CN109712116B - Fault identification method for power transmission line and accessories thereof - Google Patents

Abstract

The application discloses a fault identification method for a power transmission line and accessories thereof, which comprises the following steps: forming a power transmission line image by extracting and connecting a plurality of line segments shot with the power transmission line image; after identifying the tower in the image of the power transmission line, removing the image of the tower area; and identifying the power transmission line and the power transmission line accessory in the remaining image area, and carrying out defect inspection on the power transmission line accessory and the power transmission line. Compared with the prior art, the method for line characteristic connection based on matrix operation solves the defects of low efficiency and poor fitting effect of solving a large number of line characteristic connection problems by using circulation, and overcomes the problem that the photographed power transmission line cannot be automatically identified in the prior art by respectively identifying faults of the obtained power transmission line image and the power transmission line accessory image after performing image semantic segmentation on the power transmission line and accessories thereof, thereby saving a large amount of human resources.

Description

Fault identification method of transmission line and its accessories

technical field

The present application relates to the technical field of image recognition, in particular to a method for fault recognition of power transmission lines and their accessories.

Background technique

Due to the possibility of faults in the transmission line due to the influence of the natural environment, regular inspections of the transmission line are required. For a long time, the inspection of transmission lines has often been completed by manual inspection, but this line inspection method not only has low inspection efficiency, but also has certain safety risks.

In order to solve the above problems, unmanned aerial vehicle technology is used in the prior art to replace manual transmission line inspection. However, when inspecting transmission lines with existing technologies, it was found that drones can only take photos, and a large amount of manpower is still required to check the pictures taken by drones to find faults in transmission lines.

Contents of the invention

The technical problem to be solved by the embodiment of the present application is how to identify the fault of the transmission line, so as to save human resources.

In order to solve the above problems, an embodiment of the present application provides a fault identification method for a transmission line and its accessories, which is suitable for execution in a computing device, and at least includes the following steps:

Collect and shoot the color image of the power transmission line, and after converting the color image data into a grayscale image, extract all the line segments in the grayscale image one by one as intermittent line segments, and convert all the line segments that can be connected between the two Fit the intermittent line segments in pairs to obtain the transmission line image;

According to the preset image area division number, after dividing the power transmission line image into multiple image areas of the same size, each of the image areas exceeds the first preset length and the included angle with the horizontal axis is at a preset Set the line segment of the included angle interval, mark it as the line segment to be processed, and divide the image area whose number of the line segment to be processed exceeds the preset number into the tower area, and divide the plurality of the image areas that are not divided into the tower area Image regions, forming the first set of regions to be processed;

Extracting the line segment whose concentrated length, end point, and horizontal angle with the horizontal axis of the first area to be processed all reach a preset range, and after obtaining the transmission line image of the transmission line, according to the attachment template of the transmission line attachment, in the The first area to be processed is concentrated, and the image of the component parts of the power transmission line accessories located below the power transmission line image is searched;

After combining the plurality of images of the found components, according to the combination result between the images of the plurality of components, it is judged whether there is a defect in the attachment of the transmission line, and all the components are removed The first set of regions to be processed after the image is marked as the second set of regions to be processed;

After the line segments that do not form the power line image are collected in the second area to be processed, and marked as line segments to be identified, the line segments to be identified are extended by the seed growth method, and extended to exceed the second preset length And the to-be-recognized line segment judged to be the foreground by the background filter is marked as a broken strand of the transmission line.

Further, it also includes:

According to the endpoint distance between each line segment in the second area to be processed, cluster all the line segments in the second area to be processed into a plurality of line feature clusters, which will be judged by the background filter The linear feature blobs that are foreground are marked as foreign object images.

Further, extracting all the line segments in the grayscale image one by one as intermittent line segments, and fitting the intermittent line segments that can be connected in pairs to obtain the transmission line image, specifically:

According to Gestalt's law, perform matrix transformation on all the discontinuous line segments, after obtaining the approximation matrix, collinearity matrix and continuity matrix of all the discontinuous line segments, the approximation matrix, collinearity matrix and continuity matrix Take the logical sum of the elements at the same position in to get the adjacency matrix;

According to the graph theory method, after the adjacency matrix is divided into a plurality of connected domains, the endpoints of the intermittent line segments corresponding to each of the elements in the connected domain are fitted by a weighted least square method to obtain The transmission line image.

