CN112991303B - Automatic extraction method of electric tower insulator string based on three-dimensional point cloud - Google Patents

Automatic extraction method of electric tower insulator string based on three-dimensional point cloud Download PDF

Info

Publication number
CN112991303B
CN112991303B CN202110300702.1A CN202110300702A CN112991303B CN 112991303 B CN112991303 B CN 112991303B CN 202110300702 A CN202110300702 A CN 202110300702A CN 112991303 B CN112991303 B CN 112991303B
Authority
CN
China
Prior art keywords
point
point cloud
layer
tower
dimensional
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110300702.1A
Other languages
Chinese (zh)
Other versions
CN112991303A (en
Inventor
黄玉春
欧阳凝晖
彭祖铭
王浩宇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan University WHU
Original Assignee
Wuhan University WHU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Wuhan University WHU filed Critical Wuhan University WHU
Priority to CN202110300702.1A priority Critical patent/CN112991303B/en
Publication of CN112991303A publication Critical patent/CN112991303A/en
Application granted granted Critical
Publication of CN112991303B publication Critical patent/CN112991303B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T19/00—Manipulating 3D models or images for computer graphics
    • G06T19/20—Editing of 3D images, e.g. changing shapes or colours, aligning objects or positioning parts
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/06—Topological mapping of higher dimensional structures onto lower dimensional surfaces
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20112—Image segmentation details
    • G06T2207/20132—Image cropping
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30181—Earth observation
    • G06T2207/30184—Infrastructure
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Computer Graphics (AREA)
  • Architecture (AREA)
  • Software Systems (AREA)
  • Quality & Reliability (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an automatic extraction method of an electric tower insulator string based on three-dimensional point cloud. Firstly, preprocessing an original three-dimensional point cloud according to a roughly set electric tower center coordinate, wherein the preprocessing comprises clipping of the point cloud so as to reduce data volume and reduce interference of irrelevant point clouds. For the point cloud after the pretreatment, a slip angle kappa with respect to the coordinate axis XY is calculated, and then the point cloud is rotated according to the kappa angle so that the Y coordinate axis coincides with the electric tower direction. Layering the rotated point cloud according to a certain height, judging the type of each layer of wire insulator string, and then treating the top layer, the non-top layer vertical strings and the non-top layer horizontal strings by adopting different methods to respectively obtain top layer hanging points, vertical insulator strings and horizontal insulator strings. In each link of point cloud processing, the three-dimensional problem is converted into the two-dimensional image processing problem by a projection method, so that the operation efficiency is improved, and the method has higher robustness and accuracy for various different types of electric towers.

