Image analysis method and system for judging state of power transmission line

Abstract

The invention discloses an image analysis method and system for judging the state of a power transmission line, and relates to the technical field of image analysis, and the method comprises the steps: dividing image features into stable features and non-stable features through statistical values, dividing feature types through combining the stability of the image features and the correlation between the image features, and carrying out the classification of the features; and a reliable basis is provided for subsequent power transmission line state judgment. The foreground and background regions of the current image are separated by means of the stable features and the non-stable features, and the edge is adjusted by combining the distribution conditions of the stable features and the non-stable features, so that the foreground and background regions are accurately separated. The abnormal condition of the current state of the power transmission line is determined based on the current state and the abnormal state of the power transmission line, so that the specific category and condition of the abnormal state of the power transmission line are realized, and the accuracy and adaptability of power transmission line state image analysis are improved; therefore, an administrator can carry out maintenance, management and prevention work on the power transmission line in a targeted manner.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and particularly to an image analysis method and system for judging the state of a transmission line. Background Art

[0002] With the increasing requirements of the power system for safe and stable operation, the traditional manual inspection method is difficult to meet the large-scale, remote, and real-time diagnosis needs. Therefore, using image analysis technology to achieve automatic monitoring and fault diagnosis of transmission line status has become a research hotspot. This solution uses a helicopter or a robot to carry a high-definition camera to capture images of the transmission line, and uses technologies such as image preprocessing, feature extraction, target recognition and classification to intelligently detect and diagnose key components of the transmission line (such as power lines, poles, and insulators). At the same time, in response to challenges such as complex natural backgrounds and changing light conditions, adaptive exposure technology, noise filtering algorithms, etc. are used to optimize the image quality and improve the recognition accuracy and diagnosis efficiency. This technology not only improves the efficiency and safety of transmission line inspection, but also provides effective technical support for the rapid location of faults.

[0003] In the prior art, only image analysis is used to determine whether the transmission line is abnormal, but the specific abnormal category or other information cannot be determined, resulting in poor accuracy and adaptability of the image analysis of the transmission line status, which is not conducive to the maintenance and use of equipment such as transmission lines.

[0004] Therefore, how to improve the accuracy and adaptability of the image analysis of the transmission line status is a technical problem to be solved at present. Summary of the Invention

[0005] The purpose of the present invention is to solve the problem of poor accuracy and adaptability in the image analysis of the transmission line status in the prior art, and to propose an image analysis method for judging the state of a transmission line, which includes, Collect historical images of the transmission line, extract image features, and calculate the statistical values of each type of image feature. The image features are divided into two categories: stable features and unstable features through the statistical values; Define the abnormal state of the transmission line according to the stable features and unstable features; Collect the current image of the transmission line, separate the current image into foreground and background regions by virtue of the stable features and unstable features, and generate the current state of the transmission line according to the features of the foreground region; Based on the current state and abnormal state of the transmission line, determine the abnormal situation of the current state of the transmission line, so as to assist in the maintenance and management of the transmission line.

[0006] In some embodiments of the present application, extracting image features includes, Distinguish and label the foreground area, background area, normal state, and abnormal state in the historical images of the transmission line, determine the image features of the foreground area and background area, and extract the image features in the foreground area and background area respectively through image processing algorithms and tools.

[0007] In some embodiments of the present application, calculate the statistical values of each type of image feature, and classify the image features into two categories: stable features and unstable features through the statistical values, including, The statistical values include variance and mutual information; Calculate the variance of each type of image feature in the foreground area and background area, and determine the typical stable features of the foreground area and background area respectively through the variance of each type of image feature; Calculate the mutual information between other features and each typical stable feature respectively, obtain multiple mutual informations, sort them according to the size of the mutual information, and combine the variance of the typical stable feature ranked first and the variance of this other feature to obtain the new variance of this other feature; Use the new variance of other features to regard the other features and typical stable features that meet the requirements as stable features, and regard the other features that do not meet the requirements as unstable features.

