[go: up one dir, main page]

CN108256517B - Insulator string identification method based on lidar - Google Patents

Insulator string identification method based on lidar Download PDF

Info

Publication number
CN108256517B
CN108256517B CN201611234430.5A CN201611234430A CN108256517B CN 108256517 B CN108256517 B CN 108256517B CN 201611234430 A CN201611234430 A CN 201611234430A CN 108256517 B CN108256517 B CN 108256517B
Authority
CN
China
Prior art keywords
insulator string
depth
image
insulator
characteristic curve
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
CN201611234430.5A
Other languages
Chinese (zh)
Other versions
CN108256517A (en
Inventor
姜勇
刘国伟
王洪光
朱正国
宋屹峰
陈鹏
何斌斌
刘澈
胡冉
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.)
Shenyang Institute of Automation of CAS
Shenzhen Power Supply Bureau Co Ltd
Original Assignee
Shenyang Institute of Automation of CAS
Shenzhen Power Supply Bureau Co Ltd
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 Shenyang Institute of Automation of CAS, Shenzhen Power Supply Bureau Co Ltd filed Critical Shenyang Institute of Automation of CAS
Priority to CN201611234430.5A priority Critical patent/CN108256517B/en
Publication of CN108256517A publication Critical patent/CN108256517A/en
Application granted granted Critical
Publication of CN108256517B publication Critical patent/CN108256517B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
    • G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Image Analysis (AREA)

Abstract

本发明涉及基于激光雷达的绝缘子串识别方法，包括以下步骤：深度图像的获取：通过机器人上安装的激光雷达设备获得深度图像；深度图像的预处理：对带有绝缘子串的深度图像进行预处理，获得感兴趣的区域；深度图像的特征提取：在感兴趣的区域中，建立深度特征曲线模型，构建深度周期性特征矩阵和宽度向量，提取绝缘子串段数特征和盘径变化特征；绝缘子串类型识别：根据提取出的特征对绝缘子串类型进行识别。本发明根据绝缘子串本身的特性，利用灰度周期性特征和灰度阈值特征，对绝缘子串进行识别。识别结果准确性高，提高了工作效率，对之后的变电站巡检机器人完成冲洗任务提供依据。

Figure 201611234430

The invention relates to a laser radar-based insulator string identification method, which includes the following steps: acquisition of a depth image: obtaining a depth image through a laser radar device installed on a robot; preprocessing of the depth image: preprocessing the depth image with the insulator string , to obtain the region of interest; feature extraction of depth images: in the region of interest, establish a depth characteristic curve model, construct a depth periodic feature matrix and a width vector, and extract the characteristics of the number of insulator strings and the variation of the disk diameter; the type of insulator strings Identification: Identify the type of insulator string according to the extracted features. According to the characteristics of the insulator string itself, the invention uses the grayscale periodic feature and the grayscale threshold value feature to identify the insulator string. The recognition result has high accuracy, improves work efficiency, and provides a basis for the subsequent substation inspection robot to complete the flushing task.

