A method and apparatus for detecting icing thickness of overhead power lines based on images

Abstract

The present invention discloses an image-based method and device for detecting ice thickness of overhead cables. The method comprises the following steps: step S1, obtaining an ice-free image of an overhead cable, detecting the ice-free image of the overhead cable, detecting the ice-free cable, performing edge extraction, calculating the left and right boundaries of the overhead cable, and determining the center line; step S2, obtaining a current image of the overhead cable, taking the ice-free image in step S1 as a reference image, dividing the current image into a plurality of segmented areas, marking each segmented area as an inner area, an outer area, and a middle area based on the left and right boundaries and the center line of the cable identified by the reference image, taking the inner area as the foreground of the image, the outer area as the background, and the middle area as the unknown area, calculating the boundary of the middle area, and thereby obtaining the ice thickness of the overhead cable using the boundary of the middle area.

Description

Technical Field

[0001] The present invention relates to the technical field of power transmission and distribution, and in particular to an image-based method and device for detecting ice thickness of overhead cables. Background Art

[0002] Most of the existing power overhead insulated cables are composed of conductors and insulating layers and are used for long-distance power transmission. Overhead insulated cables with this structure can meet the requirements of safe power transmission under weather conditions without ice, rain, wind and snow. However, in actual operation, when encountering heavy snowfall and freezing rain, frost, ice and icicles will form on the surface of the cables, and even a thicker ice sleeve will form on the surface of the overhead insulated cables, increasing the weight of the overhead insulated cables themselves, resulting in broken wires or towers, circuit interruptions, inconvenience to people's lives, and huge losses to the country. Therefore, in ice, snow and freezing rain weather, it is particularly important to timely detect the thickness of the overhead insulated cables so as to defrost and de-ice them in time to prevent ice on their surfaces. Summary of the invention

[0003] In order to overcome the shortcomings of the above-mentioned prior art, the purpose of the present invention is to provide an image-based overhead cable ice thickness detection method and device, so as to timely detect the ice thickness of the overhead cable based on the un-iced reference picture and current picture of the overhead cable, so as to remind the staff to defrost and de-ice in time to prevent the surface from freezing and improve safety.

[0004] To achieve the above object, the present invention proposes an image-based method for detecting ice thickness of overhead cables, comprising the following steps:

[0005] Step S1, obtaining an ice-free image of an overhead cable, detecting the ice-free image of the overhead cable, detecting the ice-free cable, performing edge extraction, calculating the left and right boundaries of the overhead cable, and determining the center line;

[0006] Step S2, obtaining the current image of the overhead cable, taking the ice-free figure in step S1 as the reference image, dividing the current image into several segmented areas, marking each segmented area as an inner area, an outer area, and a middle area based on the left and right boundaries and the center line of the cable identified by the reference image, taking the inner area as the foreground of the image, the outer area as the background, and the middle area as the unknown area to calculate the boundary of the middle area, thereby using the boundary of the middle area to derive the thickness of the ice covering the overhead cable.

[0007] Preferably, step S1 further comprises:

[0008] Step S100, obtaining an ice-free image of the overhead cable, dividing the image into several blocks, calculating the color mean of each block and the maximum value of the color difference within the block, and setting a cable mark;

[0009] Step S101, marking the block where the midpoint of the bottom of the ice-free image is located as a cable;

[0010] Step S102, for each pixel block marked as a cable, a color mean difference between the surrounding blocks and the pixel block marked as a cable is calculated, and a cable mark of each surrounding block is determined according to a comparison result between the obtained color mean difference and the maximum color difference value within the pixel block marked as a cable;

[0011] Step S103, for a cable pixel block, if all its surrounding blocks are cables, it is marked as an internal block; if there are non-cable blocks in its surrounding blocks, it is marked as a boundary block;

[0012] Step S104 , edge extraction is performed on the cable pixel block marked as the boundary block, and the left and right boundaries of the cable are calculated using Hough transform, and the center line is determined.