[0008] In some embodiments of the present application, define the abnormal state of the transmission line according to stable features and unstable features, including, Determine the stable features and unstable features of the foreground area, and construct normal samples and abnormal samples of the foreground area through the normal state and abnormal state marked on the historical images; Under the normal samples and abnormal samples of the foreground area, capture the intersections of stable features and unstable features of each type to obtain a normal feature intersection and an abnormal feature intersection; Compare the normal feature intersection and the abnormal feature intersection to obtain an abnormal feature difference set, and describe the abnormal state of the transmission line according to the abnormal feature difference set; Among them, the abnormal feature difference set includes a fuzzy feature difference set and a specific feature difference set.

[0009] In some embodiments of the present application, separate the current image into foreground and background areas by virtue of stable features and unstable features, including, Identify the initial edges in the current image through the Canny edge detection algorithm, extract stable features and unstable features in the area near the initial edges, identify the positions of stable features and unstable features, divide the area near the initial edges into multiple sub-regions, and calculate the stable feature density of each sub-region; ; Among them, is the stable feature density of the th sub-region, is the number of stable features of the th sub-region, is the area size of the th sub-region, is the number of non-stable features of the th sub-region, is a preset constant; Statistically calculate the density of stable features for each sub-region, screen out the standard sub-regions by comparing the density of stable features, determine the center points of the standard sub-regions, and calculate the distances between the center points of each standard sub-region and the initial edge. Adjust the initial edge based on the distances to separate the foreground and background regions of the current image.

[0010] In some embodiments of the present application, based on the current state and abnormal state of the transmission line to determine the abnormal situation of the current state of the transmission line, including, Normalize the stable features and non-stable features, calculate the matching degrees of the stable features and non-stable features in the current state of the transmission line and the fuzzy feature difference set, and thereby calculate the abnormal degree of the current state of the transmission line; ; Among them, is the abnormal degree of the current state of the transmission line, , are respectively the numbers of stable features and non-stable features, , are respectively the th stable feature and the th non-stable feature combination weights, , are respectively the th stable feature and the th non-stable feature matching degree sizes, represents and the comprehensive value of the respective maximum values in, is a preset constant; Determine the abnormal situation of the current state of the transmission line according to the abnormal degree of the current state of the transmission line.

[0011] In some embodiments of the present application, determine the abnormal situation of the current state of the transmission line according to the abnormal degree of the current state of the transmission line, including, If the abnormal degree of the current state of the transmission line is less than the preset value, then calculate the state transition probability of the current state of the transmission line and the specific feature difference set, and output the abnormal situation through the state transition probability; Otherwise, calculate the compliance of the current state of the transmission line and the specific feature difference set, and output the abnormal situation through the compliance.

[0012] Correspondingly, the present application also provides an image analysis system for judging the state of a transmission line, including: A first module for collecting historical images of the transmission line, extracting image features, and calculating statistical values of each type of image feature, and classifying the image features into two categories: stable features and unstable features based on the statistical values; A second module for defining the abnormal state of the transmission line according to the stable features and unstable features; A third module for collecting the current image of the transmission line, separating the current image into foreground and background regions by virtue of the stable features and unstable features, and generating the current state of the transmission line according to the features of the foreground region; A fourth module for determining the abnormal situation of the current state of the transmission line based on the current state and abnormal state of the transmission line, so as to assist in the maintenance and management of the transmission line.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By classifying the image features into stable features and unstable features through statistical values, and combining the stability and correlation of the image features to divide the feature types, a reliable basis is provided for the subsequent determination of the state of the transmission line.

[0014] 2. By virtue of the stable features and unstable features, the current image is separated into foreground and background regions, and the edges are adjusted in combination with the distribution of the stable features and unstable features, so as to accurately separate the foreground and background regions. Based on the current state and abnormal state of the transmission line, the abnormal situation of the current state of the transmission line is determined, realizing the specific categories and situations of the abnormal state of the transmission line, improving the accuracy and adaptability of the image analysis of the transmission line state, and enabling the administrator to carry out targeted maintenance, management and prevention work on the transmission line. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a schematic flowchart of an image analysis method for judging the state of a transmission line proposed by the present invention; Figure 2 is a schematic structural diagram of an image analysis system for judging the state of a transmission line proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.

