CN109785352B - Intelligent efficient airborne radar point cloud analysis method - Google Patents
Intelligent efficient airborne radar point cloud analysis method Download PDFInfo
- Publication number
- CN109785352B CN109785352B CN201811571777.8A CN201811571777A CN109785352B CN 109785352 B CN109785352 B CN 109785352B CN 201811571777 A CN201811571777 A CN 201811571777A CN 109785352 B CN109785352 B CN 109785352B
- Authority
- CN
- China
- Prior art keywords
- point cloud
- matrix
- cloud data
- information
- crown
- 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
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 11
- 239000011159 matrix material Substances 0.000 claims abstract description 42
- 238000000034 method Methods 0.000 claims abstract description 6
- 230000004888 barrier function Effects 0.000 claims abstract description 5
- 238000013507 mapping Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 2
- 230000001105 regulatory effect Effects 0.000 claims 1
- 238000007689 inspection Methods 0.000 abstract description 6
- 238000005516 engineering process Methods 0.000 abstract description 2
- 230000005540 biological transmission Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 239000011241 protective layer Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an intelligent high-efficiency airborne radar point cloud analysis method, which comprises the steps of respectively obtaining first crown information and second crown information by using point cloud data images formed at different time aiming at the same area needing to be detected, then matching the two times of crown information, selecting corresponding points of the two point cloud data images after matching is successful, and carrying out difference comparison to obtain the growth condition and the growth rate of a tree. The method and the device combine information of the tree crowns twice to judge the growth condition of the tree, so as to evaluate the danger of the tree to the high-voltage wire and overcome the problem that the existing unmanned aerial vehicle inspection technology cannot generate a unified electronic report. In addition, the proper matrix A and matrix B are selected, so that the properties of the point cloud data can be accurately described, and meanwhile, the time complexity is within an acceptable range. Finally, aiming at the application scene, a series of reasonable parameter settings are set when the point cloud of the barrier tree is subjected to image matching, and the simple operation is adopted to replace the complex root-opening and index operation as far as possible in the algorithm flow, so that the operation rate is greatly improved.
Description
Technical Field
The invention relates to the technical field of module identification and image processing, in particular to an intelligent efficient airborne radar point cloud analysis method.
Background
It is known that the conservative estimation of the transmission line in China is 45 ten thousand kilometers, and the data will be further increased along with the development and the improvement of the power system. How to regularly patrol and maintain these lines has been a huge project. After all, most of the high-voltage transmission lines are arranged in rare and complicated field environments, and trees which cannot grow under the high-voltage transmission line are difficult to avoid.
Such cases are not uncommon. When the high-voltage wire is very close to the branch, strong wind blows, and when the branch swings, the insulation protective layer of the wire is possibly worn, so that the electric leakage phenomenon is easily caused. If in thunderstorm weather, sparks are easily generated when the wires are contacted with branches, and forest fire is caused. The importance of power patrols can be seen.
Therefore, in power patrol, a tree is also an object to be observed with emphasis. Most of the existing implementation schemes only stop in manual inspection, the unmanned aerial vehicle inspection is in a perfect stage, and particularly, the harmfulness judgment and research on the electric wire from the fine branch to the tree is less.
Although many units have studied through unmanned aerial vehicle to carry out electric power inspection, also made progress a lot, to each electric power inspection data can generate the electronic report of dangerous condition such as a barrier of a tree to reflect the result of patrolling and examining. But the report generation rate is slow, and in actual production, if the tree obstacles in each report need to be cleared by attendance, huge manpower and material resources are consumed.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides an intelligent high-efficiency airborne radar point cloud analysis method. Thereby reflecting the dangerous situation in a certain time period of the region, rather than one report per time period in isolation. Therefore, the growth condition of the tree obstacle is reflected, and the danger of the tree obstacle is further evaluated.
In order to realize the purpose, the technical scheme provided by the invention is as follows:
a point cloud analysis method for an intelligent high-efficiency airborne radar includes the steps of respectively obtaining first crown information and second crown information by point cloud data images formed in different time aiming at the same area needing to be detected, matching the first crown information and the second crown information, selecting corresponding points of two corresponding point cloud data images after matching is successful, comparing the difference between the two corresponding points to obtain the growth condition and growth rate of a tree, and accordingly evaluating dangerousness.
