CN113658120A - Defect identification method and device for electric power equipment, computer equipment and storage medium - Google Patents
Defect identification method and device for electric power equipment, computer equipment and storage medium Download PDFInfo
- Publication number
- CN113658120A CN113658120A CN202110903217.3A CN202110903217A CN113658120A CN 113658120 A CN113658120 A CN 113658120A CN 202110903217 A CN202110903217 A CN 202110903217A CN 113658120 A CN113658120 A CN 113658120A
- Authority
- CN
- China
- Prior art keywords
- feature vector
- image
- recognized
- determining
- vector set
- 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
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
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06T2207/30164—Workpiece; Machine component
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The application relates to a defect identification method and device for power equipment, computer equipment and a storage medium. The method comprises the following steps: the method comprises the steps that an image to be identified is sent by a receiver nest, wherein the image to be identified is obtained by shooting power equipment through an unmanned aerial vehicle configured by the nest; acquiring a reference feature vector library corresponding to the image to be recognized, and determining a feature vector set to be recognized based on the image to be recognized, wherein the reference feature vector library comprises reference feature vector sets of a plurality of reference images, and the reference images are images of the power equipment when parts are missing; and determining a recognition result based on the reference feature vector library and the feature vector set to be recognized, and sending the recognition result to the cloud server. The method improves the defect identification efficiency and can realize the real-time identification of the defects of the power equipment.
Description
Technical Field
The present application relates to the field of image processing technologies, and in particular, to a method and an apparatus for identifying defects of an electrical device, a computer device, and a storage medium.
Background
The application of the multi-rotor unmanned aerial vehicle technology in the aspect of power transmission line inspection business in the power industry tends to be mature. The mode is patrolled and examined to present many rotor unmanned aerial vehicle mainly is flying control application APP control unmanned aerial vehicle by removing the end and carrying out automatic tour to power line according to the three-dimensional air traffic that plans in advance. Because the flight time that many rotor unmanned aerial vehicle battery power supported is not many, lead to once to accomplish the tour of whole circuit, the latest technique of solving this problem adopts the mode of "machine nest-unmanned aerial vehicle" to carry out the circuit and patrols and examines, promptly installs the machine nest of unmanned aerial vehicle according to the circuit district section, every machine nest fixed configuration unmanned aerial vehicle, unmanned aerial vehicle can charge at the machine nest, the machine nest sends out the tour instruction to unmanned aerial vehicle and receives unmanned aerial vehicle tour data, a plurality of "machine nest-unmanned aerial vehicle" tours to a plurality of circuit district sections and constitutes the tour to whole circuit.
At present, a server performs defect identification on an image of an electrical device by using a defect identification method, and the existing defect identification method generally needs to compare a reference image of the electrical device with a shot image, for example, to realize defect identification through deep learning. However, a large number of images to be defect-identified are usually located on the patrol road, the existing defect identification method is complex in process, high in requirements on computing capacity, memory and the like of a server, and long in time required for performing defect identification once, so that the efficiency of performing defect identification on the images of the electric power equipment is low, and real-time identification on the defects of the electric power equipment cannot be realized.
Disclosure of Invention
In view of the above, it is desirable to provide a method, an apparatus, a computer device, and a storage medium for identifying defects of an electric power device, which are efficient and can identify defects in real time.
A defect identification method of power equipment is applied to an edge server and comprises the following steps:
the method comprises the steps that an image to be identified is sent by a receiver nest, wherein the image to be identified is obtained by shooting power equipment through an unmanned aerial vehicle configured by the nest;
acquiring a reference feature vector library corresponding to the image to be recognized, and determining a feature vector set to be recognized based on the image to be recognized, wherein the reference feature vector library comprises reference feature vector sets of a plurality of reference images, and the reference images are images of the power equipment when parts are missing;
and determining a recognition result based on the reference feature vector library and the feature vector set to be recognized, and sending the recognition result to the cloud server.
In one embodiment, the image to be recognized is configured with a first image identifier; the obtaining of the reference feature vector library corresponding to the image to be recognized includes:
in the memory of the edge server, a reference feature vector library corresponding to the image to be recognized is obtained based on the first image identifier, wherein the reference feature vector library is determined by the cloud server in advance based on a plurality of reference images of the power equipment and is sent to the edge server.
In one embodiment, the determining a feature vector set to be recognized based on the image to be recognized includes:
determining a plurality of characteristic points in the image to be identified by adopting a FAST detection algorithm;
and determining the feature vector to be recognized of each feature point by adopting an ORB algorithm, and determining a feature vector set to be recognized based on the feature vector to be recognized of each feature point.
In one embodiment, the determining a recognition result based on the reference feature vector library and the feature vector set to be recognized includes:
determining similarity corresponding to any reference feature vector set based on any reference feature vector set and the feature vector set to be identified, wherein the similarity is used for reflecting the number of identical feature vectors in any reference feature vector set and the feature vector set to be identified;
and determining the recognition result according to the similarity corresponding to any reference feature vector set and a preset threshold.