[0017] Referring to Figure 1 , an image analysis method for judging the state of a transmission line includes the following steps: Step S101, collect historical images of the transmission line, extract image features, and calculate the statistical value of each type of image features, and divide the image features into two categories: stable features and unstable features according to the statistical values.

[0018] In this embodiment, the statistical values ​​here are the variance and mutual information of the features. Variance is an important statistical indicator to measure the degree of data dispersion, which reflects the degree of deviation between each data and the mean. In image processing, variance can be used to describe the fluctuation of feature values. Mutual information is an information theory concept that measures the degree of correlation and dependence between two features.

[0019] In some embodiments of the present application, extracting image features includes: In the historical images of the transmission line, the foreground area and the background area as well as the normal state and the abnormal state are distinguished and marked, the image features of the foreground area and the background area are determined, and the image features are extracted in the foreground area and the background area respectively through image processing algorithms and tools.

[0020] In this embodiment, the collected images are subjected to preprocessing operations such as denoising and enhancement to improve the image quality. The image format and resolution are unified to facilitate subsequent processing and analysis. Foreground area: In the historical images of the transmission line, the foreground area usually refers to the transmission line and its related components, such as conductors, insulators, transmission towers, etc. These components are the objects that need to be focused on and analyzed in the image. Background area: The background area refers to the part outside the foreground area, which may include the sky, the ground, vegetation, buildings, etc. These parts usually play the role of setting off the foreground area in the image. Annotation method: In the image, different colors or lines can be used to distinguish and annotate the foreground area and the background area. For example, a red rectangular frame can be used to annotate the foreground area (transmission line and its components), while the background area remains the same or is filled with other colors. According to the characteristics of the transmission line, representative image features are selected, such as color features, texture features, shape features, etc. Color features can be extracted by methods such as color histograms and color moments; texture features can be extracted by methods such as grayscale co-occurrence matrices and wavelet transforms; shape features can be obtained by methods such as edge detection and contour extraction.

[0021] It can be understood that the normal state: indicates that the state quantities of the line are stable and within the warning value and attention value (referred to as the standard limit) specified in the regulations, and can operate normally. Under normal conditions, the appearance and performance of the transmission line and its components meet the design requirements, and there is no abnormality or damage. Abnormal state: indicates that some important state quantities of the line are close to or slightly exceed the standard value, and it is necessary to monitor the operation and arrange maintenance in a timely manner. Abnormal conditions may include various types, such as broken wire strands, self-explosion of insulators, and tilting of transmission towers. These abnormal conditions may pose a threat to the safe operation of transmission lines. Marking method: For abnormal conditions, it is necessary to clearly mark the type, location and other information of the abnormality in the image. For example, different colors of circles, triangles and other shapes can be used to mark different types of abnormal conditions, and text descriptions of the type and location of the abnormality can be attached next to them. At the same time, the abnormal conditions can also be graded and marked to better assess their impact on the safe operation of the transmission line.

[0022] In some embodiments of the present application, the statistical value of each type of image feature is calculated, and the image features are divided into two categories: stable features and unstable features by the statistical value, including: Statistical values ​​include variance and mutual information; Calculate the variance of each type of image features in the foreground area and the background area, and determine the typical stable features of the foreground area and the background area through the variance of each type of image features; Calculate the mutual information between other features and each typical stable feature to obtain multiple mutual information, sort them according to the mutual information, combine the variance of the first ranked typical stable feature with the variance of the other feature, and obtain the new variance of the other feature; Through the new variance of other features, other features that meet the requirements and typical stable features are regarded as stable features, and other features that do not meet the requirements are regarded as unstable features.

[0023] In this embodiment, typical stable features include basic features such as color, material, and shape of the line. The correlation dependence (mutual information) between other features and these typical stable features is calculated, and the variance of the typical stable feature with the maximum mutual information is selected and combined with the variance of the other features to obtain the new variance of the other features (weighted sum or average). If the new variance meets the requirements, the other features are used as stable features. Stable features are those that have small changes and are easy to identify, or those that have strong correlation and little change. These features play an important role in helping image analysis. Unstable features include features that have large changes and are difficult to identify, such as shadows and changes in lighting.