Further, the specific steps of obtaining crown information from the point cloud data image are as follows:
s1: collecting a point cloud image of a detection area;
s2: preprocessing point cloud data;
s3: based on the time invariant information, selecting a tower channel and carrying out threshold segmentation to obtain ideal crown point cloud;
s4: and observing the tree barrier condition between two adjacent towers, and recording crown information.
Further, the step S2 of preprocessing the point cloud data includes the specific steps of:
firstly, a down-sampling algorithm is carried out on the point cloud to reduce the density of the point cloud to one third, and then data noise is removed through an outlier algorithm.
Further, step S3 is based on the time invariant information, selects a tower channel and performs threshold segmentation to obtain an ideal crown point cloud, and specifically includes: and (3) importing known tower coordinates in advance, performing threshold cutting on a channel with the width of about 20 m between adjacent towers, storing point cloud data with the height of more than 2 m, and recording as ideal crown point cloud information.
Further, the two times of crown information are matched by corresponding to whether the top points and the edges in the two point cloud data images are similar, specifically as follows:
modeling yields the following formula:
c 1+ c 2 =1
in the above formula, X is a Markov matrix, and the element p in the X matrix ij =0 or p ij =1; a and A' are adjacency matrices describing the properties of edges of the two graphs, respectively; b and B' are matrixes of description vertexes of the two graphs respectively; c. C 1 And c 2 Weights that consider edges and points, respectively; if is a Frobeniu matrix; when the value is taken as the minimum, the difference value of the edge and the vertex properties of the two graphs is the minimum under the mapping of X, namely X is the mapping of the two graphsThe relationship between the beams.
Further, in order to improve the operation efficiency and facilitate the solution, the formula obtained by modeling is subjected to identity change, vector differential operator and projection to obtain the following formula:
wherein the content of the first and second substances,
M(X)=...M 2 M 1 M 2 M 1 (X)
in the above formula, the matrix A is a distance matrix, and in order to avoid complex operation of the root-opening number, A ij The element represents the inverse of the square of the distance from the ith point to the jth point in a graph;
in order to reduce large-scale point cloud data and improve convergence speed, a set of Harris angular points of found ideal crown point clouds is selected as a B matrix; b is i1 、B i2 、B i3 Respectively represent xyz information of the ith point in the graph; if the two graphs are successfully matched, X is converged, and the formula value is minimum;
the specific operation process is as follows:
s1: inputting A, A ', B, B';
s2: x and M are initialized, wherein,
M is a zero matrix of n multiplied by n;
s3: and (3) calculating: m = AXA '+ kBB';
s4: and (3) calculating:
s5: judging whether M is converged, if not, returning to the step S4; if M has converged, then calculate:
X=(1-α)X+αM;
s6: judging whether X is converged, if not, returning to the step S3; until M converges;
after the graphs are successfully matched, converting the X matrix into a permutation matrix by using a greedy algorithm, and if X is successful, converting the X matrix into the permutation matrix ij If =1, it means that the ith point of the first graph corresponds to the jth point of the second graph; and selecting corresponding points of the two point cloud data images, and comparing the corresponding points with the corresponding points to obtain the growth condition and the growth rate of the tree.
Compared with the prior art, the principle and the advantages of the scheme are as follows:
1. and judging the growth condition of the tree by combining the information of the tree crowns twice, so as to evaluate the danger of the tree to the high-voltage wire. The problem that unified electronic reports cannot be generated by the existing unmanned aerial vehicle inspection technology is solved.
2. And selecting proper matrix A and matrix B to accurately describe the properties of the point cloud data, wherein the time complexity is in an acceptable range.
3. Aiming at the application scene, a series of reasonable parameter settings are carried out when the obstacle tree point cloud is subjected to image matching, and simple operation is adopted to replace complicated root-opening and index operation as much as possible in the algorithm flow, so that the operation speed is greatly improved.
4. For the application scenario, by modifying the iteration condition, convergence can be faster.
Drawings
FIG. 1 is a working schematic diagram of an intelligent high-efficiency airborne radar point cloud analysis method of the invention.
Detailed Description
The invention is further illustrated by the following specific examples:
according to the point cloud analysis method for the intelligent efficient airborne radar, for the same area needing to be detected, first crown information and second crown information are obtained by point cloud data images formed at different time respectively, then the two times of crown information are matched, after matching is successful, corresponding points of two corresponding point cloud data images are selected, difference comparison is conducted between the two corresponding points, the growth condition and the growth rate of a tree are obtained, and therefore the danger is evaluated.