[0013] Preferably, in step S100, the cable label is set to label ij , when label ij =1 indicates that the image is a cable; when label ij = -1 indicates that the image is non-cable; when label ij =0, it means that the block image is unknown, and the cable label of all block images is initially set to 0.

[0014] Preferably, in step S102, for each pixel block B marked as a cable ij , find its surrounding blocks B pq , get the color mean c of each surrounding block pq and compare it with the pixel block B marked as a cable ij The color mean c ij Perform difference calculation, if ||c pq -c ij || <d ij , then set the surrounding block B pq Cable marking label pq =1, otherwise set the cable label pq =0, where d ij is the maximum value of the color difference within the pixel block marked as the cable.

[0015] Preferably, step S2 further comprises:

[0016] Step S200, obtaining a current image of the overhead cable, dividing the current image into a number of small segmented areas, and calculating the average color of each segmented area as the area color;

[0017] Step S201, selecting a boundary from the left and right boundaries of the cable determined in step S1, marking a first clustering area of ​​the current image according to the selected boundary and its center line, clustering the area colors of all the segmented areas of the first clustering area, and generating a plurality of cluster centers;

[0018] Step S202, defining a detection boundary outside the selected boundary;

[0019] Step S203, taking the area between the detected boundary and the boundary selected in step S201 as a second clustering area, clustering the area colors of all the segmented areas of the second clustering area to generate a plurality of cluster centers;

[0020] Step S204, for the segmented area completely located in the first cluster area, record it as an internal area;

[0021] Step S205, for each internal region, expand it toward the selected boundary direction and mark the unmarked segmented regions;

[0022] Step S206, setting all unmarked areas between the detection boundary and step S201 as external areas;

[0023] Step S207, scanning all segmented areas between the detection boundary and the center line, and marking the middle area;

[0024] Step S208, taking the inner area as the foreground of the image, the outer area as the background, and the middle area as the unknown area, and calculating the boundary of the middle area;

[0025] Step S209, calculating the thickness of ice covering the overhead cables according to the boundary of the middle area, the boundary selected in step S201 and the center line.

[0026] Preferably, in step S201, the regional colors of the segmented regions of all the first cluster centers are clustered using k-means to generate a plurality of cluster centers. In step S203, the regional colors of all segmented regions between the detected boundary and the boundary selected in step S201 are clustered using k-means to generate a number of cluster centers.

[0027] Preferably, in step S202, a detection boundary is defined outside the selected boundary to ensure that the image of the snow-covered part can be covered.

[0028] Preferably, in step S205, the expansion method is as follows:

[0029] a) For a region x in the interior region i , find an unlabeled region x in its adjacent region j ;

[0030] b) Calculate x j Distance to internal regional clusters and x j Distance to the second clustering area cluster

[0031] c) If Then the unmarked area x j as an inner zone; otherwise, mark it as an outer zone.

[0032] Preferably, in step S209, the thickness of the ice coating is calculated according to the following formula:

[0033]

[0034] Among them, y represents the calculation in the vertical direction of the image, IL A is the boundary of the middle area, L A It is the boundary selected in step S201, and M represents the center line.

[0035] In order to achieve the above object, the present invention also provides an image-based overhead cable ice thickness detection device, comprising:

[0036] An image acquisition unit, used for acquiring an ice-free image of the overhead cable and a current image;

[0037] An ice-free cable detection unit is used to obtain an ice-free image of an overhead cable, detect the ice-free image of the overhead cable, detect the ice-free cable, perform edge extraction, calculate the left and right boundaries of the cable, and determine the center line;

[0038] The ice thickness detection unit is used to obtain the current image of the overhead cable, and use the ice-free pattern in the ice-free cable detection unit as a reference image to divide the current image into a number of segmented areas. Based on the left and right boundaries and the center line of the cable identified by the reference image, each segmented area is marked as an inner area, an outer area, and a middle area. The inner area is used as the foreground of the image, the outer area is used as the background, and the middle area is used as the unknown area to calculate the boundary of the middle area, so as to use the boundary of the middle area to derive the thickness of the ice on the overhead cable.