In one embodiment, the reference feature vector set comprises a plurality of reference feature vectors, the feature vector set to be identified comprises a plurality of feature vectors to be identified, and the plurality of reference feature vectors and the plurality of feature vectors to be identified are in one-to-one correspondence; the determining the similarity corresponding to any reference feature vector set based on any reference feature vector set and the feature vector set to be recognized includes:
determining a Hamming distance between any reference characteristic vector and a characteristic vector to be identified corresponding to any reference characteristic vector;
and determining the number of Hamming distances smaller than a preset distance, and taking the determined number as the corresponding similarity of any reference feature vector set.
In one embodiment, the determining the recognition result according to the similarity corresponding to any one of the reference feature vector sets and a preset threshold includes:
if the similarity corresponding to any reference feature vector is smaller than the preset threshold, determining that the identification result is non-defective information;
if the similarity corresponding to any reference feature vector is greater than or equal to the preset threshold, taking a reference feature vector set corresponding to the similarity greater than or equal to the preset threshold as a target feature vector set, and determining defect information according to the target feature vector set.
In one embodiment, the defect information includes defect identifiers and defect pixel point coordinates, and each reference vector set is configured with a second image identifier; determining defect information according to the target feature vector set, including:
determining the coordinates of defective pixel points according to the target feature vector set, wherein the coordinates of the defective pixel points are the coordinates of pixel points corresponding to the same feature vectors in the target feature vector set, and the same feature vectors are the same feature vectors in the target feature vector set as the feature vector set to be identified;
and taking a second image identifier of the target feature vector set as a defect identifier, wherein the defect identifier is used for the cloud server to determine the defect level and the defect description information of the image to be recognized.
A defect identifying apparatus of an electric power device, comprising:
the image receiving module is used for receiving an image to be identified sent by a nest, wherein the image to be identified is obtained by shooting power equipment by an unmanned aerial vehicle configured by the nest;
the characteristic vector set determining module is used for acquiring a reference characteristic vector library corresponding to the image to be recognized and determining a characteristic vector set to be recognized based on the image to be recognized, wherein the reference characteristic vector library comprises reference characteristic vector sets of a plurality of reference images, and the reference images are images of the power equipment when parts are missing;
and the identification result determining module is used for determining an identification result based on the reference feature vector library and the feature vector set to be identified and sending the identification result to the cloud server.
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:
the method comprises the steps that an image to be identified is sent by a receiver nest, wherein the image to be identified is obtained by shooting power equipment through an unmanned aerial vehicle configured by the nest;
acquiring a reference feature vector library corresponding to the image to be recognized, and determining a feature vector set to be recognized based on the image to be recognized, wherein the reference feature vector library comprises reference feature vector sets of a plurality of reference images, and the reference images are images of the power equipment when parts are missing;
and determining a recognition result based on the reference feature vector library and the feature vector set to be recognized, and sending the recognition result to the cloud server.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
the method comprises the steps that an image to be identified is sent by a receiver nest, wherein the image to be identified is obtained by shooting power equipment through an unmanned aerial vehicle configured by the nest;
acquiring a reference feature vector library corresponding to the image to be recognized, and determining a feature vector set to be recognized based on the image to be recognized, wherein the reference feature vector library comprises reference feature vector sets of a plurality of reference images, and the reference images are images of the power equipment when parts are missing;
and determining a recognition result based on the reference feature vector library and the feature vector set to be recognized, and sending the recognition result to the cloud server.
According to the defect identification method of the electric power equipment, the edge server does not store the reference image and does not extract the reference feature vector set of the reference image, the reference feature vector library corresponding to the image to be identified is directly obtained, only the feature vector set to be identified of the image to be identified needs to be extracted, and the identification result is determined according to the reference feature vector library and the feature vector set to be identified, so that the defect identification efficiency of the edge server is improved, and the identification result can be obtained in real time; therefore, the defect identification method based on the electric power equipment improves the defect identification efficiency and can realize the real-time identification of the defects of the electric power equipment.
Drawings
FIG. 1 is a diagram illustrating an application scenario of a defect identification method for an electrical device according to an embodiment;
FIG. 2 is a flow chart illustrating a method for identifying defects in an electrical device according to one embodiment;
FIG. 3 is a schematic view of an exemplary mounting nest;
FIG. 4 is a schematic diagram of an embodiment of configuring an edge server for a cell;
FIG. 5 is a diagram of corner criteria in one embodiment;
FIG. 6 is a schematic structural diagram of a defect identifying apparatus of an electric power device according to an embodiment;
FIG. 7 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.
The defect identification method for the power equipment can be applied to an application scene shown in fig. 1, wherein the nest 102 sends an image to be identified, which is shot by the unmanned aerial vehicle, to the edge server 104, the edge server 104 obtains a reference feature vector library issued by the cloud server 106, the edge server 104 determines an identification result according to the reference feature vector library and the image to be identified, and sends the identification result to the cloud server 106.