Referring to fig. 1, the specific steps are as follows:
s1: collecting a point cloud image of an area needing to be detected;
s2: carrying out down-sampling algorithm on the point cloud data to reduce the density of the point cloud to one third, and then removing data noise through an outlier algorithm;
s3: leading in known tower coordinates in advance, performing threshold cutting on a channel with the width of about 20 meters between adjacent towers, storing point cloud data with the height of more than 2 meters, and recording as ideal crown point cloud information;
s4: observing the condition of the tree barrier between two adjacent towers, and recording the information of the first crown;
s5: circulating the steps S1-S4 at different time periods, and recording the information of the second crown;
s6: matching of the crown information twice is carried out by judging whether the top points and the edges in the two corresponding point cloud data images are similar, which specifically comprises the following steps:
modeling yields the following formula:
c 1+ c 2 =1
in the above formula, X is a Markov matrix, and the element p in the X matrix ij =0 or p ij =1; a and A' are respectively adjacency matrices of the two figures describing the properties of the edges; b and B' are the matrixes of the description vertexes of the two graphs respectively; c. C 1 And c 2 Weights that consider edges and points, respectively; if is a Frobeniu matrix; when the value is taken as the minimum, the difference value of the properties of the edge and the vertex of the two graphs is the minimum under the mapping of X, namely X is the mapping relation of the two.
In order to improve the operation efficiency and facilitate the solution, the formula obtained by modeling is subjected to identity change, vector differential operator and projection to obtain the following formula:
wherein the content of the first and second substances,
M(X)=...M 2 M 1 M 2 M 1 (X)
in the above formula, the matrix A is a distance matrix, and in order to avoid complex operation of root-opening numbers, A ij The element represents the inverse of the square of the distance from the ith point to the jth point in a graph;
for reducing large-scale point cloud data and increasing convergence speed, selecting Harris corner point of ideal crown point cloudCollecting as a B matrix; b is i1 、B i2 、B i3 Respectively represent xyz information of the ith point in the graph; if the two graphs are successfully matched, X is converged, and the formula has the minimum value;
the specific operation process is as follows:
s1: inputting A, A ', B, B';
s2: initializing X and M, wherein,
M is a zero matrix of n multiplied by n;
s3: and (3) calculating: m = AXA + kBB';
s4: and (3) calculating:
s5: judging whether M is converged, if not, returning to the step S4; if M has converged, then calculate:
X=(1-α)X+αM;
s6: judging whether X is converged, if not, returning to the step S3; until M converges;
after the graphs are successfully matched, converting the X matrix into a permutation matrix by using a greedy algorithm, and if X is successful, converting the X matrix into the permutation matrix ij =1, it means that the ith point of the first graph corresponds to the jth point of the second graph; and selecting corresponding points of the two point cloud data images, and comparing the corresponding points with the corresponding points to obtain the growth condition and the growth rate of the tree.
This embodiment combines twice crown information, judges the trees growth condition to this aassessment trees are to the danger of high-tension line, overcome the problem that current unmanned aerial vehicle patrols and examines the technique and can't generate unified electronic report. And when the crown information is matched twice, selecting a proper matrix A and a proper matrix B, so that the property of the point cloud data is accurately described, and meanwhile, the time complexity is in an acceptable range. Finally, aiming at the application scene, a series of reasonable parameter settings are set when the obstacle tree point cloud is subjected to image matching, and simple operation is adopted to replace complicated root-opening and index operation as much as possible in the algorithm flow, so that the operation speed is greatly improved.
The above-mentioned embodiments are merely preferred embodiments of the present invention, and the scope of the present invention is not limited thereto, so that variations based on the shape and principle of the present invention should be covered within the scope of the present invention.