[0039] Compared with the prior art, the present invention provides an image-based overhead cable ice thickness detection method and device. By acquiring an un-iced image of the overhead cable, the un-iced image of the overhead cable is detected, the un-iced cable is detected, and edge extraction is performed. The left and right boundaries of the overhead cable are calculated and the center line is determined. Then, the current image of the overhead cable is acquired, and the un-iced image is used as a reference image. The current image is segmented into a plurality of segmented areas. Based on the left and right boundaries and the center line of the cable identified by the reference image, each segmented area is marked as an inner area, an outer area, and a middle area. The inner area is used as the foreground of the image, the outer area is used as the background, and the middle area is used as the unknown area to calculate the boundary of the middle area, so as to use the boundary of the middle area to derive the ice thickness of the overhead cable and timely detect the ice thickness of the overhead cable. Based on the un-iced reference image and the current image of the overhead cable, the present invention can timely detect the ice thickness of the overhead cable to remind the staff to perform defrosting and de-icing in time to prevent the surface from icing, thereby improving safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A flowchart of the steps of an image-based method for detecting ice thickness of overhead cables according to the present invention;

[0041] Figure 2 This is a system architecture diagram of an image-based overhead cable ice thickness detection device of the present invention. DETAILED DESCRIPTION

[0042] The following describes the implementation of the present invention through specific examples and in conjunction with the accompanying drawings. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific examples, and the details in this specification can also be modified and changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention.

[0043] Figure 1 FIG. 1 is a flowchart of a method for detecting ice thickness of overhead cables based on an image according to the present invention. Figure 1 As shown, the present invention provides an image-based method for detecting ice thickness of overhead cables, comprising the following steps:

[0044] Step S1, obtaining an ice-free image of an overhead cable, detecting the ice-free image of the overhead cable, detecting ice-free cables, performing edge extraction, and calculating left and right boundaries of the cable.

[0045] Specifically, step S1 further includes:

[0046] Step S100, obtaining an ice-free image of the overhead cable, dividing the image into several blocks, calculating the color mean of each block and the maximum value of the color difference within the block, and setting a cable mark.

[0047] In a specific embodiment of the present invention, the ice-free image is divided into pixel blocks B of size 8×8. ij , and calculate the color mean c of each block ij (i.e., the color mean of all pixels in the block), and the maximum color difference within the block is where c xy It is B ij The color value of the pixel (x, y) in the cable and set the cable label to label ij In a specific embodiment of the present invention, when label ij =1 indicates that the block is a cable; when label ij = -1 indicates that the block is non-cable; when label ij = 0 means the block is unknown. Initially, the cable labels of all blocks are set to 0.

[0048] Step S101: Mark the block where the midpoint of the bottom of the ice-free image is located as a cable.

[0049] Specifically, the block where the bottom midpoint of the ice-free image is located (when the camera is installed, it is ensured that the lower edge of the image is on the cable) is recorded as B bc , mark the block as a cable, that is, the cable label of the block bc =1,

[0050] Step S102, for each pixel block marked as a cable, the color mean difference between the surrounding blocks and the pixel block marked as a cable is calculated, and the cable mark of each surrounding block is determined based on the comparison result of the obtained color mean difference and the maximum color difference within the pixel block marked as a cable.

[0051] Specifically, for each pixel block B marked as a cable ij , find the surrounding blocks B pq , get the color mean c of each surrounding block pq and compare it with the pixel block B marked as a cable ij The color mean c ij Perform difference calculation, if ||c pq -c ij || <d ij , then set the surrounding block B pq Cable marking label pq =1, otherwise set the cable label pq =0.

[0052] Step S103: for a cable pixel block, if all its surrounding blocks are cables, it is marked as an internal block; if any of its surrounding blocks are non-cable blocks, it is marked as a boundary block.

[0053] Step S104 , edge extraction is performed on the cable pixel block marked as the boundary block, and the left and right boundaries of the cable are calculated using Hough transform.