In one embodiment, as shown in fig. 2, a defect identification method for an electrical device is provided, which is described by taking the method as an example applied to the edge server in fig. 1, and includes the following steps:
The image to be identified is obtained by shooting power equipment through the unmanned aerial vehicle configured in the nest. The nest is installed on the transformer substation, unmanned aerial vehicle and nest one-to-one, unmanned aerial vehicle with the nest passes through remote control communication connection, and the nest provides to charge for unmanned aerial vehicle, and it can be wired charging to charge, perhaps wireless charging. The nest sends a patrol work order to the unmanned aerial vehicle through remote control, and the unmanned aerial vehicle patrols the power equipment of the overhead transmission line according to the patrol work order, specifically shoots the image of the power equipment at a preset position (described by longitude and latitude and ground height) in a patrol range according to the patrol work order, obtains an image to be identified, and sends the image to be identified to the nest through remote control.
The method comprises the steps that a nest sends an image to be identified to an edge server, and the edge server receives the image to be identified sent by the nest. The corresponding relation exists between the edge server and the machine nest, and the corresponding relation can be set according to the computing power of the edge server and the size of the inspection range of the machine nest. For example, the patrol range of the machine nest is small, the computing power of the edge server is small, and one edge server can be set to correspond to one machine nest; the patrol range of the machine nest is small, the calculation power of the edge server is large, and one edge server can be set to correspond to two machine nests. The edge server establishes a communication connection with its corresponding cell. The image to be identified sent by the edge server receiving nest means that the edge server receives the image to be identified sent by the only nest corresponding to the edge server, or when the edge server corresponds to two nests, the edge server receives the image to be identified sent by any nest of the two nests.
Wherein the reference feature vector library comprises a reference feature vector set of a plurality of reference images, and the reference images are images of the power equipment when parts are missing. The power equipment refers to the power equipment shot in the image to be identified.
The reference image is an image of the power equipment shot by the image to be recognized when a part is missing, the reference image is shot at a reference position (described by longitude and latitude and ground height), and a preset position corresponding to the image to be recognized is the same as the reference position corresponding to the reference image, so that the framing size and the framing content of the reference image to the power equipment are the same as the framing size and the framing content of the image to be recognized to the power equipment.
The power equipment for shooting the image to be identified comprises a plurality of parts, and the reference image is an image with one missing part or two or more missing parts, so that the reference images corresponding to the power equipment are multiple, and the missing parts of each reference image are different.
In order to reduce the computing pressure of the edge server, the edge server does not store the reference image and does not execute the process of extracting the reference feature vector set of the reference image, but directly obtains the reference feature vector library corresponding to the image to be identified from the cloud server.
Specifically, the feature vector set to be recognized of the image to be recognized may be extracted by a conventional feature extraction method, for example, the feature vector set to be recognized of the image to be recognized may be extracted by Harris, SIFT, SURF, or the like, or the image to be recognized may be input to a trained feature classifier by a deep learning method, so as to obtain the feature vector of the image to be recognized.
Specifically, a defect identification result is obtained by determining the difference between a reference image and an image to be identified, and the difference between the reference image and the image to be identified is determined by each reference feature vector set and each feature vector set to be identified in the reference feature vector library, so that the identification result is obtained. The identification result comprises non-defective information or defective information, the non-defective information is used for reflecting that the image to be identified has no missing part, and the defective information is used for reflecting that the image to be identified has the missing part. The edge server and the cloud server have larger bandwidth, the bandwidth required for transmitting the identification result is smaller, network congestion cannot be caused in the transmission process, and the identification result can be transmitted to the cloud server in real time.
In the defect identification method for the power equipment, the edge server does not store the reference image, does not extract the reference feature vector set of the reference image, directly obtains the reference feature vector library corresponding to the image to be identified, and only needs to extract the feature vector set to be identified of the image to be identified, and determines the identification result according to the reference feature vector library and the feature vector set to be identified, so that the defect identification efficiency of the edge server is improved, and the identification result can be obtained in real time; therefore, the defect identification method based on the electric power equipment improves the defect identification efficiency and can realize the real-time identification of the defects of the electric power equipment.
In one embodiment, prior to step 201, the nests are pre-installed and an edge server is configured for each nest installed.
Specifically, referring to fig. 3, the machine nest is installed on a transformer substation, it is determined that a transformer substation is installed with a first machine nest 1, a patrol range of the machine nest 1 is obtained through field testing, a radius of a patrol range of the machine nest 1 is determined, a next transformer substation is determined according to the radius of the patrol range of the machine nest 1, a second machine nest 2 is installed on the next transformer substation, a radius of the patrol range of the machine nest 2 is determined, a third machine nest 3 is installed according to the radius of the patrol range of the machine nest 2, and the process of installing the machine nest is repeated until the patrol range of the machine nest covers the whole identification area.
