CN109598772B - Automatic labeling picture data source expansion method based on single defect of overhead transmission line - Google Patents
Automatic labeling picture data source expansion method based on single defect of overhead transmission line Download PDFInfo
- Publication number
- CN109598772B CN109598772B CN201811406897.2A CN201811406897A CN109598772B CN 109598772 B CN109598772 B CN 109598772B CN 201811406897 A CN201811406897 A CN 201811406897A CN 109598772 B CN109598772 B CN 109598772B
- Authority
- CN
- China
- Prior art keywords
- transmission line
- overhead transmission
- background
- defect
- shooting
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/60—Editing figures and text; Combining figures or text
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/40—Filling a planar surface by adding surface attributes, e.g. colour or texture
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration by the use of local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
Abstract
The invention discloses an automatic labeling picture data source expansion method based on single defect of an overhead transmission line, which comprises the following steps: s1, building a defect-adjustable physical model by referring to a physical object of the overhead transmission line, and meanwhile, arranging a background for the physical model of the overhead transmission line by using green cloth; s2, simulating normal and various defect states of the overhead transmission line on the built model; s3, acquiring a data sample by means of a shooting device, adjusting a shooting distance and a shooting angle during shooting, and performing multi-angle shooting at different positions away from a target to acquire a visible light picture of a green single background; s4, processing the green background picture by adopting traditional image processing methods such as expansion, corrosion and the like, and extracting a front scene; and S5, adding a background shot by an unmanned aerial vehicle to the extracted foreground object to obtain a final picture data sample. The method can provide sufficient image data sources for automatic marking of the single defects of the overhead transmission line based on machine learning so as to realize batch marking of the defects.
Description
Technical Field
The invention relates to the technical field of machine learning, in particular to an automatic labeling picture data source expansion method based on single defects of an overhead transmission line.
Background
The overhead transmission line is widely applied to power transmission due to the advantages of simple structure, convenience in construction, overhaul and maintenance and the like. The operation condition of the system directly represents the operation condition of the whole power grid to a great extent. In order to ensure safe and stable operation of the power grid, personnel must perform regular inspections of the overhead transmission lines in accordance with relevant regulations. The traditional detection mode mainly depends on manual work, and a large amount of manpower and material resources are consumed when the problem of false detection or missed detection exists. The high-voltage transmission line is large in scale and wide in range, and the surrounding environment has a lot of potential dangers, so that the manual detection is not facilitated, and the application range of the traditional maintenance and inspection method is greatly reduced. In order to solve the problem, the helicopter inspection technology and the unmanned aerial vehicle inspection technology are widely applied successively, carry photographing or camera shooting equipment to fly along a power transmission line corridor, shoot lines and key components thereof in a close range, acquire aerial images and manually mark the defect parts of the overhead power transmission line in the aerial images. The traditional image labeling is manually completed, and the understanding and labeling of the image are relatively accurate, but in the image labeling in a big data environment, because the existence of massive image data sources and the defect types are complicated and complicated, the manual labeling workload is huge, the manual labeling is easily influenced by subjective experience, and the labeling of the same image is inconsistent.
Therefore, some researchers have proposed an intelligent detection method for realizing automatic labeling of images by using computer technology. The intelligent detection method is gradually paid attention as a new line inspection mode due to the advantages of excellent performance, wide application range, conformity with modern intelligentization and automation requirements and the like.
Machine learning is an important branch of artificial intelligence, is a fundamental way for enabling a computer to have intelligence, mainly researches how the computer simulates and realizes the learning behavior of human beings, identifies the existing knowledge, acquires new knowledge, continuously improves the performance and perfects the computer, and uses data or past experience so as to optimize the performance standard of a computer program. In contrast, deep learning plays an extremely important role in defect detection as a branch of machine learning. Therefore, the deep learning and modern inspection technology are combined, and the method has high convenience and superiority in the aspect of power defect detection. In recent years, the problem of defect detection based on images has achieved considerable achievement in defect detection with deep learning, and meanwhile, the defect detection technology based on machine learning still has a lot of key problems to be solved.
On one hand, under the background of large image data, although image resources are massive and abundant, and the number of some internet-based image data set samples is sufficient, for a specific field of a power system, the number of labeled image samples is not large, which causes great difficulty to deep learning; on the other hand, the deep convolutional neural network has a complex structure and huge parameter quantity, needs a large amount of data for training, and is very easy to generate an under-fitting phenomenon.
