Insulator rod core extraction method based on skeleton extraction and shortest path
By employing skeleton extraction and shortest path methods, the core of the insulator is accurately extracted from infrared insulator images, solving the problem of core extraction in complex scenarios and improving the accuracy and efficiency of insulator temperature anomaly detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANDONG ZHIYANG ELECTRIC
- Filing Date
- 2023-07-03
- Publication Date
- 2026-06-16
AI Technical Summary
Existing technologies struggle to accurately extract the core rod from infrared insulator images in complex scenarios, affecting the accuracy of insulator temperature anomaly detection.
A method based on skeleton extraction and shortest path is adopted. The insulator skeleton is extracted by Zhang-Suen thinning algorithm, and the core path is searched by improved A* algorithm. The starting point is determined by the search algorithm of the center point of the first and last discs. The shortest path is found by improved A* algorithm, and the midpoint of the short side of the detection box is connected to form the core line.
It achieves accurate fitting of insulator rod cores in complex scenarios, improves the accuracy and efficiency of detection, adapts to bending and shading conditions, has fast calculation speed, and wide applicability.
Smart Images
Figure CN116862857B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent power inspection technology, and in particular to a method for extracting the core of an insulator rod from an infrared image based on skeleton extraction and the shortest path. Background Technology
[0002] Abnormal temperature of insulators in power transmission lines is a serious hidden danger. Because abnormal insulator temperature can lead to serious consequences such as short circuits and tripping of equipment, it is crucial to conduct regular inspections and temperature checks on insulators.
[0003] Currently, the mainstream method for detecting insulator temperature anomalies is insulator target detection and temperature measurement based on infrared images. Two methods are commonly used to determine insulator temperature anomalies: temperature difference discrimination and relative temperature difference discrimination.
[0004] Temperature difference discrimination only requires measuring the highest and lowest temperatures of the insulator target area. Traditional methods directly calculate the extreme temperature points within the detection frame; however, this method is easily affected by background temperature interference within the detection frame, presenting significant limitations. One solution is to use an image segmentation algorithm to segment the insulator within the detection frame, calculating only the masked area and eliminating background interference.
[0005] Relative temperature difference is used to identify anomalies by calculating the percentage of the temperature rise difference between two corresponding measuring points relative to the higher temperature point. However, since the normal temperature of each insulator varies slightly—with insulators near the conductor and tower ends having slightly higher normal temperatures and those in the middle of the insulator string having slightly lower normal temperatures—selecting the appropriate measuring points on the insulator string is crucial. One solution is to extract the core of the insulator string and use the points on the core as measuring points to calculate the relative temperature difference.
[0006] A common method for extracting insulator cores is to use a rotating target detection model to inspect the insulator, and then draw the line connecting the midpoints of the short sides of the rotating frame as the insulator core line. However, the effectiveness of this method for core extraction depends on the accuracy of the angle regression of the rotating target detection frame. Furthermore, when the insulator string is long, it will have a certain curvature, which further reduces the fitting effect of this method on the core, and the extracted core line may even lie in the background between the insulator discs, making it difficult to use for calculating relative temperature differences.
[0007] In summary, how to accurately extract the core of an insulator from an infrared insulator image in complex scenarios, so as to lay the foundation for calculating the relative temperature difference of the insulator and judging temperature anomalies, is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0008] The technical problem to be solved by the present invention is to provide a method for extracting the core of an insulator rod from an infrared image based on skeleton extraction and shortest path, so as to accurately extract the core of the insulator rod from the infrared insulator image in complex scenes.
[0009] To solve the above-mentioned technical problems, the present invention provides the following technical solution:
[0010] A method for extracting the core of an insulator rod from infrared images based on skeleton extraction and shortest path extraction includes:
[0011] Step 1: Extract the skeleton of the insulator mask to obtain the insulator skeleton;
[0012] Step 2: Use the first and last disc center point search algorithm to find the center points of the first and last insulator discs in the insulator frame as the starting points for core path retrieval;
[0013] Step 3: Use the improved A* algorithm to search for the shortest path between the starting points of the core path retrieval as the main insulator core wire;
[0014] Step 4: Connect the main insulator rod core wire to the midpoint of the short side of the detection frame to form the final insulator rod core wire.
[0015] Furthermore, prior to step 1, the following is included:
[0016] Step 10: Obtain the detection frame of the insulator and the insulator mask within the detection frame from the output of the rotating target detection.
[0017] Furthermore, in step 1, the skeleton extraction employs the Zhang-Suen thinning algorithm.