Further, extract the line segment whose concentrated length, end point, and horizontal angle with the horizontal axis of the second area to be processed all reach a preset range, and obtain the transmission line image of the transmission line, specifically:

Extracting all line segments whose length is greater than a second preset length in the second to-be-processed area set as the first set of line segments;

Extracting the first line segment set, all line segments whose minimum distance from the end point to the boundary of the transmission line image is less than a preset distance are used as the second line segment set;

Extracting, as the transmission line image, contours formed by all line segments in the second set of line segments whose included angles with the horizontal axis are smaller than a preset included angle.

Further, the power transmission line accessories at least include a shockproof hammer.

Further, according to the accessory template of the transmission line accessory, in the first area to be processed, the image of the components of the transmission line accessory located under the transmission line image is searched, specifically:

According to the kAS semantic model, after extracting the 2AS template and the 3AS template of the anti-vibration hammer, extract a plurality of 2AS images from the first area to be processed according to the 2AS template in the area below the power line image, and After combining each two of the multiple 2AS images into a first 3AS image, filter the multiple first 3AS images through background filtering to obtain multiple second 3AS images, and combine them with the Each of the second 3AS images that are successfully matched by the 3AS template serves as a component image of the anti-vibration hammer.

Further, the multiple first 3AS images are filtered through the background to obtain multiple second 3AS images, specifically:

After each of the first 3AS images is framed, the first 3AS images whose pixels are non-foreground pixels are filtered out through the Grabcut algorithm to obtain multiple second 3AS images.

Further, after combining the found images of the components, it is judged whether there is a defect in the transmission line accessories, specifically:

Combining a plurality of component images in pairs according to a preset distance range, and judging the component component images combined in pairs as complete anti-vibration hammers, and determining the component component images that cannot be combined as There are defects in the anti-vibration hammer.

Further, it also includes:

Judging whether there is a pair of line segment groups that are equal in length, parallel to each other, and capable of vertically projecting more than 80% of themselves to each other in the first area to be processed; wherein, if there are, two connecting line segments are generated, Form a quadrilateral with the set of line segments.

Further, the image area is a rectangular area;

The first preset length is 1/3 min(a, b); wherein, a is the length of the rectangular area, and b is the width of the rectangular area.

Implementing the embodiment of the present application has the following beneficial effects:

A fault identification method for a transmission line and its accessories provided by an embodiment of the present application includes: forming a transmission line image by extracting and connecting multiple line segments with pictures of the transmission line; after identifying the towers in the transmission line image, removing Image of the tower area; identification of transmission lines and transmission line accessories in the remaining image area, and defect inspection of transmission line accessories and transmission lines. Compared with the prior art, this application adopts the method of fault identification for the obtained transmission line image and transmission line attachment image after image semantic segmentation of the transmission line and its accessories, which overcomes the inability to identify the faults in the prior art The problem of automatic fault identification of the captured transmission lines has achieved the effect of automatic identification of transmission line faults.

Description of drawings

Fig. 1 is a schematic flow chart of a fault identification method for a power transmission line and its accessories provided by an embodiment of the present application;

Fig. 2 is a schematic flow diagram of a system for semantic segmentation of transmission lines and their accessories images provided by another embodiment of the present application;

Fig. 3 is a schematic diagram of Gestalt's law of perception;

Fig. 4 is a schematic diagram of line feature connection effect;

Figure 5 is a schematic diagram of 2AS and 3AS;

Figure 6 is a schematic diagram of extracting 2AS and 3AS templates;

Fig. 7 is a schematic diagram of the anti-vibration hammer 2AS and the transmission line;

Fig. 8 is a schematic diagram of identifying broken strands by the seed growth method;

Figure 9 is a speed comparison diagram of two kinds of line feature connections.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the application with reference to the drawings in the embodiments of the application. Apparently, the described embodiments are only some of the embodiments of the application, not all of them. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the scope of protection of this application.

Please refer to Figure 1 and Figure 3-9.

Referring to FIG. 1 , it is a schematic flowchart of a fault identification method for a transmission line and its accessories provided by an embodiment of the present application. As shown in FIG. 1 , the task processing method includes steps S11 to S15. Each step is as follows:

Step S11, collect and shoot the color image of the transmission line, and after converting the color image data into a grayscale image, extract all the line segments in the grayscale image one by one as intermittent line segments, and make the intermittent line segments that can be connected between two pairs The line segments are fitted in pairs to obtain the transmission line image.