Therefore, it is necessary to provide an expansion method for single-defect automatic annotation picture data source of overhead transmission line in order to solve the above problems.
Disclosure of Invention
The invention aims to overcome the problem that the existing picture data source samples for the defects in the electric power field for deep learning training are insufficient, provides an automatic labeling picture data source expansion method based on single defects of an overhead transmission line, and can provide sufficient image data sources for automatic labeling of the single defects of the overhead transmission line based on machine learning so as to realize batch labeling of the defects.
The purpose of the invention can be achieved by adopting the following technical scheme:
an expansion method for single-defect automatic labeling picture data source of an overhead transmission line comprises the following steps:
s1, referring to an overhead transmission line real object, building a real object model of an electric power component with adjustable defects, and meanwhile, using green cloth to arrange a background for the overhead transmission line real object model;
s2, analyzing common defects of the overhead transmission line on the built physical model by combining data provided by a power inspection department, simulating, and simulating the defects of different degrees according to each type of defects when necessary;
s3, shooting a picture by a shooting device, adjusting a shooting distance and a shooting angle during shooting, and shooting at multiple angles at different positions away from the target defect to obtain a visible light picture of a green single background;
s4, processing a green background picture data source by adopting traditional image processing methods such as expansion, corrosion and the like, and extracting a scene in front of the overhead transmission line;
and S5, adding a background with strong interference to the picture on the basis of the obtained foreground object, wherein the background mainly comprises trees, buildings, roads, towers and the like, and obtaining a final picture data source.
As a preferred technical solution, the specific process of step S1 is as follows:
the method comprises the steps of referring to an overhead transmission line real object, building a real object model with adjustable defects of power parts such as transmission conductors, insulators, wire clamps and connecting hardware fittings, and meanwhile, adjusting the positions where green cloth is placed and the thickness of the green cloth to enable a shot picture background to be completely covered by the green cloth.
As a preferred technical solution, the specific process of step S2 is as follows:
on the built physical model, the common defects of the overhead transmission line are analyzed and simulated by combining the data provided by the power inspection department, and the defects of different degrees need to be simulated aiming at each type of defects when necessary.
According to the actual condition of power inspection, various defect types of the overhead transmission line are simulated by adopting modes of replacing and dismantling components and the like.
As a preferred technical solution, the specific process of step S3 is as follows:
selecting a time period suitable for illumination and temperature, aiming at each simulated defect, firstly adjusting a shooting device to be a long focal length, then aiming at a target object to carry out downward shooting, upward shooting and side shooting, wherein the positions and angles during shooting are diversified as much as possible, 5-8 pictures are shot at each position and each angle, and for some angles, in order to ensure that the background is completely covered by green cloth, the hanging position of the green cloth is readjusted.
As a preferred technical solution, the specific process of step S4 is as follows:
reading a simple background picture shot by an experiment on OpenCV, and then carrying out color space conversion, namely converting an RGB color space to an HSV color space; then, according to the HSV value range of the background, making a mask of the front scenery, and performing expansion and corrosion operations on the mask; and finally, extracting an interested area by taking the mask as a tool, and separating a foreground object (referred to as an overhead transmission line) from a green background.
As a preferred technical solution, the specific process of step S5 is as follows:
the unmanned aerial vehicle carries the shooting equipment used in the step S3 to collect various backgrounds, wherein the backgrounds mainly comprise towers, house buildings, trees and the like, and finally the backgrounds and foreground objects are superposed to obtain a final data source.
Compared with the prior art, the invention has the following advantages and effects:
1. the method of the invention gives full play to the advantages of the traditional image processing method in processing simple background pictures;
2. according to the method, on one hand, various defects of the overhead transmission line are simulated and the occurrence degree of the defects is considered, and on the other hand, various backgrounds are added to each picture based on the green screen technology, so that conditions are created for acquiring massive picture data sources in short time due to specific defects.