Considering that the patrol range of the machine nests may have slight changes in actual operation, setting that the patrol ranges of any two adjacent machine nests have an overlapping area, wherein the overlapping area comprises a preset number of towers; the preset number may be set to 2, that is, the overlapping area of the patrol ranges of any two adjacent nests includes 2 towers. For example, when the next substation installs a second nest 2, the distance between the nest 1 and the nest 2 is smaller than the radius of the patrol range of the nest 1, which is determined according to the radius of the patrol range of the nest 1.
Referring to fig. 4, the edge server should be configured for a nest in consideration of the defect recognition efficiency and cost. If the patrol range of each nest is large, the number of images to be identified processed by the edge server corresponding to the nest is large, and in consideration of defect identification efficiency, one edge server can be independently configured for each nest. Or, if the patrol range of each nest is smaller, it indicates that the number of images to be identified processed by the edge server corresponding to the nest is less, and on the premise of not affecting the defect identification efficiency, considering the cost, one edge server may be configured for two nests adjacent to the patrol range.
The nest sends a patrol work order to the unmanned aerial vehicles through remote control, the unmanned aerial vehicles patrol the power equipment of the overhead power transmission line according to the patrol work order, the patrol line of each unmanned aerial vehicle is fixed, and the images to be identified received by the nest are images of the fixed power equipment; because the machine nest is fixedly arranged in the transformer substation, after the corresponding relation between the machine nest and the edge server is determined, the image to be identified received by the edge server is also the image of the fixed power equipment. Therefore, the edge server only needs to query the reference feature vector library corresponding to the image to be identified in the reference feature vector library corresponding to the image of the fixed power equipment, so that the time for the edge server to acquire the reference feature vector library corresponding to the image to be identified is shortened, and the defect identification efficiency is improved.
The edge server is arranged in a preset range of the machine nest, the preset range can be a range with the radius of 1m to 5m, namely, the transmission distance between the edge server and the machine nest is short, the time required for the machine nest to transmit the image to be recognized to the edge server is short, and the defect recognition efficiency is improved.
In one embodiment, step 202 comprises:
step 301, in the memory of the edge server, obtaining a reference feature vector library corresponding to the image to be recognized based on the first image identifier.
The reference feature vector library is determined by the cloud server in advance based on a plurality of reference images of the electric power equipment and is sent to the edge server. The first image identifier is an image identifier configured by the image to be recognized, and the first image identifier is used for representing the shooting position of the image to be recognized.
Specifically, the cloud server extracts reference feature vector sets corresponding to all reference images in advance, assembles a plurality of corresponding reference feature vector sets at the same shooting position into a reference feature vector library corresponding to the shooting position, allocates a position identifier to each reference feature vector library, and issues the reference feature vector libraries (allocated with the position identifiers) corresponding to the plurality of shooting positions processed by the edge server to the edge processor.
The edge processor stores a plurality of reference feature vector libraries sent by a cloud server in an internal memory, and when the edge server receives an image to be recognized sent by a nest, the edge processor searches the reference feature vector library corresponding to the image to be recognized in the plurality of reference feature vector libraries according to a first image identifier of the image to be recognized, wherein the position identifier of the reference feature vector library corresponding to the image to be recognized is consistent with the first image identifier of the image to be recognized.
In another implementation, step 202 includes:
step 310, sending an acquisition request to the cloud server, where the acquisition request includes the first image identifier.
Step 311, receiving a reference feature vector library obtained by the cloud server based on the first image identifier.
Specifically, the edge server responds to the received image to be recognized and sends an acquisition request to the cloud server, and the cloud server acquires a reference feature vector library according to a first image identifier in the acquisition request and sends the reference feature vector library to the edge server.
In this implementation, the edge server does not store the reference image and the reference feature vector library, and the hardware requirement on the edge server is low, but compared with step 301, the edge server needs to take longer time to request the reference feature vector library from the cloud server, and the defect identification efficiency is not as good as that in step 301 to some extent, but the hardware requirement on the edge server is low in this implementation. Considering the cost and the defect recognition efficiency together, the processing mode of step 301 or the processing modes of step 310 to step 311 may be selected according to actual requirements.
In one embodiment, step 202 further comprises:
step 302, determining a plurality of feature points in the image to be recognized by using a FAST detection algorithm.
The edge server has low computing capacity, and in order to improve the defect identification speed, a feature vector set to be identified can be determined through feature detection and feature description.
Specifically, determining a plurality of feature points of the image to be identified by using a FAST corner detection algorithm includes:
step 3021, excluding non-corner points.
Referring to fig. 5, fig. 5 is a schematic diagram of a corner criterion. Specifically, step 3021 includes:
step 30211, recording IpAnd (3) drawing a circle by taking the P as the center of the circle and taking 3 pixels as the radius for the gray value of any pixel point P, and numbering the pixels on the circumference in a clockwise sequence to obtain the pixels with the numbers of 1 to 16.