Step S12, after dividing the transmission line image into multiple image areas of the same size according to the preset number of image area divisions, divide each image area, which exceeds the first preset length and has an included angle with the horizontal axis at a preset Mark the line segment in the included angle interval as the line segment to be processed, and divide the image area with the number of line segments to be processed exceeding the preset number into the tower area, and then divide the multiple image areas that are not divided into the tower area into the first area to be processed set.

Step S13, extracting the line segment whose concentrated length, end point, and horizontal angle with the horizontal axis of the first area to be processed all reach the preset range, and after obtaining the transmission line image of the transmission line, according to the attachment template of the transmission line attachment, in the first to-be-processed area Processing area concentration, finds component part images of transmission line accessories that lie below the transmission line image.

Step S14, after combining the found images of multiple components, according to the result of combining the images of multiple components, it is judged whether there is a defect in the transmission line accessories, and the first image after removing all the images of the components is determined. The set of regions to be processed, marked as the second set of regions to be processed.

Step S15: After marking the line segments that do not form the power line image in the second area to be processed as line segments to be identified, the line segments to be identified are extended by the seed growth method, and the line segments are extended to exceed the second preset length and filtered by the background The to-be-recognized line segment judged by the detector as the foreground is marked as a broken strand of the transmission line.

For step S11, specifically, according to Gestalt's law, perform matrix transformation on all discontinuous line segments, after obtaining the approximation matrix, collinearity matrix and continuity matrix of all discontinuous line segments, the approximation matrix, collinearity matrix and continuity matrix The elements in the same position in the matrix are logically summed to obtain the adjacency matrix; according to the graph theory method, after the adjacency matrix is divided into multiple connected domains, the weighted least squares method is used to fit the segment corresponding to each element in the connected domain. The end points of the continuation line segment are obtained to obtain the transmission line image.

In this embodiment, the color picture with the transmission line is first converted into a grayscale picture, and then the LSD algorithm is called to extract all line segments from the picture. Due to the weak anti-noise ability of the LSD algorithm, a complete linear profile often generates a lot of broken and separated discontinuous line segments. Line segments are processed.

As shown in Figure 3, approximation is obtained, collinearity is obtained, and continuity is obtained. Among them, is the distance between the endpoints of the two line segments close to each other, is the given threshold, is the angle between the two line segments and the horizontal axis, is the given threshold, is the vertical distance from point S ₂ to L ₁ , and is the given threshold set threshold.

Since LSD can extract thousands of line segments from a high-resolution image, in this embodiment, the matrix transformation is used to fit the thousands of line segments, specifically:

Express the i-th line by its endpoint coordinates (x ₁ ⁽ⁱ⁾ , y ₁ ⁽ⁱ⁾ , x ₂ ⁽ⁱ⁾ , y ₂ ⁽ⁱ⁾ ), define

After defining Y ₁ , X ₂ , and Y ₂ similarly to X ₁ , calculate

D _s1,s2 =min(D ₁ ,D ₂ ,D ₃ ,D ₄ )

in,

D _{s1, s2} are distances between endpoints of two line segments that are close to each other among the plurality of line segments.

夹角阈值设定为θ_thr1＝[θ_thr1]_n×n。多条线段中每两条线段之间的近似性可表示为：In this embodiment, the endpoint distance threshold between two line segments is set as D _thr1 =[d _thr1 ] _n×n , and the angle between the line segment and the horizontal axis is

The included angle threshold is set as θ _thr1 =[θ _thr1 ] _n×n . The approximation between every two of the multiple line segments can be expressed as:

in,

In this embodiment, the given threshold for judging the proximity is D _thr2 =[d _thr2 ] _n×n . Therefore, the Gestalt perception law of multiple line segments can be expressed in the form of a matrix, specifically:

D _s1,s2 <D _thr1

|θ-θ ^T |<θ _thr1

D _v < D _thr2

After obtaining the approximation matrix M _a , the collinearity matrix M _b and the continuity matrix M _c from the above three matrix inequalities, the logical sum operation is performed on the elements in the same position of the three Boolean matrices to obtain the leading matrix M, specifically expressed for:

M = M _a · M _b · M _c .

Using the method of graph theory, the connected domain can be extracted from the adjacency matrix, and the elements of each connected domain represent the sequence numbers of the line segments that can be connected with each other. Therefore, through the weighted least squares method, the endpoints of the line segments that can be connected to each other are fitted, and the weight is proportional to the length of the corresponding line segment, and finally the complete line features after the connection are obtained. The effect is shown in Figure 4.