Figure 201611234430

Description

Laser radar-based insulator string identification method
Technical Field
The invention relates to an autonomous identification method based on laser radar for a substation insulator string, in particular to an automatic identification method of the insulator string based on machine vision.
Background
Insulators operating on-line, in the natural environment, subject to SO2Nitrogen oxides, particulate dust and the like, and a layer of fouling substances is gradually deposited on the surface of the substrate. In dry weather, the insulators can keep a higher insulation level, and the discharge voltage of the insulators is close to that of the insulators in a clean and dry state; when the insulator is in humid weather such as fog, dew and rain, and when ice and snow are melted, the electrolyte in a filthy layer is dissolved due to the fact that the filthy matter on the surface of the insulator absorbs water, the insulation level of the insulator is reduced, leakage current is increased, and flashover accidents occur in severe cases. Therefore, the insulator string needs to be cleaned periodically.
At the present stage, the insulator string is cleaned by professional personnel of a transformer substation. When cleaning, the problems of high labor intensity, high danger and the like exist.
As an advanced technical means, the transformer substation inspection robot can replace manpower to carry out insulator string washing tasks. In order to complete the task, the robot needs to automatically identify the insulator string. Therefore, an insulator string automatic identification method with high automation degree needs to be provided.
Because the insulator string is in the outdoor environment, the illumination change is complex, and the structure built in the transformer substation is complex. Through laser radar detection, the influence of illumination on the sensor can be avoided, the influence of a complex background can be overcome, and the outdoor environment can be well adapted. And at present, an insulator string automatic identification method based on laser radar is not reported.
Disclosure of Invention
Aiming at the problems, the invention aims to provide a laser radar-based insulator string automatic identification method with high identification precision.
The technical scheme adopted by the invention for solving the technical problems is as follows: the laser radar-based insulator string identification method comprises the following steps of:
acquiring a depth image: obtaining a depth image through laser radar equipment arranged on the robot;
preprocessing of the depth image: preprocessing the depth image with the insulator string to obtain an interested area;
feature extraction of the depth image: in the interested region, establishing a depth characteristic curve model, constructing a depth periodic characteristic matrix and a width vector, and extracting insulator string segment number characteristics and disc diameter variation characteristics;
identifying the type of an insulator string: and identifying the type of the insulator string according to the extracted features.
The depth image preprocessing comprises the following steps:
(2.1) carrying out binarization processing on the depth image, separating a foreground and a background in the depth image, and eliminating background interference;
(2.2) carrying out expansion operation on the processed binary image, and then carrying out corrosion operation to eliminate interference pixel points on the binary image;
and (2.3) selecting the maximum connected region, extracting the region of interest, and finishing pretreatment.
The depth characteristic curve model is established as follows:
extracting the left edge of the target area:
Figure BDA0001195051140000021
lik＝k,
Figure BDA0001195051140000022
extracting the right edge of the target area:
Figure BDA0001195051140000023
rik＝k,
Figure BDA0001195051140000024
where D (x, y) is a binary image, k represents the number of rows in the image, t represents the number of columns in the image, lk，rkPixel point locations on the edge;
determining the position on the line in the target area by the left and right edges:
Figure BDA0001195051140000025
mik＝k,
Figure BDA0001195051140000031
in the formula mkAs coordinates of the depth profile pixel points, mikIs a depth profile pixel point row coordinate, mjkThe pixel point row coordinates of the depth characteristic curve are obtained; n isIThe number of image lines;
establishing a depth characteristic curve model:
f(xk,yk)＝0
wherein:
xk＝mik
Figure BDA0001195051140000032