Description

Automatic extraction method of electric tower insulator string based on three-dimensional point cloud
Technical Field
The invention belongs to the field of power line inspection, and particularly relates to an automatic extraction method of an electric tower insulator string based on three-dimensional point cloud.
Background
Along with unmanned aerial vehicle's wide use in the power line inspection, changed the outdoor electric power inspection in the past and relied on the manpower walking, climbing, the manpower occupation is big and the problem of overall efficiency is not high. The automatic inspection at the present stage mainly adopts a route flight mode, and the inspection precision is ensured in an auxiliary closed loop process. And the unmanned aerial vehicle automatically patrols and examines, need to plan corresponding flight path according to the space position information of power line part to the automatic part that needs to patrol and examine is shot.
The insulator string of the electric tower is an important content of power line inspection, and three-dimensional coordinate information of the insulator string is an important basis for unmanned aerial vehicle inspection line planning. Because of the numerous electric towers in the power transmission line, the distribution is wide and the topography is complex, an automatic method is needed to extract insulator strings in the electric towers, thereby obtaining relevant three-dimensional information.
Disclosure of Invention
Aiming at the existing background demand, the invention provides an automatic extraction method of an electric tower insulator string based on three-dimensional point cloud, thereby solving the technical problems that the electric towers are large in number, wide in distribution and complex in topography, and the insulator string is difficult to extract.
In order to achieve the above object, according to one aspect of the present invention, an automatic extraction method for an insulator string of an electric tower based on a three-dimensional point cloud is provided, and the biggest innovation point of the present invention is that a large amount of projection methods are used, so that three-dimensional point cloud processing is converted into two-dimensional image processing contents, the problem of insulator string extraction is simplified, and the operation efficiency is greatly improved. Comprising the following steps:
Step 1: acquiring three-dimensional point cloud data of the electric tower and the center of the electric tower, and further cutting the three-dimensional point cloud data of the electric tower in the horizontal XY direction and in the vertical Z direction to obtain cut point clouds;
preferably, the three-dimensional point cloud data of the electric tower in the step 1 is:
data 0 (x,y,z)＝{data 0 k (x k ,y k ,z k ),k∈[1,K]}
wherein, data 0 (x, y, z) is three-dimensional point cloud data of the electric tower, data 0 k (x k ,y k ,z k ) Is the pixel of the kth point in the three-dimensional point cloud data of the electric tower, x k Is the X-direction coordinate, y of the kth point in the three-dimensional point cloud data of the electric tower k Is the Y-direction coordinate, z of the kth point in the three-dimensional point cloud data of the electric tower k The Z-direction coordinate of the kth point in the three-dimensional point cloud data of the electric tower is obtained, and K is the number of the three-dimensional points in the three-dimensional point cloud data of the electric tower;
the horizontal XY direction cutting in the step 1 is as follows:
based on the centre of the tower 0 (x 0 ,y 0 ,z 0 )；
Data of three-dimensional point cloud data of electric tower 0 (x, y, z) search for abs (x) k -x 0 )＜threshold 0 And abs (y) k -y 0 )＜threshold 0 Constructing point cloud after horizontal clipping as data 1 (x,y,z)，threshold 0 Is a threshold parameter;
the vertical Z-direction cutting in the step 1 is as follows:
point cloud, namely data after being combined with horizontal cutting 1 (x, y, z), searching for a satisfying abs (x k -x 0 )＜threshold 1 And abs (y) k -y 0 )＜threshold 1 The minimum value of the Z-direction coordinate in the point cloud in the range is defined as Z min The method comprises the steps of carrying out a first treatment on the surface of the Searching for satisfying abs (x k -x 0 )＜threshold 1 And abs (y) k -y 0 )＜threshold 1 The maximum value of the Z-direction coordinate in the point cloud in the range is defined as Z max ，threshold 1 Is a threshold parameter;
defining a layering interval as z_interval=1.5;
the number of layers is calculated as follows:
z_num＝ceil[(z max -z min )/z_interval]
wherein ceil represents an upward rounding function, and z_num represents the number of layers;
defining the point cloud density of each layer as follows:
z_density＝{z_density l ,l∈[1,z_num]}
and initializes z-density l =0, and layer 1 is the highest layer;
traversing data 1 (x, y, z) according to z_now=ceil [ (z) max -z k )/z_interval]Calculating a layer z_now to which a kth point belongs, and counting the number of points of each layer;
from layer 1, find a satisfying z_density a+s ＞10，s＝[1,6]Calculating a reserved value z_up in the Z direction according to z_up=z_max- (a-1) z_interval;
point cloud data 1 (x, y, z) satisfies z k ≥z_min+threshold 2 And z k Point reservation of z_up, updating the highest point z_max=z_up, and the lowest point z_min=z_min+threshold of the point cloud 2 ，threshold 2 Is a threshold parameter.
Finally obtaining the tailored point cloud data 2 (x,y,z)
Step 2: for the point cloud data after clipping 2 (x, y, z) acquiring point cloud data of a certain range of the topmost layer 3 (x, y, z) calculating the slip angle kappa, and then adding the point cloud data 2 The (x, Y, z) rotation enables the tower direction of the electric tower to be consistent with the Y direction of the coordinate axis, and the point cloud data is obtained 4 (x,y,z)；
Preferably, step 2 includes:
step 2.1: from point cloud data 2 The (x, y, z) highest point goes down, set the layering interval z_interval 1 =0.1, calculate the number of layers z_num 1 ＝ceil[(z max -z min )/z_interval 1 ]Defining the density z-density of each layer of point cloud 1 ＝{z_density 1 k ,k∈[1,z_num 1 ]And initialize z_density 1 k =0, and layer 1 is the highest layer. Traversing point cloud data 2 (x, y, z) according to z_now=ceil [ (z) max -z k )/z_interval 1 ]And calculating the layer to which the kth point belongs, and counting the number of points of each layer. From layer 1, find a satisfying z_density 1 k Layer k greater than 100, calculating the maximum value Z of the top layer point cloud in the Z direction top Max, point cloud data 2 (x, y, z) reservation satisfies z k ≥z top Max-3 and z k ≤z top The point of _max is taken as top level point cloud data 3 (x,y,z)。
Step 2.2: for the top layer point cloud data 3 (x, y, z) downward projection to obtain XY projection map (the projections mentioned in this invention are all binarized projections, where the XY projections are generated as follows, otherwise similar 3 The maximum and minimum of the XY coordinates of (x, y, z) are set to a two-dimensional grid interval of grid_interval=0.1, and the grid_interval is set according to row=ceil [ (x_max-x_min)/grid_interval ]]Calculating the number of XY projection lines, col=ceil [ (y_max-y_min)/grid_interval ]]Calculating XY projectionsDrawing column number, traversing point cloud data 3 (x, y, z) according to row_now=ceil [ (x_max-x) k )/grid_interval]Calculating the line number of the current point falling on the projection graph according to col_now=ceil [ (y_max-y) k )/grid_interval]Calculating the column number of the current point falling in the projection graph, if a pixel has a point cloud falling in the range, the pixel value is 255, otherwise 0), carrying out morphological filling and corrosion operation on the XY projection, and extracting the maximum outline. The rotation angle kappa is calculated from the first principal direction of the maximum profile using PCA.
Step 2.3: according to the calculated rotation deflection angle kappa, the clipped point cloud data 2 (x, y, z) centering the tower with the center 0 (x 0 ,y 0 ,z 0 ) As a rotation center, a coordinate system Z axis is taken as a rotation axis, and a rotation angle is calculated according to a rotation deflection angle kappa to rotate, so that the direction of a tower is consistent with the direction of a coordinate system Y axis, and a rotated point cloud data is obtained 4 (x,y,z)。
Step 3: for the rotated point cloud data 4 (x, y, Z) projecting to obtain ZX projection diagram, layering according to wire height to obtain approximate Z position z_layer of crossing wire at two sides of each layer of electric tower k K is the number of layers obtained by layering;
preferably, step 3 includes:
step 3.1: and (5) projecting to obtain a ZX projection map. And finding out the column at the center of the tower and the column at the left and right boundaries of the tower according to the fact that the number of the points of the column at which the electric towers are located is greater than the number of the points of the column at which the electric wires are located, and setting the pixel value to 0 to remove the electric towers.
Step 3.2: and taking the lowest point of the ZX image as a seed point, growing to obtain the interference of ground residual tree barriers and the like, and setting the pixel value of the part of points to 0, so that the rest part of the image basically only comprises electric wires.
Step 3.3: and extracting edges of the ZX projection image, detecting a Hough straight line of the edge image, and screening according to the vertical distance of the slope of the straight line, so that the position of each layer of electric wire on the line where the central column of the tower is located is obtained.
Step 4: based on the position Z of each layer obtained by layering, cutting to obtain point clouds of each layer, wherein the uppermost layer is firstly used as a top layer, and judging the type of an insulator string in the non-top layer to be divided into a vertical string layer and a horizontal string layer;
preferably, step 4 includes:
step 4.1: the uppermost layer is regarded as the top layer and does not participate in type judgment. And for the non-top layer, respectively taking a small-range point cloud from top to bottom at the calculated Z position, and projecting downwards to obtain an XY projection graph.
Step 4.2: and counting the continuous condition of the pixel points by using the tower center to be located in a certain range of the upper row and the lower row, and if the continuous condition indicates that the layer has a tower, belonging to a horizontal string layer, otherwise belonging to a vertical string layer.
