An AI image recognition-based power line tree barrier detection method and system

Abstract

The application relates to the technical field of image processing, in particular to an electric power line tree barrier detection method and system based on AI image recognition, which specifically comprises the following steps: acquiring conductor suspected lines in each frame of image based on edge line features in the frame of image; calculating an optimized energy function of each iteration of a Snake model based on the trend of all conductor suspected lines in the current frame of image and the proximity of the contour of the Snake model iteration to the conductor suspected lines, improving the Snake model, conducting conductor contour detection through the improved Snake model, conducting electric power line tree barrier detection based on the appearance of vegetation pixels near the detected conductor, reducing the possibility of deviation from the actual conductor position when the Snake model obtains the conductor contour, improving the extraction accuracy and speed of the conductor, and further improving the accuracy of tree barrier detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a power line tree barrier detection method and system based on AI image recognition. BACKGROUND

[0002] In the power system, the power line is the key channel for power transmission, and its safe and stable operation is crucial to the protection of social production and life. However, the power line often faces the threat of tree barrier hazards, and the contact between tree branches and conductors may cause the insulation performance between conductors to decrease, and then short circuit phenomenon occurs.

[0003] The document "Transmission line tree barrier analysis technology research based on monocular vision" proposes a monocular vision detection method, which uses online monitoring and image processing technology to realize real-time early warning and effectively reduce the inspection cost. However, when extracting the conductor, although the improved Canny edge detection algorithm is used, irrelevant straight lines may still be detected, which interferes with the accuracy of the point detection, and then affects the accuracy of the tree barrier detection. SUMMARY

[0004] In order to solve the above technical problems, the purpose of the present application is to provide a power line tree barrier detection method and system based on AI image recognition, and the technical solution adopted is as follows:

[0005] In the first aspect, the present application provides a power line tree barrier detection method based on AI image recognition, which comprises the following steps:

[0006] Collecting video images of conductors between towers to obtain each frame of image, and the image includes tower images connected on both sides of the conductor;

[0007] Obtaining the conductor suspected line in each frame of image based on the edge line features in each frame of image;

[0008] In the current frame of image, the initial contour of the Snake model is determined according to the position of the conductor in the image; the straight line prior energy of the Snake model is determined based on the direction of all conductor suspected lines in the current frame of image; the scene constraint energy of the current iteration of the Snake model is calculated based on the proximity of the contour of the current iteration of the Snake model to the conductor suspected line;

[0009] Based on the straight line prior energy and the scene constraint energy, the optimization energy function of the current iteration is obtained, and the gradient of the total energy of the current iteration is calculated; the evolution equation of each iteration is determined based on the gradient of the total energy, and the Snake model is combined to obtain the contour of each conductor in the current frame of image, and the neighborhood of the conductor contour is taken as the detection area;

[0010] Obtaining vegetation pixels in each frame image through a deep learning model; and performing power line tree barrier detection based on the occurrence of the vegetation pixels in a detection area of each frame image.

[0011] In one embodiment, the process of obtaining conductor suspected lines in each frame image is as follows:

[0012] Performing edge detection on the gray-scale image of each frame image to obtain each edge image, and performing straight line detection on each edge image to obtain each straight line in each edge image;

[0013] Setting a standard straight line in each frame image;

[0014] Taking a straight line with a length greater than a preset length threshold and an included angle with the standard straight line less than a preset angle threshold as a conductor suspected line.

[0015] In one embodiment, the process of setting the standard straight line is as follows:

[0016] Randomly selecting a pixel point in the region of any tower in the first frame image, denoted as A1(u1, v1); obtaining a pixel point corresponding to the position of A1(u1, v1) in the region of another tower in the first frame image, denoted as A2(u2, v2), connecting the straight line between the points A1(u1, v1) and A2(u2, v2), denoted as a standard straight line; and setting the position of the standard straight line in the first frame image as the position of the standard straight line in each frame image.