xkas the abscissa of the characteristic curve, i.e. the number of image lines, ykIs the ordinate of the characteristic curve, namely the depth value; l (m)k) Representing a depth image.
The construction of the depth periodic feature matrix comprises the following steps:
Figure BDA0001195051140000033
firstly, the depth characteristic curve f (x) is obtainedk,yk) Maximum and minimum values on 0; building (2)Vertical matrix
Figure BDA0001195051140000034
In the formula:
Figure BDA0001195051140000035
v2kis a depth characteristic curve f (x)k,yk) 0 on the abscissa of the extreme point, v3kIs a depth characteristic curve f (x)k,yk) The ordinate of the extreme point on 0, namely the depth value; k is 1,2 … nI；
Setting a threshold lambdac1,λc2And lambdac1﹤λc2(ii) a Then to
Figure BDA0001195051140000036
Traversing by column to obtain V3,k+1-V3,k＞λc1And V3,k+1-V3,k＜λc2Time V1,kAnd V2,kRecording; when the traversal is finished, the values meeting the conditions are formed into a matrix according to the recording sequence
Figure BDA0001195051140000037
V is to be1,kRecorded in a matrix
Figure BDA0001195051140000038
First row of (v)2,kRecorded in the second row of the matrix, resulting in:
Figure BDA0001195051140000041
n3is the number of matrix columns;
then to
Figure BDA0001195051140000042
Go through the rows by row, order
c1,k＝abs(v11,k+1-v11,k)
c2,k＝v12,k+1-v12,k
c3,k＝v12,k
Obtaining a depth periodic feature matrix
Figure BDA0001195051140000043
Constructing a width vector:
Figure BDA0001195051140000047
in the formula:
Figure BDA0001195051140000044
i＝1,2…nI
(xi1,yi1,zi1) Is the spatial coordinate of the left edge point, (x)i2,yi2,zi2) Is the spatial coordinate of the right edge point; n isIRepresenting the number of image lines.
The method for extracting the insulator string number characteristics and the disc diameter variation characteristics comprises the following steps:
(1) by
Figure BDA0001195051140000045
Extracting insulator string segment number characteristics
First, a threshold value alpha is setcInitializing the flag bit deltacIs 0, the segment number information num (C) is 0;
then to
Figure BDA0001195051140000046
Traversing by columns:
when C is present1,k＝1，C2,k＜αcAnd deltacWhen equal to 0, set the flag bit deltacIs 1, and C is3,kStored in vector T as upper edge of insulator string；
When deltac1 and C1,k0 or C2,k＞αcSetting a flag bit deltacIs 0, adding C3,kStoring the lower edge of the insulator string into the vector T, and adding 1 to the number information num (C);
vector T is the upper and lower edges of the insulator string, where the odd-term element T2n-1The even-numbered element t being the upper edge of the insulator string2nThe lower edge of the insulator string; n is 1,2 … k1；k1Is the length of the vector T.
(2) Extracting disc diameter change characteristics of insulator string from W
And calculating the width variance of the insulator string according to the vector T:
Figure BDA0001195051140000051
Figure BDA0001195051140000052
when d is2< threshold value alphawThe insulator is equal in disc diameter, otherwise, the insulator is variable in disc diameter.
The method for identifying the insulator string type according to the extracted features comprises the following steps:
if num (C) is 1, and d2< threshold value alphawIf the insulator string is a single-section insulator string with the same disc diameter; if num (C) is 2, and d2< threshold value alphawIf num (c) is 1 and d is equal to d2> threshold value alphawAnd the insulator string is a single-section disc-changing insulator string.
The invention has the following beneficial effects and advantages:
1. the invention utilizes the laser depth periodicity characteristic, the insulator string spacing characteristic and the disc diameter characteristic to identify, has high insulator identification accuracy, improves the working efficiency, and provides a basis for the completion of tasks of the transformer substation inspection robot.
2. The invention adopts the laser radar as the sensor, and can accurately identify the insulator string under the backlight condition. In the depth image, the foreground and the background can be accurately separated, and the influence of the complex background on the identification precision is overcome.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a laser depth map;
FIG. 3 is a depth binary map after preprocessing;
FIG. 4 is a depth signature graph;
fig. 5 is a recognition effect diagram.
Detailed Description
The present invention will be described in further detail with reference to examples.
A laser radar-based insulator string automatic identification method comprises the following steps:
(1) acquiring a depth image: installing laser radar equipment on the transformer substation inspection robot, fixing the laser radar equipment on a holder, and rotating the laser radar by a fixed angle to obtain a depth image;
(2) preprocessing of the depth image: and preprocessing the depth image with the insulator string to obtain the interested region.
(3) Feature extraction of the depth image: in the interested region, a depth characteristic curve model is established, a depth periodic characteristic matrix and a width vector are established, and insulator string segment number characteristics and disc diameter variation characteristics are extracted.
(4) Identifying the type of an insulator string: and identifying the type of the insulator string according to the extracted features.
The depth image preprocessing comprises the following steps:
and (2.1) separating the foreground and the background in the depth image through depth binarization processing, and eliminating background interference in the depth image.
And (2.2) processing the depth binary image by adopting image morphology, firstly adopting expansion operation, and then adopting corrosion operation to eliminate interference pixel points on the depth binary image.
And (2.3) selecting the maximum connected domain from the depth binary image, extracting the interested region and finishing the pretreatment.
The feature extraction method of the depth image comprises the following steps:
(3.1) establishing a depth characteristic curve model, wherein the defined depth characteristic curve can reflect the characteristics of the insulator string, and the method comprises the following steps:
extracting the left edge of the target area:
Figure BDA0001195051140000061
lik＝k,
Figure BDA0001195051140000062
extracting the right edge of the target area:
Figure BDA0001195051140000063
rik＝k,
Figure BDA0001195051140000071
where D (x, y) is a depth binary image, k represents the number of rows of the image, t represents the number of columns of the image, lk，rkPixel point locations on the edge. n isIRepresented as the maximum number of lines of the image. From these two sets, a set of positions on the line in the target area can be determined.
Figure BDA0001195051140000072
mik＝k,
Figure BDA0001195051140000073
In the formula mkAs coordinates of the depth profile pixel points, mikIs a depth profile pixel point row coordinate, mjkIs the pixel point column coordinate of the depth characteristic curve.
Establishing a depth characteristic curve model:
f(xk,yk)＝0
wherein:
xk＝mik
Figure BDA0001195051140000074
xkas the abscissa of the characteristic curve, i.e. the number of image lines, ykIs the ordinate of the characteristic curve, namely the depth value; l (m)k) Representing a depth image.
And (3.2) constructing a depth periodic feature matrix and a width vector through a certain rule, and extracting the insulator string number features and the disc diameter variation features.
And analyzing the number characteristics and the width characteristics of the insulator string segments extracted from the depth image to finish the identification of the insulator string.
The flow chart of the laser radar-based insulator string automatic identification method is shown in figure 1. The specific process is as follows:
1. laser radar data acquisition
The laser radar scans the insulator string region to obtain a laser depth image L (x, y), as shown in fig. 2.
2. Image pre-processing
Binarizing the depth image according to formula (1):
Figure BDA0001195051140000081
d (x, y) represents binarization, and M is a threshold value. And then carrying out expansion corrosion operation and marking the maximum connected domain. After pretreatment, the foreground and background can be separated to obtain the insulator string and the erection thereof. As shown in fig. 3.
3. Establishing a depth characteristic curve model
According to the depth periodic variation of the region where the insulator string is located in the depth image, a depth characteristic curve model is established, and the method comprises the following steps:
extracting the left edge of the target area:
Figure BDA0001195051140000082
lik＝k,
Figure BDA0001195051140000083
extracting the right edge of the target area:
Figure BDA0001195051140000084
rik＝k,
Figure BDA0001195051140000085
where D (x, y) is a binary image, k represents the number of rows in the image, t represents the number of columns in the image, lk，rkPixel point locations on the edge. The left edge and the right edge are respectively the left edge and the right edge of the insulator string. From these two sets, a set of positions of the lines in the target area can be determined.