Step 5: based on the point cloud of each layer of the electric tower and the category of the electric tower, different methods are adopted for processing, finally, a top layer obtains suspension points, a vertical string layer obtains a corresponding vertical insulator string, and a horizontal layer obtains a corresponding horizontal insulator string;
preferably, step 5 includes:
step 5.1: the uppermost layer of the layering result of the electric tower is regarded as the top layer, and suspension points are extracted by processing
Step 5.2: for the non-top layer of the electric tower layering result, judging the type as the part of the vertical string layer, and processing to extract the horizontal insulator string
Step 5.3: for the non-top layer of the electric tower layering result, judging the type as the part of the horizontal string layer, and processing to extract the vertical insulator string
Preferably, step 5.1 comprises:
step 5.1.1: and obtaining a ZX projection graph by top-layer point cloud projection, counting the number of points in each row, finding out the boundary row of the area with the maximum density, and calculating the X coordinate of the hanging point according to the average value of the boundary row.
Step 5.1.2: starting from the boundary columns found in step 5.1.1, respectively, to the two sides, until a column with a sufficiently small width is found, calculating the Z coordinate of the hanging point according to the starting row of the column.
Step 5.1.3: and (5) carrying out point cloud projection to obtain an XY projection map. The Y coordinate of the hanging point is calculated by moving to the row of the hanging point along the electric wire from the upper end and the lower end.
Step 5.1.4: and (3) rotating the calculated three-dimensional coordinates of the hanging points by a kappa angle according to the rotating direction of the step (2) to obtain the three-dimensional coordinates of the hanging points under the original coordinate system.
Preferably, step 5.2 comprises:
step 5.2.1: and the point cloud is projected downwards to obtain an XY projection graph, the number of the wires is counted according to the intervals among different wires in the same layer, and the range of each wire Y is obtained.
Step 5.2.2: and (3) carrying out point cloud projection on each electric wire to obtain a ZX projection graph, filling, thinning and deburring the ZX projection graph, and then searching a tri-bifurcation point for the thinned graph, wherein the bifurcation point is an intersection point of the insulator string and the electric wire, so that the ZX coordinate of the intersection point is calculated.
Step 5.2.3: and taking a certain range of point clouds by rays corresponding to the intersection point ZX, and taking the average value of the cluster point clouds Y as the Y coordinate of the intersection point, so that the three-dimensional coordinate of the intersection point is obtained.
Step 5.2.4: and according to the three-dimensional coordinates of the intersection point, a certain range of point clouds are upwards taken to obtain ZY projection, a row with a certain width, namely the position of a hanging point Z, is found in the row starting direction from the intersection point, and the XY coordinates of the hanging point are consistent with the XY coordinates of the intersection point, so that the three-dimensional coordinates of the hanging point are obtained.
Step 5.2.5: and (3) rotating the three-dimensional coordinates of the vertical strings under the original coordinate system according to the rotation-kappa angle of the rotation direction of the step (2) according to the three-dimensional coordinates of the vertical string intersection point and the suspension point calculated by each electric wire.
Preferably, step 5.3 comprises:
step 5.3.1: and the point cloud projects downwards to obtain an XY projection graph, and the part outside a certain range above and below the central line of the tower is cut and removed in the XY projection.
Step 5.3.2: and analyzing the rest of cutting, namely solving 4 corner points on the inner sides of the H-shaped pattern which are corner points on two sides of the two connected domains, and slightly expanding the corner point coordinates to obtain the XY coordinates of the suspension points.
Step 5.3.3: and cutting according to the XY coordinates of the suspension points and the position of the center X of the tower to obtain a point cloud, wherein a part of the point cloud only comprises the tower without the electric wire and the insulator string, projecting to obtain a ZY projection graph, and finding out the line with the number of 255 points and jumping as the line where the Z coordinates of the horizontal suspension points are located.
Step 5.3.4: the layer of point cloud projection obtains a ZX projection graph, and the line-by-line traversal and recording of the highest line of the electric wire, the lowest line of the electric wire and the highest line of the suspended object are carried out according to the principle that the left suspension point is found left and the right suspension point is found right; and taking the difference between the wire highest behavior reference condition, the wire lowest line and the suspension highest line as judgment conditions, and carrying out classified discussion and fitting to obtain the ZX value.
Step 5.3.5: cutting the layer of point cloud according to the ZX coordinates of the cross-string intersection points, taking a certain range of point cloud by rays corresponding to the intersection points ZX, dividing the point cloud into two clusters according to the Y coordinates of the cluster point cloud, and taking the average value of the point cloud Y of each cluster as the Y coordinates of each cross-string intersection point.
Step 5.3.6: and (3) according to the three-dimensional coordinates of each cross string intersection point and the suspension point calculated by the layer of point cloud, rotating by a kappa angle according to the rotating direction of the step (2) to obtain the three-dimensional coordinates of the vertical string under the original coordinate system.
Step 6: in view of the problem of incomplete data, to enhance robustness, a series of post-processing is performed on the lateral strings to complement the horizontal insulator strings that may have insufficient extraction.
Preferably, step 6 includes:
step 6.1: for step 5.3.4, the intersection point may not be found in the ZX projection, and the post-processing method is adopted for perfecting. If the layer has a transverse string extracted, the length of the transverse string in the three-dimensional space is calculated. And calculating the length of the insulator string in the ZX projection according to the insulator string direction vector. The intersection ZX coordinates are directly calculated from this length.
Step 6.2: the method is almost the same as the processing method in the step 6.1, except that when the same layer does not extract the transverse strings, if the electric tower has the extracted transverse strings, the length of the insulator strings extracted by the electric tower is taken as the length of the transverse strings in the three-dimensional space, and the coordinates of the intersection point ZX are complemented.
Step 6.3: the method is similar to the processing method in the step 6.1, except that when the electric tower does not extract the transverse string, the intersection point ZX coordinate is complemented by taking a predefined length as the length of the transverse string in the three-dimensional space.
The method has the advantages that the three-dimensional point cloud problem is converted into the two-dimensional image processing problem by a projection method, so that the operation efficiency is improved, and the method has higher robustness and accuracy for various electric towers.
Drawings
Fig. 1: the method is a schematic flow chart of a method provided by the embodiment of the invention;
fig. 2: is a schematic of the calculated kappa angle provided by embodiments of the present invention;
fig. 3: the Y result diagram of the center of the calculation tower is provided by the embodiment of the invention;
fig. 4: the residual wiring diagram provided by the embodiment of the invention is used for eliminating the interference of the tower and the tree obstacle;
fig. 5: the method is a top layer hanging point X coordinate result graph provided by the embodiment of the invention;
fig. 6: the Z coordinate result diagram of the top hanging point is provided by the embodiment of the invention;
Fig. 7: the method is a vertical string ZX projection refinement diagram provided by the embodiment of the invention;
fig. 8: the method is a ZY projection diagram for calculating the Y coordinates of the vertical strings;
fig. 9: the method is a transverse string suspension point XY coordinate result graph provided by the embodiment of the invention;
fig. 10: the Z coordinate result diagram of the cross string suspension point provided by the embodiment of the invention;
fig. 11: the method is a cross string intersection point ZX coordinate result diagram provided by the embodiment of the invention;
fig. 12: the embodiment of the invention provides a projection incomplete condition diagram of a cross string intersection point ZX.
Fig. 13: the embodiment of the invention provides a vertical string extraction result diagram;
fig. 14: the embodiment of the invention provides a vertical string extraction result graph.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. In addition, the technical features of the embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
In the examples of the present invention, "first," "second," etc. are used to distinguish between different objects and are not necessarily used to describe a particular order or sequence.
Fig. 1 is a schematic flow chart of a method according to a first embodiment of the present invention, which specifically includes the following steps:
(1) Preprocessing point cloud data: because the obtained three-dimensional point cloud has a large number of ground points, the three-dimensional point cloud has no effect on the extraction of the insulator string and can slow down the operation efficiency; on the other hand, the tree obstacle and other interferences exist, so that the pretreatment of the input three-dimensional point cloud is very necessary. The pretreatment mainly comprises cutting in the XY direction and cutting in the Z direction.