Step 30212, examining the number of the pixels 1 and 9 on the circumference, and if the gray values of the number of the pixels 1 and 9 are both in the range of Ip+εdAnd Ip-εdIn between, the possibility of P being a corner point can be excluded, where εdIs a set threshold;
step 30213, if the possibility that P is an angular point cannot be eliminated according to step 30212, then examining the pixels numbered 5 and 13 on the circumference, and if the gray values of at least 3 pixels are not smaller than I among the pixels numbered 1, 9, 5, and 13p+εdOr not more than Ip-εdThe possibility of P being a corner is excluded.
And (3) executing step 30212 and step 30213 on each pixel point of the image to be recognized to exclude non-corner points in the image to be recognized, so as to determine a plurality of pixel points to be processed, wherein none of the pixel points to be processed are corner points.
And step 3022, determining corner points in the plurality of pixel points to be processed.
And for any pixel point to be processed, determining a plurality of circumferential pixel points of the any pixel point to be processed by taking the any pixel point to be processed as a circle center, and if n pixel points in the plurality of circumferential pixel points meet the formula (1), determining the any pixel point to be processed as a corner point.
|I(i)-Ip|>εd (1)
Wherein, I (I) represents the gray value of the ith circumferential pixel point of any pixel point P to be processed, IpN may be set to 9, which is the gray level of the pixel P.
And (3) executing step 3021 and step 3022 on each pixel point to be processed to determine a plurality of corner points of the image to be recognized, and using the plurality of corner points as a plurality of feature points of the image to be recognized.
Step 303, determining the feature vector to be recognized of each feature point by using an ORB algorithm, and determining a feature vector set to be recognized based on the feature vector to be recognized of each feature point.
The ORB algorithm has rotation invariance, and the feature points are described by binary descriptors, so that the ORB algorithm conforms to the computing capacity of the edge server. After a plurality of feature points of the image to be recognized are determined, a feature vector set to be recognized of the image to be recognized is determined by adopting a feature description algorithm in an ORB algorithm.
Specifically, step 303 includes:
step 3031, determining a circular area by taking any characteristic point Q as a circle center and d as a radius, selecting t groups of point pairs in the circular area, wherein each group of point pairs is marked as: q1(u1,v1)，Q2(u2,v2)，……，Qt(ut,vt) (ii) a t may be set to 512.
Step 3032, determining the reference value T of each group of point pairs according to the formula (2), and determining the descriptor according to the reference value T of each group of point pairs.
Wherein the content of the first and second substances,
is a point uiIs determined by the gray-scale value of (a),
is point viThe gray value of (a).
For example, T (Q)1(u1,v1))＝0,T(Q2(u2,v2))＝1,T(Q3(u3,v3))＝1,……,T(Qt(ut,vt) 0), the descriptor is: 011, … …, 0.
Step 3033, calculating the position of the center of mass in the neighborhood S of the characteristic point Q.
The moments of the domain S are defined according to equation (3).
Wherein, I (x, y) is the gray expression of the pixel (x, y), and (x, y) belongs to S.
The centroid position of the moment is determined by equation (4).
And step 3034, determining the direction angle of the feature vector to be identified.
The direction angle of the feature vector to be identified is determined by formula (5).
And determining the feature vector to be identified of each feature point through the process.
The method comprises the steps of determining a plurality of feature points of an image to be recognized by adopting a FAST corner detection algorithm, determining a feature vector to be recognized of each feature point by adopting a feature description algorithm in an ORB algorithm, and determining a feature vector set to be recognized according to the feature vector to be recognized of each feature point.
In one embodiment, step 103 comprises:
step 401, determining a similarity corresponding to any reference feature vector set based on any reference feature vector set and the feature vector set to be recognized.
And the similarity is used for reflecting the number of the same feature vectors in any reference feature vector set and the feature set to be identified.
Specifically, the reference feature vector set includes a plurality of reference feature vectors, and the feature vector set to be recognized includes a plurality of feature vectors to be recognized; the feature vector to be identified is a feature vector of the feature point. The reference feature vectors and the feature vectors to be identified correspond one to one. In one embodiment, step 401 comprises:
step 4011, determining a hamming distance between the any reference feature vector and a feature vector to be identified corresponding to the any reference feature vector.
Specifically, the coordinates of the reference feature point corresponding to any reference feature vector are the same as the coordinates of the feature point of the feature vector to be identified corresponding to any reference feature vector. The feature vector to be identified is represented by a binary string of 0 and 1, and similarly, the reference feature vector is also represented by a binary string of 0 and 1, so that the hamming distance between the feature vector to be identified and the reference feature vector can be determined.
Step 4012, determining the number of hamming distances smaller than the preset distance, and using the determined number as the similarity corresponding to any one of the reference feature vector sets.
Specifically, the preset distance may be set to 25, and if a hamming distance between any reference feature vector and the feature vector to be identified corresponding to the reference feature vector is smaller than the preset distance, it is determined that the reference feature vector is the same as the feature vector to be identified, and the number of the same reference feature vector and the number of the feature vector to be identified are determined to obtain the similarity.
And 402, determining a recognition result according to the similarity corresponding to any reference feature vector set and a preset threshold.