[0017] In one embodiment, the process of obtaining the straight line prior energy is as follows:

[0018] Taking the included angle between the i-th conductor suspected line and the standard straight line as the direction angle of the i-th conductor suspected line, denoted as Taking the straight line prior energy as E line , the expression of E line is as follows: In the formula, M is the number of conductor suspected lines in the current frame image, cos 2 () is the square of the calculated cosine function.

[0019] In one embodiment, the expression of the scene constraint energy of the current iteration is as follows:

[0020]

[0021] In the formula, E prior is the scene constraint energy of the current iteration; M is the number of conductor suspected lines in the current frame image; N is the number of pixel points on the contour of the current iteration; D[] is a calculated Euclidean distance function; (x k ,y k ) is the coordinate of the k-th pixel point on the contour of the current iteration; (xHi , y Hi is the coordinate of the midpoint on the i-th suspected line of the current frame image.

[0022] In one embodiment, the expression of the optimization energy function of the current iteration is:

[0023] E total = α × E grad + β × E curve + γ × E line + δ × E prior

[0024] In the formula, E total is the optimization energy function of the current iteration, α, β, γ and δ are preset energy weights, E grad , E curve are the gradient energy and the curvature energy in the energy function of the original Snake model respectively, E line is the straight line prior energy, E prior is the scene constraint energy of the current iteration; wherein α > β > γ > δ, and the sum of α, β, γ and δ is 1.

[0025] In one embodiment, the process of obtaining the gradient of the total energy of the current iteration is:

[0026] The optimization energy function of the current iteration is taken as the energy function of the current iteration in the Snake model, and the gradient of the total energy of the current iteration is calculated.

[0027] In one embodiment, the evolution equation of each iteration is:

[0028]

[0029] In the formula, x represents the k-th point on the contour in the t+1-th iteration; represents the k-th point on the contour in the t-th iteration; η represents the learning rate; is the gradient of the total energy of the current iteration.

[0030] In one embodiment, the tree barrier detection of the power line is based on the occurrence of the vegetation pixels in the detection area of each frame image, and specifically:

[0031] For each detection area in each frame image, the pixel at the average value of the preset number of vegetation pixel coordinates closest to the edge of the detection area is taken as a detection point, if the detection point is in the detection area, a tree barrier appears in the power line corridor, and an alarm is issued; otherwise, no tree barrier appears in the power line corridor.

[0032] In a second aspect, the embodiments of the present application also provide an AI image recognition-based power line tree barrier detection system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the method in any one of the preceding aspects when executing the computer program.

[0033] The embodiments of the present application have at least the following beneficial effects:

[0034] The embodiments of the present application have at least the following beneficial effects: The embodiments of the present application have at least the following beneficial effects:

[0035] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0036] Figure 1 A step flow chart of a power line tree barrier detection method based on AI image recognition provided by an embodiment of the present application is shown in the following figure:

[0037] Figure 2 An acquisition process diagram of the conductor suspected line is shown in the following figure. DETAILED DESCRIPTION

[0038] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined purposes, the following describes in detail the specific implementation, structure, features and effects of the AI image recognition-based power line tree barrier detection method and system according to the present application, with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0039] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0040] The application provides a kind of based on AI image recognition's electric power line tree barrier detection method and system, and the specific scheme of the method is specifically described below in conjunction with the drawings.

[0041] Please refer to Figure 1 It shows the step flow chart of a kind of based on AI image recognition's electric power line tree barrier detection method provided by an embodiment of the application, and the method comprises the following steps:

[0042] Step S1, the video image of the conductor between the tower is collected, and each frame image is obtained, and the image includes the tower image connected on both sides of the conductor.

[0043] When detecting the tree barrier in the electric power line corridor range, an optical camera is installed at the highest point of each tower, and the video image of the conductor between the tower is collected. The video image of the conductor includes the tower image connected on both sides of the conductor. It should be noted that for the placement position of the optical camera, the application only provides a placement method, and the implementer can also set the placement method of the camera according to the actual situation, and the application does not make specific limitations.