[0054] Step S2, obtaining the current image of the overhead cable, taking the ice-free figure in step S1 as the reference image, dividing the current image into several segmented areas, marking each segmented area as an inner area, an outer area, and a middle area based on the left and right boundaries and the center line of the cable identified by the reference image, taking the inner area as the foreground of the image, the outer area as the background, and the middle area as the unknown area to calculate the boundary of the middle area, thereby using the boundary of the middle area to derive the thickness of the ice covering the overhead cable.

[0055] Specifically, step S2 further includes:

[0056] Step S200, obtaining a current image of the overhead cable, dividing the current image into small segmented areas, and calculating the average color of each segmented area as the area color.

[0057] In the present invention, a current image I of the overhead cable is obtained, and the current image I is divided into small segmented areas. For each segmented area x i , calculate its average color c i As the region color.

[0058] Step S201, selecting a boundary from the left and right boundaries of the cable determined in step S1, marking a first clustering area of ​​the current image according to the selected boundary and its center line, clustering the area colors of all the segmented areas of the first clustering area, and generating a plurality of cluster centers.

[0059] In a specific embodiment of the present invention, the ice-free image in step S1 is used as the reference image A, and the left and right boundaries of the cable in the reference image A are respectively denoted as L A and R A , and mark the midline of the two as M, and select the left boundary L A Then the current image I is completely located at the left boundary L A The area between the center line M is marked as the left area (the first clustering area). The regional colors of all the segmented areas in the left area are clustered using k-means to generate several cluster centers. In the specific embodiment of the present invention, it is assumed that 24 cluster centers are generated. n=1,2,…,24. Of course, you can also choose the right boundary R AThe area between the middle line M is recorded as the right area as the first clustering area, which will not be described in detail here.

[0060] Step S202, defining a detection boundary outside the selected boundary to ensure that the image of the snow-covered part can be covered, for example, defining a detection boundary outside the selected boundary so that the distance between the detection boundary and the selected boundary is 5 times the distance between the selected boundary and the center line.

[0061] In the present invention, the outer side of the selected boundary refers to the side away from the cable. In a specific embodiment of the present invention, the outer side of the selected boundary is the side away from the cable. A The left side defines a detection left boundary DL A , so that ||DL A -L A ||=5||L A -M||.

[0062] Step S203: taking the area between the detected boundary and the boundary selected in step S201 as a second clustering area, clustering the area colors of all the segmented areas of the second clustering area to generate a plurality of cluster centers.

[0063] In a specific embodiment of the present invention, the left boundary DL is detected. A and the left boundary L A The regional colors of all segmented areas between are clustered using k-means to generate 24 cluster centers

[0064] Step S204: For the segmented area completely located in the first cluster area, it is recorded as an internal area.

[0065] In a specific embodiment of the present invention, if a region is completely located at the left boundary L A The area between the center line M is referred to as the inner area.

[0066] Step S205: for each internal region, expand it toward the selected boundary direction and mark the unmarked segmented regions.

[0067] Specifically, for the left boundary L A The inner area between the center line M is expanded to the left according to the following method:

[0068] a) For a region x in the interior region i , find an unlabeled region x in its adjacent region j ;

[0069] b) Calculate x j Distance to internal regional clusters and x jDistance to the detection area (second clustering area) cluster

[0070] c) If Then the unmarked area x j as an inner zone; otherwise, mark it as an outer zone.

[0071] Step S206, setting all unmarked areas between the detection boundary and step S201 as external areas.

[0072] Step S207, scanning all segmented areas between the detection boundary and the center line, and marking the middle area.

[0073] In a specific embodiment of the present invention, scanning the detection boundary DL A For all segmented regions between the middle line M, if a segmented region is surrounded by both external and internal regions, it will be marked as the middle region.

[0074] Step S208, taking the inner area as the foreground of the image, the outer area as the background, and the middle area as the unknown area, and calculating the boundary of the middle area.

[0075] In a specific embodiment of the present invention, the inner area is taken as the foreground of the image, the outer area is taken as the background, and the middle area is taken as the unknown area. The boundary IL of the middle area is calculated by using the Matting technique. A In a specific embodiment of the present invention, α>0.5 is set as foreground, and α≤0.5 is set as background, where α is a parameter calculated according to the Matting algorithm for identifying the transparency of the region.