For step S12, specifically, after dividing the transmission line image into a plurality of rectangular image areas of the same size according to the preset number of image area divisions, in each image area, a length exceeding 1/3min(a, b), and the line segment whose angle with the horizontal axis is in the preset angle interval is marked as a line segment to be processed, and after the image area whose number of line segments to be processed exceeds the preset number is divided into the tower area, it will not be divided into the tower area A plurality of image regions of the region form a first set of regions to be processed.

Among them, a is the length of the image area, and b is the width of the image area.

In this embodiment, since the transmission tower has the characteristics of a densely staggered steel structure, the area of the transmission tower is located according to the length, angle, and spatial arrangement information of the line feature.

Specifically, according to the angle interval shown in Table 1, after dividing the angle between each line segment and the horizontal axis in the transmission line image into four groups A, B, C and D, the transmission line image is divided into 8X6 regions, and Determine whether there are more than three line segments that exceed 1/3min(a, b) and belong to the two groups B and C in each area, and if there are more than three. Then the area is divided into the tower area. Among them, Table 1 is:

category A B C D. angle (-5,5) (5,85) (-85,-5) (-90,-85),(85,90)

For step S13, specifically, extract all line segments whose length is greater than the second preset length in the second area to be processed as the first line segment set; extract the first line segment set, the minimum distance between the endpoint and the boundary of the transmission line image is less than the preset Set all the line segments of the distance as the second line segment set; extract the outline formed by all the line segments in the second line segment set and the horizontal axis whose angle is smaller than the preset angle, and use it as the transmission line image, and extract the transmission line accessories according to the kAS semantic model After the 2AS template and 3AS template, according to the 2AS template, extract multiple 2AS images from the first area to be processed, located in the area below the power line image, and combine the 2AS images of each two of the multiple 2AS images into the first After the 3AS image, multiple first 3AS images are filtered through the background to obtain multiple second 3AS images, and each second 3AS image that successfully matches the 3AS template is used as a component image of the transmission line accessory. Multiple component images are combined in pairs according to the preset distance range, and the combined component images are determined as complete transmission line accessories, and the component component images that cannot be combined are determined as defective transmission line accessories. After the fault judgment of the transmission line accessory is completed, the image of the transmission line accessory is removed from the first area set to be processed.

It should be noted that, in this embodiment, the power transmission line accessories include shockproof hammers and spacer rods.

In this embodiment, according to the three characteristics that the transmission line has a long length, is basically horizontal, and will extend to the boundary of the image, the transmission line is extracted from the first area to be processed, and the method of obtaining the line segments that form the transmission line is as follows: by obtaining Line segment length l _i >l _thr is used as the first line segment set; the line segment d _i <d _thr3 in the first line segment set is obtained as the second line segment set; the line segment in the second line segment set θ _i <θ _thr2 is obtained to form a transmission line .

where l _i is the length of the line feature, d _i is the shortest distance from the endpoint of the line segment to the image boundary, and θ _i is the angle between the line feature and the horizontal axis. l _thr , d _thr3 , θ _thr2 are preset thresholds.

In this embodiment, the identification of the anti-vibration hammer or spacer is applied to the kAS semantic model, where k represents the number of line segments of the cluster. Taking the anti-vibration hammer as an example, the sketches of 2AS and 3AS are shown in Figure 5, where (a) is the sketch of 2AS, and (b) is the sketch of 3AS.

In this embodiment, the vector of 2AS is P _2as =(l ₁ ,l ₂ ,α,θ). where l ₁ is the length of line segment L ₁ , l ₂ is the length of line segment L ₂ , and L ₁ is longer than L ₂ ; α is the angle between line segments L ₁ and L ₂ ; θ is the angle between L ₁ and the horizontal axis . The vector of 3AS is P _3as =(l ₁ ,l ₂ ,α ₁ ,α ₂ ). Among them, l ₃ is the length of line segment L ₃ , and L ₁ is longer than L ₃ , α ₁ is the angle between line segments L ₁ and L ₂ , and α ₂ is the angle between line segments L ₂ and L ₃ .

In this embodiment, the calculation method of the difference degree of two 2AS vectors is:

The calculation method of the difference degree of two 3AS vectors is:

Among them, ω _θ is an artificially given weight coefficient, which is used to balance the proportion of length difference and angle difference in the overall difference degree.