Wherein the identification result comprises non-defective information or defective information, and the non-defective information indicates that the power equipment has no missing part. The non-defective information comprises a non-defective mark and a compressed image to be identified. The defect information comprises defect identification and defect pixel point coordinates, and the defect information also comprises a compressed image to be identified.
Specifically, step 402 includes:
step 4021, if the similarity corresponding to any one of the reference feature vectors is smaller than the preset threshold, determining that the identification result is non-defective information.
And the preset threshold is equal to 90% of the number of all pixel points of the image to be identified.
Specifically, if the similarity corresponding to any one of the reference feature vectors is smaller than the preset threshold, it indicates that the image to be recognized is not similar to the reference image, and since the reference image is an image of the power device when the part is missing, it may be determined that the image to be recognized is not defective (the power device corresponding to the image to be recognized is not missing the part).
Step 4022, if the similarity corresponding to any one of the reference feature vectors is greater than or equal to the preset threshold, taking the reference feature vector set corresponding to the similarity greater than or equal to the preset threshold as a target feature vector set, and determining defect information according to the target feature vector set.
Specifically, if the similarity corresponding to any one of the reference feature vectors is greater than or equal to the preset threshold, it indicates that the image to be recognized is similar to the reference image, and therefore, it may be determined that the image to be recognized has a defect (the electric device corresponding to the image to be recognized has a part missing).
Determining defect information according to the target feature vector set, including:
determining the coordinates of defective pixel points according to the target feature vector set, wherein the coordinates of the defective pixel points are the coordinates of pixel points corresponding to the same feature vectors in the target feature vector set, and the same feature vectors are the same feature vectors in the target feature vector set as the feature vector set to be identified; and taking a second image identifier of the target feature vector set as a defect identifier. The defect identification is used for the cloud server to determine the defect grade and the defect description information of the image to be recognized.
Specifically, if the similarity corresponding to the reference feature vector S1 is greater than or equal to the preset threshold, S1 is taken as a target feature vector, the reference feature vectors a1, a2, … … and a20 in S1 are the same as b1, b2, … … and b20 in the feature vector to be identified, pixel points corresponding to a1, a2, … … and a20 are taken as defective pixel points, the defective pixel points are also pixel points corresponding to b1, b2, … … and b20, and the coordinates of the pixel points corresponding to a1, a2, … … and a20 are the coordinates of the defective pixel points.
And each reference vector set is configured with a second image identifier, and the second image identifier is used for reflecting the defect grade and the defect description information of the reference image corresponding to the feature vector set. For example, the reference image P is an image when a pin of the power device is missing, the defect level corresponding to the pin missing is 2 level, the defect description information is "pin missing", the second image identification of the reference image P is determined as t1 according to the defect level and the defect description information, that is, the defect level and the defect description information can be determined according to t 1.
If the similarity corresponding to the reference feature vector S1 is greater than or equal to the preset threshold, the second image identifier t1 corresponding to S1 is used as a defect identifier, and the defect identifier t1 and the coordinates of the defect pixel (the coordinates of the pixel corresponding to a1, a2, … …, a 20) are used as defect information.
The edge server sends the defect information to the cloud server, the cloud server can determine the defect type and the defect description information according to the defect identification t1 in the defect information, and the defect position can be marked according to the coordinates of the defect pixel points.
According to the defect identification method for the power equipment, the edge server does not store the reference image, does not extract the reference feature vector set of the reference image, directly obtains the reference feature vector library corresponding to the image to be identified, only needs to extract the feature vector set to be identified of the image to be identified, and has low requirement on the storage capacity of the edge server;
the edge server extracts the feature vector set to be recognized of the image to be recognized by adopting an FAST algorithm and an ORB algorithm, the algorithm is not complex, the required time is short, the requirement on the computing capacity of the edge server is not high, and the defect recognition speed is improved;
the defect identification efficiency and cost are comprehensively considered, the edge servers and the machine nests can be set to be one-to-one, or one edge server bears defect identification tasks corresponding to the two machine nests, the task of the edge server for identifying the defects is not heavy, network congestion cannot be caused, and the identification result of defect detection can be obtained in real time;
the machine nests are fixedly arranged in the transformer substation, the edge server corresponds to one or two machine nests, and the edge server only needs to query the reference feature vector library corresponding to the image to be identified in the reference feature vector library corresponding to the image of the fixed power equipment, so that the time for the edge server to acquire the reference feature vector library corresponding to the image to be identified is reduced, and the defect identification efficiency is improved;
the transmission distance between the edge server and the machine nest is short, the time required for the machine nest to transmit the image to be identified to the edge server is short, and the defect identification efficiency is improved;
the bandwidth between the edge server and the cloud server is wide, the edge server transmits the identification result to the cloud server and network congestion cannot occur, and the cloud server can obtain the identification result in real time;
in summary, the defect identification method for the power equipment improves the defect identification efficiency of the power equipment, realizes real-time defect identification, improves the timeliness of defect finding and processing on the whole, and reduces the potential safety hazard of circuit operation.