[0076] Step S209, calculating the thickness of ice covering the overhead cables according to the boundary of the middle area, the boundary selected in step S201 and the center line.

[0077] Specifically, the ice thickness (in cable diameter) is calculated according to the following formula:

[0078]

[0079] Among them, y indicates that the calculation is performed in the vertical direction of the image.

[0080] Figure 2 This is a system architecture diagram of an image-based overhead cable ice thickness detection device of the present invention.

[0081] like Figure 2 As shown, the present invention provides an image-based overhead cable ice thickness detection device, comprising:

[0082] The image acquisition unit 201 is used to acquire an ice-free image of the overhead cable and a current image.

[0083] The ice-free cable detection unit 202 is used to obtain an ice-free image of the overhead cable, detect the ice-free image of the overhead cable, detect the ice-free cable, perform edge extraction, and calculate the left and right boundaries of the cable.

[0084] Specifically, the ice-free cable detection unit 202 further includes:

[0085] The image segmentation module is used to obtain an ice-free image of the overhead cable, divide the image into several blocks, calculate the color mean of each block and the maximum color difference within the block, and set cable marks.

[0086] In a specific embodiment of the present invention, the image segmentation module segments the ice-free image into pixel blocks B of size 8×8. ij , and calculate the color mean c of each block ij , the maximum color difference within the block is d ij , and set the cable label to label ij In a specific embodiment of the present invention, when label ij =1 indicates that the block is a cable; when label ij = -1 indicates that the block is non-cable; when label ij = 0 means the block is unknown. Initially, the cable labels of all blocks are set to 0.

[0087] The first block marking module is used to mark the block where the bottom midpoint of the ice-free image is located as a cable.

[0088] Specifically, the block where the midpoint of the bottom of the ice-free image is located is denoted as B bc , mark the block as a cable, that is, the cable label of the block bc =1.

[0089] Each pixel block marking module is used to calculate the color mean difference between the surrounding blocks of each pixel block marked as a cable and the pixel block marked as a cable, and determine the cable mark of each surrounding block according to the comparison result of the obtained color mean difference and the maximum color difference within the block of the pixel block marked as the cable.

[0090] Specifically, for each pixel block B marked as a cable ij , find the surrounding blocks B pq , get the color mean c of each surrounding block pq and compare it with the pixel block B marked as a cable ij The color mean c ijPerform difference calculation, if ||c pq -c ij || <d ij , then set the surrounding block B pq Cable marking label pq =1, otherwise set the cable label pq =0.

[0091] The classification and marking module is used to mark a cable pixel block as an internal block if all its surrounding blocks are cables, and mark it as a boundary block if there are non-cable blocks around it.

[0092] The edge extraction module is used to extract the edge of the cable pixel block marked as the boundary block, and calculate the left and right boundaries of the cable by using Hough transform.

[0093] The ice thickness detection unit 203 is used to obtain the current image of the overhead cable, and use the ice-free figure in the ice-free cable detection unit 202 as a reference image to divide the current image into a number of segmented areas. Based on the left and right boundaries and the center line of the cable identified by the reference image, each segmented area is marked as an inner area, an outer area, and a middle area. The inner area is used as the foreground of the image, the outer area is used as the background, and the middle area is used as the unknown area to calculate the boundary of the middle area, so as to use the boundary of the middle area to derive the thickness of the ice on the overhead cable.

[0094] Specifically, the ice thickness detection unit 203 further includes:

[0095] The current image segmentation module is used to obtain the current image of the overhead cable, segment the current image into small segmented areas, and for each segmented area, calculate its average color as the area color.

[0096] In the present invention, a current image I of the overhead cable is obtained, and the current image I is divided into small segmented areas. For each segmented area x i , calculate its average color c i As the region color.

[0097] The first clustering module is used to select a boundary from the left and right boundaries of the cable determined by the ice-free cable detection unit 202, mark the first clustering area of ​​the current image according to the selected boundary and the line therein, cluster the area colors of all the segmented areas of the first clustering area, and generate a plurality of cluster centers.