If D(a,b)<D _thr , it is considered that the two kAS are similar, where D _thr is a preset threshold.

In this embodiment, taking the anti-vibration hammer as an example, as shown in FIG. 6 , after extracting the 2AS template and the 3AS template of the anti-vibration hammer, multiple 2AS images are extracted from the first region to be processed. The extraction method of the 2AS image is shown in Figure 7. It needs to be screened by shape constraints, angle constraints, position constraints, and size constraints. The shape constraint is that the 2AS image extracted from the first area to be processed must be similar to the 2AS template, and the angle constraint The direction of the 2AS image M must be basically parallel to the power line, that is, the horizontal angle θ _t is equal to the horizontal angle θ _v , the position constraint is that the 2AS image must be within the preset distance threshold d ₁ below the power line image, and the size constraint is The length of the long side l of 2AS needs to be within a preset range.

It should be noted that since the LSD algorithm will extract two parallel line segments from the two edges of the transmission line, after extracting the line segments that make up the transmission line, the two parallel line segments need to be fitted by the least square method as A line segment, and record the distance between the two line segments, that is, the width w of the transmission line, and use this distance as the reference distance for the preset distance threshold adjustment. Since the width of the transmission line is proportional to the size of its attachments, the automatic adjustment of the preset distance threshold is realized.

In this embodiment, it is judged whether any two 2AS images screened out have shared edges, and if so, the two 2AS images with shared edges are combined into a 3AS, and compared with the 3AS template to see if they are similar, thereby identifying The 3AS image of the anti-vibration hammer, that is, the image of the components of the anti-vibration hammer.

Since the shooting position of the image may not be directly in front of the power transmission line accessories, it is impossible to linearly extract the contours on both sides of the power transmission line accessories. Therefore, in this embodiment, taking the anti-vibration hammer as an example, it shows the The contour completion method is specifically: judging whether there is a pair of line segment groups in the first area to be processed that are equal in length, parallel to each other, and capable of vertically projecting a part greater than 80% of itself to the other; wherein, if there is, Then two connecting line segments are generated to form a quadrilateral with the line segment group, thus completing the missing line features of the anti-vibration hammer.

In an embodiment, multiple first 3AS images are filtered through the background to obtain multiple second 3AS images, specifically:

After each first 3AS image is framed, the Grabcut algorithm is used to filter out the first 3AS images whose pixels are non-foreground pixels to obtain a plurality of second 3AS images.

Taking the background ditch as an example, since the background ditch is a non-foreground image, when each first 3AS image is framed, and the Grabcut algorithm is called to process each first 3AS image, the 3AS image whose image is the background ditch is excluded , so as to realize the filtering of complex background.

For step S14, specifically, multiple component images are combined in pairs according to the preset distance range, and after the combined component images are determined as complete transmission line accessories, the component component images that cannot be combined are determined as There are defective transmission line accessories, and after the determination is completed, all component images are removed from the first area set to be processed, and the remaining areas are marked as the second area set to be processed.

In this embodiment, taking the anti-vibration hammer as an example, after the 3AS image of the anti-vibration hammer is identified, the 3AS extracted by clustering, if the distance between the two 3AS is within the preset distance range, such as d ₂ shown in Figure 7 , then they will form a 3AS pair, and the position of the 3AS pair corresponds to a possible anti-vibration hammer, and the position of a single 3AS that has not been combined successfully corresponds to the possible existence of a shock-proof hammer with one side missing.

For step S15, in this embodiment, the seed growth method is adopted, as shown in FIG. 8 . Set the line segment to be identified as the seed, and grow the seed according to the rules of α _i,j <α _thr and d _si,sj <d _thr6 . Wherein α _i,j is the angle between line segment i and line segment j, d _si,sj is the shortest end-point distance between two line segments, α _thr and d _thr6 are preset thresholds. If a seed can grow to a length greater than 5 times the width of the power line according to the above rules, and it is judged as the foreground by the background filter, it is judged to be a broken strand of the power line.

The embodiment of the present application provides a fault identification method for a transmission line and its accessories. By extracting and connecting multiple line segments with pictures of the transmission line and connecting them, an image of the transmission line is formed; image; identify transmission lines and transmission line accessories in the remaining image area, and carry out defect inspection on transmission line accessories and transmission lines; compared with the prior art, this application adopts a line feature connection method based on matrix operation, which solves the problem of Using loops to solve a large number of line feature connection problems has the disadvantages of low efficiency and poor fitting effect. After image segmentation is performed on the tower area, transmission lines and transmission line accessories, fault detection is performed on the transmission lines and transmission line accessories respectively, which overcomes the In the existing technology, it is impossible to identify the faults of the captured transmission lines, which saves a lot of human resources, and uses the background filter to reduce the adverse interference of the complex background, and through the automatic completion of the contour line features, it improves the accuracy of the transmission lines. Recall for attachment identification.