[0018] Furthermore, step 2 includes:
[0019] Step 21: In the insulator skeleton, traverse each pixel and check its eight neighborhoods. If there are more than two other skeleton points in the neighborhood, it is a branch point, and then mark the point as a candidate point for the center point of the insulator disc.
[0020] Step 22: Among the candidate points of the center point of the insulator disc, find the two points closest to the midpoint of the short side of the detection frame as the starting points for the core path retrieval.
[0021] Furthermore, in step 3, the heuristic function used for the total cost function of the improved A* algorithm is:
[0022] h(x,y)=(x1-x2)2+(y1-y2)2
[0023] Here, (x1,y1) and (x2,y2) represent the coordinates of node x and endpoint y, respectively.
[0024] Furthermore, step 3 includes:
[0025] The improved A* algorithm combines jump point search to reduce the number of nodes that need to be traversed.
[0026] Furthermore, step 4 is followed by:
[0027] Step 5: Output the final insulator rod core wire.
[0028] The present invention has the following beneficial effects:
[0029] 1) This invention designs a method for extracting the core of an insulator rod from infrared images based on skeleton extraction and shortest path extraction. This includes using the Zhang-Suen thinning algorithm to extract the insulator skeleton and an improved A* algorithm to search for the insulator rod core wires. This method can accurately fit the core of the insulator in the infrared image and also performs well in cases of insulator string bending and occlusion.
[0030] 2) This invention optimizes the cost function of the A* algorithm and uses a jump point search algorithm to make the shortest path search more efficient.
[0031] In summary, compared with existing technologies, this invention offers advantages such as more accurate insulator core fitting, higher computational speed, and adaptability to complex insulator testing conditions. These advantages will provide a more reliable and efficient solution for detecting insulator heating defects in power systems. Attached Figure Description
[0032] Figure 1 This is a flowchart of the infrared image insulator core extraction method based on skeleton extraction and shortest path according to the present invention;
[0033] Figure 2 This is a visual image of the insulator frame in step 1 of this embodiment of the invention;
[0034] Figure 3 This is a visualization image of the candidate center point of the insulator disc in step 21 of the embodiment of the present invention;
[0035] Figure 4 This is a visualization image of the starting point of the core path retrieval in step 22 of this embodiment of the invention;
[0036] Figure 5 This is a visualization of the shortest path searched by the improved A* algorithm in step 3 of this embodiment of the invention.
[0037] Figure 6 This is a visualization image of the final insulator rod core wire in step 4 of the embodiment of the present invention. Detailed Implementation
[0038] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.
[0039] This invention provides a method for extracting the core of an insulator rod from an infrared image based on skeleton extraction and shortest path extraction, such as... Figure 1 As shown, it includes:
[0040] Step 1: Extract the skeleton of the insulator mask to obtain the insulator skeleton;
[0041] In this step, the skeleton extraction can employ the Zhang-Suen thinning algorithm. The Zhang-Suen thinning algorithm gradually erodes the pixels in the image to obtain a skeleton that reflects the main shape and structure of the insulator, such as... Figure 2 As shown.
[0042] As an optional embodiment, step 1 may include the following:
[0043] Step 10: Obtain the detection frame of the insulator and the insulator mask within the detection frame from the output of the rotating target detection.
[0044] In this step, you can input the detection frame of the insulator and the insulator mask within the detection frame, which are output by the rotating target detection (detecting the infrared image of the insulator).
[0045] Step 2: Use the first and last disc center point search algorithm to find the center points of the first and last insulator discs in the insulator frame as the starting points for core path retrieval;
[0046] In this step, the algorithm for searching the center point of the first and last insulator discs can be implemented using conventional techniques in this field. For example, the algorithm can first identify the position of the first and last insulator discs based on the disc shape, and then use a conventional center point algorithm to find the center point of the first and last insulator discs. The specific implementation algorithm will not be described in detail here.
[0047] As an optional embodiment, step 2 may include:
[0048] Step 21: In the insulator skeleton, traverse each pixel and check its eight neighborhoods. If there are more than two other skeleton points in the neighborhood, it is a branch point, and then mark the point as a candidate point for the center point of the insulator disc.
[0049] The candidate points for the center point of the insulator disc obtained in this step are as follows: Figure 3 The white dot in the image is shown.
[0050] Step 22: Among the candidate points of the center point of the insulator disc, find the two points closest to the midpoint of the short side of the detection frame as the starting points for the core path retrieval.
[0051] The starting point for core path retrieval obtained in this step is as follows: Figure 4 As shown by the black dots in the image.
[0052] Step 3: Use the improved A* algorithm to search for the shortest path between the starting points of the core path retrieval as the main insulator core wire;
[0053] The shortest path between the starting points of the core path retrieval in this step is as follows: Figure 5 As shown.