[0098] In a specific embodiment of the present invention, the ice-free image of the ice-free cable detection unit 202 is used as the reference image A, and the left and right boundaries of the cable in the reference image A are respectively denoted as L A and R A, and mark the midline of the two as M, and select the left boundary L A Then the current image I is completely located at the left boundary L A The area between the left area and the center line M is marked as the left area (the first clustering area). The regional colors of all the segmented areas of the left area are clustered using k-means to generate several cluster centers. In the specific embodiment of the present invention, it is assumed that 24 cluster centers are generated. Of course, you can also choose the right boundary R A The area between the middle line M is recorded as the right area as the first clustering area, which will not be described in detail here.

[0099] The detection boundary determination module is used to define a detection boundary outside the selected boundary to ensure that the image of the snow-covered part can be covered. For example, a detection boundary is defined outside the selected boundary so that the distance between the detection boundary and the selected boundary is 5 times the distance between the selected boundary and the center line.

[0100] In the present invention, the outer side of the selected boundary refers to the side away from the cable. In a specific embodiment of the present invention, the outer side of the selected boundary is the side away from the cable. A The left side defines a detection left boundary DL A , so that ||DL A -L A ||=5||L A -M||.

[0101] The second clustering module is used to take the area between the detection boundary and the boundary selected by the first clustering module as the second clustering area, cluster the area colors of all the segmented areas of the second clustering area, and generate a plurality of cluster centers.

[0102] In a specific embodiment of the present invention, the left boundary DL is detected. A and the left boundary L A The regional colors of all segmented regions between are clustered using k-means to generate several cluster centers. In the specific embodiment of the present invention, it is assumed that 24 cluster centers are generated.

[0103] The internal region marking module is used to mark the segmented region completely located in the first cluster region as an internal region.

[0104] In a specific embodiment of the present invention, if a region is completely located at the left boundary L A The area between the center line M is referred to as the inner area.

[0105] The internal region expansion and marking module is used to expand each internal region toward the selected boundary direction and mark the unmarked segmented regions.

[0106] Specifically, for the left boundary L A The inner area between the center line M is expanded to the left according to the following method:

[0107] a) For a region x in the interior region i , find an unlabeled region x in its adjacent region j ;

[0108] b) Calculate x j Distance to internal regional clusters and x j Distance to the detection area (second clustering area) cluster

[0109] c) If Then the unmarked area x j as an inner zone; otherwise, mark it as an outer zone.

[0110] The external region determination module is used to set all unmarked regions between the detection boundary and the boundary selected by the first clustering module as external regions.

[0111] The middle area marking module is used to scan all segmented areas between the detection boundary and the center line and mark the middle area.

[0112] In a specific embodiment of the present invention, scanning the detection boundary DL A For all segmented regions between the middle line M, if a segmented region is surrounded by both external and internal regions, it will be marked as the middle region.

[0113] The middle area boundary calculation module is used to take the inner area as the foreground of the image, the outer area as the background, and the middle area as the unknown area, and calculate the boundary of the middle area.

[0114] In a specific embodiment of the present invention, the inner area is taken as the foreground of the image, the outer area is taken as the background, and the middle area is taken as the unknown area. The boundary IL of the middle area is calculated by using the Matting technique. A In a specific embodiment of the present invention, α>0.5 is set as foreground, and α≤0.5 is set as background, where α is a parameter calculated according to the Matting algorithm for identifying the transparency of the region.

[0115] The ice thickness calculation module is used to calculate the thickness of ice on the overhead cable according to the boundary of the middle area, the boundary selected by the first clustering module and the center line.

[0116] Specifically, the ice thickness (in cable diameter) is calculated according to the following formula:

[0117]

[0118] Among them, y represents the calculation in the vertical direction of the image

[0119] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Any person skilled in the art may modify and alter the above embodiments without departing from the spirit and scope of the present invention.