[0044] The collected video image is frame processed to obtain each frame image. The video frame is a known technology, and the specific process will not be repeated.

[0045] Since the image captured by the camera is perspective deformation, there will be a large error in directly judging the distance and position on the image captured by the camera, so the inverse perspective transformation is performed on each frame image to obtain the corrected each frame image and the corresponding relationship between the world coordinate system and the coordinate system of each frame image. The inverse perspective transformation is a known technology, and the specific process will not be repeated. It should be noted that in the embodiment of the application, the corrected each frame image can be regarded as the overhead view of the conductor between the adjacent towers.

[0046] Step S2, based on the edge line features in each frame image, the conductor suspected line in each frame image is obtained; in the current frame image, the initial contour of the Snake model is determined according to the position of the conductor in the image; the straight line prior energy of the Snake model is determined based on the trend of all conductor suspected lines in the current frame image; the scene constraint energy of the current iteration of the Snake model is calculated based on the proximity of the contour of the current iteration of the Snake model and the conductor suspected line.

[0047] The shape of the conductor in the power line is not completely straight due to the influence of gravity, wind and other factors, and may have a certain bending. The Snake model can be used to extract the conductor contour. However, in the actual power line scene image, the background is relatively complex, in addition to the conductor, there may also be trees, towers, other power equipment and the like. When facing a complex background, the traditional Snake model may be disturbed by the edges of other objects in the background, resulting in inaccurate contour extraction. For example, some parts of the tower may partially overlap with the conductor in the image, and the traditional Snake model may mistakenly include the tower edge into the conductor contour. Therefore, the Snake model is improved according to the actual application scene.

[0048] (1) For the corrected image, since it is a color image, a gray image of the color image is obtained by gray image conversion; further, an edge image of the gray image is obtained by a Canny edge detection algorithm, and each straight line in the edge image is obtained by a Hough detection algorithm. The gray image conversion, the Canny edge detection algorithm and the Hough detection algorithm are known technologies, and the specific process will not be described again.

[0049] It should be noted that for the acquisition of the edge image and the straight line in the image, the present application only provides an edge detection algorithm and a straight line detection algorithm, and there are many existing edge detection algorithms and straight line detection algorithms, and the implementer can also use other edge detection algorithms and straight line detection algorithms to acquire the edge image and the straight line in the image, respectively, and the present application does not make specific limitations.

[0050] (2) The straight lines in the edge image are screened to obtain conductor suspected lines, specifically:

[0051] In the region of any tower in the first image, a pixel point is randomly selected, denoted as A1(u1, v1); in the region of another tower in the first image, a pixel point corresponding to A1(u1, v1) is obtained, denoted as A2(u2, v2). Preferably, in the present application, the connecting points of the same conductor and the two side towers are respectively taken as the point A1(u1, v1) and the point A2(u2, v2); a straight line connected with the point A1(u1, v1) and the point A2(u2, v2) is obtained, denoted as a standard straight line; since the positions of the towers in the video image are basically unchanged, the position of the standard straight line in the first image is taken as the position of the standard straight line in each image.

[0052] For each straight line in the edge image, a straight line with a length greater than a preset length threshold and an included angle with the standard straight line less than a preset angle threshold is obtained as a conductor suspected line. Preferably, in the present application, the length threshold is set to 200 pixels, and the angle threshold is set to 15°. As other embodiments of the present application, the implementer can set the length threshold and the angle threshold according to the actual situation.

[0053] The conductors in the power line have a certain length feature in the image. The length greater than 200 pixels is set to exclude the interference of some short lines in the image, which can be noise or the edge of other small objects. Because the real conductors will present relatively long line segments in the camera field of view, through this length screening, the line segments that are more likely to be conductors can be preliminarily screened, and the detection accuracy is improved. The conductors of the power line are usually laid approximately horizontally, and the straight line with an angle deviation limited within 15° further narrows down the screening range and excludes the interference of line segments in other directions.