See Figure 2.

Referring to FIG. 2, it is a schematic structural diagram of a system for semantic segmentation of transmission lines and their accessories images provided by an embodiment of the present application. In addition to the steps shown in FIG. 1, it also includes:

Step S16, according to the endpoint distance between each line segment in the second area to be processed, cluster all the line segments in the second area to be processed into a plurality of line feature clusters, which will be judged as foreground by the background filter Linear feature blobs are marked as foreign object images.

As an effect comparison of this embodiment, the original cycle-based method and this embodiment are used to perform semantic segmentation of the transmission line and its accessories on the photos with transmission lines taken by the UAV, as shown in Figure 9. comparison results. When the number of line segments contained in the 29 groups of line segments increases from 1 to 5052, the time-consuming of the two connection methods are tested respectively. The results show that when the number of online features is 5052, the original method takes 169.4 seconds. This embodiment proposes The method takes 1.984 seconds, which has an improvement of dozens of times. Wherein, n=32 means that among the 29 groups of line segments, a certain group includes 32 line segments.

An embodiment of the present application provides a fault identification method for a transmission line and its accessories, including: extracting and connecting multiple line segments with pictures of the transmission line to form a transmission line image; after identifying the towers in the transmission line image, removing the towers The image of the area; use the remaining image area as, and concentrate on identifying the transmission line and the transmission line accessories in the first area to be processed, and carry out defect inspection to the transmission line accessories and the power line; use the image of the transmission line accessories from the first area to be processed Concentrated removal is performed to obtain a second set of regions to be processed, and foreign object identification is performed in the second set of regions to be processed. Compared with the prior art, this application adopts a line feature connection method based on matrix operation, which solves the problems of inefficiency and poor fitting effect of solving a large number of line feature connection problems by using loops, and has a good understanding of tower areas, transmission lines and transmission lines. After the image segmentation of the line accessories, the fault detection is performed on the transmission line and the transmission line accessories respectively, which overcomes the problem that the fault identification of the captured transmission line cannot be performed in the prior art, thereby saving a lot of human resources, and using the background filter , which reduces the adverse interference of complex backgrounds, and improves the recall rate of transmission line accessory recognition through automatic completion of contour line features.

In addition, by dividing the line segments concentrated in the second area to be processed into multiple line feature groups, and judging the multiple line feature groups through the background filter, foreign object detection on the transmission line is realized.

Another embodiment of the present application also provides a configurable terminal device of a motion control device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processing When the computer executes the computer program, the fault identification method of the transmission line and its accessories as described in the above embodiments is realized.

The above description is the preferred implementation mode of the present application. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the application, some improvements and modifications can also be made, and these improvements and modifications are also considered For the scope of protection of this application.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented through computer programs to instruct related hardware, and the programs can be stored in a computer-readable storage medium. During execution, it may include the processes of the embodiments of the above-mentioned methods. Wherein, the storage medium may be a magnetic disk, an optical disk, a read-only memory (Read-Only Memory, ROM) or a random access memory (Random Access Memory, RAM) and the like.

Claims (10)

1. A fault identification method for a power transmission line and accessories thereof is characterized by at least comprising the following steps:

collecting and shooting a color image of the power transmission line, converting the data of the color image into a gray image, extracting all line segments in the gray image one by one to be used as intermittent line segments, and fitting the connectable intermittent line segments in pairs to obtain a power transmission line image;

dividing the image of the power transmission line into a plurality of image areas with the same size according to the preset image area dividing quantity, marking the line segments which exceed the first preset length and have the included angles with the horizontal axis in the preset included angle interval in each image area as the line segments to be processed, dividing the image areas of which the number of the line segments to be processed exceeds the preset number into tower areas, and then dividing the plurality of the image areas which are not divided into the tower areas into a first set of areas to be processed;

extracting line segments of which the lengths, end points and horizontal included angles with a horizontal axis of the first to-be-processed areas reach preset ranges, obtaining a power transmission line image of the power transmission line, and searching component part images of the power transmission line accessories positioned below the power transmission line image in the first to-be-processed areas according to accessory templates of the power transmission line accessories;

after the searched images of the components are combined, judging whether the power transmission line accessory has defects according to the combination result between every two images of the components, and marking the first to-be-processed area set with all the images of the components removed as a second to-be-processed area set;

and after marking the line segments which do not form the power transmission line image in the second to-be-processed area as the to-be-recognized line segments, extending the to-be-recognized line segments by a seed growing method, and marking the to-be-recognized line segments which are extended to exceed a second preset length and are judged as foreground by a background filter as broken strands of the power transmission line.