(2) And (3) rotating the point cloud: in order to accelerate the operation efficiency, a projection method is largely used for projecting the three-dimensional point cloud to the two-dimensional plane. In order to make projection fast and effective, the point cloud needs to be rotated so that the direction of the electric tower is more standard, and the direction of the electric tower after rotation is consistent with the direction of the Y coordinate axis.
(3) Electric tower layering: because of different electric towers, the number, the types and the distribution of the insulator strings are different, the whole electric tower is projected, and all point clouds are mixed together and cannot be effectively distinguished. Therefore, layering the electric towers can reduce the number of point clouds processed each time, and information such as the number of electric towers can be acquired, so that the number and type information of the insulator strings can be acquired, and the method is beneficial to subsequent insulator string extraction.
(4) Judging the layer-by-layer type: because of the different types of insulator strings, the spatial distribution forms of the insulator strings are different, and the characteristics of surrounding point clouds are also different. It is difficult to extract by one method for various types of insulator strings, and it is much easier to extract by a different method according to the type of insulator string. Therefore, after layering the electric towers, the invention can judge the type of each layer of insulator string.
(5) Extracting insulator strings layer by layer: based on the point cloud of each layer of the electric tower and the category of the electric tower, different methods are adopted for processing, finally, the top layer obtains suspension points, the vertical string layer obtains corresponding vertical insulator strings, and the horizontal layer obtains corresponding horizontal insulator strings.
(6) And (3) performing post-treatment of horizontal string extraction: considering the problem of incomplete data itself, the cross string features may not be obvious enough, and the method of extracting the cross string in step (5) may be disabled. To enhance robustness, a series of post-processing is performed on the lateral strings to complement the possibly existing horizontal insulator strings with insufficient extraction;
in the embodiment of the present invention, the step (1) may be implemented by:
(1.1) cutting in the horizontal direction:
according to the central coordinates (x) c ,y c ,z c ) Will satisfy abs (x i -x c )＜thre x ，abs(y i -y c )＜thre y Is retained.
(1.2) vertical cutting:
and obtaining a Z value of the lowest point in the small XY range of the center of the electric tower, adding a certain height to the Z value as a clipping threshold, and removing the point cloud lower than the height. And certain intervals are arranged between other power transmission line wires and the electric towers, so that the wires are layered at certain intervals from the highest point, positions with a certain number of point clouds on a certain continuous layer are found, the point clouds higher than the positions are removed, and interference from other power transmission line wires is eliminated.
In the embodiment of the present invention, the step (2) may be implemented by:
(2.1) acquiring a top-level point cloud:
the kappa angle calculated by the invention is the direction of the first principal component calculated by the PCA method through the XY projection of the top view of the electric tower. The reason for taking the top-level point cloud is that the top-level wires are generally thinner and are easy to separate from the towers. In addition, because the shapes of the electric towers are different, the point clouds with the highest points in a certain range are not robust enough, the method for acquiring the top-layer point clouds is to send downwards from the highest points, layer the point clouds at certain intervals, calculate the number of the point clouds of each layer, and acquire the point clouds with a sufficient number of layers below Z in a certain range as the top-layer point clouds.
(2.2) calculating kappa angle:
and (3) obtaining an XY projection graph through top layer point cloud projection, performing filling and corrosion operation, extracting the maximum outline, and calculating the angle between the maximum outline and the Y axis as a rotation deflection angle kappa. As shown in fig. 2, the top XY projection of the tower is shown, filled and etched.
(2.3) rotating the point cloud:
and rotating the cut point cloud according to the calculated rotation deflection angle, wherein the selection center is the approximate electric tower center, and the rotation axis is the Z axis, so that the tower direction is consistent with the Y axis direction of the coordinate system.
In the embodiment of the present invention, the step (3) may be implemented by:
(3.1) digging out the electric Tower
Since the initial tower center may not be accurate enough, the tower center column and the column with the left and right boundaries of the tower are found and the pixel value of 0 is set to eliminate the tower according to the number of points in the column of the tower being greater than the number of points in the column of the wire. The calculated center column of the tower on the ZX projection is shown in fig. 3.
(3.2) tree barrier removal:
the electric wires are basically separated from the tree barriers after the electric towers are excavated, the lowest point of the ZX image is taken as a seed point, the ground residual tree barriers and other interferences are obtained through growth, the pixel value of the point of the part is set to be 0, and therefore the rest part of the image basically only comprises the electric wires. FIG. 4 is a ZX projection of the tower removed and the tree barrier removed.
(3.3) electric tower layering:
and extracting edges of the ZX projection image, detecting a Hough straight line of the edge image, and screening according to the vertical distance of the slope of the straight line, so that the position of each layer of electric wire on the line where the central column of the tower is located is obtained.
In the embodiment of the present invention, step (4) may be implemented by:
(4.1) generating XY projections layer by layer:
the uppermost layer of the electric tower layering result is regarded as the top layer and does not participate in type judgment. And for the non-top layer, respectively taking a small-range point cloud from top to bottom at the calculated Z position, and projecting downwards to obtain an XY projection graph.
(4.2) judging the layer-by-layer type:
since the column in which the column center is in the ZX projection is calculated in step (3.1), it corresponds to the column center line in the known XY projection. And counting the continuous condition of the pixel points by using the tower center to be located in a certain range of the upper row and the lower row, and if the continuous condition indicates that the layer has a tower, belonging to a horizontal string layer, otherwise belonging to a vertical string layer.
In the embodiment of the present invention, the step (5) may be implemented by:
(5.1) top layer treatment:
the uppermost layer of the layering result of the electric tower is regarded as the top layer, and suspension points are extracted by processing
(5.2) vertical string layer processing:
for the non-top layer of the electric tower layering result, judging the type as the part of the vertical string layer, and processing to extract the horizontal insulator string
(5.3) horizontal-serial layer processing:
for the non-top layer of the electric tower layering result, judging the type as the part of the horizontal string layer, and processing to extract the vertical insulator string
In an embodiment of the present invention, step (5.1) may be implemented by:
(5.1.1) calculating the suspension point X coordinate:
and obtaining a ZX projection graph by top-layer point cloud projection, counting the number of points in each row, finding out the boundary row of the area with the maximum density, and calculating the X coordinate of the hanging point according to the average value of the boundary row. Fig. 5 shows the top-level X coordinate calculation result.
(5.1.2) calculating a suspension point Z coordinate:
starting from the boundary columns found in the step (5.1.1), respectively starting from two sides, until a column with a sufficiently small width is found, and calculating the Z coordinate of the hanging point according to the starting row of the column. Fig. 6 shows the top-level X coordinate calculation result.
(5.1.3) calculating a hanging point Y coordinate:
and (5) carrying out point cloud projection to obtain an XY projection map. The Y coordinate of the hanging point is calculated by moving to the row of the hanging point along the electric wire from the upper end and the lower end.
(5.1.4) calculating the original three-dimensional coordinates of the suspension points:
and (3) rotating the calculated three-dimensional coordinates of the hanging points by a kappa angle according to the rotating direction of the step (2) to obtain the three-dimensional coordinates of the hanging points under the original coordinate system.
In an embodiment of the present invention, step (5.2) may be implemented by:
(5.2.1) counting the number and the range of the wires in each layer:
firstly, the layer of point cloud is projected downwards to obtain an XY projection diagram, the number of wires is counted according to the intervals among different wires in the same layer, and the range of each wire column, namely the range of Y, is obtained.
(5.2.2) calculating the ZX coordinates of the vertical string intersection point:
and (3) carrying out point cloud projection on each electric wire to obtain a ZX projection graph, filling, thinning and deburring the ZX projection graph, and then searching a tri-bifurcation point for the thinned graph, wherein the bifurcation point is an intersection point of the insulator string and the electric wire, so that the ZX coordinate of the intersection point is calculated. Fig. 7 is a vertical string intersection ZX.
(5.2.3) calculating the Y coordinate of the vertical string intersection point:
and taking a certain range of point clouds by rays corresponding to the intersection point ZX, and taking the average value of the cluster point clouds Y as the Y coordinate of the intersection point, so that the three-dimensional coordinate of the intersection point is obtained.
(5.2.4) calculating vertical string hanging point coordinates:
and according to the three-dimensional coordinates of the intersection point, a certain range of point clouds are upwards taken to obtain ZY projection, a row with a certain width, namely the position of a hanging point Z, is found in the row starting direction from the intersection point, and the XY coordinates of the hanging point are consistent with the XY coordinates of the intersection point, so that the three-dimensional coordinates of the hanging point are obtained. As shown in FIG. 8, a vertical string hanging point Z diagram
(5.2.5) calculating the original three-dimensional coordinates of the vertical strings:
and (3) rotating the vertical string according to the rotation direction of the step (2) by a kappa angle according to the three-dimensional coordinates of the vertical string intersection point and the hanging point calculated by each electric wire to obtain the three-dimensional coordinates of the vertical string under the original coordinate system.
In an embodiment of the present invention, step (5.3) may be implemented by:
(5.3.1) cutting the center part of the XY projection tower:
and the point cloud projects downwards to obtain an XY projection graph, and the part outside a certain range above and below the central line of the tower is cut and removed in the XY projection.
(5.3.2) calculating the suspension point XY coordinates:
and analyzing the rest of cutting, solving corner points (namely 4 corner points on the inner side of the H-shaped pattern) on two sides of the two connected domains, and slightly expanding the corner point coordinates to obtain suspension point XY coordinates. Fig. 9 is a graph showing the horizontal suspension points XY.
(5.3.3) calculating a suspension point Z coordinate:
and cutting according to the XY coordinates of the suspension points and the position of the center X of the tower to obtain a point cloud, wherein a part of the point cloud only comprises the tower without the electric wire and the insulator string, projecting to obtain a ZY projection graph, and finding out the line with the number of 255 points and jumping as the line where the Z coordinates of the horizontal suspension points are located. FIG. 10 is a cross-string hanging point Z-chart
(5.3.4) calculating the intersection ZX coordinates:
the layer of point cloud projection obtains a ZX projection graph, and the line-by-line traversal and recording of the highest line of the electric wire, the lowest line of the electric wire and the highest line of the suspended object are carried out according to the principle that the left suspension point is found left and the right suspension point is found right; and taking the difference between the wire highest behavior reference condition, the wire lowest line and the suspension highest line as judgment conditions, and carrying out classified discussion and fitting to obtain the ZX value. Fig. 11 is a cross-string intersection ZX.
(5.3.5) calculating the intersection Y coordinate:
cutting the layer of point cloud according to the ZX coordinates of the cross-string intersection points, taking a certain range of point cloud by rays corresponding to the intersection points ZX, dividing the point cloud into two clusters according to the Y coordinates of the cluster point cloud, and taking the average value of the point cloud Y of each cluster as the Y coordinates of each cross-string intersection point.
(5.3.6) calculating the raw three-dimensional coordinates of the transverse strings:
and (3) rotating the three-dimensional coordinates of the vertical strings under the original coordinate system according to the rotation-kappa angle of the step (2) according to the three-dimensional coordinates of each horizontal string intersection point and the suspension point calculated by the layer point cloud.
In the embodiment of the present invention, step (6) may be implemented by:
(6.1) Cross string post-treatment 1:
for the step (5.3.4), the intersection point may not be found in the ZX projection, and the post-processing method is adopted for perfecting. If the layer has a transverse string extracted, the length of the transverse string in the three-dimensional space is calculated. And calculating the length of the insulator string in the ZX projection according to the insulator string direction vector. The intersection ZX coordinates are directly calculated from this length.
(6.2) horizontal string post-treatment 2:
the method is almost the same as the processing method in the step (6.1), except that when the same layer does not extract the transverse strings, if the electric tower has the extracted transverse strings, the length of the insulator strings extracted by the electric tower is taken as the length of the transverse strings in the three-dimensional space, and the coordinates of the intersection point ZX are complemented.
(6.3) horizontal string post-treatment 3:
the method is similar to the processing method in the step (6.1), except that when the electric tower does not extract the transverse string, the intersection point ZX coordinate is complemented by taking a predefined length as the length of the transverse string in the three-dimensional space.
The second embodiment of the invention is an automatic extraction method of an electric tower insulator string based on three-dimensional point cloud, which comprises the following specific steps:
step 1: acquiring three-dimensional point cloud data of the electric tower and the center of the electric tower, and further cutting the three-dimensional point cloud data of the electric tower in the horizontal XY direction and in the vertical Z direction to obtain cut point clouds;
the three-dimensional point cloud data of the electric tower in the step 1 are as follows:
data 0 (x,y,z)＝{data 0 k (x k ,y k ,z k ),k∈[1,K]}
wherein, data 0 (x, y, z) is three-dimensional point cloud data of the electric tower, data 0 k (x k ,y k ,z k ) Is the pixel of the kth point in the three-dimensional point cloud data of the electric tower, x k Is the X-direction coordinate, y of the kth point in the three-dimensional point cloud data of the electric tower k Is the Y-direction coordinate, z of the kth point in the three-dimensional point cloud data of the electric tower k Z-direction coordinates of a kth point in the three-dimensional point cloud data of the electric tower, wherein K is the three-dimensional point cloud data of the electric towerThe number of medium three-dimensional points;
the horizontal XY direction cutting in the step 1 is as follows:
based on the centre of the tower 0 (x 0 ,y 0 ,z 0 )；
Data of three-dimensional point cloud data of electric tower 0 (x, y, z) search for abs (x) k -x 0 )＜threshold 0 And abs (y) k -y 0 )＜threshold 0 Constructing point cloud after horizontal clipping as data 1 (x,y,z)，threshold 0 Is a threshold parameter;
the vertical Z-direction cutting in the step 1 is as follows:
point cloud, namely data after being combined with horizontal cutting 1 (x, y, z), searching for a satisfying abs (x k -x 0 )＜threshold 1 And abs (y) k -y 0 )＜threshold 1 The minimum value of the Z-direction coordinate in the point cloud in the range is defined as Z min The method comprises the steps of carrying out a first treatment on the surface of the Searching for satisfying abs (x k -x 0 )＜threshold 1 And abs (y) k -y 0 )＜threshold 1 The maximum value of the Z-direction coordinate in the point cloud in the range is defined as Z max ，threshold 1 Is a threshold parameter;
defining a layering interval as z_interval=1.5;
the number of layers is calculated as follows:
z_num＝ceil[(z max -z min )/z_interval]
wherein ceil represents an upward rounding function, and z_num represents the number of layers;
defining the point cloud density of each layer as follows:
z_density＝{z_density l ,l∈[1,z_num]}
and initializes z-density l =0, and layer 1 is the highest layer;
traversing data 1 (x, y, z) according to z_now=ceil [ (z) max -z k )/z_interval]Calculating a layer z_now to which a kth point belongs, and counting the number of points of each layer;
from layer 1, find a satisfying z_density a+s ＞10，s＝[1,6]Calculating a reserved value z_up in the Z direction according to z_up=z_max- (a-1) z_interval;
point cloud data 1 (x, y, z) satisfies z k ≥z_min+threshold 2 And z k Point reservation of z_up, updating the highest point z_max=z_up, and the lowest point z_min=z_min+threshold of the point cloud 2 ，threshold 2 Is a threshold parameter.
Finally obtaining the tailored point cloud data 2 (x,y,z)
Step 2: for the point cloud data after clipping 2 (x, y, z) acquiring point cloud data of a certain range of the topmost layer 3 (x, y, z) calculating the slip angle kappa, and then adding the point cloud data 2 The (x, Y, z) rotation enables the tower direction of the electric tower to be consistent with the Y direction of the coordinate axis, and the point cloud data is obtained 4 (x,y,z)；
Preferably, step 2 comprises:
step 2.1: from point cloud data 2 The (x, y, z) highest point goes down, set the layering interval z_interval 1 =0.1, calculate the number of layers z_num 1 ＝ceil[(z max -z min )/z_interval 1 ]Defining the density z-density of each layer of point cloud 1 ＝{z_density 1 k ,k∈[1,z_num 1 ]And initialize z_density 1 k =0, and layer 1 is the highest layer. Traversing point cloud data 2 (x, y, z) according to z_now=ceil [ (z) max -z k )/z_interval 1 ]And calculating the layer to which the kth point belongs, and counting the number of points of each layer. From layer 1, find a satisfying z_density 1 k Layer k greater than 100, calculating the maximum value Z of the top layer point cloud in the Z direction top Max, point cloud data 2 (x, y, z) reservation satisfies z k ≥z top Max-3 and z k ≤z top The point of _max is taken as top level point cloud data 3 (x,y,z)。
Step 2.2: for the top layer point cloud data 3 (x, y, z) downward projection to obtain XY projection map (the projections mentioned in this invention are all binarized projections, where the XY projections are generated as follows, otherwise similar 3 Maximum and minimum of XY coordinates of (x, y, z)The value, set the two-dimensional grid interval as grid_interval=0.1, according to row=ceil [ (x_max-x_min)/grid_interval ]]Calculating the number of XY projection lines, col=ceil [ (y_max-y_min)/grid_interval ] ]Calculating the number of XY projection image columns and traversing point cloud data 3 (x, y, z) according to row_now=ceil [ (x_max-x) k )/grid_interval]Calculating the line number of the current point falling on the projection graph according to col_now=ceil [ (y_max-y) k )/grid_interval]Calculating the column number of the current point falling in the projection graph, if a pixel has a point cloud falling in the range, the pixel value is 255, otherwise 0), carrying out morphological filling and corrosion operation on the XY projection, and extracting the maximum outline. The rotation angle kappa is calculated from the first principal direction of the maximum profile using PCA.
