49,-7.23,-4.23,-1.23,2.23,5.23,5.25,5.27,5.26,5.34,5.31,5.36,5.35,5.34,5.33,5.33,5.39,......],......};
[0124] The height threshold e2 is set to 0.5, and each point in the terrain reference interval is divided.
[0125] After ground division, there are 1,104,383 ground points and 1,348,443 non-ground points.
[0126] The points whose x-axis coordinates are in the interval [-13.18, 12.36] and whose y-axis coordinates are in the interval [-2.41, 4.05] are classified as tower points;
[0127] The points whose x-axis coordinates are in the interval [-13.57, 12.17] and whose y-axis coordinates are in the interval [262.92, 268.9] are classified as tower points;
[0128] After the pole tower division, there are 21,359 pole tower points and 1,327,084 non-pole tower points.
[0129] Preliminary filtering is performed based on the x-axis coordinates of the tower area. Among the non-tower points, if the x-axis coordinates are less than -13.57 or greater than 12.36, they are classified as other points, and the points classified as other points are removed from the non-tower point cloud.
[0130] In the y-axis direction, the non-tower point cloud after step filtering is divided into multiple non-tower point intervals with a step length of 1 meter. The non-tower point interval set K can be expressed as:
[0131] K={[-5.29,-4.29),[-4.29,-3.29),[-3.29,-2.29),[-2.29,-1.29),[-1.29,- 0.29),[-0.29,0.71),[0.71,1.71),[1.71,2.71),[2.71,3.71),[3.71,4.71),[ 4.71,5.71),[5.71,6.71),[6.71,7.71),[7.71,8.71),[8.71,9.71),[9.71,10. 71),[10.71,11.71),[11.71,12.71),[12.71,13.71),[13.71,14.71),...}.
[0132] Find the maximum height point of each non-tower interval and establish the conductor value array C, C = [50.89, 50.62, 50.37, 50.08, 49.83, 49.57, 49.32, 49.06, 48.78, 48.55, 48.29, 48.05, 47.8, 47.55, 47.32, 47.07, 46.83, 46.6, 46.37, 46.12, ...];
[0133] Set the boundary threshold e3 to 2 to determine whether the wire value array C changes continuously;
[0134] Under the premise of considering the attached components, some values in the wire value array C are assigned. The wire value array C after the assignment is expressed as, C = [48.89, 48.62, 48.37, 48.08, 47.83, 47.57, 47.32, 47.06, 46.78, 46.55, 46.29, 46.05, 45.8, 45.55, 45.32, 45.07, 44.83, 44.6, 44.37, 44.12, ...];
[0135] The above steps are executed cyclically starting from the division of the non-tower point interval until the conductor value array C is determined to be discontinuously changing, and the last set of conductor value array C with continuous changes is taken as the conductor dividing line D.
[0136] Wire dividing line D = [38.58, 38.29, 38.01, 37.72, 37.42, 37.16, 36.88, 36.62, 36.35, 36.09, 35.82, 35.57, 35.31, 35.03, 34.78, 34.54, 34.27, 34.04, 33.77, 33.52, ...]
[0137] The linear interpolation method is used to assign the invalid value (None) in the wire segmentation line array D to obtain the corrected wire segmentation line array D'.
[0138] The wire points and non-wire points are divided based on the corrected wire segmentation line array D'.
[0139] After the wires are divided, there are 23,805 wire points and 1,303,595 other points.
[0140] The overall effect after classification is as follows Figure 7 As shown, the ground point cloud is Figure 8 As shown, the tower point cloud is as follows Fig. 9 As shown, the wire point cloud is shown in the figure, and the other point clouds are shown in the figure. Through the three-dimensional point cloud classification of the transmission channel performed in this embodiment, the classification effect achieved is basically close to the ground, towers, and wires in the real scene. The running time to complete this embodiment is 25 seconds, which greatly improves the efficiency compared to methods such as three-dimensional voxel model and PointNet deep learning model, and can meet the requirements of immediacy and accuracy at the same time.
[0141] The present application also provides a system that applies the above-mentioned transmission channel three-dimensional point cloud classification method. The specific steps of the transmission channel three-dimensional point cloud classification method refer to the detailed description of the above-mentioned embodiment and the detailed description of the drawings in the specification. No further description is given here. The system that adopts the above-mentioned transmission channel three-dimensional point cloud classification method can achieve the same technical effect.
[0142] An embodiment of the present application also provides a computer storage medium, wherein the computer storage medium stores a computer program for electronic data exchange, and the computer program enables a processor to execute part or all of the steps of any method recorded in the above method embodiments.