[0054] (3) In the current frame image, an initial contour of the Snake model is constructed according to the transverse length of the tower and the distance between adjacent towers, wherein the transverse length of the tower is taken as the width of the initial contour, and the distance between adjacent towers is taken as the length of the initial contour, so as to obtain the initial contour of the Snake model. It should be noted that the width and length of the initial contour need to be set to ensure that the initial contour contains all the conductors between adjacent towers. As other embodiments of the present application, the implementer can set the width and length of the initial contour according to the actual situation.

[0055] (4) In the image, the edge of an object is usually represented as a sharp change in pixel value, which can be quantified by the image gradient. The image gradient is a vector, and its size reflects the sharpness of the change in pixel value, and the direction points to the direction in which the pixel value changes most quickly. In each frame image, the difference in pixel value between the conductor and the background forms an obvious gradient change at the edge of the conductor.

[0056] In the power line scene, although the conductors can be bent to a certain extent due to gravity, wind and other factors, they still have an approximately straight trend as a whole. In the model, an energy term related to the theoretical trend of the conductor is introduced, so that the contour can refer to this prior information during the evolution process and optimize in the direction consistent with the trend of the conductor.

[0057] In the current frame image, first, for each conductor suspected line, the included angle between the conductor suspected line and the standard straight line is taken as the direction angle of the conductor suspected line; further, the straight line prior energy of the Snake model is obtained, and the expression is:

[0058]

[0059] In the formula, E line is the straight line prior energy of the Snake model, M is the number of conductor suspected lines in the current frame image, cos 2 () is the square of the calculated cosine function, represents the direction angle of the i-th conductor suspected line.

[0060] the value of the direction angle is closer to 1, otherwise, when the direction angle is larger, the value of E line is closer to 0. When the direction of a part of the contour deviates from the theoretical direction of the conductor, in order to minimize the total energy, the model will adjust the contour, so that the direction of this part of the contour changes to a direction close to the theoretical direction of the conductor, thereby guiding the evolution of the entire contour to a direction consistent with the conductor direction.

[0061] The straight line prior energy can fully utilize the prior knowledge of the conductor direction in the power line, can provide strong constraints for the model, reduce the uncertainty of the model in searching the conductor contour in a complex image environment, and improve the convergence speed and accuracy of the algorithm. On the other hand, it enhances the robustness of the model to noise and local interference. Even in the case of image noise, conductor occlusion or poor image quality, the straight line prior information can help the model to maintain the judgment of the overall direction of the conductor, so that the extracted conductor contour is closer to the real situation, and the reliability and stability of the model in practical application are improved.

[0062] (5) In the power line tree barrier detection, the Hough transform can quickly detect straight lines in the image, and can preliminarily locate the approximate position of the conductor in the conductor detection. However, the Hough transform has limitations in detecting curved conductors and cannot accurately outline the conductor contour. The present application fuses the coarse positioning result of the Hough transform by introducing a scene constraint term, so that the model can refer to the existing preliminary position information when optimizing the contour, avoid deviating too far from the real conductor position in the search process, effectively utilize the advantages of the Hough transform, and improve the overall detection accuracy.

[0063] Therefore, in the current frame image, the scene constraint energy of the current iteration of the Snake model is determined by the proximity of the conductor suspected line to the contour of the current iteration of the Snake model, and the expression is:

[0064]

[0065] In the formula, E prior is the scene constraint energy of the current iteration of the Snake model; M is the number of conductor suspected lines in the current frame image; N is the number of pixel points on the contour of the current iteration of the Snake model; D[] is a calculation function of the Euclidean distance; (x k ,y k ) is the coordinate of the kth pixel point on the contour of the current iteration of the Snake model; (x Hi ,y Hi ) is the coordinate of the midpoint of the ith conductor suspected line in the current frame image. In which the midpoint coordinate of the conductor suspected line is used instead of the whole position of the conductor suspected line.​

