CN112541478A - Insulator string stain detection method and system based on binocular camera - Google Patents

Insulator string stain detection method and system based on binocular camera Download PDF

Info

Publication number
CN112541478A
CN112541478A CN202011558906.7A CN202011558906A CN112541478A CN 112541478 A CN112541478 A CN 112541478A CN 202011558906 A CN202011558906 A CN 202011558906A CN 112541478 A CN112541478 A CN 112541478A
Authority
CN
China
Prior art keywords
insulator
insulator string
infrared
image
determining
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.)
Pending
Application number
CN202011558906.7A
Other languages
Chinese (zh)
Inventor
张凯
魏豪
杨爽
郑磊
曹源
张丁文
樊家树
姜姝宇
孙晓楠
朴艺芳
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.)
Information and Telecommunication Branch of State Grid Jilin Electric Power Co Ltd
Original Assignee
Information and Telecommunication Branch of State Grid Jilin Electric Power 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 Information and Telecommunication Branch of State Grid Jilin Electric Power Co Ltd filed Critical Information and Telecommunication Branch of State Grid Jilin Electric Power Co Ltd
Priority to CN202011558906.7A priority Critical patent/CN112541478A/en
Publication of CN112541478A publication Critical patent/CN112541478A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/50Context or environment of the image
    • G06V20/52Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00Machine learning
    • G06N20/10Machine learning using kernel methods, e.g. support vector machines [SVM]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00Image enhancement or restoration
    • G06T5/70Denoising; Smoothing
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/194Segmentation; Edge detection involving foreground-background segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/20Image preprocessing
    • G06V10/26Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G06V10/267Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/56Extraction of image or video features relating to colour
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10048Infrared image
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20024Filtering details
    • G06T2207/20032Median filtering
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20084Artificial neural networks [ANN]

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Software Systems (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)

Abstract

The invention relates to a binocular camera-based insulator string fouling detection method and system. The method comprises the following steps: calibrating an infrared camera and a high-definition camera; correcting the insulator infrared image and the insulator color image, and determining the corrected insulator infrared image and the corrected insulator color image; determining the matching relation between the corrected insulator infrared image and the corresponding pixel points of the corrected insulator color image; respectively preprocessing the corrected insulator infrared image and the corrected insulator color image, and extracting an insulator string region image in the preprocessed insulator infrared image by using a YOLOv4-tiny network; determining an insulator string segmentation image on the preprocessed insulator color image according to the matching relation of the corresponding pixel points; and determining the pollution grade of the insulator string according to the insulator string segmentation image. The invention overcomes the problems of a single camera and improves the detection accuracy of insulator fault detection.

Description

Insulator string stain detection method and system based on binocular camera
Technical Field
The invention relates to the field of insulator string fouling detection, in particular to a binocular camera-based insulator string fouling detection method and system.
Background
Insulators are very important power system devices, and the operating condition of the insulators directly affects the stability of the power system operation. Generally, insulator equipment is exposed to outdoor environment for a long time and always works under extreme conditions such as a strong electric field and strong mechanical load, and the problem of pollution caused by coverage of foreign matters such as bird droppings, silt and leaves or internal defects of the insulator equipment is solved. If the problems can not be found and solved in time, the stable operation of the power system can be influenced, and faults such as short circuit, open circuit and the like of the power system can be caused in serious cases, so that great economic loss is caused. For this reason, national grid companies have strictly performed periodic inspection tasks on power systems, and have strengthened the inspection work on the above hidden dangers.
Currently, common insulator detection means include five types of methods, such as manual detection based, ultraviolet light detection based, infrared light detection based, and computer vision detection based. Based on the fact that manual detection is the most original and simple detection method, a power grid inspection worker can directly observe the surface state of the insulator from the ground through a telescope or a special ultrasonic instrument, or measure the discharge distance of two ends of the insulator through a small ball operated manually. The method based on manual detection can well check the hidden dangers of obvious falling, large-area dirt covering and the like of the insulator, but due to the subjective factors of people, the problems of incomplete observation and the like exist, and the missing detection risk exists for the damage conditions of the insulator surface which is slightly damaged or faces away from inspection personnel. The ultraviolet-based light detection is that ultraviolet radiation is generated in partial discharge of the insulator, so that the discharge conditions of the insulators with different voltage levels are monitored, and whether the insulator has a discharge fault or not is judged. The ultraviolet-based light detection has the advantages of all weather and high accuracy, but the equipment cost is high, and the detection can be carried out only when partial discharge exists in the insulator. The infrared light detection method is used for judging whether the insulator has a contamination fault according to the change degree of the surface temperature of the insulator. The insulator with good performance has good insulating performance and does not have the problem of electric leakage, but if the insulator is stained, the surface of the insulator possibly has the problem of electric leakage, the surface of the insulator can be abnormally heated, infrared radiation energy can be captured through infrared imaging, and the fault detection of the insulator is realized. The infrared-based optical detection method belongs to non-contact detection and has high reliability, but the method is easily influenced by the atmosphere, so that the detection of the fault of the insulator is inaccurate. The method based on computer vision mainly collects the visual image of the insulator through visible light imaging (different from ultraviolet light detection and infrared light detection), and further realizes the fouling detection of the surface of the insulator by adopting methods such as shallow machine learning, deep learning and the like. The method based on computer vision has the characteristics of low cost, simple and convenient operation, high reliability and the like, and is a main research method for the current insulator fouling detection. There are two types of common vision-based insulator detection: shallow machine learning detection and deep learning detection. The detection method based on shallow machine learning firstly needs to extract visual low-level features from the insulator stained image, wherein the existing features comprise edge features, color features, texture features, shape features, scale invariant features, wavelet features and the like; secondly, training the features by using machine learning (SVM, Boosting and the like) to obtain a classifier; and finally, detecting the acquired image to be detected by using the obtained classifier. The method has higher requirements on the quality of the collected image, is limited by the quality of manual setting characteristic selection, and is not high in insulator detection precision in practical application scenes. The method based on deep learning does not need to calculate the bottom layer characteristics of the insulator in advance, and the characteristic extraction and the characteristic detection are automatically completed by a deep learning framework, so that the problem of manually selecting the characteristics is effectively solved. Common deep learning frameworks are YOLO, SSD, CNN, Darknet, etc. The method has high detection precision and good adaptability to the environment, but due to the complex deep learning framework, the requirement on an embedded hardware platform is high, the insulator fouling condition differentiation is large, a comprehensive training sample is difficult to construct, and the final effect of the deep learning algorithm on insulator fouling detection is influenced.
Disclosure of Invention
The invention aims to provide a method and a system for detecting fouling of an insulator string based on a binocular camera, which aim to solve the problem that the fouling degree of the surface of the insulator string cannot be measured by using a detection method based on an infrared single camera; the detection method based on the visible light camera has the problems that due to the complex acquisition background, the accurate positioning of the insulator is difficult to a certain degree, and the detection accuracy of the fault detection of the insulator is low.
In order to achieve the purpose, the invention provides the following scheme:
a binocular camera-based insulator string fouling detection method comprises the following steps: the device comprises an infrared camera and a high-definition camera; the infrared camera is used for identifying the temperature change of the surface of the insulator string; the high-definition camera is used for detecting dirty substances on the surface of the insulator;
calibrating the infrared camera and the high-definition camera, and determining infrared camera parameters and high-definition camera parameters; the infrared camera parameters comprise an infrared internal parameter, an infrared external parameter and an infrared distortion coefficient of the infrared camera; the high-definition camera parameters comprise high-definition internal parameters, high-definition external parameters and high-definition distortion coefficients of the high-definition camera;
correcting the insulator infrared image collected by the infrared camera by using the infrared camera parameters, correcting the insulator color image collected by the high-definition camera by using the high-definition camera parameters, and determining the corrected insulator infrared image and the corrected insulator color image; determining the matching relation between the corrected insulator infrared image and corresponding pixel points of the corrected insulator color image;
respectively preprocessing the corrected insulator infrared image and the corrected insulator color image, and determining a preprocessed insulator infrared image and a preprocessed insulator color image;
extracting an insulator string region image in the preprocessed insulator infrared image by using a YOLOv4-tiny network;
according to the matching relation of corresponding pixel points, mapping the insulator string region images to the preprocessed insulator color image one by one, and determining an insulator string segmentation image on the preprocessed insulator color image;
and determining the pollution grade of the insulator string according to the insulator string segmentation image.
Optionally, the determining the matching relationship between the corrected insulator infrared image and the corresponding pixel points of the corrected insulator color image specifically includes:
and determining the matching relation of corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image by using a polar line constraint method.
Optionally, the extracting, by using a YOLOv4-tiny network, an insulator string region image in the preprocessed insulator infrared image specifically includes:
classifying all insulator infrared images, and taking the infrared insulator strings on all the insulator infrared images as an infrared insulator string data set; the infrared insulator string data set comprises a normal insulator string, a low-value insulator string and a zero-value insulator string;
dividing the infrared insulator string data set to determine a training set and a testing set;
training the YOLOv4-tiny network according to the training set to generate a classification model; the network layer of the YOLOv4-tiny network is deleted from 162 layers to 38 layers, and different scale features of the insulator infrared image are automatically extracted by adopting a feature pyramid network;
and extracting an insulator string region image in the preprocessed insulator infrared image according to the classification model.
Optionally, the mapping the insulator string region images to the preprocessed insulator color image one by one according to the matching relationship of the corresponding pixel points, and determining the insulator string segmentation images on the preprocessed insulator color image specifically include:
carrying out binarization processing on the insulator string region image by using a maximum between-class variance method, segmenting an insulator string region and a background region in the insulator string region image, and determining a binarized insulator string region image;
and mapping the insulator string region image to the preprocessed insulator color image according to the matching relation of the corresponding pixel points, and determining an insulator string segmentation image on the preprocessed insulator color image.
Optionally, the determining the pollution level of the insulator string according to the insulator string segmentation image specifically includes:
acquiring a plurality of insulator string sample pictures with different fouling grades;
performing median filtering denoising and histogram equalization processing on each insulator string sample picture to determine a processed insulator string sample picture;
calculating a histogram feature vector of the processed insulator string sample picture;
calculating histogram feature vectors of the insulator string segmentation images;
calculating Euler distances between histogram feature vectors of the insulator string segmentation images and histogram feature vectors of all processed insulator string sample pictures;
sequencing all the Euler distances in a sequence from near to far, and determining histogram feature vectors of the processed insulator string sample pictures corresponding to the Euler distances within Euler distance thresholds;
and determining the grade of each corresponding processed insulator string sample picture feature vector, and taking the grade with the largest number of grades as the pollution grade of the insulator string.
A stained detecting system of insulator chain based on binocular camera includes: the device comprises an infrared camera and a high-definition camera; the infrared camera is used for identifying the temperature change of the surface of the insulator string; the high-definition camera is used for detecting dirty substances on the surface of the insulator;
the calibration module is used for calibrating the infrared camera and the high-definition camera and determining parameters of the infrared camera and the high-definition camera; the infrared camera parameters comprise an infrared internal parameter, an infrared external parameter and an infrared distortion coefficient of the infrared camera; the high-definition camera parameters comprise high-definition internal parameters, high-definition external parameters and high-definition distortion coefficients of the high-definition camera;
the correction module is used for correcting the insulator infrared image acquired by the infrared camera by using the infrared camera parameters, correcting the insulator color image acquired by the high-definition camera by using the high-definition camera parameters, and determining the corrected insulator infrared image and the corrected insulator color image; determining the matching relation between the corrected insulator infrared image and corresponding pixel points of the corrected insulator color image;
the preprocessing module is used for respectively preprocessing the corrected insulator infrared image and the corrected insulator color image and determining a preprocessed insulator infrared image and a preprocessed insulator color image;
the insulator string region image extraction module is used for extracting an insulator string region image in the preprocessed insulator infrared image by using a YOLOv4-tiny network;
the insulator string segmentation image determining module is used for mapping the insulator string region images to the preprocessed insulator color images one by one according to the matching relation of the corresponding pixel points, and determining the insulator string segmentation images on the preprocessed insulator color images;
and the insulator string pollution grade determining module is used for determining the pollution grade of the insulator string according to the insulator string segmentation image.
Optionally, the determining, in the correction module, a matching relationship between corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image specifically includes:
and the matching relation determining unit is used for determining the matching relation of corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image by using an extreme line constraint method.
Optionally, the insulator string region image extraction module specifically includes:
the infrared insulator string data set determining unit is used for classifying all insulator infrared images and taking the infrared insulator strings on all the insulator infrared images as an infrared insulator string data set; the infrared insulator string data set comprises a normal insulator string, a low-value insulator string and a zero-value insulator string;
the dividing unit is used for dividing the infrared insulator string data set to determine a training set and a test set;
the classification model generation unit is used for training the YOLOv4-tiny network according to the training set to generate a classification model; the network layer of the YOLOv4-tiny network is deleted from 162 layers to 38 layers, and different scale features of the insulator infrared image are automatically extracted by adopting a feature pyramid network;
and the insulator string region image determining unit is used for extracting the insulator string region image in the preprocessed insulator infrared image according to the classification model.
Optionally, the insulator string segmentation image determining module specifically includes:
the binarization processing unit is used for carrying out binarization processing on the insulator string region image by using a maximum inter-class variance method, segmenting an insulator string region and a background region in the insulator string region image and determining a binarized insulator string region image;
and the insulator string segmentation image determining unit is used for mapping the insulator string region image to the preprocessed insulator color image according to the matching relation of the corresponding pixel points, and determining the insulator string segmentation image on the preprocessed insulator color image.
Optionally, the module for determining the pollution level of the insulator string specifically includes:
the insulator string sample picture acquisition unit is used for acquiring a plurality of insulator string sample pictures with different fouling grades;
the processed insulator string sample picture determining unit is used for performing median filtering denoising and histogram equalization processing on each insulator string sample picture to determine a processed insulator string sample picture;
the first histogram feature vector calculation unit is used for calculating the histogram feature vector of the processed insulator string sample picture;
the second histogram feature vector calculation unit is used for calculating the histogram feature vectors of the insulator string segmentation images;
the Euler distance calculating unit is used for calculating Euler distances between the histogram feature vectors of the insulator string segmentation images and the histogram feature vectors of all the processed insulator string sample pictures;
the pixel point determining unit is used for sequencing all the Euler distances from near to far and determining the histogram feature vector of the processed insulator string sample picture corresponding to the Euler distances within the Euler distance threshold;
and the insulator string pollution grade determining unit is used for determining the grade of each corresponding processed insulator string sample picture feature vector, and taking the grade with the largest number of grades as the pollution grade of the insulator string.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: the invention discloses a binocular camera-based insulator string contamination detection method and system, wherein the dual cameras are used for realizing the contamination detection of an insulator string, the dual cameras comprise an infrared imaging camera and a high-definition visible light camera, the two cameras are used for carrying out fusion contamination detection, the advantages of the different cameras are fully exerted, the problems existing in a single camera are solved, firstly, the two cameras are fixed, then, the three-dimensional matching of the two cameras is realized by calibrating internal and external parameters and distortion coefficients, the detection of an insulator is realized on the infrared camera by using a YOLOv4-tiny algorithm, and whether the insulator is an abnormal insulator or not, namely a zero-value insulator and a low-value insulator is distinguished; then, binaryzation processing is carried out on the infrared detection insulator region, so that the background region except the non-insulator is removed; and correspondingly searching the same area on the color image according to the target insulator pixels with the removed background and the matching result of the two cameras, and realizing the division of the surface fouling grade on the area by using a K-nearest neighbor algorithm, thereby improving the detection accuracy of the insulator fault detection and being capable of operating on an embedded platform.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
Fig. 1 is a flow chart of a binocular camera-based insulator string contamination detection method provided by the invention;
FIG. 2 is a block diagram of an insulator string contamination detection system provided in the present invention;
FIG. 3 is a flow chart of the insulator string region extraction process provided by the present invention;
fig. 4 is a structural diagram of an insulator string contamination detection system based on a binocular camera provided by the invention.
Detailed Description
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 only a part of the embodiments of the present invention, and not all of the embodiments. 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.
The invention aims to provide a binocular camera-based insulator string contamination detection method and system, which overcome the problems of a single camera and improve the detection accuracy of insulator fault detection.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a flowchart of a method for detecting fouling of an insulator string based on a binocular camera, and fig. 2 is a frame diagram of a method for detecting fouling of an insulator string, as shown in fig. 1-2, a method for detecting fouling of an insulator string based on a binocular camera includes: the device comprises an infrared camera and a high-definition camera; the infrared camera is used for identifying the temperature change of the surface of the insulator string; the high-definition camera is used for detecting dirty substances on the surface of the insulator.
Step 101: calibrating the infrared camera and the high-definition camera, and determining infrared camera parameters and high-definition camera parameters; the infrared camera parameters comprise an infrared internal parameter, an infrared external parameter and an infrared distortion coefficient of the infrared camera; the high-definition camera parameters comprise high-definition internal parameters, high-definition external parameters and high-definition distortion coefficients of the high-definition camera.
Binocular inside/outside parameter calibration
The binocular camera calibration comprises two parts, namely internal parameter calibration and external parameter calibration.
The internal parameter calibration can obtain an internal parameter matrix of the camera as follows:
Figure BDA0002859668590000081
wherein fx and fy are camera focal lengths, xc and yc are camera imaging principal points, and s is a camera tilt parameter.
External parameter calibration can obtain the relative position between the two cameras, and the translation vector T and the rotation matrix R of the high-definition camera relative to the infrared camera are calculated by taking the infrared camera as the origin of world coordinates.
Because the two cameras have different imaging principles, in order to realize the joint calibration of the binocular cameras, a standard checkerboard calibration plate containing controllable heating plates inside needs to be selected and respectively positioned at the front ends of the two cameras, the angles and the positions of the calibration plate are continuously changed, checkerboard images are synchronously obtained, and then a current common camera calibration method is utilized for calibration, so that internal/external parameters are obtained.
Step 102: correcting the insulator infrared image collected by the infrared camera by using the infrared camera parameters, correcting the insulator color image collected by the high-definition camera by using the high-definition camera parameters, and determining the corrected insulator infrared image and the corrected insulator color image; and determining the matching relation of the corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image.
The determining the matching relationship between the corrected insulator infrared image and the corresponding pixel points of the corrected insulator color image specifically includes: and determining the matching relation of corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image by using a polar line constraint method.
Infrared/color image correction
In the calibration process, besides the internal/external parameters, distortion coefficients of two cameras including a radial distortion coefficient and a tangential distortion coefficient can be obtained, distortion correction is carried out on camera imaging by using the internal/external parameters and the distortion coefficients obtained by calibration, parallax of a target point formed on imaging of the left camera and the right camera is calculated on a corrected result image, and matching of corresponding pixel points of imaging of the two cameras is achieved by using an epipolar constraint method.
Step 103: and respectively preprocessing the corrected insulator infrared image and the corrected insulator color image, and determining the preprocessed insulator infrared image and the preprocessed insulator color image.
Image preprocessing:
for infrared images, due to the imaging mode, the imaging results have the problems of poor contrast, edge blurring and low signal-to-noise ratio, which greatly affects the post-processing, so the pre-processing is necessary. And carrying out noise reduction processing on the infrared image by adopting a median filtering method.
In the actual detection of the contamination of the outdoor insulator, the gray level of an image is concentrated (too bright or too dark) due to environmental reasons, and the contrast is low. In order to solve the problem, a histogram equalization mode is used for adjusting an image imaging gray level space, and the visual imaging effect is improved.
Step 104: and extracting an insulator string region image in the preprocessed insulator infrared image by using a YOLOv4-tiny network.
The step 104 specifically includes: classifying all insulator infrared images, and taking the infrared insulator strings on all the insulator infrared images as an infrared insulator string data set; the infrared insulator string data set comprises a normal insulator string, a low-value insulator string and a zero-value insulator string; dividing the infrared insulator string data set to determine a training set and a testing set; training the YOLOv4-tiny network according to the training set to generate a classification model; the network layer of the YOLOv4-tiny network is deleted from 162 layers to 38 layers, and different scale features of the insulator infrared image are automatically extracted by adopting a feature pyramid network; and extracting an insulator string region image in the preprocessed insulator infrared image according to the classification model.
Yolov4-tiny insulator detection
The insulator target detection method adopts the YOLOv4-tiny detection operator to realize the insulator target detection with light weight, and compared with the YOLOv3-tiny detection operator, the detection performance is greatly improved. The YOLOv4-tiny operator is a simplified version of YOLOv4, the network layer is reduced to 38 layers from 162 layers of YOLOv4, different scale features of insulator images are automatically obtained by adopting a feature pyramid network, the detection speed of the insulator sub-object is improved, and the method is suitable for deployment on an embedded platform. Table 1 is a schematic table of the entire detection network structure, and the entire detection network structure is shown in table 1.
TABLE 1
Figure BDA0002859668590000101
Figure BDA0002859668590000111
Figure BDA0002859668590000121
According to the common knowledge of the power system, the voltage distribution of a normal insulator string is relatively uniform, and the voltage distribution of a defective insulator string (zero value insulator and low value insulator) presents non-uniform characteristics. The uneven voltage causes uneven heating power of each insulator, the temperature distribution on the surface of the insulator can be changed correspondingly, the surface temperature of the low-value insulator is higher than that of the normal insulator, and the surface temperature of the zero-value insulator is lower than that of the normal insulator. The temperature change of the insulator string can be detected through the infrared thermal camera, and whether the insulator has a problem or not is judged through the temperature change degree. The specific method is that manual labeling is carried out on an infrared insulator string data set acquired on site, and the data set is divided into three types: normal insulator string, low value insulator string and zero value insulator string.
After the data set classification is completed, the data set needs to be divided, one part is used as a training set, and the other part is used as a testing set. The training set is used for training the YOLOv4-tiny network to obtain a classification model, and the testing set is used for evaluating the obtained classification model.
In order to train the Yolov4-tiny network, the classified insulator string data set is labeled manually, the position and the size of the insulator string in the image are labeled, and the image size is scaled to the input size 416 × 416pixel of the Yolov4-tiny network.
And training the YOLOv4-tiny network to generate a classification model file. For the insulator image to be detected, the classification model can detect the type of the insulator and output the position, the size and the category (normal, low value and zero value) of the insulator in a result picture.
Step 105: and mapping the insulator string region images to the preprocessed insulator color images one by one according to the matching relation of the corresponding pixel points, and determining the insulator string segmentation images on the preprocessed insulator color images.
The step 105 specifically includes: carrying out binarization processing on the insulator string region image by using a maximum between-class variance method, segmenting an insulator string region and a background region in the insulator string region image, and determining a binarized insulator string region image; and mapping the insulator string region image to the preprocessed insulator color image according to the matching relation of the corresponding pixel points, and determining an insulator string segmentation image on the preprocessed insulator color image.
Insulator string region extraction:
for an insulator string region obtained by infrared image detection, the insulator string region and a background region need to be segmented so as to judge the contamination level of the insulator string, and a specific flowchart is shown in fig. 3.
Carrying out insulator string image area binarization based on OSTU:
for an insulator string image region obtained by detection of a YOLOv4-tiny operator, segmentation of the insulator string region and a background region is realized by using a maximum inter-class variance (OSTU) method, the method is a threshold-based adaptive image segmentation method, and a binarization segmentation threshold is determined by using the maximum inter-class variance of gray level change between a target and a background. The specific method comprises the following steps:
firstly, let m be the number of pixels with a gray scale value of l in the image area of the insulator stringlWherein, L is the gray scale of the insulator string image, and L belongs to [0, L-1 ]]。
Then, the probability of occurrence of each gray level is calculated according to the following formula:
Figure BDA0002859668590000131
where M is the total number of pixels.
Secondly, for a given segmentation threshold Th, the probability of the occurrence of the gray value of two segmented sub-regions (w0 and w1) is calculated separately
Figure BDA0002859668590000132
Thirdly, by calculating the mean value μ of the pixels within the two regionsw0、μw1And the mean value mu of the internal pixels of the whole area, and calculating the inter-class variance by using the following formula:
σ2=pw0w0-μ)2+pw1w1-μ)2
finally, take σ2The maximum Th is used as a threshold for region division.
Color image insulator string extraction
In the infrared/color image correction process, the pixel matching relationship between the infrared image and the corresponding color image is obtained, and the binarization result of the insulator string region obtained by the infrared image and the color image are mapped one by one, so that an insulator string segmentation image on the color image can be obtained.
Step 106: and determining the pollution grade of the insulator string according to the insulator string segmentation image.
The step 106 specifically includes: acquiring a plurality of insulator string sample pictures with different fouling grades; performing median filtering denoising and histogram equalization processing on each insulator string sample picture to determine a processed insulator string sample picture; calculating a histogram feature vector of the processed insulator string sample picture; calculating histogram feature vectors of the insulator string segmentation images; calculating Euler distances between histogram feature vectors of the insulator string segmentation images and histogram feature vectors of all processed insulator string sample pictures; sequencing all the Euler distances in a sequence from near to far, and determining histogram feature vectors of the processed insulator string sample pictures corresponding to the Euler distances within Euler distance thresholds; and determining the grade of each corresponding processed insulator string sample picture feature vector, and taking the grade with the largest number of grades as the pollution grade of the insulator string.
Insulator string contamination level detection
In order to realize the detection of the pollution grades of the insulator string, five grades are defined, namely grade 0, grade 1, grade 2, grade 3 and grade 4, and the higher the grade is, the larger the pollution degree is. Grade 0 was no insult, grade 1 was light insult, grade 2 was medium insult, and grade 3 was heavy insult. The determination of the fouling grade is realized by adopting a K-nearest neighbor algorithm, and the specific realization process is as follows:
firstly, insulator string sample pictures with different fouling grades are manually selected, the fouling grades are marked (the grade is 0 … 4), the number of samples in each grade is equal, the number of samples is recommended to be 30, and the overlarge calculation complexity is not high.
Secondly, each selected sample picture is preprocessed, and median filtering denoising and histogram equalization are included.
Again, histogram feature vectors are computed for the processed image, with each BIN in the histogram being 8 pixels in size, thus being divided from 0 to 255 into [0,7 ], [7,14 ], … [284,256), for a total of 32 BINs. And the value of each component in the characteristic vector is defined as the number of corresponding pixel values in the insulator sample image in the BIN, and the characteristic vector is subjected to normalization processing.
And finally, for the sample image to be detected, obtaining a corresponding histogram feature vector according to the steps, calculating the Euler distances between the histogram feature vector and all marked samples, sequencing the obtained Euler distances, finding out the nearest k points (k is 20), carrying out classification statistics on the k points, and determining the level as the level of the sample to be detected, wherein the number of the points in the level is the largest.
Fig. 4 is a structural diagram of an insulator string contamination detection system based on a binocular camera, as shown in fig. 4, an insulator string contamination detection system based on a binocular camera includes: the device comprises an infrared camera and a high-definition camera; the infrared camera is used for identifying the temperature change of the surface of the insulator string; the high-definition camera is used for detecting dirty substances on the surface of the insulator.
A calibration module 401, configured to calibrate the infrared camera and the high definition camera, and determine infrared camera parameters and high definition camera parameters; the infrared camera parameters comprise an infrared internal parameter, an infrared external parameter and an infrared distortion coefficient of the infrared camera; the high-definition camera parameters comprise high-definition internal parameters, high-definition external parameters and high-definition distortion coefficients of the high-definition camera.
A correction module 402, configured to correct the insulator infrared image collected by the infrared camera by using the infrared camera parameters, correct the insulator color image collected by the high-definition camera by using the high-definition camera parameters, and determine a corrected insulator infrared image and a corrected insulator color image; and determining the matching relation of the corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image.
The determining, in the correction module, a matching relationship between corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image specifically includes: and the matching relation determining unit is used for determining the matching relation of corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image by using an extreme line constraint method.
The preprocessing module 403 is configured to respectively preprocess the corrected insulator infrared image and the corrected insulator color image, and determine a preprocessed insulator infrared image and a preprocessed insulator color image.
And an insulator string region image extraction module 404, configured to extract an insulator string region image in the preprocessed insulator infrared image by using a YOLOv4-tiny network.
The insulator string region image extraction module 404 specifically includes: the infrared insulator string data set determining unit is used for classifying all insulator infrared images and taking the infrared insulator strings on all the insulator infrared images as an infrared insulator string data set; the infrared insulator string data set comprises a normal insulator string, a low-value insulator string and a zero-value insulator string; the dividing unit is used for dividing the infrared insulator string data set to determine a training set and a test set; the classification model generation unit is used for training the YOLOv4-tiny network according to the training set to generate a classification model; the network layer of the YOLOv4-tiny network is deleted from 162 layers to 38 layers, and different scale features of the insulator infrared image are automatically extracted by adopting a feature pyramid network; and the insulator string region image determining unit is used for extracting the insulator string region image in the preprocessed insulator infrared image according to the classification model.
And an insulator string segmentation image determination module 405, configured to map the insulator string region images to the preprocessed insulator color image one by one according to a matching relationship between corresponding pixel points, and determine an insulator string segmentation image on the preprocessed insulator color image.
The insulator string segmentation image determination module 405 specifically includes: the binarization processing unit is used for carrying out binarization processing on the insulator string region image by using a maximum inter-class variance method, segmenting an insulator string region and a background region in the insulator string region image and determining a binarized insulator string region image; and the insulator string segmentation image determining unit is used for mapping the insulator string region image to the preprocessed insulator color image according to the matching relation of the corresponding pixel points, and determining the insulator string segmentation image on the preprocessed insulator color image.
And the insulator string pollution grade determining module 406 is used for determining the pollution grade of the insulator string according to the insulator string segmentation image.
The insulator string contamination level determining module 406 specifically includes: the insulator string sample picture acquisition unit is used for acquiring a plurality of insulator string sample pictures with different fouling grades; the processed insulator string sample picture determining unit is used for performing median filtering denoising and histogram equalization processing on each insulator string sample picture to determine a processed insulator string sample picture; the first histogram feature vector calculation unit is used for calculating the histogram feature vector of the processed insulator string sample picture; the second histogram feature vector calculation unit is used for calculating the histogram feature vectors of the insulator string segmentation images; the Euler distance calculating unit is used for calculating Euler distances between the histogram feature vectors of the insulator string segmentation images and the histogram feature vectors of all the processed insulator string sample pictures; the pixel point determining unit is used for sequencing all the Euler distances from near to far and determining the histogram feature vector of the processed insulator string sample picture corresponding to the Euler distances within the Euler distance threshold; and the insulator string pollution grade determining unit is used for determining the grade of each corresponding processed insulator string sample picture feature vector, and taking the grade with the largest number of grades as the pollution grade of the insulator string.
The invention can realize high-precision insulator string contamination detection, is superior to the existing single-camera-based technology, mainly comes from the fact that the invention utilizes two types of cameras to carry out fusion contamination detection, and fully exerts the advantages of different types of cameras, namely, an infrared camera can well present the position of an insulator, has better adaptability to a complex background, but is insensitive to the surface details of the insulator string; the high-definition visible light camera just makes up for the point, but the high-definition visible light camera has large interference on insulator detection under a complex background. On the other hand, the YOLOv4-tiny deep learning framework is used as an algorithm for insulator detection, the algorithm is high in detection accuracy, and the algorithm can run on an embedded platform.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims (10)

1. The binocular camera-based insulator string fouling detection method is characterized by comprising the following steps of: the device comprises an infrared camera and a high-definition camera; the infrared camera is used for identifying the temperature change of the surface of the insulator string; the high-definition camera is used for detecting dirty substances on the surface of the insulator;
calibrating the infrared camera and the high-definition camera, and determining infrared camera parameters and high-definition camera parameters; the infrared camera parameters comprise an infrared internal parameter, an infrared external parameter and an infrared distortion coefficient of the infrared camera; the high-definition camera parameters comprise high-definition internal parameters, high-definition external parameters and high-definition distortion coefficients of the high-definition camera;
correcting the insulator infrared image collected by the infrared camera by using the infrared camera parameters, correcting the insulator color image collected by the high-definition camera by using the high-definition camera parameters, and determining the corrected insulator infrared image and the corrected insulator color image; determining the matching relation between the corrected insulator infrared image and corresponding pixel points of the corrected insulator color image;
respectively preprocessing the corrected insulator infrared image and the corrected insulator color image, and determining a preprocessed insulator infrared image and a preprocessed insulator color image;
extracting an insulator string region image in the preprocessed insulator infrared image by using a YOLOv4-tiny network;
according to the matching relation of corresponding pixel points, mapping the insulator string region images to the preprocessed insulator color image one by one, and determining an insulator string segmentation image on the preprocessed insulator color image;
and determining the pollution grade of the insulator string according to the insulator string segmentation image.
2. The binocular camera-based insulator string stain detection method according to claim 1, wherein the determining of the matching relationship between corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image specifically comprises:
and determining the matching relation of corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image by using a polar line constraint method.
3. The binocular camera-based insulator string fouling detection method according to claim 1, wherein the extracting of the insulator string region image in the preprocessed insulator infrared image by using a YOLOv4-tiny network specifically comprises:
classifying all insulator infrared images, and taking the infrared insulator strings on all the insulator infrared images as an infrared insulator string data set; the infrared insulator string data set comprises a normal insulator string, a low-value insulator string and a zero-value insulator string;
dividing the infrared insulator string data set to determine a training set and a testing set;
training the YOLOv4-tiny network according to the training set to generate a classification model; the network layer of the YOLOv4-tiny network is deleted from 162 layers to 38 layers, and different scale features of the insulator infrared image are automatically extracted by adopting a feature pyramid network;
and extracting an insulator string region image in the preprocessed insulator infrared image according to the classification model.
4. The binocular camera-based insulator string stain detection method according to claim 1, wherein the step of mapping the insulator string region images to the preprocessed insulator color image one by one according to the matching relationship of corresponding pixel points to determine an insulator string segmentation image on the preprocessed insulator color image specifically comprises:
carrying out binarization processing on the insulator string region image by using a maximum between-class variance method, segmenting an insulator string region and a background region in the insulator string region image, and determining a binarized insulator string region image;
and mapping the insulator string region image to the preprocessed insulator color image according to the matching relation of the corresponding pixel points, and determining an insulator string segmentation image on the preprocessed insulator color image.
5. The binocular camera-based insulator string fouling detection method according to claim 1, wherein the determining of the insulator string fouling level according to the insulator string segmentation image specifically comprises:
acquiring a plurality of insulator string sample pictures with different fouling grades;
performing median filtering denoising and histogram equalization processing on each insulator string sample picture to determine a processed insulator string sample picture;
calculating a histogram feature vector of the processed insulator string sample picture;
calculating histogram feature vectors of the insulator string segmentation images;
calculating Euler distances between histogram feature vectors of the insulator string segmentation images and histogram feature vectors of all processed insulator string sample pictures;
sequencing all the Euler distances in a sequence from near to far, and determining histogram feature vectors of the processed insulator string sample pictures corresponding to the Euler distances within Euler distance thresholds;
and determining the grade of each corresponding processed insulator string sample picture feature vector, and taking the grade with the largest number of grades as the pollution grade of the insulator string.
6. The utility model provides an insulator chain is stained detecting system based on binocular camera which characterized in that includes: the device comprises an infrared camera and a high-definition camera; the infrared camera is used for identifying the temperature change of the surface of the insulator string; the high-definition camera is used for detecting dirty substances on the surface of the insulator;
the calibration module is used for calibrating the infrared camera and the high-definition camera and determining parameters of the infrared camera and the high-definition camera; the infrared camera parameters comprise an infrared internal parameter, an infrared external parameter and an infrared distortion coefficient of the infrared camera; the high-definition camera parameters comprise high-definition internal parameters, high-definition external parameters and high-definition distortion coefficients of the high-definition camera;
the correction module is used for correcting the insulator infrared image acquired by the infrared camera by using the infrared camera parameters, correcting the insulator color image acquired by the high-definition camera by using the high-definition camera parameters, and determining the corrected insulator infrared image and the corrected insulator color image; determining the matching relation between the corrected insulator infrared image and corresponding pixel points of the corrected insulator color image;
the preprocessing module is used for respectively preprocessing the corrected insulator infrared image and the corrected insulator color image and determining a preprocessed insulator infrared image and a preprocessed insulator color image;
the insulator string region image extraction module is used for extracting an insulator string region image in the preprocessed insulator infrared image by using a YOLOv4-tiny network;
the insulator string segmentation image determining module is used for mapping the insulator string region images to the preprocessed insulator color images one by one according to the matching relation of the corresponding pixel points, and determining the insulator string segmentation images on the preprocessed insulator color images;
and the insulator string pollution grade determining module is used for determining the pollution grade of the insulator string according to the insulator string segmentation image.
7. The binocular camera-based insulator string stain detection system according to claim 6, wherein the determining of the matching relationship between the corrected insulator infrared image and the corresponding pixel points of the corrected insulator color image in the correction module specifically comprises:
and the matching relation determining unit is used for determining the matching relation of corresponding pixel points of the corrected insulator infrared image and the corrected insulator color image by using an extreme line constraint method.
8. The binocular camera-based insulator string fouling detection system according to claim 6, wherein the insulator string region image extraction module specifically comprises:
the infrared insulator string data set determining unit is used for classifying all insulator infrared images and taking the infrared insulator strings on all the insulator infrared images as an infrared insulator string data set; the infrared insulator string data set comprises a normal insulator string, a low-value insulator string and a zero-value insulator string;
the dividing unit is used for dividing the infrared insulator string data set to determine a training set and a test set;
the classification model generation unit is used for training the YOLOv4-tiny network according to the training set to generate a classification model; the network layer of the YOLOv4-tiny network is deleted from 162 layers to 38 layers, and different scale features of the insulator infrared image are automatically extracted by adopting a feature pyramid network;
and the insulator string region image determining unit is used for extracting the insulator string region image in the preprocessed insulator infrared image according to the classification model.
9. The binocular camera-based insulator string fouling detection system of claim 6, wherein the insulator string segmentation image determination module specifically comprises:
the binarization processing unit is used for carrying out binarization processing on the insulator string region image by using a maximum inter-class variance method, segmenting an insulator string region and a background region in the insulator string region image and determining a binarized insulator string region image;
and the insulator string segmentation image determining unit is used for mapping the insulator string region image to the preprocessed insulator color image according to the matching relation of the corresponding pixel points, and determining the insulator string segmentation image on the preprocessed insulator color image.
10. The binocular camera-based insulator string fouling detection system according to claim 6, wherein the insulator string fouling level determination module specifically comprises:
the insulator string sample picture acquisition unit is used for acquiring a plurality of insulator string sample pictures with different fouling grades;
the processed insulator string sample picture determining unit is used for performing median filtering denoising and histogram equalization processing on each insulator string sample picture to determine a processed insulator string sample picture;
the first histogram feature vector calculation unit is used for calculating the histogram feature vector of the processed insulator string sample picture;
the second histogram feature vector calculation unit is used for calculating the histogram feature vectors of the insulator string segmentation images;
the Euler distance calculating unit is used for calculating Euler distances between the histogram feature vectors of the insulator string segmentation images and the histogram feature vectors of all the processed insulator string sample pictures;
the pixel point determining unit is used for sequencing all the Euler distances from near to far and determining the histogram feature vector of the processed insulator string sample picture corresponding to the Euler distances within the Euler distance threshold;
and the insulator string pollution grade determining unit is used for determining the grade of each corresponding processed insulator string sample picture feature vector, and taking the grade with the largest number of grades as the pollution grade of the insulator string.
CN202011558906.7A 2020-12-25 2020-12-25 Insulator string stain detection method and system based on binocular camera Pending CN112541478A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011558906.7A CN112541478A (en) 2020-12-25 2020-12-25 Insulator string stain detection method and system based on binocular camera

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011558906.7A CN112541478A (en) 2020-12-25 2020-12-25 Insulator string stain detection method and system based on binocular camera

Publications (1)

Publication Number Publication Date
CN112541478A true CN112541478A (en) 2021-03-23

Family

ID=75017509

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011558906.7A Pending CN112541478A (en) 2020-12-25 2020-12-25 Insulator string stain detection method and system based on binocular camera

Country Status (1)

Country Link
CN (1) CN112541478A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114724042A (en) * 2022-06-09 2022-07-08 国网江西省电力有限公司电力科学研究院 Automatic detection method for zero-value insulator in power transmission line
CN115542100A (en) * 2022-11-29 2022-12-30 广东电网有限责任公司东莞供电局 Insulator fault detection method, device, equipment and medium

Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5909244A (en) * 1996-04-15 1999-06-01 Massachusetts Institute Of Technology Real time adaptive digital image processing for dynamic range remapping of imagery including low-light-level visible imagery
JP2005322139A (en) * 2004-05-11 2005-11-17 Nippon Conlux Co Ltd Paper sheet identification device and method
CN106596579A (en) * 2016-11-15 2017-04-26 同济大学 Insulator contamination condition detection method based on multispectral image information fusion
CN107833211A (en) * 2017-11-01 2018-03-23 国家电网公司 Zero resistance insulator automatic testing method and device based on infrared image
CN107917734A (en) * 2017-11-29 2018-04-17 国网吉林省电力有限公司信息通信公司 Cable's Fault Forecasting Methodology based on temperature and resistance
CN108230237A (en) * 2017-12-15 2018-06-29 同济大学 A kind of multispectral image reconstructing method for electrical equipment on-line checking
CN108345898A (en) * 2017-12-31 2018-07-31 国网山西省电力公司检修分公司 A kind of novel line insulator Condition assessment of insulation method
CN108596196A (en) * 2018-05-15 2018-09-28 同济大学 A kind of filthy state evaluating method based on insulator characteristics of image dictionary
CN108872819A (en) * 2018-07-29 2018-11-23 湖南湖大华龙电气与信息技术有限公司 Isolator detecting unmanned plane and method based on infrared thermal imagery and visible light
CN110136131A (en) * 2019-05-24 2019-08-16 贵州电网有限责任公司 A kind of zero resistance insulator detection method based on deep learning
CN111311967A (en) * 2020-03-31 2020-06-19 普宙飞行器科技(深圳)有限公司 Unmanned aerial vehicle-based power line inspection system and method
CN112001260A (en) * 2020-07-28 2020-11-27 国网湖南省电力有限公司 Cable trench fault detection method based on infrared and visible light image fusion
CN112036464A (en) * 2020-08-26 2020-12-04 国家电网有限公司 Insulator infrared image fault detection method based on YOLOv3-tiny algorithm

Patent Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5909244A (en) * 1996-04-15 1999-06-01 Massachusetts Institute Of Technology Real time adaptive digital image processing for dynamic range remapping of imagery including low-light-level visible imagery
JP2005322139A (en) * 2004-05-11 2005-11-17 Nippon Conlux Co Ltd Paper sheet identification device and method
CN106596579A (en) * 2016-11-15 2017-04-26 同济大学 Insulator contamination condition detection method based on multispectral image information fusion
CN107833211A (en) * 2017-11-01 2018-03-23 国家电网公司 Zero resistance insulator automatic testing method and device based on infrared image
CN107917734A (en) * 2017-11-29 2018-04-17 国网吉林省电力有限公司信息通信公司 Cable's Fault Forecasting Methodology based on temperature and resistance
CN108230237A (en) * 2017-12-15 2018-06-29 同济大学 A kind of multispectral image reconstructing method for electrical equipment on-line checking
CN108345898A (en) * 2017-12-31 2018-07-31 国网山西省电力公司检修分公司 A kind of novel line insulator Condition assessment of insulation method
CN108596196A (en) * 2018-05-15 2018-09-28 同济大学 A kind of filthy state evaluating method based on insulator characteristics of image dictionary
CN108872819A (en) * 2018-07-29 2018-11-23 湖南湖大华龙电气与信息技术有限公司 Isolator detecting unmanned plane and method based on infrared thermal imagery and visible light
CN110136131A (en) * 2019-05-24 2019-08-16 贵州电网有限责任公司 A kind of zero resistance insulator detection method based on deep learning
CN111311967A (en) * 2020-03-31 2020-06-19 普宙飞行器科技(深圳)有限公司 Unmanned aerial vehicle-based power line inspection system and method
CN112001260A (en) * 2020-07-28 2020-11-27 国网湖南省电力有限公司 Cable trench fault detection method based on infrared and visible light image fusion
CN112036464A (en) * 2020-08-26 2020-12-04 国家电网有限公司 Insulator infrared image fault detection method based on YOLOv3-tiny algorithm

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
LI WEI-WEI: "Contamination Grades Recognition of Insulators Under Different Humidity Using Infrared Image Features and RBPNN", 《ENGINEERING, ENVIRONMENTAL SCIENCE》, 31 December 2006 (2006-12-31), pages 1 - 10 *
SHU JIA YAN等: "Insulator Contamination Measurement Based on Infrared Thermal and Visible Image Information Fusion", 《2019 PHOTONLCS & ELECTROMAGNETICS RESEARCH SYMPOSIUM - SPRING (PIERS-SPRING)》, 2 March 2020 (2020-03-02), pages 1006 - 1011 *
曹培等: "基于多光源图像信息融合的绝缘子污秽状态识别", 《电瓷避雷器》, no. 4, 31 December 2019 (2019-12-31), pages 206 - 212 *
李华伟: "基于图像处理的绝缘子污秽检测技术研究", 《中国优秀硕士学位论文全文数据库 工程科技II辑》, no. 3, 15 March 2020 (2020-03-15), pages 042 - 162 *
金立军等: "基于红外与可见光图像信息融合的绝缘子污秽等级识别", 《中国电机工程学报》, vol. 36, no. 13, 5 July 2016 (2016-07-05), pages 3682 - 3691 *
高宇: "基于机器视觉的接触网绝缘子污秽检测技术研究", 《中国优秀硕士学位论文全文数据库 工程科技II辑》, no. 2, 15 February 2018 (2018-02-15), pages 033 - 117 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114724042A (en) * 2022-06-09 2022-07-08 国网江西省电力有限公司电力科学研究院 Automatic detection method for zero-value insulator in power transmission line
CN115542100A (en) * 2022-11-29 2022-12-30 广东电网有限责任公司东莞供电局 Insulator fault detection method, device, equipment and medium

Similar Documents

Publication Publication Date Title
Davari et al. Intelligent diagnosis of incipient fault in power distribution lines based on corona detection in UV-visible videos
CN110389127B (en) System and method for identifying metal ceramic parts and detecting surface defects
CN111784633B (en) Insulator defect automatic detection algorithm for electric power inspection video
CN111242123B (en) Power equipment fault diagnosis method based on infrared image
CN103487729B (en) Based on the power equipments defect detection method that ultraviolet video and infrared video merge
GB2569751A (en) Static infrared thermal image processing-based underground pipe leakage detection method
CN110910350B (en) Nut loosening detection method for wind power tower cylinder
CN103235830A (en) Unmanned aerial vehicle (UAV)-based electric power line patrol method and device and UAV
CN108537170A (en) A kind of power equipment firmware unmanned plane inspection pin missing detection method
CN109034272A (en) A kind of transmission line of electricity heat generating components automatic identifying method
CN110517265A (en) A kind of detection method of surface defects of products, device and storage medium
CN109712084B (en) Image restoration method, image restoration system and flat panel detector
CN112541478A (en) Insulator string stain detection method and system based on binocular camera
CN113781415B (en) Defect detection method, device, equipment and medium for X-ray image
CN115294031A (en) Photovoltaic module fault image identification method based on infrared thermal imaging analysis
CN113313107A (en) Intelligent detection and identification method for multiple types of diseases on cable surface of cable-stayed bridge
CN112991432A (en) Icing shape identification method based on image processing
CN117455843A (en) Intelligent cable head defect detection system
CN112288682A (en) Electric power equipment defect positioning method based on image registration
CN117456371B (en) Group string hot spot detection method, device, equipment and medium
CN117252840B (en) Photovoltaic array defect elimination evaluation method and device and computer equipment
CN116485802B (en) Insulator flashover defect detection method, device, equipment and storage medium
CN116758425A (en) Automatic acceptance checking method and device for large-base photovoltaic power station
CN114792328A (en) Infrared thermal imaging image processing and analyzing method
CN115546716A (en) Binocular vision-based method for positioning fire source around power transmission line

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