[0024] It should be noted that image features not only include direct features such as color, brightness, shape, texture, etc., but also can be indirect features obtained through these direct features, such as edge features, direction features, etc. An edge is a place where the brightness or color in an image changes significantly, and usually represents the contour of an object. In the image of a transmission line, edge features can help identify the line's orientation, connection points, damaged areas, etc. Directional features refer to the directionality of the texture or shape in an image. In the transmission line image, directional features can be used to identify the line's orientation, determine whether the line is distorted or offset, and detect directional defects such as cracks or scratches on the line.

[0025] Step S102, define the abnormal state of the transmission line according to the stable features and unstable features.

[0026] In this embodiment, the abnormal state of the transmission line is described by stable features and unstable features, which can include the numerical values or text descriptions of the features (such as shape features, etc.).

[0027] In some embodiments of the present application, defining the abnormal state of the transmission line according to the stable features and unstable features includes: Determine the stable features and unstable features of the foreground area, and construct normal samples and abnormal samples of the foreground area through the normal state and abnormal state marked on the historical images; Under the normal samples and abnormal samples of the foreground area, capture the intersection of the stable features and unstable features of each category to obtain the normal feature intersection and the abnormal feature intersection; Compare the normal feature intersection and the abnormal feature intersection to obtain the abnormal feature difference set, and describe the abnormal state of the transmission line according to the abnormal feature difference set; Among them, the abnormal feature difference set includes a fuzzy feature difference set and a specific feature difference set.

[0028] In this embodiment, for the features of the same category, respectively determine the common points (intersection) in the normal state and the abnormal state. For example, for numerical features, take the intersection of multiple numerical ranges as the common points. Compare the normal feature intersection and the abnormal feature intersection, and find the difference between the two, that is, the abnormal feature difference set. Integrate all the abnormal feature difference sets in the abnormal state together to form the fuzzy feature difference set of the abnormal state of the transmission line, divide the abnormal state by category, and integrate the abnormal feature difference sets according to each abnormal category to form the specific feature difference set.

[0029] It can be understood that the fuzzy feature difference set describes the comprehensive abnormal situation in all abnormal states, and the specific feature difference set describes the abnormal situation in different abnormal category states.

[0030] Step S103: Collect the current image of the transmission line, separate the current image into foreground and background regions based on stable and unstable features, and generate the current state of the transmission line according to the features of the foreground region.

[0031] In this embodiment, using image processing algorithms such as threshold segmentation, edge detection, and region growing, the current image is divided into foreground and background regions (the first separation). The foreground region mainly includes the transmission line and its key components, while the background region includes irrelevant information such as the sky and the ground. The complexity of the background region may make it difficult to accurately identify the targets in the foreground region. For example, when the background of the transmission line image is a complex natural environment (such as mountains and forests) or an urban landscape, the objects in the background and the lines and equipment in the foreground may intersect with each other, making image segmentation and target extraction difficult. This interference may cause the algorithm to misidentify the objects in the background as foreground targets or miss key information in the foreground region.

[0032] Implementation process of Canny edge detection algorithm Grayscale conversion: Convert the original image into a grayscale image because edge detection mainly depends on the brightness change of the image rather than color.

[0033] Gaussian filtering: Perform Gaussian filtering on the grayscale image to reduce the influence of noise on edge detection. Gaussian filtering smooths the image to reduce noise while retaining the main structure of the image.

[0034] Calculate the gradient: Use Sobel operator or Prewitt operator, etc. to calculate the gradient magnitude and direction of the image. The gradient magnitude represents the rate of change of the image brightness, and the gradient direction represents the direction of change.

[0035] Non-maximum suppression: In the gradient magnitude image, retain the local maximum values and suppress the non-maximum values to zero. This step helps to refine the edges and make the edges more accurate.