Figure BDA0001195051140000086
mik＝k,
Figure BDA0001195051140000087
In the formula mkAs coordinates of the depth profile pixel points, mikIs a depth profile pixel point row coordinate, mjkIs a depth characteristic curvePixel point column coordinates.
Establishing a depth characteristic curve model:
f(xk,yk)＝0
wherein:
xk＝mik
Figure BDA0001195051140000091
as shown in fig. 4, the abscissa represents the number of image lines, and the ordinate represents the depth value.
4. Constructing depth periodic feature and width feature vectors
(1) Constructing a depth periodic feature matrix:
Figure BDA0001195051140000092
firstly, the depth characteristic curve f (x) is obtainedk,yk) Maximum and minimum values on 0. Building a matrix
Figure BDA0001195051140000093
In the formula:
Figure BDA0001195051140000094
v2kis a depth characteristic curve f (x)k,yk) 0 on the abscissa of the extreme point, v3kIs a depth characteristic curve f (x)k,yk) The ordinate of the extreme point at 0, i.e. the depth value. n is1Representation matrix
Figure BDA0001195051140000095
Number of columns, n2Representation matrix
Figure BDA0001195051140000096
The number of columns.
Setting a threshold lambdac1,λc2，λc1﹤λc2In this embodiment, 3 and 10 are taken, respectively. Then to
Figure BDA0001195051140000097
Go through by column, go v3,k+1-v3,k＞λc1And v is3,k+1-v3,k＜λc2V of time1,kAnd v2,kAnd recording is carried out. When the traversal is finished, the values meeting the conditions are formed into a matrix according to the recording sequence
Figure BDA0001195051140000098
V is to be1,kRecorded in a matrix
Figure BDA0001195051140000099
First row of (v)2,kRecorded in the second row of the matrix, resulting in:
Figure BDA00011950511400000910
n3is the number of matrix columns.
Then go through V1 by column, let
c1,k＝abs(v11,k+1-v11,k)
c2,k＝v12,k+1-v12,k
c3,k＝v12,k
And obtaining a depth periodic feature matrix.
(2) The width feature vector is constructed as follows:
Figure BDA0001195051140000106
in the formula:
Figure BDA0001195051140000101
i＝1，2...nI
(xi1,yi1,zi1) Is the spatial coordinate of the left edge point, (x)i2,yi2,zi2) The spatial coordinates of the right edge point. n isIRepresenting the maximum number of lines of the image. Here, the
5. Extracting insulator string segment number characteristic and disc diameter characteristic
(1) By
Figure BDA0001195051140000102
Extracting insulator string segment number characteristics
First, a threshold value alpha is setcIn this embodiment, 15 is taken. Initialization flag bit deltacIs 0, and the segment number information num (C) is 0.
Then to
Figure BDA0001195051140000103
Traversal is performed by column. When C is present1,k＝1，C2,k＜αcAnd deltacWhen equal to 0, set the flag bit deltacIs 1, and C is3,kAnd storing the vector T as the upper edge of the insulator string. When deltac1 and C1,k0 or C2,k＞αcSetting a flag bit deltacIs 0, adding C3,kAnd storing the lower edge of the insulator string into the vector T, and adding 1 to the segment number information num (C). Traversing to the tail end of the matrix to finish the recognition of the insulator string. Vector T is the upper and lower edges of the insulator string, where the odd-term element T2n-1The even-numbered element t being the upper edge of the insulator string2nThe lower edge of the insulator string. n is 1,2 … k1；k1Is the length of the vector T.
(2) Extracting disc diameter change characteristics of insulator string from W
And calculating the width variance of the insulator string according to the vector T:
Figure BDA0001195051140000104
Figure BDA0001195051140000105
when d is2＜αwWhen is αwFor the threshold, this embodiment takes 0.01. (ii) a And extracting the disc diameter characteristic as equal disc diameter, otherwise, extracting the disc diameter characteristic as variable disc diameter.
6. Insulator string type identification
And identifying three types of pillar insulator strings in the transformer substation according to the section number information and the disc diameter change information. Combining the obtained characteristics of the number of segments and the disc diameter, if num (C) is 1 and d is2< threshold value alphawIf num (c) is 2 and d is equal to d2< threshold value alphawIf num (c) is 1 and d is equal to d2> threshold value alphawAnd the insulator string is a single-section disc-changing insulator string. And (4) identifying three insulator strings commonly used by the transformer substation. As shown in fig. 5, the insulator string is a two-section equal-disc-diameter insulator string.