[0143] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0144] In the several embodiments provided in the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the above-mentioned units or modules, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical or other forms.
[0145] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one physical space, or may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.
[0146] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0147] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for a person skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to replace some of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions claimed to be protected by the present invention.
Claims
1. Classification method of three-dimensional point cloud of power transmission channel, It is characterized in that The following steps are involved: Acquire a three-dimensional point cloud file of a power transmission channel, wherein the power transmission channel includes at least two towers; Obtaining the coordinates of the tower; Set the tower area; Determine the topography of the transmission channel, and divide the ground point cloud and the non-ground point cloud according to the determined topography of the transmission channel; Get the plane coordinates of each non-ground point in the non-ground point cloud; Divide the non-ground point cloud into the tower point cloud and the non-tower point cloud; If the plane coordinates of the non-ground point belong to any tower area, the non-ground point is divided into a tower point; if the plane coordinates of the non-ground point do not belong to any tower area, the non-ground point is divided into a non-tower point; determine the wire segmentation line in the non-tower point point cloud; Dividing the non-tower point cloud into a conductor point cloud and other point clouds according to the conductor segmentation line; Determining the conductor segmentation line in the non-tower point cloud includes the following steps: Perform preliminary filtering on the non-tower point cloud according to the x-axis coordinate of the tower area; In the y-axis direction, the filtered non-tower point cloud is divided into a plurality of non-tower point intervals; A plurality of non-tower point intervals constitute a non-tower point interval set; Sort the z-axis coordinates of the non-tower points in each non-tower point interval in the non-tower point interval set, determine the point with the largest z-axis coordinate, and record it as the maximum height point of the corresponding non-tower interval; Set the z-axis coordinate of the maximum height point in each non-tower point interval as the conductor value of the non-tower point interval; The conductor values of all non-tower point intervals form a conductor value array; Determine whether the wire value array changes continuously; If the conductor value array changes continuously, the above steps are executed cyclically starting from the step of dividing the filtered non-tower point cloud into multiple non-tower point intervals in the y-axis direction until the conductor value array changes discontinuously, and the last set of continuously transformed conductor value arrays is the conductor segmentation line; The process of determining whether the wire value array changes continuously includes the following steps: If there is no point in the non-tower point interval, the conductor value of the non-tower point interval is recorded as an invalid value; Calculate the proportion of invalid values in the wire value array; If the proportion of invalid values is greater than or equal to the set ratio, the wire value array is considered to be discontinuously changing; If the invalid value ratio is less than the set ratio, two consecutive wire values are randomly selected from the wire value array; Calculate the absolute value of the difference between two consecutive wire values; The absolute value of the difference between two consecutive wire values is compared to a boundary threshold; If the difference between two consecutive wire values is greater than the boundary threshold, the wire value array changes discontinuously; The wire value array is traversed, and if there is no case where the difference between two consecutive wire values is greater than the boundary threshold, the wire value array is considered to change continuously.
2. The three-dimensional point cloud classification method for power transmission channels according to claim 1, Features: Determining the topography of the transmission channel, and dividing the ground point cloud and the non-ground point cloud according to the determined topography of the transmission channel includes the following steps: Extract the plane composed of the x-axis and the y-axis in the 3D point cloud file and define it as the reference plane; A plurality of terrain reference intervals are divided in the reference plane, and the plurality of terrain reference intervals constitute a terrain reference interval set; Sampling the z-axis coordinates of all points in each terrain reference interval in the terrain reference interval set, sorting the z-axis coordinates of all points in each terrain reference interval, obtaining the point with the smallest z-axis coordinate in each terrain reference interval, and taking the z-axis coordinate value of the point with the smallest z-axis coordinate as the terrain value of the terrain reference interval; The terrain values of all terrain reference intervals form a two-dimensional array of terrain values; Divide ground points and non-ground points based on the terrain values in the terrain value 2D array: Set the height threshold; Calculate the sum of the terrain value and the height threshold in each terrain reference interval, and record it as the terrain determination reference; If the z-axis coordinate of a point in the terrain reference interval is less than the terrain determination reference, then the point is a ground point; If the z-axis coordinate of a point in the terrain reference interval is greater than or equal to the terrain determination reference, the point is a non-ground point; Get the ground point cloud and non-ground point cloud.