[0066] The sum of the squared Euclidean distances between each point on the current wire suspected line and the Hough transform positioning point reflects the difference between the current Snake model contour and the Hough transform rough positioning result. The greater the difference, the greater the value of E prior . During the model iteration process, when E prior is large, it means that the current contour deviates greatly from the Hough transform rough positioning result, which will have a great impact on the total energy, prompting the model to adjust the contour so that it is close to the Hough transform positioned wire position, thereby converging more quickly to the true wire contour.

[0067] The scene constraint energy effectively reduces the blindness of the model in the contour optimization process, making the finally extracted wire contour closer to the true situation. Especially when the wire part is blocked or the image quality is poor, the preliminary result of the Hough transform can be used to improve the reliability of detection. In terms of efficiency, since there is a rough positioning guide of the Hough transform, the Snake model does not need to search comprehensively in the entire image range, greatly reducing the search space, reducing the number of iterations, speeding up the contour extraction speed, improving the running efficiency of the algorithm, and being more suitable for the real-time demand of actual power line tree detection.

[0068] In step S3, the optimization energy function of the current iteration is obtained based on the straight line prior energy and the scene constraint energy, the gradient of the total energy of the current iteration is calculated, the evolution equation of each iteration is determined based on the gradient of the total energy, and the Snake model is combined to obtain the contour of each wire in the current frame image. The neighborhood of the wire contour is used as the detection area.

[0069] For the current frame image, the energy function in the original Snake model is optimized by the straight line prior energy and the scene constraint energy, and the expression is:

[0070] E total = α × E grad + β × E curve + γ × E line + δ × E prior

[0071] In the formula, E total is the optimization energy function of the current iteration, α, β, γ and δ are all preset energy weights, E grad , E curve are the gradient energy and curvature energy in the energy function of the original Snake model, E line is the straight line prior energy of the Snake model, and E prior is the scene constraint energy of the current iteration of the Snake model. The calculation of the gradient energy and the curvature energy in the energy function of the original Snake model is a known technology, and the specific process is not described again.

[0072] It should be noted that for the setting of each energy weight, preferably, in the embodiment of the present application, each energy weight is set to a = 0.4, b = 0.3, g = 0.2, d = 0.1. As other embodiments of the present application, the implementer can set the value of each energy weight according to the actual situation. Among them, a > b > g > d, and the sum of a, b, g and d is 1.

[0073] Further, the optimization energy function of the current iteration is taken as the energy function of the current iteration in the Snake model, and the gradient of the total energy of the current iteration is calculated. Wherein the gradient of the total energy of the Snake model is a known technology, and the specific process will not be repeated.

[0074] Further, the evolution equation in the contour iteration process is constructed by the gradient of the total energy, and the expression is:

[0075]

[0076] In the formula, represents the kth point on the contour of the t+1th iteration of the Snake model; represents the kth point on the contour of the tth iteration of the Snake model; and n represents the learning rate. is the gradient of the total energy of the current iteration.

[0077] Among them, for the learning rate n, the learning rate is a known content, which controls the speed and amplitude of the contour evolution, and the value range of n is 10 -6 ~1, preferably, in the embodiment of the present application, the value of n is set to 0.05; if the contour finds the correct update direction in the evolution process, but the update amplitude is too large to miss the optimal contour, the learning rate can be reduced; if the contour converges too slowly, the learning rate can be increased.

[0078] The condition for stopping iteration is set as stopping iteration when the average value of the displacement of all contour points between the current iteration and the previous iteration is less than a preset first threshold, or the maximum iteration number is reached. Preferably, in the embodiment of the present application, the first threshold is set to 0.1 pixel, and the maximum iteration number is set to 200 times. Thus, the contour of each conductor in the current frame image is obtained.