Claims (4)

1.基于激光雷达的绝缘子串识别方法，其特征在于，包括以下步骤：1. The insulator string identification method based on lidar, is characterized in that, comprises the following steps: 深度图像的获取：通过机器人上安装的激光雷达设备获得深度图像；Acquisition of depth images: obtain depth images through the lidar equipment installed on the robot; 深度图像的预处理：对带有绝缘子串的深度图像进行预处理，获得感兴趣的区域；Preprocessing of depth images: Preprocessing depth images with insulator strings to obtain regions of interest; 深度图像的特征提取：在感兴趣的区域中，建立深度特征曲线模型，构建深度周期性特征矩阵和宽度向量，提取绝缘子串段数特征和盘径变化特征；Feature extraction of depth images: In the region of interest, establish a depth feature curve model, build a depth periodic feature matrix and width vector, and extract the characteristics of insulator string segments and disk diameter variation characteristics; 绝缘子串类型识别：根据提取出的特征对绝缘子串类型进行识别；Insulator string type identification: identify the insulator string type according to the extracted features; 所述建立深度特征曲线模型如下：The described establishment of the depth characteristic curve model is as follows: 提取目标区域左边缘：Extract the left edge of the target area:
Figure FDA0003015882920000014
Figure FDA0003015882920000014
 lik＝k,li k = k,
Figure FDA0003015882920000011
Figure FDA0003015882920000011
 提取目标区域右边缘：Extract the right edge of the target area:
Figure FDA0003015882920000015
Figure FDA0003015882920000015
 rik＝k,ri k = k,
Figure FDA0003015882920000012
Figure FDA0003015882920000012
 式中D(x,y)为二值图像，k代表图像的行数，t代表图像的列数，lk，rk为边缘上的像素点位置；where D(x, y) is a binary image, k represents the number of rows of the image, t represents the number of columns of the image, l k , r k are the pixel positions on the edge; 通过左边缘和右边缘确定目标区域中线上的位置：Determine the position of the center line of the target area by the left and right edges:
Figure FDA0003015882920000016
Figure FDA0003015882920000016
 mik＝k,mi k = k,
Figure FDA0003015882920000013
Figure FDA0003015882920000013
 式中mk为深度特征曲线像素点的坐标，mik为深度特征曲线像素点行坐标，mjk为深度特征曲线像素点列坐标；nI为图像行数；where m k is the coordinate of the pixel point of the depth characteristic curve, mi k is the row coordinate of the pixel point of the depth characteristic curve, mj k is the column coordinate of the pixel point of the depth characteristic curve; n I is the number of image rows; 建立深度特征曲线模型:Build the depth characteristic curve model: f(xk,yk)＝0f(x k ,y k )=0 其中:in: xk＝mik x k = mi k
Figure FDA0003015882920000021
Figure FDA0003015882920000021
 xk为特征曲线的横坐标，即图像行数，yk为特征曲线的纵坐标，即深度值；L(mk)表示深度图像；x k is the abscissa of the characteristic curve, that is, the number of image lines, y k is the ordinate of the characteristic curve, that is, the depth value; L(m k ) represents the depth image; 所述构建深度周期性特征矩阵：Said constructing a deep periodic feature matrix:
Figure FDA0003015882920000022
Figure FDA0003015882920000022
 先求出深度特征曲线f(xk,yk)＝0上的极大值和极小值；建立矩阵First find the maximum and minimum values on the depth characteristic curve f(x k , y k )=0; establish a matrix
Figure FDA0003015882920000023
Figure FDA0003015882920000023
 式中：where:
Figure FDA0003015882920000024
Figure FDA0003015882920000024
 v2k为深度特征曲线f(xk,yk)＝0上极值点的横坐标，v3k为深度特征曲线f(xk,yk)＝0上极值点的纵坐标，即深度值；k＝1,2…nI；v 2k is the abscissa of the extreme point on the depth characteristic curve f(x k , y k )=0, v 3k is the ordinate of the extreme point on the depth characteristic curve f(x k , y k )=0, that is, the depth value; k=1,2...n I ; 设定阈值λc1,λc2且λc1﹤λc2；再对
Figure FDA0003015882920000025
进行按列遍历，将V3,k+1-V3,k＞λc1并且V3,k+1-V3,k＜λc2时的V1,k与V2,k进行记录；当遍历结束时，将满足条件的值按记录顺序组成矩阵
Figure FDA0003015882920000026
将v1,k记录在矩阵
Figure FDA0003015882920000027
的第一行，v2,k记录在矩阵的第二行，得到：Set the thresholds λ c1 , λ c2 and λ c1 <λ c2 ;
Figure FDA0003015882920000025
Perform column traversal, record V 1,k and V 2,k when V 3,k+1 -V 3,k >λ c1 and V 3,k+1 -V 3,k <λ c2 ; when At the end of the traversal, the values that meet the conditions are formed into a matrix in record order
Figure FDA0003015882920000026
record v 1,k in the matrix
Figure FDA0003015882920000027
The first row of , v 2,k is recorded in the second row of the matrix, giving:
Figure FDA0003015882920000028
Figure FDA0003015882920000028
 n3为矩阵列数；n 3 is the number of matrix columns; 再对
Figure FDA0003015882920000029
按列进行遍历，令Right again
Figure FDA0003015882920000029
Iterate over the columns, let c1,k＝abs(v11,k+1-v11,k)c 1,k =abs(v1 1,k+1 -v1 1,k ) c2,k＝v12,k+1-v12,k c 2,k =v1 2,k+1 -v1 2,k c3,k＝v12,k c 3,k = v1 2,k 得到深度周期性特征矩阵