[0036] Double-threshold detection: Set two thresholds: a high threshold and a low threshold. Mark the pixels with gradient magnitude greater than the high threshold as strong edges, mark the pixels with gradient magnitude between the low threshold and the high threshold as weak edges, and suppress the remaining pixels to zero.

[0037] Edge connection: Form continuous edge contours by connecting strong edges and weak edges. This step is usually implemented using a hysteresis tracking algorithm.

[0038] In some embodiments of the present application, the current image is separated into foreground and background regions by virtue of stable features and unstable features, including, The initial edges in the current image are identified through the Canny edge detection algorithm, and stable features and unstable features are extracted in the regions near the initial edges, and the positions of the stable features and unstable features are identified. The regions near the initial edges are divided into multiple sub-regions, and the stable feature density of each sub-region is calculated; ; Among them, is the stable feature density of the th sub-region, is the number of stable features of the th sub-region, is the area size of the th sub-region, is the number of unstable features of the th sub-region, is a preset constant; The stable feature density of each sub-region is statistically analyzed, the standard sub-regions are screened out by comparing the stable feature densities, the center points of the standard sub-regions are determined, and the distances between the center points of each standard sub-region and the initial edges are calculated. The initial edges are adjusted by the distances, so as to separate the foreground and background regions of the current image.

[0039] In this embodiment, the initial edges may not be accurate, which affects the accuracy of subsequent image analysis. Here, the initial edges are adjusted by stable features and unstable features to cover as many stable features as possible. represents the correction of the initial stable feature density (only stable features) by unstable features. The more unstable features, the smaller the stable feature density. The standard sub-regions are screened out by comparing the stable feature densities, and the relative size feature densities after comparison are obtained for region screening. The sub-regions with higher feature densities are used as the standard sub-regions, and the initial edges are adjusted by the distances between the standard sub-regions and the initial edges. The distance formula (such as Euclidean distance) is used to calculate the distances between the center points of each standard sub-region and the nearest initial edge points. More complex distance metrics, such as geodesic distance, can also be considered to handle curved edges. For example, when the center point of a certain standard sub-region is far from the initial edge, the edge is moved in that direction. An iterative method can be used to gradually adjust the edge until a certain stopping condition is met (such as the edge position no longer changes significantly). The adjusted edge is used as the dividing line to divide the image pixels into foreground and background categories. Post-processing can be performed on the separation results, such as removing small noise regions and smoothing the edges.

[0040] Step S104: Determine the abnormal conditions of the current state of the transmission line based on the current state and abnormal state of the transmission line, so as to assist in the maintenance and management of the transmission line.

[0041] In this embodiment, first, the degree of abnormality of the current state of the transmission line is judged by the fuzzy feature difference set. If the degree of abnormality is relatively high, then further judge what type of abnormality it is through the specific feature difference set. If the degree of abnormality is relatively low, calculate the state transition probability, that is, predict the risk probability of future abnormalities, so as to issue a warning.

[0042] In some embodiments of the present application, determining the abnormal conditions of the current state of the transmission line based on the current state and abnormal state of the transmission line includes: Normalize the stable features and unstable features, calculate the matching degrees of the stable features and unstable features of the current state of the transmission line and the fuzzy feature difference set, and thus calculate the degree of abnormality of the current state of the transmission line; ; Wherein, is the degree of abnormality of the current state of the transmission line, and are the numbers of stable features and unstable features respectively, and are the combination weights of the th stable feature and the th unstable feature respectively, and are the matching degree sizes of the th stable feature and the th unstable feature respectively, represents and the comprehensive value of the respective maximum values in is a preset constant; Determine the abnormal conditions of the current state of the transmission line according to the degree of abnormality of the current state of the transmission line.

[0043] Determine the abnormal conditions of the current state of the transmission line according to the degree of abnormality of the current state of the transmission line.