2. The method for identifying faults of the power transmission line and accessories thereof according to claim 1, further comprising:

and clustering all the line segments of the second to-be-processed region set into a plurality of linear feature clusters according to the end point distance between every two line segments in the second to-be-processed region set, and marking the linear feature clusters which are judged as foregrounds by the background filter as foreign body images.

3. The method for identifying faults of the power transmission line and the accessories of the power transmission line according to claim 1, wherein all line segments in the gray level image are extracted one by one to serve as intermittent line segments, and the intermittent line segments which can be connected between every two are fitted in pairs to obtain the power transmission line image, and specifically the method comprises the following steps:

performing matrix transformation on all the intermittent line segments according to the Gestalt law, acquiring an approximation matrix, a colinearity matrix and a continuity matrix of all the intermittent line segments, and then performing logical sum on elements at the same positions in the approximation matrix, the colinearity matrix and the continuity matrix to obtain an adjacent matrix;

according to a graph theory method, after the adjacency matrix is divided into a plurality of connected domains, the end points of the discontinuous line segments corresponding to the elements in the connected domains one by one are fitted through a weighted least square method, and the power transmission line image is obtained.

4. The method for identifying the faults of the power transmission line and the accessories of the power transmission line according to claim 1, wherein line segments of which the concentration length, the end points and the horizontal included angle with the horizontal axis of the second region to be processed reach preset ranges are extracted to obtain a power transmission line image of the power transmission line, and the method specifically comprises the following steps:

extracting all line segments with the length being larger than a second preset length in the second region set to be processed as a first line segment set;

extracting all line segments in the first line segment set, wherein the minimum distance from an end point to the boundary of the image of the power transmission line is smaller than a preset distance, and the all line segments are used as a second line segment set;

and extracting the outline formed by all the line segments with the included angles smaller than the preset included angle with the horizontal axis in the second line segment set as the power transmission line image.

5. The method of claim 1, wherein the transmission line accessory comprises at least a vibration damper.

6. The method according to claim 5, wherein component images of the transmission line accessory located below the transmission line image are searched for in the first set of areas to be processed according to an accessory template of the transmission line accessory, and the method specifically includes:

extracting a 2AS template and a 3AS template of the shockproof hammer according to a kAS semantic model, extracting a plurality of 2AS images from an area below the power transmission line image in a centralized manner from the first area to be processed according to the 2AS template, combining the 2AS images of every two of the 2AS images into a first 3AS image, filtering the background of the first 3AS images to obtain a plurality of second 3AS images, and taking the second 3AS images successfully matched with the 3AS templates AS component images of the shockproof hammer one by one.

7. The method according to claim 6, wherein the plurality of first 3AS images are filtered through a background to obtain a plurality of second 3AS images, and specifically:

after each first 3AS image is selected, the first 3AS image with pixels being non-foreground pixels is filtered out through a Grabcut algorithm, and a plurality of second 3AS images are obtained.

8. The method according to claim 5, wherein the step of determining whether the accessory of the power transmission line has a defect after combining the searched images of the components is specifically as follows:

combining the component images in pairs according to a preset distance range, judging the component images which are combined in pairs as complete anti-vibration hammers, and judging the component images which cannot be combined as the anti-vibration hammers with defects.

9. The method of claim 1, further comprising:

judging whether a pair of line segment groups which are equal in length, parallel to each other and capable of vertically projecting a part larger than eighty percent of the first to-be-processed area set to the opposite side exist in the first to-be-processed area set or not; if the line segment exists, two connecting line segments are generated, and a quadrangle is formed by the two connecting line segments and the line segment group.

10. The method according to any one of claims 1 to 9, wherein the image area is a rectangular area;

the first preset length is 1/3min (a, b); wherein a is the length of the rectangular region and b is the width of the rectangular region.

Images