Drawings
FIG. 1 is a flow chart of an image processing algorithm in the present invention;
FIG. 2 is a schematic diagram of the spatial relationship of objects in the present invention;
fig. 3 is a flowchart of implementation steps of the method for automatically labeling the image data source expansion based on the single defect of the overhead transmission line disclosed by the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Examples
Fig. 1 shows a flow chart of an image processing algorithm, which illustrates how to add complex background interference to a green background picture. Fig. 2 shows the spatial position relationship among the photographing device, the loose closing pin and the green screen according to the embodiment. FIG. 3 is a flowchart illustrating exemplary steps comprising:
s1, building an overhead transmission line model capable of adjusting the tightness state of a closed pin, hanging green cloth on a support behind the overhead transmission line model, adjusting the position of the green cloth to enable a suspension clamp to be located in the center position in front of the green cloth, and paying attention to the fact that the closed pin on the clamp is in a normal state at the moment;
s2, simulating the loosening defect of the closed pin, and pulling the pin upwards to enable the pin to be in a certain loosening state;
s3, adjusting the focal length of the shooting device, aligning the focal length of the shooting device to a loosened cotter pin on the suspension clamp, shooting at multiple angles, and shooting multiple pictures at each angle to ensure that the discrimination of the front scenery and the green background of the pictures meets the requirement of a good state;
s4, repeating the step S2 to place the closed pin in another loosening state (the bolt loosening degree is different from that in the step S3), repeating the step S3, and obtaining a picture data source of the closed pin in another loosening state;
s5, manually arranging a preliminary picture data source, removing unclear pictures and pictures with a background not meeting requirements, reading the rest pictures on an OpenCV, then performing color space conversion, then manufacturing a mask of the foreground suspension clamp according to the HSV value range of the background, and performing expansion and corrosion operations on the mask; finally, extracting an interested area by taking the mask as a tool;
and S6, adding backgrounds with strong interference, mainly including trees, sky, roads, house buildings and the like, and obtaining a final picture data source.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.
Claims (5)
1. An automatic labeling picture data source expansion method based on single defect of an overhead transmission line is characterized by comprising the following steps:
s1, referring to an overhead transmission line real object, building a real object model of an electric power component with adjustable defects, and meanwhile, using green cloth to arrange a background for the overhead transmission line real object model;
s2, analyzing common defects of the overhead transmission line on the built physical model by combining data provided by a power inspection department, and simulating each type of defect;
s3, shooting a picture by a shooting device, adjusting a shooting distance and a shooting angle during shooting, and shooting at multiple angles at different positions away from the target defect to obtain a visible light picture of a green single background;
s4, processing a green background picture data source by adopting an image processing method, and extracting a scene in front of the overhead transmission line;
and S5, adding a background with strong interference to the picture on the basis of the obtained foreground object to obtain a final picture data source.
2. The overhead transmission line single-defect-based automatic labeling picture data source expansion method according to claim 1, wherein the electric power components comprise transmission conductors, insulators, wire clamps and/or connecting hardware fittings.
3. The overhead transmission line single-defect-based automatic labeling picture data source expansion method according to claim 1, wherein in the step S2, various defect types of the overhead transmission line are simulated by means of replacing and dismantling components.
4. The overhead transmission line single-defect-based automatic labeling picture data source expansion method according to claim 1, wherein the step S4 specifically comprises:
reading a background picture of the power transmission line model with the defects, which is shot in an experiment, on the OpenCV, and then performing color space conversion, namely converting an RGB color space into an HSV color space; then, according to the HSV value range of the background, making a mask of the front scenery, and performing expansion and corrosion operations on the mask; and finally, taking the mask as a tool, extracting an interested area, and separating the overhead transmission line from the green background.