In one embodiment, as shown in fig. 6, there is provided a defect identifying apparatus of an electric power device, including:
the image receiving module is used for receiving an image to be identified sent by a nest, wherein the image to be identified is obtained by shooting power equipment by an unmanned aerial vehicle configured by the nest;
the characteristic vector set determining module is used for acquiring a reference characteristic vector library corresponding to the image to be recognized and determining a characteristic vector set to be recognized based on the image to be recognized, wherein the reference characteristic vector library comprises reference characteristic vector sets of a plurality of reference images, and the reference images are images of the power equipment when parts are missing;
and the identification result determining module is used for determining an identification result based on the reference feature vector library and the feature vector set to be identified and sending the identification result to the cloud server.
For specific definition of the defect identification apparatus of the electrical equipment, reference may be made to the above definition of the defect identification method of the electrical equipment, and details are not described herein again. The modules in the defect identifying device of the power equipment can be wholly or partially 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. 7. 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 a method of defect identification for an electrical power device. 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. 7 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 provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
the method comprises the steps that an image to be identified is sent by a receiver nest, wherein the image to be identified is obtained by shooting power equipment through an unmanned aerial vehicle configured by the nest;
acquiring a reference feature vector library corresponding to the image to be recognized, and determining a feature vector set to be recognized based on the image to be recognized, wherein the reference feature vector library comprises reference feature vector sets of a plurality of reference images, and the reference images are images of the power equipment when parts are missing;
and determining a recognition result based on the reference feature vector library and the feature vector set to be recognized, and sending the recognition result to the cloud server.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
the method comprises the steps that an image to be identified is sent by a receiver nest, wherein the image to be identified is obtained by shooting power equipment through an unmanned aerial vehicle configured by the nest;
acquiring a reference feature vector library corresponding to the image to be recognized, and determining a feature vector set to be recognized based on the image to be recognized, wherein the reference feature vector library comprises reference feature vector sets of a plurality of reference images, and the reference images are images of the power equipment when parts are missing;
and determining a recognition result based on the reference feature vector library and the feature vector set to be recognized, and sending the recognition result to the cloud server.
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 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-mentioned embodiments 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 defect identification method for electric equipment is applied to an edge server, and comprises the following steps:
the method comprises the steps that an image to be identified is sent by a receiver nest, wherein the image to be identified is obtained by shooting power equipment through an unmanned aerial vehicle configured by the nest;
acquiring a reference feature vector library corresponding to the image to be recognized, and determining a feature vector set to be recognized based on the image to be recognized, wherein the reference feature vector library comprises reference feature vector sets of a plurality of reference images, and the reference images are images of the power equipment when parts are missing;
and determining a recognition result based on the reference feature vector library and the feature vector set to be recognized, and sending the recognition result to the cloud server.
2. The method according to claim 1, characterized in that the image to be recognized is configured with a first image identification; the obtaining of the reference feature vector library corresponding to the image to be recognized includes:
in the memory of the edge server, a reference feature vector library corresponding to the image to be recognized is obtained based on the first image identifier, wherein the reference feature vector library is determined by the cloud server in advance based on a plurality of reference images of the power equipment and is sent to the edge server.
3. The method of claim 1, wherein the determining a set of feature vectors to be recognized based on the image to be recognized comprises:
determining a plurality of characteristic points in the image to be identified by adopting a FAST detection algorithm;
and determining the feature vector to be recognized of each feature point by adopting an ORB algorithm, and determining a feature vector set to be recognized based on the feature vector to be recognized of each feature point.
4. The method according to claim 1, wherein the determining a recognition result based on the reference feature vector library and the feature vector set to be recognized comprises:
determining similarity corresponding to any reference feature vector set based on any reference feature vector set and the feature vector set to be identified, wherein the similarity is used for reflecting the number of identical feature vectors in any reference feature vector set and the feature vector set to be identified;
and determining the recognition result according to the similarity corresponding to any reference feature vector set and a preset threshold.
5. The method according to claim 4, wherein the reference feature vector set comprises a plurality of reference feature vectors, the feature vector set to be identified comprises a plurality of feature vectors to be identified, and the plurality of reference feature vectors and the plurality of feature vectors to be identified correspond one to one; the determining the similarity corresponding to any reference feature vector set based on any reference feature vector set and the feature vector set to be recognized includes:
determining a Hamming distance between any reference characteristic vector and a characteristic vector to be identified corresponding to any reference characteristic vector;
and determining the number of Hamming distances smaller than a preset distance, and taking the determined number as the corresponding similarity of any reference feature vector set.
6. The method according to claim 5, wherein the recognition result includes defect-free information or defect information, and the determining the recognition result by the similarity corresponding to any one of the reference feature vector sets and a preset threshold comprises:
if the similarity corresponding to any reference feature vector is smaller than the preset threshold, determining that the identification result is non-defective information;
if the similarity corresponding to any reference feature vector is greater than or equal to the preset threshold, taking a reference feature vector set corresponding to the similarity greater than or equal to the preset threshold as a target feature vector set, and determining defect information according to the target feature vector set.
