CN111832398A - Unmanned aerial vehicle image distribution line pole tower ground wire broken strand image detection method - Google Patents
Unmanned aerial vehicle image distribution line pole tower ground wire broken strand image detection method Download PDFInfo
- Publication number
- CN111832398A CN111832398A CN202010490245.2A CN202010490245A CN111832398A CN 111832398 A CN111832398 A CN 111832398A CN 202010490245 A CN202010490245 A CN 202010490245A CN 111832398 A CN111832398 A CN 111832398A
- Authority
- CN
- China
- Prior art keywords
- image
- unmanned aerial
- aerial vehicle
- area
- analyzed
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/54—Testing for continuity
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/58—Testing of lines, cables or conductors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The invention relates to the technical field of image recognition, in particular to a broken-strand image detection method for a distribution line pole tower ground wire of an unmanned aerial vehicle image, which comprises the following steps: A) establishing a picture data set of broken strands of the conducting wires and the grounding wires, and marking and positioning a large target in the image; B) transmitting picture data shot by the unmanned aerial vehicle to a PC (personal computer) end from the unmanned aerial vehicle by wire, entering a typical defect intelligent analysis module, identifying an area to be analyzed, establishing an analysis frame, manually marking defects, and obtaining sample data; C) constructing a Cascade R-CNN network, and training and testing the Cascade R-CNN network by using sample data; D) and acquiring a tour shot picture of the unmanned aerial vehicle, identifying an area to be analyzed, importing the area into a Cascade R-CNN network, and acquiring a detection result. The substantial effects of the invention are as follows: the performance of detecting the small target is improved; the accuracy of detecting the broken strand fault of the ground wire of the tower of the distribution line is improved.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a broken-strand image detection method for a distribution line pole tower ground wire of an unmanned aerial vehicle image.
Background
The transmission line is an artery of the power system, and the transmission line conductor is an important component constituting the transmission line. The strand breakage and damage of the transmission conductor caused by factors such as external force damage, meteorological disasters, line aging and the like seriously threaten the stable operation of the line, and seriously influence the normal production and life of people. By carrying out image processing on the aerial photography power transmission conductor video/image of the unmanned aerial vehicle line patrol, the real-time detection of the running state of the power transmission conductor can be realized, and the safe running of a power grid is effectively guaranteed. Therefore, the research of the distribution line pole tower ground wire broken strand image detection method based on the small sample size unmanned aerial vehicle image has very important practical significance. The starting of the inspection and detection research of unmanned aerial vehicles and helicopters in China is late compared with that of developed countries in Europe and America, a plurality of problems exist in many technical aspects, and the comprehensive implementation of the inspection and detection technology of unmanned aerial vehicles and helicopters cannot be realized currently. At present, images acquired by image acquisition equipment erected in an unmanned aerial vehicle and a helicopter in China mainly depend on later-stage manual detection, detection results are affected by artificial subjective factors, and efficiency is low. The method for detecting strand breakage and damage of the existing transmission conductor in China mainly comprises the following steps: fault monitoring based on infrared images, fault monitoring based on visible light images, steel core aluminum strand wire strand breakage detection based on magnetic flux leakage and eddy current sensors, wire fatigue strength detection based on various sensors and the like. The method is disclosed in the patent of ' detection of broken strand of power transmission line and robot behavior planning based on visual method ' [ J ]. robot, 2015,37(2):204 + 211 '.
The applicant finds that the closest to the technology of the present application is the document by consulting the data: jiang Liang, Xia Yun Feng, Zhang Qiang, and the like, Power Transmission line Strand Break image detection based on optimized Gabor Filter [ J ] Power System Automation, 2011,35(15): 78-83.) among Jiang Liang and the like, a recognition method for realizing the detection of the Power Transmission line Strand Break Defect is provided by calculating the output response of the convolution energy of the intact wire image and the filter, applying a niche genetic algorithm to find the optimal Gabor filter parameter and segmentation threshold, and performing binarization processing on the detected image and the energy. The method can realize the acquisition of the strand breakage and damage conditions of the single side of the lead involved in the lead image, the strand breakage condition of the undetected back area of the lead is unknown, and the strand breakage lead with serious strand scattering is difficult to effectively detect and is easy to have false detection and missing detection; secondly, the detection result of the method is effective for the wire image of the power transmission line with a single background, but the detection result of the method has certain limitation because the complex background in the wire image of the power transmission line obtained by a helicopter, an unmanned aerial vehicle or an inspection robot always exists, and based on the color characteristic of dark wire of the power transmission line and the characteristics of the image that the gray value of pixels on the surface of the wire in the obtained wire image is influenced by illumination is uneven, the definition of the wire area is high (the dispersion degree of the gray value is high), and the definition of trees, houses, roads and the like in the background area is low.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the current method for detecting the broken strand of the front ground wire has low detection accuracy. The method for detecting the broken strand image of the distribution line pole tower ground wire of the unmanned aerial vehicle image is provided, the method carries out optimization processing on the identification of the small target characteristics, the detection accuracy is greatly improved, and meanwhile the number of required training samples is reduced.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows: a method for detecting broken strand images of a ground wire of a power distribution line tower of an unmanned aerial vehicle image comprises the following steps: A) establishing a picture data set of broken strands of the conducting wires and the grounding wires, and marking and positioning a large target in the image; B) transmitting picture data shot by the unmanned aerial vehicle to a PC (personal computer) end from the unmanned aerial vehicle through a wire, entering a typical defect intelligent analysis module, providing a model for intelligently identifying an area to be analyzed, establishing an analysis frame, manually marking defects, and obtaining sample data; C) constructing a Cascade R-CNN network, and training and testing the Cascade R-CNN network by using sample data; D) and C), acquiring a tour shot picture of the unmanned aerial vehicle, identifying an area to be analyzed, importing the area into the Cascade R-CNN network trained in the step C), and acquiring a detection result. The pictures of the picture data set for establishing the broken strands of the conducting wires and the grounding wires are from visible light pictures shot by the unmanned aerial vehicle, and the detected towers are fixed, so that the pictures can be transmitted back after the unmanned aerial vehicle shoots the pictures according to a preset line patrol route and a preset shooting machine position, and the towers in the pictures and other structural components capable of supporting the conducting wires and the grounding wires are marked as large targets manually for identifying the positions and the ranges of the conducting wires and the grounding wires. The typical defect intelligent analysis module can perform intelligent reinforcement on data, realize conversion from non-mechanism data to structured data, mark a region to be analyzed on the picture, and facilitate the subsequent steps to extract related defect information on the picture. And the typical defect intelligent analysis module simultaneously filters the background of the broken strand of the ground wire and performs rough positioning.
Preferably, in step B), the method for identifying the region to be analyzed and establishing the analysis frame includes: B1) obtaining a picture with a region to be analyzed and a large target, manually marking the region to be analyzed and the large target region, and pre-building an analysis frame; B2) manually marking a large target area, and aligning the large target area marked in the image data obtained in the step A); B3) and establishing an analysis frame in the area where the pre-established analysis frame is located, wherein the area where the manually marked area to be analyzed is located is the identified area to be analyzed. The typical defect intelligent analysis module established by the optimal scheme can quickly align images and establish an analysis frame, which is beneficial to improving the efficiency of fault analysis.
Preferably, the Cascade R-CNN network constructed in the step C) comprises a plurality of cascaded R-CNN networks, and each cascaded R-CNN network is provided with a different IOU threshold. The IOU threshold value set by each cascaded R-CNN network is gradually increased, so that the input proposal is optimized, and the calculation precision and efficiency of the user-defined convolutional neural network are improved.
Preferably, in the Cascade R-CNN network constructed in the step C), a feature map of each layer of image is obtained by using FPN feature extraction, and the feature map is fused according to the information abstraction degree of the receptive field and the expression corresponding to the feature map to obtain the feature expression of the image. And the FPN layer fusing the feature maps in different stages improves the performance of small target detection.
Preferably, the cascaded R-CNN network constructed in step C) includes a GAN network, the GAN network generates a Super-resolved Feature for the small target, and in step D), the Super-resolved Feature of the small target is superimposed on a corresponding region of the image. The GAN is used for generating a Super-resolved Feature which is similar to the large target for the small target, and then the Super-resolved Feature is superposed on the original Feature map of the small target, so that the small target Feature expression is enhanced to improve the detection performance of the small target.
Preferably, the Cascade R-CNN network constructed in the step C) comprises a dynamic Anchor frame density adjusting module, wherein if the analysis frame in the image data is smaller than a preset threshold value, the Anchor frame density is increased, a plurality of Anchor frames are generated in the analysis frame, and the plurality of Anchor frames are distributed according to the density of the boundary line of the image in the analysis frame. In the data training, the number of Anchor boxes matched with the small targets is increased by adopting the characteristics based on the data, the training weight of the small targets is increased, and the inclination of the network to the large targets is reduced. And adding an Anchor strategy in charge of the small target to make the small target more fully learned during training. The density of the anchors in the graph is made approximately equal while sensitivity to small targets is enhanced using a more relaxed matching strategy for anchors of small targets. The method for generating the Anchor box comprises the following steps: extracting the boundary of an image in an analysis frame, establishing a binary copy of the image in the analysis frame according to boundary and non-boundary division, dividing the copy into square grids, counting the area of a boundary region in each square grid, taking the square grids with the area of the boundary region larger than a first set threshold value as central grids, wherein the number of the central grids is the number of Anchor frames, the square grids with the area of the boundary region around the central grids larger than a second set threshold value and the central grids form an initial range of the Anchor frame, and the minimum external matrix of the initial range of the Anchor frame is the range of the Anchor frame.
Preferably, in the step D), the unmanned aerial vehicle tour shot picture is imported into a typical defect intelligent analysis module, the typical defect intelligent analysis module analyzes an area to be analyzed, the unmanned aerial vehicle tour shot picture is cut according to the area to be analyzed, the cut picture is imported into the Cascade R-CNN network trained in the step C), and a fault detection result of each cut area is obtained respectively.
The substantial effects of the invention are as follows: the characteristic diagram of each layer of image is obtained by using FPN characteristic extraction, so that the performance of small target detection is improved; generating a Super-resolved Feature similar to the large target for the small target by using GAN, and then superposing the Super-resolved Feature on an original Feature map of the small target so as to enhance the Feature expression of the small target to improve the detection performance of the small target; in data training, the number of Anchor frames matched with the small targets is increased by adopting the characteristics based on data, the training weight of the small targets is increased, the sensitivity to the small targets is enhanced, and the accuracy of detecting the broken strand fault of the grounding wire of the power distribution line tower is improved.
Drawings
Fig. 1 is a flow chart of a tower ground wire strand breakage image detection method according to an embodiment.
FIG. 2 is a schematic diagram of an embodiment of a Cascade R-CNN network.
FIG. 3 is a diagram illustrating an embodiment of FPN feature extraction.
Fig. 4 is a schematic diagram of an image detection process of a broken strand of a tower ground wire according to an embodiment.
Detailed Description
The following provides a more detailed description of the present invention, with reference to the accompanying drawings.
The first embodiment is as follows:
a method for detecting broken strand images of a ground wire and a conducting wire of a power distribution line tower of an unmanned aerial vehicle image comprises the following steps as shown in figure 1: A) and establishing a picture data set of the broken strands of the conducting wires and the grounding wires, and marking and positioning the large target in the image. B) The picture data that will unmanned aerial vehicle shoot through wired follow unmanned aerial vehicle is transmitted to the PC end, gets into typical defect intelligent analysis module, provides the model and carries out intelligent recognition and treats the analysis region, establishes the analysis frame, and artifical mark defect obtains sample data. The method for identifying the area to be analyzed and establishing the analysis frame comprises the following steps: B1) obtaining a picture with a region to be analyzed and a large target, manually marking the region to be analyzed and the large target region, and pre-building an analysis frame; B2) manually marking a large target area, and aligning the large target area marked in the image data obtained in the step A); B3) and establishing an analysis frame in the area where the pre-established analysis frame is located, wherein the area where the manually marked area to be analyzed is located is the identified area to be analyzed. The typical defect intelligent analysis module established by the optimal scheme can quickly align images and establish an analysis frame, which is beneficial to improving the efficiency of fault analysis.
C) And constructing a Cascade R-CNN network, and training and testing the Cascade R-CNN network by using sample data. As shown in FIG. 2, the Cascade R-CNN network constructed in the step C) comprises a plurality of cascaded R-CNN networks, and each cascaded R-CNN network is provided with a different IOU threshold. The IOU threshold value set by each cascaded R-CNN network is gradually increased, so that the input proposal is optimized, and the calculation precision and efficiency of the user-defined convolutional neural network are improved. As shown in fig. 3, a feature map of each layer of image is obtained by using FPN feature extraction, and a feature representation of the image is obtained by fusing the feature maps according to the receptive field and the information abstraction level of the representation corresponding to the feature map. And the FPN layer fusing the feature maps in different stages improves the performance of small target detection. And C), constructing a GAN network, generating a Super-resolved Feature for the small target by the GAN network, generating a Super-resolved Feature which is very similar to the large target by using the GAN network, and then superposing the Super-resolved Feature on an original Feature map of the small target so as to enhance the Feature expression of the small target to improve the detection performance of the small target.
The Cascade R-CNN network constructed in the step C) comprises a dynamic Anchor frame density adjusting module, wherein if the analysis frame in the image data is smaller than a preset threshold value, the Anchor frame density is increased, a plurality of Anchor frames are generated in the analysis frame, and the plurality of Anchor frames are distributed according to the density of the boundary line of the image in the analysis frame. In the data training, the number of Anchor boxes matched with the small targets is increased by adopting the characteristics based on the data, the training weight of the small targets is increased, and the inclination of the network to the large targets is reduced. And adding an Anchor strategy in charge of the small target to make the small target more fully learned during training. The density of the anchors in the graph is made approximately equal while sensitivity to small targets is enhanced using a more relaxed matching strategy for anchors of small targets. The method for generating the Anchor box comprises the following steps: extracting the boundary of an image in an analysis frame, establishing a binary copy of the image in the analysis frame according to boundary and non-boundary division, dividing the copy into square grids, counting the area of a boundary region in each square grid, taking the square grids with the area of the boundary region larger than a first set threshold value as central grids, wherein the number of the central grids is the number of Anchor frames, the square grids with the area of the boundary region around the central grids larger than a second set threshold value and the central grids form an initial range of the Anchor frame, and the minimum external matrix of the initial range of the Anchor frame is the range of the Anchor frame.
D) And C), acquiring a tour shot picture of the unmanned aerial vehicle, identifying an area to be analyzed, importing the area into the CascadeR-CNN network trained in the step C), and acquiring a detection result. And D), importing the unmanned aerial vehicle tour shot picture into a typical defect intelligent analysis module, analyzing an area to be analyzed by the typical defect intelligent analysis module, superposing the Super-resolved Feature of the small target to the corresponding area of the image, cutting the unmanned aerial vehicle tour shot picture according to the area to be analyzed, importing the cut picture into the Cascade R-CNN network trained in the step C), and respectively obtaining a fault detection result of each cut area.
As shown in fig. 4, by training the large target detector based on the tower and the corresponding part, the large target of the tower and the corresponding part is found from the image and is located for providing the input data detected in step B). Aiming at the algorithm development of typical defects of a power grid, a target detection framework based on multi-feature fusion is used, and aiming at the problems of small object missing detection and inaccurate detection frame in the target detection algorithm, a multi-feature map with multi-layer feature fusion is adopted, and low-layer original image information and high-layer semantic information are combined. Meanwhile, the speed of target detection is improved, and the detection precision and speed are higher. And quickly positioning the tower and the large target of the corresponding part in the image, and providing a coarse positioning image for filtering the background for the broken strand of the ground wire in the second step. The defect identification rate is 70%, the omission ratio is less than 20%, the false alarm rate is less than 20%, and the defect identification efficiency is not lower than 200 pieces/minute. The pictures of the picture data set for establishing the broken strands of the conducting wires and the grounding wires are from visible light pictures shot by the unmanned aerial vehicle, and the detected towers are fixed, so that the pictures can be transmitted back after the unmanned aerial vehicle shoots the pictures according to a preset line patrol route and a preset shooting machine position, and the towers in the pictures and other structural components capable of supporting the conducting wires and the grounding wires are marked as large targets manually for identifying the positions and the ranges of the conducting wires and the grounding wires.
In the embodiment, the characteristic diagram of each layer of image is obtained by using FPN characteristic extraction, so that the performance of small target detection is improved; generating a Super-resolved Feature similar to the large target for the small target by using GAN, and then superposing the Super-resolved Feature on an original Feature map of the small target so as to enhance the Feature expression of the small target to improve the detection performance of the small target; in data training, the number of Anchor frames matched with the small targets is increased by adopting the characteristics based on data, the training weight of the small targets is increased, the sensitivity to the small targets is enhanced, and the accuracy of detecting the broken strand fault of the grounding wire of the power distribution line tower is improved.
The above-described embodiments are only preferred embodiments of the present invention, and are not intended to limit the present invention in any way, and other variations and modifications may be made without departing from the spirit of the invention as set forth in the claims.
Claims (7)
1. A method for detecting broken strand images of a ground wire of a tower of a distribution line of an unmanned aerial vehicle image is characterized in that,
the method comprises the following steps:
A) establishing a picture data set of broken strands of the conducting wires and the grounding wires, and marking and positioning a large target in the image;
B) transmitting picture data shot by the unmanned aerial vehicle to a PC (personal computer) end from the unmanned aerial vehicle through a wire, entering a typical defect intelligent analysis module, providing a model for intelligently identifying an area to be analyzed, establishing an analysis frame, manually marking defects, and obtaining sample data;
C) constructing a Cascade R-CNN network, and training and testing the Cascade R-CNN network by using sample data;
D) and C), acquiring a tour shot picture of the unmanned aerial vehicle, identifying an area to be analyzed, importing the area into the Cascade R-CNN network trained in the step C), and acquiring a detection result.
2. The method of claim 1, wherein the image of the broken strands of the ground wires of the tower and the tower of the distribution line for the unmanned aerial vehicle images is obtained,
in the step B), the area to be analyzed is identified, and the method for establishing the analysis frame comprises the following steps:
B1) obtaining a picture with a region to be analyzed and a large target, manually marking the region to be analyzed and the large target region, and pre-building an analysis frame;
B2) manually marking a large target area, and aligning the large target area marked in the image data obtained in the step A);
B3) and establishing an analysis frame in the area where the pre-established analysis frame is located, wherein the area where the manually marked area to be analyzed is located is the identified area to be analyzed.
3. The method for detecting the broken strand image of the ground wires and the poles of the distribution line tower of the unmanned aerial vehicle image according to claim 1 or 2,
the Cascade R-CNN network constructed in the step C) comprises a plurality of cascaded R-CNN networks, and each cascaded R-CNN network is provided with a different IOU threshold.
4. The method for detecting the broken strand image of the ground wires and the poles of the distribution line tower of the unmanned aerial vehicle image according to claim 1 or 2,
in the Cascade R-CNN network constructed in the step C), the feature map of each layer of image is obtained by using FPN feature extraction, and the feature map is fused according to the information abstraction degree of the receptive field and the expression corresponding to the feature map to obtain the feature expression of the image.
5. The method of claim 3, wherein the image of the broken strands of the ground wires of the tower and the tower of the distribution line for the unmanned aerial vehicle image is obtained by a method,
the cascaded R-CNN network constructed in the step C) comprises a GAN network, the GAN network generates a Super-resolved Feature for the small target, and in the step D), the Super-resolved Feature of the small target is superposed to a corresponding area of the image.
6. The method for detecting the broken strand image of the ground wires and the poles of the distribution line tower of the unmanned aerial vehicle image according to claim 1 or 2,
the Cascade R-CNN network constructed in the step C) comprises a dynamic Anchor frame density adjusting module, wherein if the analysis frame in the image data is smaller than a preset threshold value, the Anchor frame density is increased, a plurality of Anchor frames are generated in the analysis frame, and the plurality of Anchor frames are distributed according to the density of the boundary line of the image in the analysis frame.
7. The method for detecting the broken strand image of the ground wires and the poles of the distribution line tower of the unmanned aerial vehicle image according to claim 1 or 2,
and D), importing the unmanned aerial vehicle tour shot picture into a typical defect intelligent analysis module, analyzing the area to be analyzed by the typical defect intelligent analysis module, cutting the unmanned aerial vehicle tour shot picture according to the area to be analyzed, importing the cut picture into the Cascade R-CNN network trained in the step C), and respectively obtaining the fault detection result of each cut area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010490245.2A CN111832398B (en) | 2020-06-02 | 2020-06-02 | Unmanned aerial vehicle image distribution line pole tower ground wire broken strand image detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010490245.2A CN111832398B (en) | 2020-06-02 | 2020-06-02 | Unmanned aerial vehicle image distribution line pole tower ground wire broken strand image detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111832398A true CN111832398A (en) | 2020-10-27 |
| CN111832398B CN111832398B (en) | 2022-07-01 |
Family
ID=72897983
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010490245.2A Active CN111832398B (en) | 2020-06-02 | 2020-06-02 | Unmanned aerial vehicle image distribution line pole tower ground wire broken strand image detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111832398B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112308135A (en) * | 2020-10-29 | 2021-02-02 | 哈尔滨市科佳通用机电股份有限公司 | Railway motor car sand spreading pipe loosening fault detection method based on deep learning |
| CN112508905A (en) * | 2020-12-01 | 2021-03-16 | 广东电网有限责任公司 | Hardware rust image detection method and computer readable storage medium |
| CN112557815A (en) * | 2020-11-27 | 2021-03-26 | 广东电网有限责任公司肇庆供电局 | Distribution network line tree obstacle identification and fault positioning method based on fixed and movable inspection images |
| CN112950620A (en) * | 2021-03-26 | 2021-06-11 | 国网湖北省电力公司检修公司 | Power transmission line damper deformation defect detection method based on cascade R-CNN algorithm |
| CN113052832A (en) * | 2021-04-19 | 2021-06-29 | 广东电网有限责任公司肇庆供电局 | Hardware fitting corrosion image detection method and device for power transmission line |
| CN113850827A (en) * | 2021-11-29 | 2021-12-28 | 广东电网有限责任公司肇庆供电局 | Method for detecting and processing broken strand image caused by abrasion of ground wire and horizontal iron |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2812890A1 (en) * | 2004-05-04 | 2005-11-04 | Intelliview Technologies Inc. | Mesh based frame processing and applications |
| CN109342466A (en) * | 2018-08-29 | 2019-02-15 | 乐山-菲尼克斯半导体有限公司 | A method of detecting discrete device bonding wire bank defect automatically using X-ray |
| US20190130191A1 (en) * | 2017-10-30 | 2019-05-02 | Qualcomm Incorporated | Bounding box smoothing for object tracking in a video analytics system |
| CN111046950A (en) * | 2019-12-11 | 2020-04-21 | 北京迈格威科技有限公司 | Image processing method and device, storage medium and electronic device |
-
2020
- 2020-06-02 CN CN202010490245.2A patent/CN111832398B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2812890A1 (en) * | 2004-05-04 | 2005-11-04 | Intelliview Technologies Inc. | Mesh based frame processing and applications |
| US20190130191A1 (en) * | 2017-10-30 | 2019-05-02 | Qualcomm Incorporated | Bounding box smoothing for object tracking in a video analytics system |
| CN109342466A (en) * | 2018-08-29 | 2019-02-15 | 乐山-菲尼克斯半导体有限公司 | A method of detecting discrete device bonding wire bank defect automatically using X-ray |
| CN111046950A (en) * | 2019-12-11 | 2020-04-21 | 北京迈格威科技有限公司 | Image processing method and device, storage medium and electronic device |
Non-Patent Citations (2)
| Title |
|---|
| JIANAN LI 等: "Perceptual Generative Adversarial Networks for Small Object Detection", 《2017 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 * |
| 石书山: "基于深度学习技术的输电线路缺陷智能分析系统研究与应用", 《通信电源技术》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112308135A (en) * | 2020-10-29 | 2021-02-02 | 哈尔滨市科佳通用机电股份有限公司 | Railway motor car sand spreading pipe loosening fault detection method based on deep learning |
| CN112557815A (en) * | 2020-11-27 | 2021-03-26 | 广东电网有限责任公司肇庆供电局 | Distribution network line tree obstacle identification and fault positioning method based on fixed and movable inspection images |
| CN112557815B (en) * | 2020-11-27 | 2022-05-20 | 广东电网有限责任公司肇庆供电局 | Fixed and movable inspection image-based distribution network line tree obstacle identification and fault positioning method |
| CN112508905A (en) * | 2020-12-01 | 2021-03-16 | 广东电网有限责任公司 | Hardware rust image detection method and computer readable storage medium |
| CN112950620A (en) * | 2021-03-26 | 2021-06-11 | 国网湖北省电力公司检修公司 | Power transmission line damper deformation defect detection method based on cascade R-CNN algorithm |
| CN113052832A (en) * | 2021-04-19 | 2021-06-29 | 广东电网有限责任公司肇庆供电局 | Hardware fitting corrosion image detection method and device for power transmission line |
| CN113850827A (en) * | 2021-11-29 | 2021-12-28 | 广东电网有限责任公司肇庆供电局 | Method for detecting and processing broken strand image caused by abrasion of ground wire and horizontal iron |
| CN113850827B (en) * | 2021-11-29 | 2022-03-22 | 广东电网有限责任公司肇庆供电局 | Method for detecting and processing broken strand image caused by abrasion of ground wire and horizontal iron |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111832398B (en) | 2022-07-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111832398B (en) | Unmanned aerial vehicle image distribution line pole tower ground wire broken strand image detection method | |
| CN109344753A (en) | A kind of tiny fitting recognition methods of Aerial Images transmission line of electricity based on deep learning | |
| CN109712127B (en) | Power transmission line fault detection method for machine inspection video stream | |
| CN109544501A (en) | A kind of transmission facility defect inspection method based on unmanned plane multi-source image characteristic matching | |
| CN109300118B (en) | High-voltage power line unmanned aerial vehicle inspection method based on RGB image | |
| CN107194396A (en) | Method for early warning is recognized based on the specific architecture against regulations in land resources video monitoring system | |
| CN113887412A (en) | Detection method, detection terminal, monitoring system and storage medium for pollution emission | |
| CN114743119A (en) | High-speed rail contact net dropper nut defect detection method based on unmanned aerial vehicle | |
| CN111145222A (en) | Fire detection method combining smoke movement trend and textural features | |
| CN113589837A (en) | Electric power real-time inspection method based on edge cloud | |
| CN113160184A (en) | Unmanned aerial vehicle intelligent inspection cable surface defect detection method based on deep learning | |
| CN115546664A (en) | Cascaded network-based insulator self-explosion detection method and system | |
| CN116681979A (en) | Power equipment target detection method under complex environment | |
| CN115526852A (en) | Molten pool and splash monitoring method in selective laser melting process based on target detection and application | |
| CN113239838B (en) | Unmanned aerial vehicle intelligent inspection identification method for power transmission tower | |
| CN111915565B (en) | Method for analyzing cracks of porcelain insulator of power transmission and transformation line in real time based on YOLACT algorithm | |
| CN115912183B (en) | Ecological measure inspection method and system for high-voltage transmission line and readable storage medium | |
| CN115984672B (en) | Detection method and device for small target in high-definition image based on deep learning | |
| CN116994161A (en) | Insulator defect detection method based on improved YOLOv5 | |
| CN111931577A (en) | Intelligent inspection method for specific foreign matters of power grid line | |
| CN116205905A (en) | Power distribution network construction safety and quality image detection method and system based on mobile terminal | |
| CN115100546A (en) | Mobile-based small target defect identification method and system for power equipment | |
| CN116739963A (en) | Power grid equipment defect detection method based on multi-level multi-scale feature fusion | |
| CN113504240B (en) | Power equipment fault detection method based on layered coding and visual perception | |
| CN112686105B (en) | Fog concentration grade identification method based on video image multi-feature fusion |
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 |