Figure FDA0003015882920000031
get deep periodic feature matrix
Figure FDA0003015882920000031
 构建宽度向量：Build the width vector:
Figure FDA0003015882920000032
Figure FDA0003015882920000032
 式中：where:
Figure FDA0003015882920000033
Figure FDA0003015882920000033
 (xi1,yi1,zi1)为左边缘点的空间坐标，(xi2,yi2,zi2)为右边缘点的空间坐标；(x i1 , y i1 , z i1 ) are the spatial coordinates of the left edge point, and (x i2 , y i2 , z i2 ) are the spatial coordinates of the right edge point; nI表示图像行数。n I represents the number of image lines. 2.根据权利要求1所述的基于激光雷达的绝缘子串识别方法，其特征在于，所述深度图像的预处理包括以下步骤：2. The lidar-based insulator string identification method according to claim 1, wherein the preprocessing of the depth image comprises the following steps: (2.1)对深度图像二值化处理，分离出深度图像中的前景、背景，并消除背景干扰；(2.1) Binarize the depth image, separate the foreground and background in the depth image, and eliminate background interference; (2.2)对处理后的二值图像采用膨胀运算，再采用腐蚀运算，消除二值图像上的干扰像素点；(2.2) Use dilation operation on the processed binary image, and then use erosion operation to eliminate the interference pixels on the binary image; (2.3)然后选取最大连通域，提取出感兴趣区域，完成预处理。(2.3) Then select the largest connected domain, extract the region of interest, and complete the preprocessing. 3.根据权利要求1所述的基于激光雷达的绝缘子串识别方法，其特征在于，所述提取出绝缘子串段数特征和盘径变化特征包括以下步骤：3. The laser radar-based insulator string identification method according to claim 1, wherein the extraction of the insulator string segment number feature and the disk diameter variation feature comprises the following steps: (1)由
Figure FDA0003015882920000034
提取绝缘子串段数特征(1) by
Figure FDA0003015882920000034
Extract the feature of insulator string segment number 首先设定阈值αc，初始化标志位δc为0，段数信息Num(C)为0；First set the threshold α c , the initialization flag δ c is 0, and the segment number information Num(C) is 0; 再对
Figure FDA0003015882920000035
按列进行遍历：Right again
Figure FDA0003015882920000035
Iterate by column: 当C1,k＝1，C2,k＜αc并且δc＝0时，置标志位δc为1，并将C3,k存入向量T中作为绝缘子串上边缘；When C 1,k =1, C 2,k <α c and δ c =0, the flag bit δ c is set to 1, and C 3,k is stored in the vector T as the upper edge of the insulator string; 当δc＝1并且C1,k＝0或C2,k＞αc时，置标志位δc为0，将C3,k存入向量T中作为绝缘子串下边缘，并将段数信息Num(C)加1；When δ c =1 and C 1,k =0 or C 2,k >α c , set the flag bit δ c to 0, store C 3,k in the vector T as the lower edge of the insulator string, and store the segment number information Num(C) plus 1; 向量T为绝缘子串的上、下边缘，其中奇数项元素t2n-1为绝缘子串的上边缘，偶数项元素t2n为绝缘子串的下边缘；n＝1,2…k1；k1为向量T的长度;The vector T is the upper and lower edges of the insulator string, wherein the odd-numbered element t 2n-1 is the upper edge of the insulator string, and the even-numbered element t 2n is the lower edge of the insulator string; n=1,2...k 1 ; k 1 is the length of the vector T; (2)由W提取绝缘子串的盘径变化特征(2) Extracting the variation characteristics of the disk diameter of the insulator string by W 根据向量T计算绝缘子串宽度方差：Calculate the insulator string width variance from the vector T:
Figure FDA0003015882920000041
Figure FDA0003015882920000041
Figure FDA0003015882920000042
Figure FDA0003015882920000042
 当d2＜阈值αw时，该绝缘子为等盘径，否则，为变盘径。When d 2 <threshold α w , the insulator has a constant disk diameter, otherwise, the insulator has a variable disk diameter. 4.根据权利要求3所述的基于激光雷达的绝缘子串识别方法，其特征在于，所述根据提取出的特征对绝缘子串类型进行识别包括以下步骤：4. The lidar-based insulator string identification method according to claim 3, wherein the identifying the insulator string type according to the extracted features comprises the following steps: 若Num(C)＝1，且d2＜阈值αw，则绝缘子串为单段等盘径绝缘子串；若Num(C)＝2，且d2＜阈值αw，则绝缘子串为双段等盘径绝缘子串，若Num(C)＝1，且d2＞阈值αw，则绝缘子串为单段变盘经绝缘子串。If Num(C)=1, and d 2 <threshold α w , the insulator string is a single-segment equal-disk diameter insulator string; if Num(C)=2, and d 2 <threshold α w , then the insulator string is a double-segment insulator string For an insulator string of equal disk diameter, if Num(C)=1, and d 2 >threshold α w , the insulator string is a single-segment variable disk passing insulator string.
CN201611234430.5A 2016-12-28 2016-12-28 Insulator string identification method based on lidar Active CN108256517B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201611234430.5A CN108256517B (en) 2016-12-28 2016-12-28 Insulator string identification method based on lidar