Step 2.3: according to the calculated rotation deflection angle kappa, the clipped point cloud data 2 (x, y, z) centering the tower with the center 0 (x 0 ,y 0 ,z 0 ) As a rotation center, a coordinate system Z axis is taken as a rotation axis, and a rotation angle is calculated according to a rotation deflection angle kappa to rotate, so that the direction of a tower is consistent with the direction of a coordinate system Y axis, and a rotated point cloud data is obtained 4 (x,y,z)。
Step 3: for the rotated point cloud data 4 (x, y, Z) projecting to obtain ZX projection diagram, layering according to wire height to obtain approximate Z position z_layer of crossing wire at two sides of each layer of electric tower k K is the number of layers obtained by layering;
preferably, step 3 comprises:
step 3.1: and (5) projecting to obtain a ZX projection map. And finding out the column at the center of the tower and the column at the left and right boundaries of the tower according to the fact that the number of the points of the column at which the electric towers are located is greater than the number of the points of the column at which the electric wires are located, and setting the pixel value to 0 to remove the electric towers.
Step 3.2: and taking the lowest point of the ZX image as a seed point, growing to obtain the interference of ground residual tree barriers and the like, and setting the pixel value of the part of points to 0, so that the rest part of the image basically only comprises electric wires.
Step 3.3: and extracting edges of the ZX projection image, detecting a Hough straight line of the edge image, and screening according to the vertical distance of the slope of the straight line, so that the position of each layer of electric wire on the line where the central column of the tower is located is obtained.
Step 4: based on the position Z of each layer obtained by layering, cutting to obtain point clouds of each layer, wherein the uppermost layer is firstly used as a top layer, and judging the type of an insulator string in the non-top layer to be divided into a vertical string layer and a horizontal string layer;
preferably, step 4 comprises:
step 4.1: the uppermost layer is regarded as the top layer and does not participate in type judgment. And for the non-top layer, respectively taking a small-range point cloud from top to bottom at the calculated Z position, and projecting downwards to obtain an XY projection graph.
Step 4.2: and counting the continuous condition of the pixel points by using the tower center to be located in a certain range of the upper row and the lower row, and if the continuous condition indicates that the layer has a tower, belonging to a horizontal string layer, otherwise belonging to a vertical string layer.
Step 5: based on the point cloud of each layer of the electric tower and the category of the electric tower, different methods are adopted for processing, finally, a top layer obtains suspension points, a vertical string layer obtains a corresponding vertical insulator string, and a horizontal layer obtains a corresponding horizontal insulator string;
Preferably, step 5 comprises:
step 5.1: the uppermost layer of the layering result of the electric tower is regarded as the top layer, and suspension points are extracted by processing
Step 5.2: for the non-top layer of the electric tower layering result, judging the type as the part of the vertical string layer, and processing to extract the horizontal insulator string
Step 5.3: for the non-top layer of the electric tower layering result, judging the type as the part of the horizontal string layer, and processing to extract the vertical insulator string
Preferably, step 5.1 comprises:
step 5.1.1: and obtaining a ZX projection graph by top-layer point cloud projection, counting the number of points in each row, finding out the boundary row of the area with the maximum density, and calculating the X coordinate of the hanging point according to the average value of the boundary row.
Step 5.1.2: starting from the boundary columns found in step 5.1.1, respectively, to the two sides, until a column with a sufficiently small width is found, calculating the Z coordinate of the hanging point according to the starting row of the column.
Step 5.1.3: and (5) carrying out point cloud projection to obtain an XY projection map. The Y coordinate of the hanging point is calculated by moving to the row of the hanging point along the electric wire from the upper end and the lower end.
Step 5.1.4: and (3) rotating the calculated three-dimensional coordinates of the hanging points by a kappa angle according to the rotating direction of the step (2) to obtain the three-dimensional coordinates of the hanging points under the original coordinate system.
Preferably, step 5.2 comprises:
Step 5.2.1: and the point cloud is projected downwards to obtain an XY projection graph, the number of the wires is counted according to the intervals among different wires in the same layer, and the range of each wire Y is obtained.
Step 5.2.2: and (3) carrying out point cloud projection on each electric wire to obtain a ZX projection graph, filling, thinning and deburring the ZX projection graph, and then searching a tri-bifurcation point for the thinned graph, wherein the bifurcation point is an intersection point of the insulator string and the electric wire, so that the ZX coordinate of the intersection point is calculated.
Step 5.2.3: and taking a certain range of point clouds by rays corresponding to the intersection point ZX, and taking the average value of the cluster point clouds Y as the Y coordinate of the intersection point, so that the three-dimensional coordinate of the intersection point is obtained.
Step 5.2.4: and according to the three-dimensional coordinates of the intersection point, a certain range of point clouds are upwards taken to obtain ZY projection, a row with a certain width, namely the position of a hanging point Z, is found in the row starting direction from the intersection point, and the XY coordinates of the hanging point are consistent with the XY coordinates of the intersection point, so that the three-dimensional coordinates of the hanging point are obtained.
Step 5.2.5: and (3) rotating the three-dimensional coordinates of the vertical strings under the original coordinate system according to the rotation-kappa angle of the rotation direction of the step (2) according to the three-dimensional coordinates of the vertical string intersection point and the suspension point calculated by each electric wire.
Preferably, step 5.3 comprises:
Step 5.3.1: and the point cloud projects downwards to obtain an XY projection graph, and the part outside a certain range above and below the central line of the tower is cut and removed in the XY projection.
Step 5.3.2: and analyzing the rest of cutting, namely solving 4 corner points on the inner sides of the H-shaped pattern which are corner points on two sides of the two connected domains, and slightly expanding the corner point coordinates to obtain the XY coordinates of the suspension points.
Step 5.3.3: and cutting according to the XY coordinates of the suspension points and the position of the center X of the tower to obtain a point cloud, wherein a part of the point cloud only comprises the tower without the electric wire and the insulator string, projecting to obtain a ZY projection graph, and finding out the line with the number of 255 points and jumping as the line where the Z coordinates of the horizontal suspension points are located.
Step 5.3.4: the layer of point cloud projection obtains a ZX projection graph, and the line-by-line traversal and recording of the highest line of the electric wire, the lowest line of the electric wire and the highest line of the suspended object are carried out according to the principle that the left suspension point is found left and the right suspension point is found right; and taking the difference between the wire highest behavior reference condition, the wire lowest line and the suspension highest line as judgment conditions, and carrying out classified discussion and fitting to obtain the ZX value.
Step 5.3.5: cutting the layer of point cloud according to the ZX coordinates of the cross-string intersection points, taking a certain range of point cloud by rays corresponding to the intersection points ZX, dividing the point cloud into two clusters according to the Y coordinates of the cluster point cloud, and taking the average value of the point cloud Y of each cluster as the Y coordinates of each cross-string intersection point.
Step 5.3.6: and (3) according to the three-dimensional coordinates of each cross string intersection point and the suspension point calculated by the layer of point cloud, rotating by a kappa angle according to the rotating direction of the step (2) to obtain the three-dimensional coordinates of the vertical string under the original coordinate system.
Step 6: in view of the problem of incomplete data, to enhance robustness, a series of post-processing is performed on the lateral strings to complement the horizontal insulator strings that may have insufficient extraction.
Preferably, step 6 comprises:
step 6.1: for step 5.3.4, the intersection point may not be found in the ZX projection, and the post-processing method is adopted for perfecting. If the layer has a transverse string extracted, the length of the transverse string in the three-dimensional space is calculated. And calculating the length of the insulator string in the ZX projection according to the insulator string direction vector. The intersection ZX coordinates are directly calculated from this length.
Step 6.2: the method is almost the same as the processing method in the step 6.1, except that when the same layer does not extract the transverse strings, if the electric tower has the extracted transverse strings, the length of the insulator strings extracted by the electric tower is taken as the length of the transverse strings in the three-dimensional space, and the coordinates of the intersection point ZX are complemented.
Step 6.3: the method is similar to the processing method in the step 6.1, except that when the electric tower does not extract the transverse string, the intersection point ZX coordinate is complemented by taking a predefined length as the length of the transverse string in the three-dimensional space.
It will be readily understood by those skilled in the art that the foregoing description is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or modifications made within the spirit and principles of the invention are intended to be included in the protection of the invention.