[0079] Taking each conductor contour as the center, respectively expanding 100 pixels to form a detection area, it should be noted that the size of the detection area can be set by the implementer according to the actual situation, and the present application does not make specific limitation.

[0080] Step S4, obtaining the vegetation pixels in each frame image through the deep learning model; and detecting the tree barrier of the power line based on the occurrence of the vegetation pixels in the detection area of each frame image.

[0081] A large amount of image data containing vegetation and non-vegetation is collected for labeling. The labeled data is input into a U-Net neural network model for training. The trained model can recognize vegetation in the image and segment the vegetation area in the image. The trained model is referred to as a tree barrier recognition model. The model training is a known technology, and the specific steps are not repeated. The U-Net neural network model is a known technology, and the specific process is not repeated.

[0082] It should be noted that for tree barrier recognition, the implementer can also use other deep learning models for training and tree barrier recognition, and the present application does not make specific limitations.

[0083] The tree barrier recognition model is used to recognize tree barriers in each frame of image. The recognized vegetation pixels in each frame of image are labeled in each frame of image. If a pixel labeled as a vegetation pixel appears in the detection area of each frame of image, it indicates that a tree barrier appears in the power line corridor, and an alarm is sent for removal.

[0084] In order to reduce the error of the deep learning model in detecting vegetation, such as false positives caused by mistaking non-vegetation pixels in the detection area as vegetation pixels, for each detection area in each frame of image, the average value of the coordinates of U vegetation pixels closest to the edge of the detection area is taken as a detection point. If the detection point is in the detection area, it is determined that a tree barrier appears in the power line corridor, and an alarm is sent. If the detection point is not in the detection area, no tree barrier appears in the power line corridor, and no alarm is sent. Preferably, in the present application, the value of U is set to 100. As other embodiments of the present application, the implementer can set the value of U according to actual conditions.

[0085] The acquisition process of the wire suspected line is shown in Figure 2 .

[0086] Based on the same inventive concept as the above method, the present application also provides an AI image recognition-based power line tree barrier detection system, which includes a memory, a processor, and a computer program stored in the memory and running on the processor. The processor executes the computer program to implement the steps of any one of the above AI image recognition-based power line tree barrier detection methods.

[0087] To sum up, the embodiment of the application provides a power line tree barrier detection method based on AI image recognition. The camera is arranged at the top of the tower, video monitoring is performed, video is collected through the camera, real-time detection can be realized, the detection real-time performance is improved, and the detection is not dependent on the inspection time. The image detected straight line is screened through the wire feature, the influence of irrelevant straight lines in the background information is reduced, the energy function of each iteration is optimized through the trend of all wire suspected lines in the current frame image and the proximity of the contour at each iteration of the Snake model, the Snake model is improved, the wire contour detection is performed through the improved Snake model, the power line tree barrier detection is performed based on the vegetation pixel appearance around the detected wire, the possibility that the Snake model deviates from the real wire position when the wire contour is acquired is reduced, the extraction precision and speed of the wire are improved, and then the accuracy of the tree barrier detection is improved.

[0088] It should be noted that the above-mentioned sequence of the embodiments of the application is only for description, and does not represent the advantages and disadvantages of the embodiments. The above describes the specific embodiments of the application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0089] Each of the embodiments in the application is described in a progressive manner, and the same and similar parts between the embodiments can be referred to each other. Each embodiment mainly describes the difference from other embodiments.

[0090] The above only describes the preferred embodiments of the application, and does not limit the application. Any modification, equivalent replacement, improvement, etc. within the principles of the application shall be included in the protection scope of the application.