[0054] In this step, when searching for the shortest path using the improved A* algorithm, the A* algorithm selects an optimal node for expansion. This ensures that each expansion simultaneously considers both the actual cost from the starting point to the current node and the estimated cost from the current node to the destination, thus finding the optimal solution as early as possible. The total cost function f(n) consists of two parts: the heuristic function h(n) and the path length g(n) already traversed.
[0055] f(n) = g(n) + h(n)
[0056] Based on the actual characteristics of the infrared insulator frame, this invention optimizes the heuristic function h(n), selecting the square of the Euclidean distance as the heuristic function. The heuristic function h(x,y) can be expressed by the following formula:
[0057] h(x,y)=(x1-x2)2+(y1-y2)2
[0058] Here, (x1,y1) and (x2,y2) represent the coordinates of node x and endpoint y, respectively.
[0059] Preferably, step 3 may include:
[0060] The improved A* algorithm combines jump point search to reduce the number of nodes that need to be traversed.
[0061] In this way, Jump Point Search (JPS) is applied to the A* algorithm to reduce the number of nodes that need to be traversed and improve the running speed.
[0062] Step 4: Connect the main insulator rod core wire to the midpoint of the short side of the detection frame to form the final insulator rod core wire.
[0063] The final insulator rod core obtained in this step is as follows: Figure 6 As shown in the figure, the round dots are the midpoints of the short sides of the insulator detection frame, and the square dots are the starting points for the rod core path retrieval.
[0064] As an optional embodiment, step 4 may be followed by:
[0065] Step 5: Output the final insulator rod core wire.
[0066] In summary, the infrared image insulator core extraction method based on skeleton extraction and shortest path of the present invention takes the rotating target detection frame and the insulator mask within the detection frame as input. First, the skeleton extraction algorithm is used to extract the skeleton structure of the insulator from the insulator mask. Then, the center points of the first and last insulator discs in the skeleton are found through the first and last disc center point search algorithm. Next, the improved A* algorithm is used to find the shortest path between the center points of the first and last insulator discs as the main insulator core line. Finally, the main insulator core line is connected to the midpoint of the short side of the detection frame to form the final insulator core line. Compared with the traditional method of directly connecting the middle of the detection frame as the core line, the present invention can solve the problem of core extraction of long strings of insulators that produce sag, and will not produce the case where the core line is outside the target. In addition, the present invention has lower requirements for the accuracy of the angle regression of the rotating detection frame and has wider applicability. Furthermore, thanks to the skeleton extraction method, the present invention has high robustness and is also applicable to partial occlusion.
[0067] Testing showed that the infrared image insulator core extraction / fitting method proposed in this invention has an accuracy rate of over 85%; when the input infrared image size is 640*512 pixels, the extraction time for a single insulator core is less than 30ms. This provides a more reliable and efficient solution for detecting insulator heating defects in power systems.
[0068] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.
Claims
1. A method for extracting the core of an insulator rod from infrared images based on skeleton extraction and shortest path, characterized in that, include: Step 10: Obtain the detection frame of the insulator and the insulator mask within the detection frame from the output of the rotating target detection; Step 1: Extract the skeleton of the insulator mask to obtain the insulator skeleton; in Step 1, the skeleton extraction adopts the Zhang-Suen thinning algorithm. Step 2: Use the first and last disc center point search algorithm to find the center points of the first and last insulator discs in the insulator frame as the starting points for core path retrieval; Step 3: Use the improved A* algorithm to search for the shortest path between the starting points of the core path retrieval as the main insulator core wire; In step 3, the heuristic function used for the total cost function of the improved A* algorithm is: h(x,y)=(x1-x2)²+(y1-y2)² Where (x1,y1) and (x2,y2) represent the coordinates of node x and endpoint y, respectively; Step 3 includes: The improved A* algorithm combines jump point search to reduce the number of nodes that need to be traversed. Step 4: Connect the main insulator rod core wire to the midpoint of the short side of the detection frame to form the final insulator rod core wire; Step 2 includes: Step 21: In the insulator skeleton, traverse each pixel and check its eight neighborhoods. If there are more than two other skeleton points in the neighborhood, it is a branch point, and then mark the point as a candidate point for the center point of the insulator disc. Step 22: Among the candidate points of the center point of the insulator disc, find the two points closest to the midpoint of the short side of the detection frame as the starting points for the core path retrieval.
2. The method according to claim 1, characterized in that, Step 4 is followed by: Step 5: Output the final insulator rod core wire.