[0044] In this embodiment, calculate the matching degrees of each stable feature and unstable feature of the current state with the fuzzy feature difference set, and combine the stable features and unstable features. represents and the correction of the sum of feature matches by the comprehensive value (weighted summation) of the respective maximum values in

[0045] In some embodiments of the present application, determining the abnormal conditions of the current state of the transmission line according to the degree of abnormality of the current state of the transmission line includes: If the abnormality degree of the current state of the transmission line is less than the preset value, calculate the state transition probability between the current state of the transmission line and the specific feature difference set, and output the abnormal situation through the state transition probability; Otherwise, calculate the compliance between the current state of the transmission line and the specific feature difference set, and output the abnormal situation through the compliance.

[0046] In this embodiment, if the abnormality degree of the current state of the transmission line is less than the preset value, it indicates that the abnormality degree is low or the state is normal, then calculate the possibility of the current state transitioning to the abnormal state, (such as Markov model, Bayesian network, etc.) to calculate the state transition probability. Otherwise, it indicates that the abnormality degree is large, that is, the current state is abnormal. At this time, specifically analyze which abnormal category and other situations, and use similarity measurement methods (such as cosine similarity, Jaccard similarity coefficient, etc.) to calculate the compliance.

[0047] Correspondingly, the present application also provides an image analysis system for judging the state of a transmission line, including, The first module is used to collect historical images of the transmission line, extract image features, and calculate the statistical values of each type of image feature, and classify the image features into two categories: stable features and unstable features through the statistical values; The second module is used to define the abnormal state of the transmission line according to the stable features and unstable features; The third module is used to collect the current image of the transmission line, separate the current image into foreground and background regions by virtue of stable features and unstable features, and generate the current state of the transmission line according to the features of the foreground region; The fourth module is used to determine the abnormal situation of the current state of the transmission line based on the current state and abnormal state of the transmission line, so as to assist in the maintenance and management of the transmission line.

[0048] Compared with the prior art, the beneficial effects of the present invention are: 1. Classify the image features into stable features and unstable features through statistical values, and divide the feature types by combining the stability and correlation of the image features, providing a reliable basis for the subsequent determination of the state of the transmission line.

[0049] 2. Separate the current image into foreground and background regions by virtue of stable features and unstable features, and adjust the edges by combining the distribution of stable features and unstable features, so as to accurately separate the foreground and background regions. Determine the abnormal situation of the current state of the transmission line based on the current state and abnormal state of the transmission line, realize the specific categories and situations of the abnormal state of the transmission line, improve the accuracy and adaptability of the image analysis of the transmission line state, and enable the administrator to carry out maintenance, management and prevention work on the transmission line targeted.

[0050] Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented by hardware or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various implementation scenarios of the present invention.

[0051] Those skilled in the art can understand that the drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the drawings are not necessarily essential for implementing the present invention.

[0052] Those skilled in the art can understand that the modules in the system in the implementation scenario can be distributed in the system of the implementation scenario according to the description of the implementation scenario, or can be correspondingly changed and located in one or more systems different from this implementation scenario. The modules in the above implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0053] As mentioned above, the above are only the preferred specific implementation manners of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent replacements or changes, and should be covered by the protection scope of the present invention.

Claims

1. An image analysis method for judging the state of a power transmission line, characterized in that: include, Collect historical images of transmission lines, extract image features, and calculate the statistical value of each type of image features. The image features are divided into two categories: stable features and unstable features based on the statistical values. Define abnormal states of transmission lines based on stable and unstable characteristics; Collect the current image of the transmission line, separate the foreground and background areas of the current image by using stable features and unstable features, and generate the current state of the transmission line according to the features of the foreground area; Based on the current state and abnormal state of the transmission line, the abnormal situation of the current state of the transmission line is determined to help the maintenance and management of the transmission line.

2. The image analysis method for determining the state of a power transmission line according to claim 1, characterized in that: Extract image features, including, In the historical images of the transmission line, the foreground area and the background area as well as the normal state and the abnormal state are distinguished and marked, the image features of the foreground area and the background area are determined, and the image features are extracted in the foreground area and the background area respectively through image processing algorithms and tools.