5. The overhead transmission line single-defect-based automatic labeling picture data source expansion method according to claim 1, wherein the background comprises towers, house buildings and/or trees.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811406897.2A CN109598772B (en) | 2018-11-23 | 2018-11-23 | Automatic labeling picture data source expansion method based on single defect of overhead transmission line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811406897.2A CN109598772B (en) | 2018-11-23 | 2018-11-23 | Automatic labeling picture data source expansion method based on single defect of overhead transmission line |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109598772A CN109598772A (en) | 2019-04-09 |
| CN109598772B true CN109598772B (en) | 2023-01-06 |
Family
ID=65959541
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811406897.2A Active CN109598772B (en) | 2018-11-23 | 2018-11-23 | Automatic labeling picture data source expansion method based on single defect of overhead transmission line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109598772B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110120091B (en) * | 2019-04-28 | 2023-06-16 | 深圳供电局有限公司 | Method and device for manufacturing electric power inspection image sample and computer equipment |
| CN110503146B (en) * | 2019-08-21 | 2021-12-14 | 杭州比智科技有限公司 | Data enhancement method and device, computing equipment and computer storage medium |
| CN110726725A (en) * | 2019-10-23 | 2020-01-24 | 许昌许继软件技术有限公司 | Transmission line hardware corrosion detection method and device |
| CN112102443B (en) * | 2020-09-15 | 2022-07-05 | 国网电力科学研究院武汉南瑞有限责任公司 | Labeling system and labeling method suitable for substation equipment inspection image |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102680837A (en) * | 2012-05-31 | 2012-09-19 | 广州供电局有限公司 | Method and device for assessing induction lightning trip-out risk of distributing circuit surrounding transmission line |
| CN108257114A (en) * | 2017-12-29 | 2018-07-06 | 天津市万贸科技有限公司 | A kind of transmission facility defect inspection method based on deep learning |
| CN108389197A (en) * | 2018-02-26 | 2018-08-10 | 上海赛特斯信息科技股份有限公司 | Transmission line of electricity defect inspection method based on deep learning |
| CN108563742A (en) * | 2018-04-12 | 2018-09-21 | 王海军 | The method for automatically creating artificial intelligence image recognition training material and marking file |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2006107329A (en) * | 2004-10-08 | 2006-04-20 | Noritsu Koki Co Ltd | Photographed image processor |
-
2018
- 2018-11-23 CN CN201811406897.2A patent/CN109598772B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102680837A (en) * | 2012-05-31 | 2012-09-19 | 广州供电局有限公司 | Method and device for assessing induction lightning trip-out risk of distributing circuit surrounding transmission line |
| CN108257114A (en) * | 2017-12-29 | 2018-07-06 | 天津市万贸科技有限公司 | A kind of transmission facility defect inspection method based on deep learning |
| CN108389197A (en) * | 2018-02-26 | 2018-08-10 | 上海赛特斯信息科技股份有限公司 | Transmission line of electricity defect inspection method based on deep learning |
| CN108563742A (en) * | 2018-04-12 | 2018-09-21 | 王海军 | The method for automatically creating artificial intelligence image recognition training material and marking file |
Non-Patent Citations (1)
| Title |
|---|
| 有限元法应用于电缆终端应力锥缺陷分析;刘刚 等;《高压技术》;20070531;第33卷(第5期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109598772A (en) | 2019-04-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109598772B (en) | Automatic labeling picture data source expansion method based on single defect of overhead transmission line | |
| CN112633535A (en) | Photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle image | |
| CN110807353B (en) | Substation foreign matter identification method, device and system based on deep learning | |
| CN114792319B (en) | Transformer substation inspection method and system based on transformer substation image | |
| CN111832398B (en) | Unmanned aerial vehicle image distribution line pole tower ground wire broken strand image detection method | |
| CN109712127B (en) | Power transmission line fault detection method for machine inspection video stream | |
| CN111241615A (en) | Highly realistic multi-source fusion three-dimensional modeling method for transformer substation | |
| CN112258491A (en) | Three-dimensional simulation method for defects of power equipment | |
| CN102831393A (en) | Rapid image recognizing method of power tower pole outline | |
| CN114240868A (en) | Unmanned aerial vehicle-based inspection analysis system and method | |
| CN112541389A (en) | Power transmission line fault detection method based on EfficientDet network | |
| CN115293508B (en) | Visual optical cable running state monitoring method and system | |
| CN109816643A (en) | It is a kind of based on line defct identification tree line apart from intelligent analysis method | |
| CN110059076A (en) | A kind of Mishap Database semi-automation method for building up of power transmission and transformation line equipment | |
| CN111220619B (en) | Insulator self-explosion detection method | |
| CN114387558A (en) | Transformer substation monitoring method and system based on multi-dimensional video | |
| CN113095160B (en) | Power system personnel safety behavior identification method and system based on artificial intelligence and 5G | |
| CN111915565B (en) | Method for analyzing cracks of porcelain insulator of power transmission and transformation line in real time based on YOLACT algorithm | |
| CN112651337A (en) | Sample set construction method applied to training line foreign object target detection model | |
| CN108470141B (en) | Statistical feature and machine learning-based insulator identification method in distribution line | |
| Fang et al. | A framework of power pylon detection for UAV-based power line inspection | |
| CN112465959B (en) | Transformer substation three-dimensional live-action model inspection method based on local scene updating | |
| CN212785620U (en) | Monitoring system for power transmission network | |
| CN210835242U (en) | Defect sample collection system of electric power tower | |
| CN107895385A (en) | The Forecasting Methodology to individual outdoor images position of sun based on convolutional neural networks |
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 |