Claims (6)

1. An automatic extraction method of an electric tower insulator string based on three-dimensional point cloud is characterized by comprising the following steps:
step 1: acquiring three-dimensional point cloud data of the electric tower and the center of the electric tower, and further cutting the three-dimensional point cloud data of the electric tower in the horizontal XY direction and in the vertical Z direction to obtain cut point clouds;
step 2: for the point cloud data after clipping 2 (x, y, z) acquiring point cloud data of a certain range of the topmost layer 3 (x, y, z) calculating the slip angle kappa, and then adding the point cloud data 2 The (x, Y, z) rotation enables the tower direction of the electric tower to be consistent with the Y direction of the coordinate axis, and the point cloud data is obtained 4 (x,y,z)；
Step 3: for the rotated point cloud data 4 (x, y, Z) projecting to obtain ZX projection diagram, layering according to wire height to obtain approximate Z position z_layer of crossing wire at two sides of each layer of electric tower k K is the number of layers obtained by layering;
step 4: based on the position Z of each layer obtained by layering, cutting to obtain point clouds of each layer, wherein the uppermost layer is firstly used as a top layer, and judging the type of an insulator string in the non-top layer to be divided into a vertical string layer and a horizontal string layer;
Step 5: based on the point cloud of each layer of the electric tower and the category of the electric tower, different methods are adopted for processing, finally, a top layer obtains suspension points, a vertical string layer obtains a corresponding vertical insulator string, and a horizontal layer obtains a corresponding horizontal insulator string;
step 6: considering the problem of incomplete data, in order to enhance robustness, carrying out a series of post-processing on the transverse strings to complement the existing horizontal insulator strings with incomplete extraction;
step 5.1 comprises:
step 5.1.1: the top layer point cloud projection obtains a ZX projection diagram, counts the number of points in each row, finds out the area boundary row with the maximum density, and calculates the X coordinate of the hanging point according to the average value of the boundary row;
step 5.1.2: starting from the boundary columns found in the step 5.1.1 to two sides respectively, until a column with a small enough width is found, calculating the Z coordinate of the hanging point according to the initial row of the column;
step 5.1.3: the point cloud projection obtains an XY projection map; moving to the line where the hanging point is located along the electric wire from the upper end and the lower end, so as to calculate the Y coordinate of the hanging point;
step 5.1.4: rotating the calculated three-dimensional coordinates of the hanging points by a kappa angle according to the rotating direction of the step 2 to obtain the three-dimensional coordinates of the hanging points under the original coordinate system;
step 5.2 comprises:
step 5.2.1: the point cloud projects downwards to obtain an XY projection graph, the number of the wires is counted according to the intervals among different wires in the same layer, and the range of each wire Y is obtained;
Step 5.2.2: the point cloud projection of each wire is used for obtaining a ZX projection graph, the ZX projection is filled, thinned and burr is removed, then a tri-bifurcation point is searched for on the thinned graph, and the point is the intersection point of the insulator string and the wire, so that the ZX coordinate of the intersection point is calculated;
step 5.2.3: taking a certain range of point clouds by rays corresponding to the intersection point ZX, and taking the average value of the cluster point clouds Y as the Y coordinate of the intersection point, so that the three-dimensional coordinate of the intersection point is obtained;
step 5.2.4: according to the three-dimensional coordinates of the intersection point, a certain range of point clouds are upwards taken to obtain ZY projection, a row with a certain width, namely the position of a hanging point Z, is found in the row starting direction from the intersection point, and the XY coordinates of the hanging point are consistent with the XY coordinates of the intersection point, so that the three-dimensional coordinates of the hanging point are obtained;
step 5.2.5: according to the three-dimensional coordinates of the vertical string intersection point and the hanging point calculated by each electric wire, rotating by a kappa angle according to the rotating direction of the step 2 to obtain the three-dimensional coordinates of the vertical string under the original coordinate system;
step 5.3 comprises:
step 5.3.1: projecting the point cloud downwards to obtain an XY projection image, and cutting out a part outside a certain range from the upper part and the lower part of the central line of the tower in the XY projection;
step 5.3.2: analyzing the left part of cutting, solving 4 corner points on the inner sides of the H-shaped pattern, which are corner points on two sides of the two connected domains, and slightly expanding the corner point coordinates to obtain suspension point XY coordinates;
Step 5.3.3: cutting according to the XY coordinates of the suspension points and the position of the center X of the tower to obtain a part of point cloud only comprising the tower without the electric wire and the insulator string, projecting to obtain a ZY projection graph, and finding out the line with the number of 255 points to jump as the line where the Z coordinates of the horizontal suspension points are located;
step 5.3.4: the layer of point cloud projection obtains a ZX projection graph, and the line-by-line traversal and recording of the highest line of the electric wire, the lowest line of the electric wire and the highest line of the suspended object are carried out according to the principle that the left suspension point is found left and the right suspension point is found right; taking the difference between the highest line of the electric wire and the lowest line of the electric wire as a judging condition, and carrying out classified discussion and fitting to obtain a ZX value;
step 5.3.5: cutting the layer of point clouds according to ZX coordinates of the cross-string intersection points, taking a certain range of point clouds by rays corresponding to the intersection points ZX, dividing the point clouds into two clusters according to Y coordinates of the cluster point clouds, and taking the average value of the point clouds Y of each cluster as the Y coordinates of each cross-string intersection point;
step 5.3.6: according to the three-dimensional coordinates of each cross string intersection point and the hanging point calculated by the layer of point cloud, rotating by a kappa angle according to the rotating direction of the step 2 to obtain the three-dimensional coordinates of the vertical string under the original coordinate system;
the step 6 comprises the following steps:
step 6.1: for the step 5.3.4, the intersection point may not be found in the ZX projection, and the post-processing method is adopted for perfecting; if the layer has the transverse strings extracted, calculating the length of the transverse strings in the three-dimensional space; calculating the length of the insulator string in ZX projection according to the direction vector of the insulator string; directly calculating the coordinate of the intersection point ZX according to the length;
Step 6.2: the method is approximately the same as the processing method in the step 6.1, except that when the same layer does not extract the transverse strings, if the electric tower has the extracted transverse strings, the length of the insulator strings extracted by the electric tower is taken as the length of the transverse strings in the three-dimensional space, and the intersection point ZX coordinates are complemented;
step 6.3: the method is similar to the processing method in the step 6.1, except that when the electric tower does not extract the transverse string, the intersection point ZX coordinate is complemented by taking a predefined length as the length of the transverse string in the three-dimensional space.
2. The automatic extraction method of an electric tower insulator string based on three-dimensional point cloud as claimed in claim 1, wherein the three-dimensional point cloud data of the electric tower in step 1 is:
data 0 (x,y,z)＝{data 0 k (x k ,y k ,z k ),k∈[1,K]}
wherein, data 0 (x, y, z) is three-dimensional point cloud data of the electric tower, data 0 k (x k ,y k ,z k ) Is the pixel of the kth point in the three-dimensional point cloud data of the electric tower, x k Is the X-direction coordinate, y of the kth point in the three-dimensional point cloud data of the electric tower k Is the Y-direction coordinate, z of the kth point in the three-dimensional point cloud data of the electric tower k The Z-direction coordinate of the kth point in the three-dimensional point cloud data of the electric tower is obtained, and K is the number of the three-dimensional points in the three-dimensional point cloud data of the electric tower;
the horizontal XY direction cutting in the step 1 is as follows:
based on the centre of the tower 0 (x 0 ,y 0 ,z 0 )；
Data of three-dimensional point cloud data of electric tower 0 (x, y, z) search for abs (x) k -x 0 )＜threshold 0 And abs (y) k -y 0 )＜threshold 0 Constructing point cloud after horizontal clipping as data 1 (x,y,z)，threshold 0 Is a threshold parameter;
the vertical Z-direction cutting in the step 1 is as follows:
point cloud, namely data after being combined with horizontal cutting 1 (x, y, z), searching for a satisfying abs (x k -x 0 )＜threshold 1 And abs (y) k -y 0 )＜threshold 1 The minimum value of the Z-direction coordinate in the point cloud in the range is defined asz min The method comprises the steps of carrying out a first treatment on the surface of the Searching for satisfying abs (x k -x 0 )＜threshold 1 And abs (y) k -y 0 )＜threshold 1 The maximum value of the Z-direction coordinate in the point cloud in the range is defined as Z max ，threshold 1 Is a threshold parameter;
defining a layering interval as z_interval=1.5;
the number of layers is calculated as follows:
z_num＝ceil[(z max -z min )/z_interval]
wherein ceil represents an upward rounding function, and z_num represents the number of layers;
defining the point cloud density of each layer as follows:
z_density＝{z_density l ,l∈[1,z_num]}
and initializes z-density l =0, and layer 1 is the highest layer;
traversing data 1 (x, y, z) according to z_now=ceil [ (z) max -z k )/z_interval]Calculating a layer z_now to which a kth point belongs, and counting the number of points of each layer;
from layer 1, find a satisfying z_density a+s ＞10，s＝[1,6]Calculating a reserved value z_up in the Z direction according to z_up=z_max- (a-1) z_interval;
point cloud data 1 (x, y, z) satisfies z k ≥z_min+threshold 2 And z k Point reservation of z_up, updating the highest point z_max=z_up, and the lowest point z_min=z_min+threshold of the point cloud 2 ，threshold 2 Is a threshold parameter;
finally obtaining the tailored point cloud data 2 (x,y,z)。
3. The automatic extraction method of an electric tower insulator string based on the three-dimensional point cloud as claimed in claim 1, wherein the step 2 comprises:
Step 2.1: from point cloud data 2 The (x, y, z) highest point goes down, set the layering interval z_interval 1 =0.1, calculate the number of layers z_num 1 ＝ceil[(z max -z min )/z_interval 1 ]Defining each layer of point cloud densityDegree z_Density 1 ＝{z_density 1 k ,k∈[1,z_num 1 ]And initialize z_density 1 k =0, and layer 1 is the highest layer; traversing point cloud data 2 (x, y, z) according to z_now=ceil [ (z) max -z k )/z_interval 1 ]Calculating the layer to which the kth point belongs, and counting the number of points of each layer; from layer 1, find a satisfying z_density 1 k Layer k greater than 100, calculating the maximum value Z of the top layer point cloud in the Z direction top Max, point cloud data 2 (x, y, z) reservation satisfies z k ≥z top Max-3 and z k ≤z top The point of _max is taken as top level point cloud data 3 (x,y,z)；
Step 2.2: for the top layer point cloud data 3 (x, y, z) projecting downwards to obtain an XY projection image, carrying out morphological filling on the XY projection image, carrying out corrosion operation, and extracting the maximum outline; calculating a rotation deflection angle kappa according to a first main direction of the maximum contour by using a PCA method;
step 2.3: according to the calculated rotation deflection angle kappa, the clipped point cloud data 2 (x, y, z) centering the tower with the center 0 (x 0 ,y 0 ,z 0 ) As a rotation center, a coordinate system Z axis is taken as a rotation axis, and a rotation angle is calculated according to a rotation deflection angle kappa to rotate, so that the direction of a tower is consistent with the direction of a coordinate system Y axis, and a rotated point cloud data is obtained 4 (x,y,z)。
4. The automatic extraction method of an electric tower insulator string based on the three-dimensional point cloud as claimed in claim 1, wherein the step 3 comprises:
Step 3.1: projecting to obtain a ZX projection map; according to the fact that the number of points of the column where the electric towers are located is larger than the number of points of the column where the electric wires are located, finding a column where the central column of the tower and the left and right boundaries of the tower are located, setting a pixel value to 0, and removing the electric towers;
step 3.2: taking the lowest point of the ZX image as a seed point, growing to obtain interference such as ground residual tree barriers, and setting the pixel value of the part of points to be 0, so that the rest part of the image basically only comprises electric wires;
step 3.3: and extracting edges of the ZX projection image, detecting a Hough straight line of the edge image, and screening according to the vertical distance of the slope of the straight line, so that the position of each layer of electric wire on the line where the central column of the tower is located is obtained.
5. The automatic extraction method of an electric tower insulator string based on the three-dimensional point cloud as claimed in claim 1, wherein the step 4 comprises:
step 4.1: the uppermost layer is regarded as the top layer and does not participate in type judgment; for the non-top layer, respectively taking small-range point clouds from top to bottom according to the calculated Z position, and projecting downwards to obtain an XY projection map;
step 4.2: and counting the continuous condition of the pixel points by using the tower center to be located in a certain range of the upper row and the lower row, and if the continuous condition indicates that the layer has a tower, belonging to a horizontal string layer, otherwise belonging to a vertical string layer.
6. The automatic extraction method of an electric tower insulator string based on the three-dimensional point cloud as claimed in claim 1, wherein the step 5 comprises:
step 5.1: the uppermost layer of the layering result of the electric tower is taken as the top layer, and suspension points are extracted by processing;
step 5.2: judging the type of a non-top layer of the electric tower layering result as a part of a vertical string layer, and processing to extract a horizontal insulator string;
step 5.3: and judging the type of the non-top layer of the electric tower layering result as the part of the horizontal string layer, and processing to extract the vertical insulator string.
CN202110300702.1A 2021-03-22 2021-03-22 Automatic extraction method of electric tower insulator string based on three-dimensional point cloud Active CN112991303B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110300702.1A CN112991303B (en) 2021-03-22 2021-03-22 Automatic extraction method of electric tower insulator string based on three-dimensional point cloud