3. The image analysis method for determining the state of a power transmission line according to claim 2, characterized in that: And calculate the statistical value of each type of image features, and divide the image features into two categories: stable features and unstable features through the statistical values. include, Statistical values ​​include variance and mutual information; Calculate the variance of each type of image features in the foreground area and the background area, and determine the typical stable features of the foreground area and the background area through the variance of each type of image features; Calculate the mutual information between other features and each typical stable feature to obtain multiple mutual information, sort them according to the mutual information, combine the variance of the first ranked typical stable feature with the variance of the other feature, and obtain the new variance of the other feature; Through the new variance of other features, other features that meet the requirements and typical stable features are regarded as stable features, and other features that do not meet the requirements are regarded as unstable features.

4. The image analysis method for determining the state of a power transmission line according to claim 3, characterized in that: The abnormal state of the transmission line is defined according to the stable and unstable characteristics, including: Determine the stable features and unstable features of the foreground area, and construct normal samples and abnormal samples of the foreground area through the normal and abnormal states marked on the historical images; Under the normal samples and abnormal samples in the foreground area, the intersection of stable features and unstable features of each category is captured to obtain the intersection of normal features and the intersection of abnormal features; Compare the intersection of normal features and the intersection of abnormal features to obtain the abnormal feature difference set, and describe the abnormal state of the transmission line according to the abnormal feature difference set; Among them, the abnormal feature difference set includes fuzzy feature difference set and specific feature difference set.

5. The image analysis method for determining the state of a power transmission line according to claim 3, characterized in that: Separate the foreground and background areas of the current image by using stable features and unstable features, including: The initial edge in the current image is identified by the Canny edge detection algorithm, stable features and unstable features are extracted in the area near the initial edge, and the positions of stable features and unstable features are identified. The area near the initial edge is divided into multiple sub-areas, and the stable feature density of each sub-area is calculated; ； in, For the The stable feature density of each sub-region, For the The number of stable features in each sub-region, For the The size of the sub-region, For the The number of unstable features in each sub-region, is a preset constant; The stable feature density of each sub-region is counted, and the standard sub-region is screened out by comparing the stable feature density. The center point of the standard sub-region is determined, and the distance between the center point of each standard sub-region and the initial edge is calculated. The initial edge is adjusted by the distance to separate the foreground and background areas of the current image.

6. The image analysis method for determining the state of a power transmission line according to claim 3, characterized in that: Determine the abnormality of the current state of the transmission line based on the current state and abnormal state of the transmission line, include, Normalize the stable features and unstable features, calculate the matching degree between the stable features and unstable features of the current state of the transmission line and the fuzzy feature difference set, and calculate the abnormality of the current state of the transmission line; ； in, is the abnormality degree of the current state of the transmission line, , and the number of stable and unstable features, respectively. , Respectively Stable features and The combined weights of the non-stable features, , Respectively Stable features and The matching degree of the non-stable features is express and The combined value of the respective maximum values ​​in is a preset constant; The abnormality of the current state of the power transmission line is determined according to the degree of abnormality of the current state of the power transmission line.

7. The image analysis method for determining the state of a power transmission line according to claim 6, characterized in that: Determining the abnormality of the current state of the transmission line according to the abnormality degree of the current state of the transmission line, including: If the abnormality of the current state of the transmission line is less than the preset value, the state transition probability between the current state of the transmission line and the specific feature difference set is calculated, and the abnormal situation is output through the state transition probability; Otherwise, the conformity between the current state of the transmission line and the specific feature difference set is calculated, and the abnormal situation is output through the conformity.

8. An image analysis system for judging the state of a power transmission line, characterized in that: include, The first module is used to collect historical images of transmission lines, extract image features, and calculate the statistical value of each type of image features. The image features are divided into two categories: stable features and unstable features through the statistical values; The second module is used to define the abnormal state of the transmission line according to the stable characteristics and the unstable characteristics; The third module is used to collect the current image of the transmission line, separate the foreground and background areas of the current image by using the stable features and the unstable features, and generate the current state of the transmission line according to the features of the foreground area; The fourth module is used to determine the abnormal situation of the current state of the transmission line based on the current state and abnormal state of the transmission line, so as to help the maintenance and management of the transmission line.

Images