Claims

1. An image-based method for detecting ice thickness of overhead cables, comprising the following steps: Step S1, obtaining an ice-free image of an overhead cable, detecting the ice-free image of the overhead cable, detecting the ice-free cable, performing edge extraction, calculating the left and right boundaries of the overhead cable, and determining the center line; Step S2, obtaining a current image of the overhead cable, using the ice-free figure in step S1 as a reference image, segmenting the current image into a plurality of segmented areas, marking each segmented area as an inner area, an outer area, and a middle area based on the left and right boundaries and the center line of the cable identified by the reference image, taking the inner area as the foreground of the image, the outer area as the background, and the middle area as the unknown area to calculate the boundary of the middle area, thereby using the boundary of the middle area to derive the thickness of ice covering the overhead cable; in, Step S1 further comprises: Step S100, obtaining an ice-free image of the overhead cable, dividing the image into several blocks, calculating the color mean of each block and the maximum value of the color difference within the block, and setting a cable mark; Step S101, marking the block where the midpoint of the bottom of the ice-free image is located as a cable; Step S102, for each pixel block marked as a cable, a color mean difference between the surrounding blocks and the pixel block marked as a cable is calculated, and a cable mark of each surrounding block is determined according to a comparison result between the obtained color mean difference and the maximum color difference value within the pixel block marked as a cable; Step S103, for a cable pixel block, if all its surrounding blocks are cables, it is marked as an internal block; if there are non-cable blocks in its surrounding blocks, it is marked as a boundary block; Step S104, performing edge extraction on the cable pixel block marked as the boundary block, calculating the left and right boundaries of the cable using Hough transform, and determining the center line of the cable; Step S2 further comprises: Step S200, obtaining a current image of the overhead cable, dividing the current image into a number of small segmented areas, and calculating the average color of each segmented area as the area color; Step S201, selecting a boundary from the left and right boundaries of the cable determined in step S1, marking a first clustering area of ​​the current image according to the selected boundary and its center line, clustering the area colors of all the segmented areas of the first clustering area, and generating a plurality of cluster centers; Step S202, defining a detection boundary outside the selected boundary; Step S203, taking the area between the detected boundary and the boundary selected in step S201 as a second clustering area, clustering the area colors of all the segmented areas of the second clustering area to generate a plurality of cluster centers; Step S204, for the segmented area completely located in the first cluster area, record it as an internal area; Step S205, for each internal region, expand it toward the selected boundary direction and mark the unmarked segmented regions; Step S206, setting all unmarked areas between the detection boundary and step S201 as external areas; Step S207, scanning all segmented areas between the detection boundary and the center line, and marking the middle area; Step S208, taking the inner area as the foreground of the image, the outer area as the background, and the middle area as the unknown area, and calculating the boundary of the middle area; Step S209, calculating the thickness of ice covering the overhead cables according to the boundary of the middle area, the boundary selected in step S201 and the center line.

2. The method for detecting ice thickness of overhead cables based on an image according to claim 1, characterized in that: In step S100, the cable label is set to label ij , when label ij =1 indicates that the image is a cable; when label ij = -1 indicates that the image is non-cable; when label ij =0, it means that the block image is unknown, and the cable label of all block images is initially set to 0.

3. The method for detecting ice thickness of overhead cables based on an image according to claim 2, characterized in that: In step S102, for each pixel block B marked as a cable ij , find its surrounding blocks B pq , get the color mean c of each surrounding block pq and compare it with the pixel block B marked as a cable ij The color mean c ij Perform difference calculation, if ‖c pq -c ij ‖ <d ij , then set the surrounding block B pq Cable marking label pq =1, otherwise set the cable label pq =0, where d ij is the maximum value of the color difference within the pixel block marked as the cable.

4. The method for detecting ice thickness of overhead cables based on an image according to claim 1, characterized in that: In step S201, the regional colors of all the segmented regions of the first cluster center are clustered using k-means to generate a number of cluster centers. In step S203, the regional colors of all segmented regions between the detected boundary and the boundary selected in step S201 are clustered using k-means to generate a number of cluster centers.