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201611234430.5A CN108256517B (en) 2016-12-28 2016-12-28 Insulator string identification method based on lidar

Publications (2)

Publication Number Publication Date
CN108256517A CN108256517A (en) 2018-07-06
CN108256517B true CN108256517B (en) 2021-05-28

Family

ID=62719369

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201611234430.5A Active CN108256517B (en) 2016-12-28 2016-12-28 Insulator string identification method based on lidar

Country Status (1)

Country Link
CN (1) CN108256517B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112037137A (en) * 2020-07-21 2020-12-04 国网湖北省电力有限公司电力科学研究院 Method and device for eliminating fuzzy region of insulator disc surface edge in infrared image
CN112444522B (en) * 2020-11-16 2021-09-28 中国科学院沈阳自动化研究所 Method for detecting defects of insulator string of power system
CN113538454B (en) * 2021-06-03 2025-03-07 北京道亨软件股份有限公司 A method for automatically dividing insulation areas of multi-insulator strings

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20090019535A (en) * 2007-08-21 2009-02-25 금오테크(주) Insulation deterioration and disconnection measurement system for non-contact transmission
CN103714342A (en) * 2013-12-20 2014-04-09 华北电力大学（保定） An aerial-photo insulator chain automatic positioning method based on binary image shape features

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20090019535A (en) * 2007-08-21 2009-02-25 금오테크(주) Insulation deterioration and disconnection measurement system for non-contact transmission
CN103714342A (en) * 2013-12-20 2014-04-09 华北电力大学（保定） An aerial-photo insulator chain automatic positioning method based on binary image shape features

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
航拍绝缘子图像的检测与识别方法研究;严凯 等;《贵州电力技术》;20151029;第18卷(第10期);第51-53页 *

Also Published As

Publication number Publication date
CN108256517A (en) 2018-07-06

Similar Documents

Publication Publication Date Title
CN106980816B (en) Automatic insulator string identification method based on optical image
CN109685788B (en) Automatic detection method for floor defect image based on morphological characteristics
CN111652857B (en) A kind of infrared detection method of insulator defect
CN108256517B (en) Insulator string identification method based on lidar
CN102520401B (en) Building area extraction method based on LiDAR data
CN108596165B (en) Road traffic marking detection method and system based on unmanned plane low latitude Aerial Images
CN103886594B (en) Road surface line laser rut detection and recognition methods and disposal system
CN108921813B (en) Unmanned aerial vehicle detection bridge structure crack identification method based on machine vision
CN101976437A (en) High-resolution remote sensing image variation detection method based on self-adaptive threshold division
CN106127728B (en) A kind of infrared image sea horizon under sea and sky background is connected to area detecting method
CN105957081B (en) A kind of glass insulator falls to go here and there fault detection method
CN115601379B (en) A precise surface crack detection technology based on digital image processing
JP4189061B2 (en) Crack detection method for concrete surface
CN110807766A (en) Double-anti-vibration hammer displacement identification method based on visible light image
CN117593300B (en) A PE pipe crack defect detection method and system
CN111046950A (en) Image processing method and device, storage medium and electronic device
CN101576957B (en) A Fast River Extraction Method for Remote Sensing Images
CN110009633A (en) A method of rail surface defect detection based on inverse Gaussian difference
CN119168859B (en) Temperature curve splicing method and system for superlong string disc-shaped suspension porcelain insulator
CN107527331A (en) Polar region iceberg remote sensing recognition method based on double bubbling methods
CN113420670A (en) Environment-friendly supervision method for changing power transmission and transformation line migration based on high-resolution remote sensing
CN110909751B (en) Visual identification method, system and medium for transformer substation insulator cleaning robot
CN104504702A (en) Cement notch pavement crack recognition method based on square lattice searching method
CN119098452A (en) An automatic detection method for pipe water spraying based on image recognition
CN117571833A (en) Ultrasonic phased array defect extraction method based on self-adaptive region growth

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