Claims (3)
1. An intelligent high-efficiency airborne radar point cloud analysis method is characterized in that point cloud data images formed in different time are used for the same area needing to be detected, first and second crown information is obtained respectively, then the two crown information are matched, after matching is successful, corresponding points of two corresponding point cloud data images are selected, difference comparison is carried out between the two corresponding point cloud data images, the growth condition and the growth rate of a tree are obtained, and therefore dangerousness is evaluated;
the specific steps for acquiring crown information from the point cloud data image are as follows:
s1: collecting a point cloud image of a detection area;
s2: preprocessing point cloud data;
s3: based on the time invariant information, selecting a tower channel and carrying out threshold segmentation to obtain ideal crown point cloud;
s4: observing the tree barrier condition between two adjacent towers, and recording crown information;
the two times of crown information are matched by corresponding whether the top points and the edges in the two point cloud data images are similar or not, and the method specifically comprises the following steps:
modeling yields the following equation:
C 1 +C 2 =1
in the above formula, X is a Markov matrix, and the element p in the X matrix ij =0 or p ij =1; a and A' are adjacency matrices describing the properties of edges of the two graphs, respectively; b and B' are the matrixes of the description vertexes of the two graphs respectively; c. C 1 And c 2 Weights that consider edges and points, respectively; | | | F denotes being a Frobeniu matrix; when in use When the value is minimum, the difference value of the properties of the edge and the vertex of the two graphs is minimum under the mapping of X, namely X is the mapping relation of the two graphs;
in order to improve the operation efficiency and facilitate the solution, the formula obtained by modeling is subjected to identity change, vector differential operator and projection to obtain the following formula:
wherein the content of the first and second substances,
M(X)=...M 2 M 1 M 2 M 1 (X)
in the above formula, to avoid complicated operation of the root-opening number, A ij The element represents the inverse of the square of the distance from the ith point to the jth point in a graph; 1 is expressed as a full 1 matrix, I is expressed as a unit matrix, n is the number of point clouds, and t is the iteration number; m 1 ,M 2 ,Are all temporary variables; alpha is a regulating parameter;
in order to reduce large-scale point cloud data and improve convergence speed, a set of Harris angular points of found ideal crown point clouds is selected as a B matrix; b is i1 、B i2 、B i3 Respectively represent xyz information of the ith point in the graph; if the two images match successfully, X converges and at the same timeThe formula has the minimum value;
the specific operation process is as follows:
s1: inputting A, A ', B, B';
s2: x and M are initialized, wherein,
M is a zero matrix of n x n;
s3: and (3) calculating: m = AXA '+ kBB';
s4: and (3) calculating:
s5: judging whether M is converged, if not, returning to the step S4; if M has converged, then calculate:
X=(1-α)X+αM;
s6: judging whether X is converged, if not, returning to the step S3; until M converges;
after the graphs are successfully matched, converting the X matrix into a permutation matrix by using a greedy algorithm, and if X is successful, converting the X matrix into the permutation matrix ij =1, it means that the ith point of the first graph corresponds to the jth point of the second graph; and selecting corresponding points of the two point cloud data images, and comparing the corresponding points with the corresponding points to obtain the growth condition and the growth rate of the tree.
2. The point cloud analysis method for the intelligent efficient airborne radar according to claim 1, wherein the step S2 of preprocessing the point cloud data comprises the following specific steps:
firstly, a down-sampling algorithm is carried out on the point cloud to reduce the density of the point cloud to one third, and then data noise is removed through an outlier algorithm.
3. The intelligent high-efficiency airborne radar point cloud analysis method according to claim 1, wherein the step S3 is based on time invariant information, a tower channel is selected and threshold segmentation is carried out to obtain ideal crown point cloud, and the steps are as follows: and (3) importing known tower coordinates in advance, performing threshold cutting on a channel with the width of about 20 m between adjacent towers, and storing point cloud data with the height of more than 2 m as ideal crown point cloud information.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811571777.8A CN109785352B (en) | 2018-12-21 | 2018-12-21 | Intelligent efficient airborne radar point cloud analysis method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811571777.8A CN109785352B (en) | 2018-12-21 | 2018-12-21 | Intelligent efficient airborne radar point cloud analysis method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109785352A CN109785352A (en) | 2019-05-21 |
CN109785352B true CN109785352B (en) | 2022-11-22 |
Family
ID=66497539