5. The method for detecting ice thickness of overhead cables based on an image according to claim 4, characterized in that: In step S202 , a detection boundary is defined outside the selected boundary such that the distance between the detection boundary and the selected boundary is 5 times the distance between the selected boundary and the center line.

6. The method for detecting ice thickness of overhead cables based on an image according to claim 5, characterized in that: In step S205, the expansion method is as follows: a) For a region x in the interior region i , find an unlabeled region x in its adjacent region j ; b) Calculate x j Distance to internal regional clusters and x j Distance to the second clustering area cluster c) If Then the unmarked area x j as an inner zone; otherwise, mark it as an outer zone.

7. The method for detecting ice thickness of overhead cables based on an image according to claim 6, characterized in that: In step S209, the thickness of the ice is calculated according to the following formula: Among them, y represents the calculation in the vertical direction of the image, IL A is the boundary of the middle area, L A It is the boundary selected in step S201, and M represents the center line.

8. An image-based overhead cable ice thickness detection device, comprising: An image acquisition unit, used for acquiring an ice-free image of the overhead cable and a current image; An ice-free cable detection unit is used to obtain an ice-free image of an overhead cable, detect the ice-free image of the overhead cable, detect the ice-free cable, perform edge extraction, calculate the left and right boundaries of the cable, and determine the center line; An ice thickness detection unit is used to obtain a current image of the overhead cable, and use the ice-free pattern in the ice-free cable detection unit as a reference image to segment the current image into a plurality of segmented areas. Based on the left and right boundaries and the center line of the cable identified by the reference image, each segmented area is marked as an inner area, an outer area, and a middle area. The inner area is used as the foreground of the image, the outer area is used as the background, and the middle area is used as the unknown area to calculate the boundary of the middle area, so as to obtain the thickness of ice on the overhead cable using the boundary of the middle area. The ice-free cable detection unit further comprises: The image segmentation module is used to obtain an ice-free image of the overhead cable, segment the image into several blocks, calculate the color mean of each block and the maximum value of the color difference within the block, and set the cable mark; A first block marking module, used for marking the block where the bottom midpoint of the ice-free image is located as a cable; Each pixel block marking module is used to calculate the color mean difference between the surrounding blocks of each pixel block marked as a cable and the pixel block marked as the cable, and determine the cable mark of each surrounding block according to the comparison result of the obtained color mean difference and the maximum color difference value within the pixel block marked as the cable; A classification and marking module is used to mark a cable pixel block as an internal block if all its surrounding blocks are cables, and mark it as a boundary block if there are non-cable blocks in its surrounding blocks; The edge extraction module is used to extract the edge of the cable pixel block marked as the boundary block, calculate the left and right boundaries of the cable by using Hough transform, and determine the center line; The ice thickness detection unit further comprises: The current image segmentation module is used to obtain the current image of the overhead cable, segment the current image into a number of small segmented areas, and for each segmented area, calculate its average color as the area color; A first clustering module is used to select a boundary from the left and right boundaries of the cable determined by the ice-free cable detection unit, mark a first clustering area of ​​the current image according to the selected boundary and the middle line, cluster the area colors of all the segmented areas of the first clustering area, and generate a plurality of cluster centers; A detection boundary determination module, used to define a detection boundary outside the selected boundary; A second clustering module is used to take the area between the detection boundary and the boundary selected by the first clustering module as a second clustering area, cluster the area colors of all the segmented areas of the second clustering area, and generate a plurality of cluster centers; An internal region marking module, used for marking a segmented region completely located in the first cluster region as an internal region; An internal region expansion and marking module is used to expand each internal region toward the selected boundary direction and mark the unmarked segmented regions; an external region determination module, configured to set all unmarked regions between the detection boundary and the boundary selected by the first clustering module as external regions; A middle area marking module, used to scan all segmented areas between the detection boundary and the midline, and mark the middle area; The middle area boundary calculation module is used to take the inner area as the foreground of the image, the outer area as the background, and the middle area as the unknown area, and calculate the boundary of the middle area; The ice thickness calculation module is used to calculate the thickness of ice on the overhead cable according to the boundary of the middle area, the boundary selected by the first clustering module and the center line.

Images