3. The three-dimensional point cloud classification method for power transmission channels according to claim 2, Features: Determining the topography of the transmission channel, and dividing the ground point cloud and the non-ground point cloud according to the determined topography of the transmission channel also includes the following steps: Set the correction threshold; Randomly select two adjacent terrain values in the two-dimensional array of terrain values; Calculate the difference between two adjacent terrain values; If the difference between two adjacent terrain values is less than or equal to the correction threshold, the terrain value in the two-dimensional array of terrain values is kept unchanged; If the difference between two adjacent terrain values is greater than the correction threshold, the larger one of the two adjacent terrain values is corrected: the difference between the larger terrain value and the correction value is calculated, the correction value is the difference between the absolute value of the difference between the two adjacent terrain values and the correction threshold, and the larger terrain value is replaced by the difference between the larger terrain value and the correction value; Repeat the above steps until all terrain values in the terrain value two-dimensional array satisfy that the difference between two adjacent terrain values is less than or equal to the correction threshold, thereby obtaining a corrected terrain value two-dimensional array; Classify ground points and non-ground points based on the terrain values in the 2D array of corrected terrain values: Set the height threshold; Calculate the sum of the terrain value and the height threshold in each terrain reference interval, and record it as the terrain determination reference; If the z-axis coordinate of a point in the terrain reference interval is less than the terrain determination reference, then the point is a ground point; If the z-axis coordinate of a point in the terrain reference interval is greater than or equal to the terrain determination reference, the point is a non-ground point; Get the ground point cloud and non-ground point cloud.
4. The three-dimensional point cloud classification method for power transmission channels according to claim 3, Features: If there is no point in one of the terrain reference intervals, the terrain value of the terrain reference interval is recorded as an invalid value; If there is an invalid value between two adjacent terrain values, the terrain value remains unchanged.
5. The three-dimensional point cloud classification method for power transmission channels according to claim 1, Features: The preliminary filtering of the non-tower point cloud based on the x-axis coordinates of the tower area includes the following steps: Determine the minimum x-axis coordinate of the tower area; Determine the maximum x-axis coordinate of the tower area; Compare the x-axis coordinate of the non-tower point in the non-tower point cloud with the minimum value; Compare the x-axis coordinate of the non-tower point in the non-tower point cloud with the maximum value; If the x-axis coordinate of the non-tower point is greater than the maximum value or less than the minimum value, the non-tower point is divided into other points and removed from the non-tower point cloud.
6. The three-dimensional point cloud classification method for power transmission channels according to claim 1, Features: Dividing the non-tower point cloud into a conductor point cloud and other point clouds according to the conductor segmentation line comprises the following steps: Using linear interpolation method to assign invalid values in the wire segmentation line array, to obtain a corrected wire segmentation line array; Dividing the wire points and non-wire points based on the corrected wire segmentation line array; Set the wire threshold; Calculate the sum of the conductor value and the conductor threshold in each non-tower point interval and record it as the conductor determination benchmark; If the z-axis coordinate of a point in the non-tower point interval is less than or equal to the conductor determination datum, then the point is another point; If the z-axis coordinate of a point in the non-tower point interval is greater than the conductor determination benchmark, then the point is a conductor point.
7. The three-dimensional point cloud classification method for power transmission channels according to claim 1, Features: The following steps are also included: If the wire value array is continuously changing, set the threshold value of the attached component; Calculate the difference between each wire value in the wire value array and the threshold value of the attached component to obtain a modified wire value array; Determine whether there is a non-tower point in the corresponding non-tower interval whose z-axis coordinate is between the value of the corrected conductor value array and the corresponding value of the conductor value array; If there is, replace the corresponding value in the wire value array with the value corresponding to the modified wire value array; If not, leave the array of wire values unchanged; The above process is repeated until there is no non-tower point whose z-axis coordinate is between the value of the corrected conductor value array and the corresponding value of the conductor value array. The reassigned conductor value array is recorded as the conductor value array.
8. The three-dimensional point cloud classification method for power transmission channels according to claim 1, It is characterized in that Before determining the topography of the transmission channel, the following steps are also included: Set the origin tower; Set origin tower area D O , origin tower area D O Denoted as [o x1 , o x2 , o y1 , o y2 ], the center point of the origin tower D O 'The coordinates are [o x , o y ] and meet , ; Set the tower adjacent to the origin tower O along the positive direction of the y-axis as the secondary tower; Set the secondary tower area T T , secondary tower area T T Denoted as [t x1 , t x2 , t y1 , t y2 ], the center point T of the secondary tower T T 'The coordinates are [t x , t y ] and meet , .
Citation Information
Cited By
Classification method, system and equipment for three-dimensional point clouds of power transmission channels
CN121259377A