7. The method of claim 6, wherein the defect information comprises defect identifiers and defect pixel coordinates, and each reference vector set is configured with a second image identifier; determining defect information according to the target feature vector set, including:
determining the coordinates of defective pixel points according to the target feature vector set, wherein the coordinates of the defective pixel points are the coordinates of pixel points corresponding to the same feature vectors in the target feature vector set, and the same feature vectors are the same feature vectors in the target feature vector set as the feature vector set to be identified;
and taking a second image identifier of the target feature vector set as a defect identifier, wherein the defect identifier is used for the cloud server to determine the defect level and the defect description information of the image to be recognized.
8. A defect identifying apparatus for an electric power device, comprising:
the image receiving module is used for receiving an image to be identified sent by a nest, wherein the image to be identified is obtained by shooting power equipment by an unmanned aerial vehicle configured by the nest;
the characteristic vector set determining module is used for acquiring a reference characteristic vector library corresponding to the image to be recognized and determining a characteristic vector set to be recognized based on the image to be recognized, wherein the reference characteristic vector library comprises reference characteristic vector sets of a plurality of reference images, and the reference images are images of the power equipment when parts are missing;
and the identification result determining module is used for determining an identification result based on the reference feature vector library and the feature vector set to be identified and sending the identification result to the cloud server.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method for defect identification of an electrical power device according to any one of claims 1 to 7.
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 for defect identification of an electrical power device according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110903217.3A CN113658120A (en) | 2021-08-06 | 2021-08-06 | Defect identification method and device for electric power equipment, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110903217.3A CN113658120A (en) | 2021-08-06 | 2021-08-06 | Defect identification method and device for electric power equipment, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113658120A true CN113658120A (en) | 2021-11-16 |
Family
ID=78478590
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110903217.3A Pending CN113658120A (en) | 2021-08-06 | 2021-08-06 | Defect identification method and device for electric power equipment, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113658120A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116309554A (en) * | 2023-05-12 | 2023-06-23 | 广东奥普特科技股份有限公司 | Defect detection network construction and defect detection method, device and equipment |
-
2021
- 2021-08-06 CN CN202110903217.3A patent/CN113658120A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116309554A (en) * | 2023-05-12 | 2023-06-23 | 广东奥普特科技股份有限公司 | Defect detection network construction and defect detection method, device and equipment |
| CN116309554B (en) * | 2023-05-12 | 2023-08-22 | 广东奥普特科技股份有限公司 | Defect detection network construction and defect detection method, device and equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110703800A (en) | Unmanned aerial vehicle-based intelligent identification method and system for electric power facilities | |
| EP4072173A1 (en) | Data transmission method and device | |
| CN106547814A (en) | A kind of power transmission line unmanned machine patrols and examines the structuring automatic archiving method of image | |
| CN105739512A (en) | Unmanned aerial vehicle automatic tour inspection system and method | |
| CN111739184A (en) | Power transmission line inspection system based on power transmission line tower pole | |
| CN112751414A (en) | Transmission line inspection system and method based on 5G link | |
| CN112364843A (en) | Plug-in aerial image target positioning detection method, system and equipment | |
| CN111222375A (en) | Vehicle driving track generation method, system and server | |
| CN113506261A (en) | Road disease detection method, device, equipment and system | |
| CN113658120A (en) | Defect identification method and device for electric power equipment, computer equipment and storage medium | |
| CN115562340A (en) | Distribution network line unmanned aerial vehicle inspection fault discrimination system | |
| CN111065044B (en) | Big data based data association analysis method and device and computer storage medium | |
| CN111046705A (en) | Image recognition method, device and system and computing equipment | |
| CN114157829A (en) | Model training optimization method and device, computer equipment and storage medium | |
| CN113359829B (en) | Unmanned aerial vehicle power plant intelligent inspection method based on big data | |
| CN113252045B (en) | Device positioning method and device, electronic device and readable storage medium | |
| CN114564049A (en) | Unmanned aerial vehicle wide area search device and method based on deep learning | |
| CN113326752B (en) | Unmanned aerial vehicle-based photovoltaic power station identification method and system | |
| CN114037895A (en) | Unmanned aerial vehicle pole tower inspection image identification method | |
| CN112859109A (en) | Unmanned aerial vehicle panoramic image processing method and device and electronic equipment | |
| CN114092771A (en) | Multi-sensing data fusion method, target detection device and computer equipment | |
| CN116912716A (en) | Target positioning method, target positioning device, electronic equipment and storage medium | |
| CN111582296A (en) | Remote sensing image comprehensive matching method and device, electronic equipment and storage medium | |
| Wang et al. | Unmanned aerial vehicle (UAV) vision-based detection and location of power towers for transmission line maintenance | |
| Wang et al. | Recognition and locating of damaged Poles in distribution network through images shot by unmanned aerial vehicle (UA V) |
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 |