Claims

1. An AI image recognition-based power line tree barrier detection method, characterized in that, The method comprises the following steps: Collecting video images of conductors between towers to obtain each frame image, and the image comprises tower images connected to both sides of the conductors; Obtaining conductor suspected lines in each frame image based on edge line features in the frame image; in the current frame image, determining an initial contour of a Snake model according to the position of the conductors in the image; determining a straight line prior energy of the Snake model based on the direction of all conductor suspected lines in the current frame image; calculating a scene constraint energy of the current iteration of the Snake model based on the proximity of the contour of the current iteration of the Snake model to the conductor suspected lines; Obtaining an optimization energy function of the current iteration based on the straight line prior energy and the scene constraint energy, calculating the gradient of the total energy of the current iteration; determining an evolution equation of each iteration based on the gradient of the total energy, and combining the Snake model to obtain each conductor contour in the current frame image, and taking the neighborhood of the conductor contour as a detection area; Obtaining vegetation pixels in each frame image through a deep learning model; and performing power line tree barrier detection based on the occurrence of the vegetation pixels in the detection area of each frame image; The process of obtaining the straight line prior energy is as follows: The included angle between the i-th wire suspected line and the standard straight line is taken as the direction angle of the i-th wire suspected line, denoted as ; the straight line prior energy is denoted as , The expression of the straight line prior energy is as follows: , wherein, is the number of wire suspected lines in the current frame image, is the square of the calculated cosine function; The expression of the scene constraint energy of the current iteration is as follows: wherein, is the scene constraint energy for the current iteration; is the number of wire suspect lines in the current frame image; is the number of contour pixels for the current iteration; is the Euclidean distance function; is the coordinate of the kth pixel on the contour for the current iteration; is the coordinate of the midpoint on the ith wire suspect line in the current frame image; The expression of the optimization energy function of the current iteration is as follows: wherein, is the optimization energy function of the current iteration, , , and are preset energy weights, , are gradient energy and curvature energy in the energy function in the original Snake model respectively, is the straight line prior energy, is the scene constraint energy of the current iteration; wherein, , and , , and the sum of which is 1.

2. The power line tree barrier detection method based on AI image recognition according to claim 1, wherein, The process of obtaining the conductor suspected lines in each frame image is as follows: Performing edge detection on the gray-scale images of each frame image to obtain each edge image, and performing straight line detection on each edge image to obtain each straight line in each edge image; Setting standard straight lines in each frame image; Taking a straight line with a length greater than a preset length threshold and an included angle with the standard straight line less than a preset angle threshold as a conductor suspected line. 3.The AI image recognition-based power line tree barrier detection method of claim 2, wherein, The process of setting the standard straight line is as follows: In the region of any tower in the first frame image, a pixel point is randomly selected, denoted as ; in the region of another tower in the first frame image, a pixel point corresponding to the position of is obtained, denoted as . A straight line connecting the points and is denoted as a standard straight line; the position of the standard straight line in the first frame image is taken as the position of the standard straight line in each frame image. 4.The AI image recognition-based power line tree barrier detection method of claim 1, wherein The process of obtaining the gradient of the total energy of the current iteration is as follows: Taking the optimization energy function of the current iteration as the energy function of the current iteration in the Snake model, and calculating the gradient of the total energy of the current iteration. 5.The AI image recognition-based power line tree barrier detection method of claim 1, wherein, The evolution equation of each iteration is as follows: wherein, denotes the kth point on the contour at the t+1th iteration; denotes the kth point on the contour at the tth iteration; denotes the learning rate; is the gradient of the total energy for the current iteration.

6. The AI image recognition-based power line tree barrier detection method of claim 1, wherein, The power line tree barrier detection based on the occurrence of the vegetation pixels in the detection area of each frame image is specifically as follows: For each detection area in each frame image, taking the average value of a preset number of vegetation pixel coordinates closest to the edge of the detection area as a detection point, if the detection point is in the detection area, a tree barrier appears in the power line corridor, and an alarm is issued; Otherwise, no tree barrier appears in the power line corridor.

7. An AI image recognition-based power line tree barrier detection system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, The processor executes the computer program to realize the steps of the method of any one of claims 1-6.

Images