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811571777.8A Active CN109785352B (en) | 2018-12-21 | 2018-12-21 | Intelligent efficient airborne radar point cloud analysis method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109785352B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113516744B (en) * | 2020-06-01 | 2022-09-27 | 阿里巴巴集团控股有限公司 | Virtual scene generation method, interface interaction method, commodity display method and equipment |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2010032495A1 (en) * | 2008-09-16 | 2010-03-25 | 株式会社アドイン研究所 | Tree information measuring method, tree information measuring device, and program |
CN107103599A (en) * | 2017-04-06 | 2017-08-29 | 云南电网有限责任公司电力科学研究院 | A kind of transmission line of electricity trees hidden danger prediction analysis method based on LiDAR |
CN107330926A (en) * | 2017-08-25 | 2017-11-07 | 上海嘉奥信息科技发展有限公司 | Non-marked medical figure registration system and method in a kind of art in navigation system |
CN108460779A (en) * | 2018-02-12 | 2018-08-28 | 浙江大学 | A kind of mobile robot image vision localization method under dynamic environment |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FI117309B (en) * | 2003-03-31 | 2006-08-31 | Geodeettinen Laitos | Procedure, device and program for determining the growth of trees |
WO2013038818A1 (en) * | 2011-09-12 | 2013-03-21 | 日産自動車株式会社 | Three-dimensional object detection device |
-
2018
- 2018-12-21 CN CN201811571777.8A patent/CN109785352B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2010032495A1 (en) * | 2008-09-16 | 2010-03-25 | 株式会社アドイン研究所 | Tree information measuring method, tree information measuring device, and program |
CN107103599A (en) * | 2017-04-06 | 2017-08-29 | 云南电网有限责任公司电力科学研究院 | A kind of transmission line of electricity trees hidden danger prediction analysis method based on LiDAR |
CN107330926A (en) * | 2017-08-25 | 2017-11-07 | 上海嘉奥信息科技发展有限公司 | Non-marked medical figure registration system and method in a kind of art in navigation system |
CN108460779A (en) * | 2018-02-12 | 2018-08-28 | 浙江大学 | A kind of mobile robot image vision localization method under dynamic environment |
Non-Patent Citations (1)
Title |
---|
"Change Detection Techniques for Canopy Height Growth Measurements Using Airborne Laser Scanner Data";Xiaowei Yu等;《Photogrammetric Engineering & Remote Sensing》;20061231;第27卷(第12期);第1339-1346页 * |
Also Published As
Publication number | Publication date |
---|---|
CN109785352A (en) | 2019-05-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Sohn et al. | Automatic powerline scene classification and reconstruction using airborne lidar data | |
US20210124855A1 (en) | Methods and apparatus for automatically defining computer-aided design files using machine learning, image analytics, and/or computer vision | |
CN109829199B (en) | Power line rapid layered extraction method based on LiDAR point cloud | |
CN110189412A (en) | More floor doors structure three-dimensional modeling methods and system based on laser point cloud | |
CN104298998A (en) | 3D point cloud data processing method | |
CN114114314A (en) | Power transmission line inspection detection system and detection method based on laser point cloud | |
US10768303B2 (en) | Method for identifying individual trees in airborne lidar data and corresponding computer program product | |
CN110910440B (en) | Power transmission line length determination method and system based on power image data | |
CN114004938B (en) | Urban scene reconstruction method and device based on mass data | |
US20210405253A1 (en) | Systems and methods for distributed-solar power forecasting using parameter regularization | |
CN110675441B (en) | Laser point cloud-based power transmission line ground wire modeling extraction method | |
Vemula et al. | Mask R-CNN Powerline Detector: A Deep Learning approach with applications to a UAV | |
Ahmed et al. | Inspection and identification of transmission line insulator breakdown based on deep learning using aerial images | |
CN114972631A (en) | Online reconstruction method and system for laser point cloud model of power transmission line | |
CN113947555A (en) | Infrared and visible light fused visual system and method based on deep neural network | |
CN115240093A (en) | Automatic power transmission channel inspection method based on visible light and laser radar point cloud fusion | |
CN115562348A (en) | Unmanned aerial vehicle image technology method based on transformer substation | |
CN109785352B (en) | Intelligent efficient airborne radar point cloud analysis method | |
CN116843636A (en) | Insulator defect detection method based on improved YOLOv7 algorithm in foggy weather scene | |
CN117132915B (en) | Power transmission line tree obstacle hidden danger analysis method based on automatic classification of point cloud | |
Zhao et al. | Classification method of LiDAR point cloud based on threedimensional convolutional neural network | |
Che et al. | Inspection of Electric Transmission Line Fault Based on UAV Lidar. | |
Awrangjeb et al. | Classifier-free extraction of power line wires from point cloud data | |
CN113284103A (en) | Substation equipment defect online detection system based on space transformation fast R-CNN model | |
CN117823741B (en) | Pipe network non-excavation repairing method and system combined with intelligent robot |
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 |