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110300702.1A CN112991303B (en) 2021-03-22 2021-03-22 Automatic extraction method of electric tower insulator string based on three-dimensional point cloud

Publications (2)

Publication Number Publication Date
CN112991303A CN112991303A (en) 2021-06-18
CN112991303B true CN112991303B (en) 2023-08-18

Family

ID=76332783

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110300702.1A Active CN112991303B (en) 2021-03-22 2021-03-22 Automatic extraction method of electric tower insulator string based on three-dimensional point cloud

Country Status (1)

Country Link
CN (1) CN112991303B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103744661A (en) * 2013-12-23 2014-04-23 广东电网公司电力科学研究院 Multi-sensor data integration processing method and system of minimum altitude unmanned aerial vehicle
CN107680102A (en) * 2017-08-28 2018-02-09 国网甘肃省电力公司电力科学研究院 A kind of airborne cloud data electric force pole tower extraction method based on space constraint
CN110780681A (en) * 2019-11-26 2020-02-11 贵州电网有限责任公司 Unmanned aerial vehicle autonomous inspection insulator path planning method based on laser point cloud
CN111830528A (en) * 2020-06-29 2020-10-27 西安交通大学 Tower characteristic point automatic identification and inclination parameter automatic measurement method based on laser point cloud

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103744661A (en) * 2013-12-23 2014-04-23 广东电网公司电力科学研究院 Multi-sensor data integration processing method and system of minimum altitude unmanned aerial vehicle
CN107680102A (en) * 2017-08-28 2018-02-09 国网甘肃省电力公司电力科学研究院 A kind of airborne cloud data electric force pole tower extraction method based on space constraint
CN110780681A (en) * 2019-11-26 2020-02-11 贵州电网有限责任公司 Unmanned aerial vehicle autonomous inspection insulator path planning method based on laser point cloud
CN111830528A (en) * 2020-06-29 2020-10-27 西安交通大学 Tower characteristic point automatic identification and inclination parameter automatic measurement method based on laser point cloud

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于交比不变性的绝缘子串中轴线提取方法;马志学 等;电子测量技术;第40卷(第6期);全文 *

Also Published As

Publication number Publication date
CN112991303A (en) 2021-06-18

Similar Documents

Publication Publication Date Title
CN109829199B (en) Power line rapid layered extraction method based on LiDAR point cloud
CN112070769B (en) Layered point cloud segmentation method based on DBSCAN
CN106709946B (en) LiDAR point cloud-based automatic multi-split conductor extraction and fine modeling method
CN111079611B (en) Automatic extraction method for road surface and marking line thereof
CN107292276B (en) Vehicle-mounted point cloud clustering method and system
CN108898672A (en) A kind of semi-automatic cloud method making three-dimensional high-definition mileage chart lane line
CN105701798B (en) The data reduction method and device of columnar object
CN114488073A (en) Method for processing point cloud data acquired by laser radar
CN111325138B (en) Road boundary real-time detection method based on point cloud local concave-convex characteristics
CN112200171B (en) Road point cloud extraction method based on scanning lines
CN113569915B (en) Multi-strategy rail transit obstacle recognition method based on laser radar
CN111783721B (en) Lane line extraction method of laser point cloud and electronic equipment
CN109657686A (en) Lane line generation method, device, equipment and storage medium
CN113920134B (en) Slope ground point cloud segmentation method and system based on multi-line laser radar
WO2023216470A1 (en) Drivable area detection method and apparatus, and device
CN109870458B (en) Pavement crack detection and classification method based on three-dimensional laser sensor and bounding box
CN115205690B (en) Method and device for extracting street tree in monomer mode based on MLS point cloud data
Zhang et al. Filtering photogrammetric point clouds using standard lidar filters towards dtm generation
CN108765478A (en) It is a kind of to build the density clustering algorithm that separating monomer is built in point cloud
CN115690081A (en) Tree counting method, system, storage medium, computer equipment and terminal
CN111524127A (en) Urban road surface extraction method for low-altitude airborne laser radar data
CN115063555A (en) Method for extracting vehicle-mounted LiDAR point cloud street tree growing in Gaussian distribution area
CN117115012A (en) Road surface point cloud marking segmentation denoising method
CN116524219A (en) Barrier detection method based on laser radar point cloud clustering
CN112991303B (en) Automatic extraction method of electric tower insulator string based on three-dimensional point cloud

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant