CN113192071B - Insulator binding wire missing identification method and device and computer equipment - Google Patents
Insulator binding wire missing identification method and device and computer equipment Download PDFInfo
- Publication number
- CN113192071B CN113192071B CN202110733778.3A CN202110733778A CN113192071B CN 113192071 B CN113192071 B CN 113192071B CN 202110733778 A CN202110733778 A CN 202110733778A CN 113192071 B CN113192071 B CN 113192071B
- Authority
- CN
- China
- Prior art keywords
- insulator
- image
- binding wire
- training
- original image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/32—Normalisation of the pattern dimensions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Abstract
The application relates to a method and a device for identifying the loss of an insulator binding wire and computer equipment. The method comprises the following steps: acquiring an original image of the insulator; inputting the original image into a trained insulator detection model to obtain the type of the insulator and the position area of the insulator in the original image; if the type of the insulator is a preset type, based on the position area, cutting an original image of the insulator to obtain an insulator area image; and inputting the insulator region image into a trained binding wire classification model to obtain a binding wire missing identification result of the insulator. By adopting the method, the efficiency of insulator binding wire missing identification can be improved, and the defects of high labor intensity, more time consumption and labor cost in the traditional manual identification are overcome.
Description
Technical Field
The application relates to the technical field of insulators, in particular to a method and a device for identifying missing of an insulator binding wire, computer equipment and a storage medium.
Background
The distribution network overhead line is long, the coverage area is wide, and in order to prevent the circuit from drooping, fire or human and animal casualties are caused, so that the line needs to be fixed on parts such as porcelain insulators and the like, and is regularly patrolled and examined to check whether the line is in an abnormal fixed condition or not.
Along with the rapid development of unmanned aerial vehicle technology, at present, although unmanned aerial vehicle can be adopted to patrol and examine the distribution network line, the efficiency of operation and maintenance of the distribution network line is improved. However, after the picture is patrolled by the unmanned aerial vehicle acquisition machine, whether the line is fixed on the porcelain insulator or not still needs to be patrolled by the machine acquired by manual identification in the later stage, and the labor intensity of the manual identification method is high, and more time and labor cost need to be consumed.
Disclosure of Invention
Based on this, it is necessary to provide a method and an apparatus for identifying the missing of the binding wire of the insulator, a computer device and a storage medium for the above technical problems of high labor intensity, and much time and labor cost in manually identifying whether the distribution network line is fixed on the porcelain insulator.
A method of insulator binding wire loss identification, the method comprising:
acquiring an original image of the insulator;
inputting the original image into a trained insulator detection model to obtain the type of the insulator and the position area of the insulator in the original image;
if the type of the insulator is a preset type, based on the position area, cutting an original image of the insulator to obtain an insulator area image;
and inputting the insulator region image into a trained binding wire classification model to obtain a binding wire missing identification result of the insulator.
In one embodiment, before the step of cropping an insulator region image from the original image of the insulator based on the position region, the method further includes:
acquiring the length of each boundary line forming the insulator based on the position area;
determining the length of the maximum boundary line from the lengths of the boundary lines;
and when the length of the maximum boundary line is greater than a length threshold value, cutting the original image of the insulator to obtain an insulator region image.
In one embodiment, after determining that the length of the maximum boundary line is greater than the length threshold, the method further includes:
expanding the position area of the insulator in the original image by a set distance to obtain an expanded position area;
and based on the expanded position area, cutting the original image of the insulator to obtain an insulator area image.
In one embodiment, the training process of the insulator detection model includes:
acquiring sample images of various types of insulators to serve as training samples;
and training the pre-constructed insulator detection model by adopting the training sample to obtain the trained insulator detection model.
In one embodiment, before training the pre-constructed insulator detection model by using the training sample, the method further includes:
optimizing the sample images in the training samples to obtain processed training samples;
and training the pre-constructed insulator detection model by adopting the processed training sample.
In one embodiment, the optimization process includes at least one of an image scaling process, an image color space adjustment process, and a combination process of a plurality of sample images;
the optimizing the sample image in the training sample to obtain the processed training sample includes:
if the optimization processing is image scaling processing, reducing the sample images with a first set number in the training samples to obtain reduced training samples, and/or amplifying the sample images with a second set number in the training samples to obtain amplified training samples;
if the optimization processing is image color space adjustment, fusing a sample image in the training sample with a preset background image to obtain a fused training sample;
and if the optimization processing is the combination processing of a plurality of sample images, splicing the plurality of sample images in the training sample, and adjusting the resolution of the obtained spliced image back to the resolution of the sample image before splicing to obtain the training sample of the spliced image after resolution adjustment.
In one embodiment, the training of the binding wire classification model comprises:
acquiring an insulator image containing a binding wire and an insulator image not containing the binding wire as training samples of the binding wire classification model;
and training a pre-constructed binding wire classification model by adopting the training sample of the binding wire classification model to obtain the trained binding wire classification model.
An insulator binding wire loss identification device, the device comprising:
the acquisition module is used for acquiring an original image of the insulator;
the detection module is used for inputting the original image into the trained insulator detection model to obtain the type of the insulator and the position area of the insulator in the original image;
the segmentation module is used for cutting an insulator region image from an original image of the insulator based on the position region if the type of the insulator is a preset type;
and the identification module is used for inputting the insulator region image into the trained binding wire classification model to obtain the binding wire missing identification result of the insulator.
A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
acquiring an original image of the insulator;
inputting the original image into a trained insulator detection model to obtain the type of the insulator and the position area of the insulator in the original image;
if the type of the insulator is a preset type, based on the position area, cutting an original image of the insulator to obtain an insulator area image;
and inputting the insulator region image into a trained binding wire classification model to obtain a binding wire missing identification result of the insulator.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
acquiring an original image of the insulator;
inputting the original image into a trained insulator detection model to obtain the type of the insulator and the position area of the insulator in the original image;
if the type of the insulator is a preset type, based on the position area, cutting an original image of the insulator to obtain an insulator area image;
and inputting the insulator region image into a trained binding wire classification model to obtain a binding wire missing identification result of the insulator.
According to the insulator binding wire missing identification method, the insulator binding wire missing identification device, the computer equipment and the storage medium, after the original image of the insulator is obtained, the original image is input into the trained insulator detection model, the type of the insulator and the position area of the insulator in the original image are obtained, when the type of the insulator is judged to be the preset type, the insulator area image is obtained by cutting the insulator from the original image of the insulator based on the position area, and finally the insulator sub-area image is input into the trained binding wire classification model, so that the binding wire missing identification result of the insulator is obtained. The method adopts a cascade detection strategy, detects the type and the insulator position area of the insulator by adopting a trained insulator detection model in the first step, and detects whether the insulator binding wire is lost or not by adopting a trained binding wire classification model on the insulator position area in the second step, so that on one hand, the identification efficiency is improved, the defects of high labor intensity, more time consumption and labor cost in the traditional manual identification are overcome, on the other hand, the accuracy of the identification of the insulator binding wire loss can be improved, and the problems that the traditional identification method depends on user experience, the identification of the binding wire loss is not timely and comprehensive, and false detection and missed detection are easy to occur are solved.
Drawings
Fig. 1 is a schematic flow chart of a method for identifying missing of an insulator binding wire in one embodiment;
FIG. 2 is a schematic diagram of insulator detection in one embodiment;
fig. 3 is a schematic flow chart of a method for identifying the absence of an insulator binding wire in another embodiment;
fig. 4 is a block diagram showing the structure of an insulator binding wire missing identification apparatus according to an embodiment;
FIG. 5 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
In an embodiment, as shown in fig. 1, a method for identifying missing of an insulator binding wire is provided, and this embodiment is illustrated by applying the method to a terminal, it is to be understood that the method may also be applied to a server, and may also be applied to a system including a terminal and a server, and is implemented by interaction between the terminal and the server. In this embodiment, the method includes the steps of:
and step S102, acquiring an original image of the insulator.
Wherein, the original image can contain at least one insulator.
In specific implementation, the method for acquiring the original image of the insulator by the terminal may be as follows: unmanned aerial vehicle is patrolling and examining the in-process, and after the camera equipment collection that installs through oneself obtained the original image of insulator, transmit the original image for the terminal, handle by the terminal. Or after the unmanned aerial vehicle acquires the original image of the insulator, the original image is transmitted to a database or a server for storage, and the terminal acquires the original image of the insulator from the database or the server and processes the original image.
And step S104, inputting the original image into the trained insulator detection model to obtain the type of the insulator and the position area of the insulator in the original image.
The trained insulator detection model can be used for detecting the type of the insulator and the position area of the insulator in the original image.
The position area is understood to be the area occupied by the position of the insulator image in the original image.
The type of the insulator can be divided into two types, namely, the line needs to be fixed through a binding wire and the line does not need to be fixed through the binding wire.
The insulator needing to fix the circuit through the binding wire is a porcelain insulator and specifically comprises a ceramic strut type porcelain insulator, a ceramic pin type insulator, a ceramic suspension insulator, a rod type porcelain cross arm and the like.
Wherein, the insulator that need not to fix the circuit through the ligature can be composite insulator and glass insulator.
In the concrete realization, this application is to the problem that whether the circuit is fixed on porcelain insulator, whether have the ligature to carry out discernment on the insulator, consequently to be the discernment that needs the porcelain insulator of ligature fixed line. Therefore, the insulator detection model needs to have a function of identifying whether the insulator in the original image is a porcelain insulator, and if the insulator in the original image is identified to be the porcelain insulator, whether the binding wire on the porcelain insulator is lost can be further identified; if the insulator in the original image is identified to be a composite insulator or a glass insulator, the insulator does not need to be fixed on the line through a binding wire, so that whether the binding wire is lost or not does not need to be further identified. Based on this, the original images of the insulators of different types need to be adopted to train the insulator detection model, so that after the original images are input into the trained insulator detection model, not only the position area of the insulator in the original images can be obtained, but also the type of the insulator can be obtained.
In practical application, the insulator detection model may be a model constructed based on a YOLOv3 detection algorithm (a deep learning target detection algorithm that can locate a position of a target in an image and identify a type of the target), referring to fig. 2, for a schematic diagram of insulator detection, first, the size of an original image is adjusted to 608 × 608, and the adjusted image is input into a YOLOv3 target detection algorithm to obtain position and type information of a target frame, where a large number of redundant frames exist in the frames, and need to be filtered out by Non-Maximum Suppression (NMS), and finally, a detection frame of a ceramic insulator component shown in a last diagram in fig. 2 is obtained. The non-maximum suppression filtering method is to remove the detection frames with the overlapping area larger than the threshold, such as 50%, from each detection frame.
And step S106, if the type of the insulator is a preset type, based on the position area, cutting the original image of the insulator to obtain an insulator area image.
The preset type indicates a porcelain insulator that needs to fix a line by a binding wire, for example, a porcelain post type porcelain insulator, a porcelain pin type insulator, a porcelain suspension type insulator, a rod type porcelain cross arm, and the like.
In the concrete implementation, if the insulator in the original image is recognized to be a porcelain insulator, whether the binding wire on the porcelain insulator is missing or not can be further recognized, specifically, the original image of the insulator can be cut based on the position area of the insulator, and the image of the area where the insulator is located is cut out, so that the interference of the background partial image is removed, the obtained insulator area image is input into the trained binding wire classification model, when the binding wire is recognized to be missing or not, the recognition area can be reduced, and the recognition efficiency is improved.
And S108, inputting the insulator region image into the trained binding wire classification model to obtain a binding wire missing identification result of the insulator.
The trained binding wire classification model is used for classifying insulators in the insulator sub-region images into two types, namely binding wires and non-binding wires.
In the specific implementation, after the insulator region image is input into the trained binding wire classification model, an identification result of whether the insulator in the insulator region image has the binding wire is obtained, and further when the identification result of the insulator without the binding wire is obtained, the line fixing abnormality at the insulator is judged, an abnormality alarm is given, and operation and maintenance personnel are prompted to re-bind and fix the line at the abnormal insulator.
According to the insulator binding wire missing identification method, after an original image of an insulator is obtained, the original image is input into a trained insulator detection model to obtain the type of the insulator and the position area of the insulator in the original image, when the type of the insulator is judged to be a preset type, an insulator area image is obtained by cutting the original image of the insulator based on the position area, and finally the insulator area image is input into the trained binding wire classification model to obtain the binding wire missing identification result of the insulator. The method adopts a cascade detection strategy, detects the type and the insulator position area of the insulator by adopting a trained insulator detection model in the first step, and detects whether the insulator binding wire is lost or not by adopting a trained binding wire classification model in the insulator position area in the second step, so that on one hand, the efficiency of identifying the loss of the insulator binding wire is improved, and the defects of high labor intensity, more time consumption and labor cost consumption existing in the traditional manual identification are overcome, on the other hand, the accuracy of identifying the loss of the insulator binding wire can be improved, and the problems that the traditional identification method depends on user experience, the loss of the binding wire is not timely and comprehensively identified, and false detection and missed detection are easy to occur are solved.
In one embodiment, before cropping the insulator region image from the original image of the insulator based on the location region, the method further includes: acquiring the length of each boundary line forming the insulator based on the position area; determining the length of the maximum boundary line from the lengths of the boundary lines; and when the length of the maximum boundary line is greater than the length threshold value, cutting the original image of the insulator to obtain an insulator region image.
Wherein the length of the boundary line may be determined by the number of pixels of the boundary line.
In the specific implementation, the unmanned aerial vehicle may include a plurality of insulator components in the acquired original image, wherein a region of a part of the insulator components may be small, and the resolution is insufficient, so that it is difficult to identify whether a binding wire exists, and therefore, before an insulator region image is cut from the original image of the insulator, the insulator which needs to be subjected to missing identification needs to be filtered based on the size of the insulator, so as to reduce the waste of computing resources. More specifically, the number of pixels of each boundary line of the insulator may be obtained according to the position area of the insulator in the original image, and the number of pixels of the maximum boundary line may be determined therefrom. The number of pixels of the maximum boundary line is compared with a pixel threshold (for example, 60 pixels), and if the number of pixels of the maximum boundary line is greater than the pixel threshold, an insulator region image can be further cut from the original image of the insulator. On the contrary, if the number of pixels of the maximum boundary line is less than or equal to the pixel threshold, the corresponding insulator does not need to be cut.
It can be understood that, if the original image includes a plurality of insulators, a plurality of insulator images with the length of the maximum boundary line being greater than the length threshold are cut out from the original image according to a comparison result between the length of the maximum boundary line of each insulator and the length threshold, so as to obtain a plurality of insulator region images.
In this embodiment, the length of the largest boundary line in each boundary line of the insulator is compared with the length threshold, and when the length of the largest boundary line is greater than the length threshold, the insulator region image is obtained by cutting from the original image of the insulator, so that the problem of invalid calculation and recognition of the trained binding wire classification model on the insulator region image of which the length of the largest boundary line is less than or equal to the length threshold can be solved.
In one embodiment, after determining that the length of the maximum boundary line is greater than the length threshold, the method further includes: expanding the position area of the insulator in the original image by a set distance to obtain an expanded position area; and based on the expanded position area, cutting the original image of the insulator to obtain an insulator area image.
In particular, the location area of the insulator in the original image may be expanded 1/10 outward.
In this embodiment, the problem of incomplete insulator component detection can be prevented by extending the position area of the insulator in the original image by a set distance.
In one embodiment, the training process of the insulator detection model includes: acquiring sample images of various types of insulators; marking the sample image by adopting a preset shape, and taking the marked sample image as a training sample of the insulator detection model; and training the pre-constructed insulator detection model by adopting the training sample to obtain the trained insulator detection model.
In the concrete implementation, the insulator detection neural network structure is built: CSPNet (cross-stage local network) + Darknet53 (backbone network of YOLO V3) can be used as a backbone network to extract training sample characteristics, so that the problem of gradient disappearance can be effectively relieved, and the number of network parameters can be reduced; and (3) generating a network by using the PANET (path aggregation network) as a feature pyramid to improve the propagation of bottom-layer features, and outputting the target category and position by using the single-layer convolution as a prediction network. After the sample image is obtained, an image labeling tool, such as a labelme labeling tool, may be used to label the sample image, generate an xml file in a VOC format as a training sample, and divide the training sample into a training set, a test set, and a verification set, specifically, the division ratio of the training set, the test set, and the verification set may be 6:1: 1. When the insulator detection model is trained, the GIOU _ Loss can be adopted as a Loss function of the regression box so as to obtain a more accurate target positionThe calculation formula is as follows:
wherein IoU is the intersection area of two regions divided by the union area of two regions,
is the minimum closure area of the two rectangular boxes, and U is the union area of the two rectangular boxes. After training is complete, the test set is used to evaluate the model effect.
In this embodiment, the insulator detection model is trained through sample images of multiple types of insulators, so that the obtained trained insulator detection model can accurately determine the position area of the insulator in the original image and the type of the insulator.
In one embodiment, before training the pre-constructed insulator detection model with the training samples, the method further includes: carrying out optimization processing on sample images in the training samples to obtain processed training samples; and training the pre-constructed insulator detection model by adopting the processed training sample.
Further, in one embodiment, the optimization process includes at least one of an image scaling process, an image color space adjustment process, and a combining process of the plurality of sample images;
the step of performing optimization processing on the sample image in the training sample to obtain the processed training sample specifically includes:
if the optimization processing is image scaling processing, reducing the sample images with the first set number in the training samples to obtain reduced training samples, and/or amplifying the sample images with the second set number in the training samples to obtain amplified training samples;
if the optimization processing is image color space adjustment, fusing a sample image in the training sample with a preset background image to obtain a fused training sample;
and if the optimization processing is the combination processing of the plurality of sample images, splicing the plurality of sample images in the training sample, and adjusting the resolution of the obtained spliced image back to the resolution of the sample image before splicing to obtain the training sample of the spliced image after resolution adjustment.
In a specific implementation, the optimization processing on the sample image can be considered from the aspects of size and background, the optimization processing on the size can comprise image scaling processing and combination processing of a plurality of sample images, and the optimization processing on the image background can be image color space adjustment. More specifically, if the image scaling processing is performed, the reduced training samples are obtained by performing reduction processing on a first set number of sample images in the training samples, and/or the enlarged training samples are obtained by performing enlargement processing on a second set number of sample images in the training samples. And if the plurality of sample images are combined, splicing the plurality of sample images in the training sample, and adjusting the resolution of the obtained spliced image back to the resolution of the sample image before splicing to obtain the training sample of the spliced image after resolution adjustment. If the optimization processing is image color space adjustment, the sample image in the training sample is fused with a preset background image to obtain a fused training sample, wherein the preset background image can be various complex background images containing strong light reflection, bird feces, dirt and the like.
In the embodiment, the diversity of the sample image in the aspect of size is expanded through the image scaling processing; through the multi-sample image combination processing, the occupation ratio of the porcelain insulator in the sample image is reduced, and the accuracy of the recognition result under the condition that the porcelain insulator belongs to a small target can be improved; by adding negative samples of complex background images such as strong light reflection, bird droppings, stains and the like, the robustness of the insulator detection model can be enhanced, and by the three optimization processing methods, the diversity of training samples is enriched, so that the anti-interference performance and robustness of the insulator detection model are improved, and the recognition effect of the model is enhanced.
In one embodiment, the training of the ligature classification model comprises: acquiring an insulator image containing a binding wire and an insulator image not containing the binding wire as training samples of a binding wire classification model; and training the pre-constructed binding wire classification model by adopting the training sample of the binding wire classification model to obtain the trained binding wire classification model.
In specific implementation, a binding wire classification model can be trained by using a MobilenetV3 network, after training samples of the binding wire classification model are obtained, the size or resolution of each sample image can be correspondingly adjusted to a set size or resolution, and subsequently, when an insulator region image is obtained by cutting from an original image of an insulator, the size or resolution of the insulator region image can be cut to be consistent with the set size or resolution, so that when the insulator region image is subjected to missing identification through the binding wire classification model, the identification efficiency and accuracy can be improved.
In this embodiment, when the binding wire classification model is trained, the size or resolution of each sample image is adjusted to a set size or resolution, and the set size or resolution is made to be consistent with the size or resolution of an insulator region image cut from an original image of an insulator, so that the recognition efficiency and accuracy when the binding wire classification model performs missing recognition on the insulator region image can be improved.
In another embodiment, as shown in fig. 3, there is provided a method for identifying the absence of an insulator binding wire, in this embodiment, the method includes the steps of:
step S302, acquiring an original image of the insulator;
step S304, inputting the original image into the trained insulator detection model to obtain the type of the insulator and the position area of the insulator in the original image;
step S306, if the type of the insulator is a preset type, based on the position area, obtaining the length of each boundary line forming the insulator, and determining the length of the maximum boundary line from the length of each boundary line;
step S308, when the length of the maximum boundary line is greater than the length threshold, the position area of the insulator in the original image is expanded outwards by a set distance to obtain an expanded position area;
step S310, based on the expanded position area, cutting an original image of the insulator to obtain an insulator area image;
and step S312, inputting the insulator region image into the trained binding wire classification model to obtain a binding wire missing identification result of the insulator.
It can be understood that the distribution network machine patrols that the image insulator binding wire lacks defect target area little, and it is big directly to look for the degree of difficulty on the whole picture, easily lou examines, easily appears the error identification condition in the outer region of insulator part moreover, because the shooting angle causes sheltering from easily, screen visual error, very easily causes the false retrieval simultaneously. According to the insulator binding wire missing identification method provided by the embodiment, a cascade detection strategy is adopted, the missing defects of the insulator components and the insulator binding wires are detected by adopting a deep learning target detection algorithm in the first step, the missing defects of the porcelain insulator binding wires are detected again on the specific insulator components by adopting the deep learning target detection algorithm in the second step, and the deduction rate of the binding wires is greatly increased. And finally, through non-maximum value inhibition, removing repeated detection results, and adding a sample without loss of the binding wire when training the binding wire classification model, thereby greatly enhancing the robustness of the model.
It should be understood that although the steps in the flowcharts of fig. 1 and 3 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 1 and 3 may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, which are not necessarily performed in sequence, but may be performed in turn or alternately with other steps or at least some of the other steps.
In one embodiment, as shown in fig. 4, there is provided an insulator binding wire missing recognition device including: an acquisition module 402, a detection module 404, a segmentation module 406, and an identification module 408, wherein:
an obtaining module 402, configured to obtain an original image of an insulator;
a detection module 404, configured to input the original image into the trained insulator detection model, so as to obtain a type of the insulator and a position area of the insulator in the original image;
a segmentation module 406, configured to, if the type of the insulator is a preset type, based on the location area, cut an insulator area image from an original image of the insulator;
and the identification module 408 is configured to input the insulator region image into the trained binding wire classification model to obtain a binding wire missing identification result of the insulator.
In an embodiment, the dividing module 406 is further configured to obtain lengths of boundary lines forming the insulator based on the position area; determining the length of the maximum boundary line from the lengths of the boundary lines; and when the length of the maximum boundary line is greater than the length threshold value, cutting the original image of the insulator to obtain an insulator region image.
In an embodiment, the segmentation module 406 is further configured to expand the position area of the insulator in the original image by a set distance to obtain an expanded position area; and based on the expanded position area, cutting the original image of the insulator to obtain an insulator area image.
In one embodiment, the apparatus further includes a detection model training module, configured to obtain sample images of multiple types of insulators as training samples; and training the pre-constructed insulator detection model by adopting the training sample to obtain the trained insulator detection model.
In one embodiment, the apparatus further includes an optimization module, configured to perform optimization processing on a sample image in a training sample to obtain a processed training sample; and training the pre-constructed insulator detection model by adopting the processed training sample.
In one embodiment, the optimization process includes at least one of an image scaling process, an image color space adjustment process, and a combining process of the plurality of sample images; the optimization module is further configured to, if the optimization processing is image scaling processing, perform reduction processing on a first set number of sample images in the training samples to obtain reduced training samples, and/or perform amplification processing on a second set number of sample images in the training samples to obtain amplified training samples; if the optimization processing is image color space adjustment, fusing a sample image in the training sample with a preset background image to obtain a fused training sample; and if the optimization processing is the combination processing of the plurality of sample images, splicing the plurality of sample images in the training sample, and adjusting the resolution of the obtained spliced image back to the resolution of the sample image before splicing to obtain the training sample of the spliced image after resolution adjustment.
In one embodiment, the apparatus further includes a classification model training module, configured to obtain an insulator image including the binding wire and an insulator image not including the binding wire as training samples of the binding wire classification model; and training the pre-constructed binding wire classification model by adopting the training sample of the binding wire classification model to obtain the trained binding wire classification model.
It should be noted that the device for identifying missing insulator binding wires and the method for identifying missing insulator binding wires of the present application correspond to each other one-to-one, and the technical features and the advantages thereof described in the embodiments of the method for identifying missing insulator binding wires are all applicable to the embodiments of the device for identifying missing insulator binding wires, and specific contents can be referred to the description in the embodiments of the method of the present application, which is not repeated herein, and thus the present application claims.
In addition, all or part of the modules in the insulator binding wire missing identification device can be realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 5. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless communication can be realized through WIFI, an operator network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement an insulator binding wire missing identification method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.
Those skilled in the art will appreciate that the architecture shown in fig. 5 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is further provided, which includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above method embodiments when executing the computer program.
In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above examples only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method for identifying the loss of an insulator binding wire is characterized by comprising the following steps:
acquiring an original image of the insulator;
inputting the original image into a trained insulator detection model to obtain the type of the insulator and the position area of the insulator in the original image; the insulator detection model takes a cross-stage local network and a Darknet53 network as a backbone network to extract training sample characteristics, takes a path aggregation network as a characteristic pyramid to generate a network to improve bottom layer characteristic propagation, and takes single-layer convolution as a prediction network to output target categories and positions; the Darknet53 network is a backbone network of a Yolov3 network;
if the type of the insulator is a preset type, extending the position area of the insulator in the original image by a set distance, and cutting the original image of the insulator to obtain an insulator area image based on the extended position area;
inputting the insulator region image into a trained binding wire classification model to obtain a binding wire missing identification result of the insulator; the trained binding wire classification model is used for classifying the insulators in the insulator region image into two types, namely binding wires and non-binding wires.
2. The method of claim 1, further comprising, before cropping an insulator region image from the original image of the insulator based on the location region:
acquiring the length of each boundary line forming the insulator based on the position area;
determining the length of the maximum boundary line from the lengths of the boundary lines;
and when the length of the maximum boundary line is greater than a length threshold value, cutting the original image of the insulator to obtain an insulator region image.
3. The method of claim 1, wherein the training process of the insulator detection model comprises:
acquiring sample images of various types of insulators to serve as training samples;
and training the pre-constructed insulator detection model by adopting the training sample to obtain the trained insulator detection model.
4. The method of claim 3, wherein before training the pre-constructed insulator testing model using the training samples, further comprising:
optimizing the sample images in the training samples to obtain processed training samples;
and training the pre-constructed insulator detection model by adopting the processed training sample.
5. The method according to claim 4, wherein the optimization process includes at least one of an image scaling process, an image color space adjustment process, and a combination process of a plurality of sample images;
the optimizing the sample image in the training sample to obtain the processed training sample includes:
if the optimization processing is image scaling processing, reducing the sample images with a first set number in the training samples to obtain reduced training samples, and/or amplifying the sample images with a second set number in the training samples to obtain amplified training samples;
if the optimization processing is image color space adjustment, fusing a sample image in the training sample with a preset background image to obtain a fused training sample;
and if the optimization processing is the combination processing of a plurality of sample images, splicing the plurality of sample images in the training sample, and adjusting the resolution of the obtained spliced image back to the resolution of the sample image before splicing to obtain the training sample of the spliced image after resolution adjustment.
6. The method of claim 1, wherein the training of the ligature classification model comprises:
acquiring an insulator image containing a binding wire and an insulator image not containing the binding wire as training samples of the binding wire classification model;
and training a pre-constructed binding wire classification model by adopting the training sample of the binding wire classification model to obtain the trained binding wire classification model.
7. An insulator binding wire loss recognition device, the device comprising:
the acquisition module is used for acquiring an original image of the insulator;
the detection module is used for inputting the original image into the trained insulator detection model to obtain the type of the insulator and the position area of the insulator in the original image; the insulator detection model takes a cross-stage local network and a Darknet53 network as a backbone network to extract training sample characteristics, takes a path aggregation network as a characteristic pyramid to generate a network to improve bottom layer characteristic propagation, and takes single-layer convolution as a prediction network to output target categories and positions; the Darknet53 network is a backbone network of a Yolov3 network;
the segmentation module is used for extending the position area of the insulator in the original image by a set distance if the type of the insulator is a preset type, and cutting the original image of the insulator to obtain an insulator area image based on the extended position area;
the identification module is used for inputting the insulator region image into the trained binding wire classification model to obtain a binding wire missing identification result of the insulator; the trained binding wire classification model is used for classifying the insulators in the insulator region image into two types, namely binding wires and non-binding wires.
8. The apparatus of claim 7, wherein the segmentation module is further configured to obtain lengths of boundary lines constituting the insulator based on the location area; determining the length of the maximum boundary line from the lengths of the boundary lines; and when the length of the maximum boundary line is greater than a length threshold value, cutting the original image of the insulator to obtain an insulator region image.
9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 6 when executing the computer program.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110733778.3A CN113192071B (en) | 2021-06-29 | 2021-06-29 | Insulator binding wire missing identification method and device and computer equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110733778.3A CN113192071B (en) | 2021-06-29 | 2021-06-29 | Insulator binding wire missing identification method and device and computer equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113192071A CN113192071A (en) | 2021-07-30 |
| CN113192071B true CN113192071B (en) | 2021-11-23 |
Family
ID=76976704
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110733778.3A Active CN113192071B (en) | 2021-06-29 | 2021-06-29 | Insulator binding wire missing identification method and device and computer equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113192071B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108537170A (en) * | 2018-04-09 | 2018-09-14 | 电子科技大学 | A kind of power equipment firmware unmanned plane inspection pin missing detection method |
| CN108647588A (en) * | 2018-04-24 | 2018-10-12 | 广州绿怡信息科技有限公司 | Goods categories recognition methods, device, computer equipment and storage medium |
| CN112651988A (en) * | 2021-01-13 | 2021-04-13 | 重庆大学 | Finger-shaped image segmentation, finger-shaped plate dislocation and fastener abnormality detection method based on double-pointer positioning |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7548649B2 (en) * | 2005-01-25 | 2009-06-16 | Siemens Medical Solutions Usa, Inc. | Multidimensional segmentation based on adaptive bounding box and ellipsoid models |
| JP2009259676A (en) * | 2008-04-18 | 2009-11-05 | Chugoku Electric Power Co Inc:The | Cable support, and method for supporting overhead cable using the same |
| CN104951990B (en) * | 2014-03-28 | 2018-06-05 | 国际商业机器公司 | Power grid data processing method and equipment |
| CN105719305B (en) * | 2016-01-25 | 2018-10-19 | 成都国铁电气设备有限公司 | Component falls off defect identification method and system in contact net |
| CN109255776B (en) * | 2018-07-23 | 2023-12-05 | 中国电力科学研究院有限公司 | Automatic identification method for cotter pin defect of power transmission line |
| CN110569762A (en) * | 2019-08-27 | 2019-12-13 | 许昌许继软件技术有限公司 | pin falling detection method and device based on multistage neural network |
| CN111311560B (en) * | 2020-02-10 | 2023-09-12 | 中国铁道科学研究院集团有限公司基础设施检测研究所 | Method and device for detecting state of steel rail fastener |
| CN111899216A (en) * | 2020-06-16 | 2020-11-06 | 西安交通大学 | Abnormity detection method for insulator fastener of high-speed rail contact network |
| CN111915558B (en) * | 2020-06-30 | 2023-12-01 | 成都思晗科技股份有限公司 | Pin state detection method for high-voltage transmission line |
| CN112699925A (en) * | 2020-12-23 | 2021-04-23 | 国网安徽省电力有限公司检修分公司 | Transformer substation meter image classification method |
| CN112634254A (en) * | 2020-12-29 | 2021-04-09 | 北京市商汤科技开发有限公司 | Insulator defect detection method and related device |
| CN112614130A (en) * | 2021-01-04 | 2021-04-06 | 东华大学 | Unmanned aerial vehicle power transmission line insulator fault detection method based on 5G transmission and YOLOv3 |
-
2021
- 2021-06-29 CN CN202110733778.3A patent/CN113192071B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108537170A (en) * | 2018-04-09 | 2018-09-14 | 电子科技大学 | A kind of power equipment firmware unmanned plane inspection pin missing detection method |
| CN108647588A (en) * | 2018-04-24 | 2018-10-12 | 广州绿怡信息科技有限公司 | Goods categories recognition methods, device, computer equipment and storage medium |
| CN112651988A (en) * | 2021-01-13 | 2021-04-13 | 重庆大学 | Finger-shaped image segmentation, finger-shaped plate dislocation and fastener abnormality detection method based on double-pointer positioning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113192071A (en) | 2021-07-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11276162B2 (en) | Surface defect identification method and apparatus | |
| CN108009543B (en) | License plate recognition method and device | |
| CN111507958B (en) | Target detection method, training method of detection model and electronic equipment | |
| CN110705405B (en) | Target labeling method and device | |
| CN111179253A (en) | Product defect detection method, device and system | |
| CN111680746B (en) | Vehicle damage detection model training, vehicle damage detection method, device, equipment and medium | |
| CN110346699B (en) | Insulator discharge information extraction method and device based on ultraviolet image processing technology | |
| CN117893817A (en) | Wafer defect classification method, device and system thereof, electronic equipment and storage medium | |
| CN112418155B (en) | Method for detecting position and type of workpiece on metro vehicle side inspection image | |
| WO2020062088A1 (en) | Image identification method and device, storage medium, and processor | |
| CN112449147B (en) | Video cluster monitoring system of photovoltaic power station and image processing method thereof | |
| CN113888514A (en) | Method and device for detecting defects of ground wire, edge computing equipment and storage medium | |
| CN116883801A (en) | YOLOv8 target detection method based on attention mechanism and multi-scale feature fusion | |
| CN115239646A (en) | Defect detection method and device for power transmission line, electronic equipment and storage medium | |
| CN114581419A (en) | Transformer insulating sleeve defect detection method, related equipment and readable storage medium | |
| CN113192071B (en) | Insulator binding wire missing identification method and device and computer equipment | |
| CN117351472A (en) | Tobacco leaf information detection method and device and electronic equipment | |
| WO2019071663A1 (en) | Electronic apparatus, virtual sample generation method and storage medium | |
| CN116579998A (en) | Defect detection method, device, equipment, medium and program product of transformer equipment | |
| CN115690505A (en) | Photovoltaic module fault detection method and device, computer equipment and storage medium | |
| KR20210037199A (en) | Apparatus for dividing, tagging an image and for detecting defect of facilities using the same | |
| CN110569717A (en) | partial discharge detection method, device, system, equipment and readable storage medium | |
| CN114241354A (en) | Warehouse personnel behavior identification method and device, computer equipment and storage medium | |
| CN111931721A (en) | Method and device for detecting color and number of annual inspection label and electronic equipment | |
| US20200167908A1 (en) | Determination device and determining method thereof |
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 |