Electrical equipment defect identification method, system and equipment and storage medium

Abstract

The invention discloses a power equipment defect identification method, system and device and a storage medium, and belongs to the technical field of power equipment defect identification. The method comprises the following steps: acquiring and processing a target infrared light image and a target visible light image of target power equipment to obtain an infrared light sub-image set and a visible light sub-image set; obtaining a feature similarity matrix, obtaining a target element set with the maximum sum of similarity values in the feature similarity matrix, carrying out key point calculation and optimization processing to obtain a target key point set, and constructing a mapping matrix of the target key point set; registering and fusing the target infrared light image and the target visible light image to obtain a target detection image; inputting the target detection image into the trained defect detection model for detection, and outputting a defect detection result of the target detection image; by combining the information of the infrared light and visible light images, the state of the power equipment can be reflected more comprehensively, so that the defect detection precision is improved.

Description

Technical Field

[0001] The present application belongs to the technical field of power equipment defect detection, and specifically relates to a power equipment defect identification method, system, device and storage medium. Background Art

[0002] Electricity is an indispensable element for economic development, social progress and industrial production, and the stable operation of the power system is of great significance. Once a power equipment fails, it will affect the safe and stable operation of the power system and even cause power supply interruption; therefore, it is of great and necessary significance to identify power equipment defects.

[0003] The existing technology for identifying defects in power equipment mainly relies on a single type of defect identification method (such as infrared light images or visible light images). This defect identification method may have low identification accuracy for defects in power equipment in complex environments, resulting in untimely maintenance and causing power system failures. Summary of the invention

[0004] The purpose of the embodiments of the present application is to provide a method, system, device and storage medium for identifying defects in power equipment, which can solve the technical problem of low recognition accuracy of existing methods for identifying defects in power equipment in complex environments.

[0005] In order to solve the above technical problems, this application is implemented as follows: In a first aspect, an embodiment of the present application provides a method for identifying defects in electric power equipment, the method comprising: Acquire a target infrared light image and a target visible light image of a target electric power device, and perform image processing on the target infrared light image and the target visible light image to obtain an infrared photon image set and a visible light photon image set; Obtaining a feature similarity matrix between the infrared photon image set and the visible photon image set, and obtaining a target element set having the largest sum of similarity values ​​in the feature similarity matrix according to the Hungarian algorithm; Performing key point calculation on the infrared photon image and the visible photon image corresponding to each element in the target element set to obtain a key point set corresponding to the target element set; Optimizing the key point set to obtain a target key point set, and constructing a mapping matrix of the target key point set; Registering the target infrared light image and the target visible light image through the mapping matrix, and fusing the registered target infrared light image and the target visible light image to obtain a target detection image; The target detection image is input into a trained defect detection model for detection, and the defect detection result of the target detection image is output.

[0006] As an optional implementation manner of the first aspect of the present application, the target infrared light image and the target visible light image of the target power equipment are acquired, and the target infrared light image and the target visible light image are image processed to obtain an infrared photon image set and a visible light sub-image set; specifically: Simultaneously photographing the target infrared light image and the target visible light image at the same position of the target power equipment by a dual-light coaxial camera; Inputting the target infrared light image and the target visible light image into the trained power equipment recognition model for detection, respectively, to identify the positions of the power equipment in the target infrared light image and the target visible light image; The target infrared light image and the target visible light image are segmented according to the position of the electric power equipment to obtain the corresponding infrared photon image set and the visible light sub-image set.

[0007] As an optional implementation of the first aspect of the present application, the obtaining of a feature similarity matrix between the infrared photon image set and the visible photon image set, and obtaining a target element set having the largest sum of similarity values ​​in the feature similarity matrix according to the Hungarian algorithm; specifically: Performing feature encoding processing on the infrared photon image set and the visible light sub-image set through an improved residual network to obtain an infrared light feature set and a visible light feature set corresponding to the infrared photon image set and the visible light sub-image set respectively; Constructing a feature similarity matrix between the infrared light feature set and the visible light feature set, wherein an element in the feature similarity matrix represents a similarity between an infrared light feature in the infrared light feature set and a visible light feature in the visible light feature set; Calculate the similarity value of each element in the feature similarity matrix by using the cosine similarity formula; According to the Hungarian algorithm, a target element set having the largest sum of similarity values ​​in the feature similarity matrix is ​​obtained. As an optional implementation of the first aspect of the present application, the infrared photon image and the visible light sub-image corresponding to each element in the target element set are subjected to key point calculation to obtain a key point set corresponding to the target element set; specifically: Using the SIFT feature matching algorithm to respectively calculate a subset of key points between the infrared photon image and the visible photon image corresponding to each element in the target element set; The key point subsets corresponding to each calculated element are merged to form the key point set. As an optional implementation of the first aspect of the present application, the key point set is optimized to obtain a target key point set, and a mapping matrix of the target key point set is constructed; specifically: The key point set is screened according to a preset ratio threshold to obtain a key point set that meets the requirements; Perform outlier filtering on the key point set that meets the requirements according to the RANSAC algorithm to obtain the target key point set; The target key point set is subjected to homography transformation according to the Homography function to obtain a mapping matrix. As an optional implementation of the first aspect of the present application, the target infrared light image and the target visible light image are registered by the mapping matrix, and the registered target infrared light image and the target visible light image are fused to obtain a target detection image; specifically: Acquire a matching position between the target visible light image and the target infrared light image according to a mapping matrix; transforming the matching position of the target visible light image according to an affine transformation function to obtain a registered visible light image that matches the target infrared light image; The registered visible light image and the target infrared light image are feature fused according to a feature fusion network to obtain a target detection image. As an optional implementation of the first aspect of the present application, the mapping matrix is ​​expressed by the following formula: , in, represents the mapping matrix, Represents the Homography function, represents the i-th infrared photon image, represents the jth visible light sub-image; The registered visible light image is expressed by the following formula: , in, represents the registered visible light image, represents a target visible light image; The target detection image is represented by the following formula: , in, represents the target detection image, represents the feature fusion network, Represents the target infrared image. In a second aspect, an embodiment of the present application provides a system for identifying defects in electric power equipment, the system comprising: An acquisition module, used for acquiring a target infrared image and a target visible light image of a target power device; A first processing module, used for performing image processing on the target infrared light image and the target visible light image to obtain a corresponding infrared photon image set and a visible light sub-image set; A second processing module is used to process the infrared photon image set and the visible photon image set to obtain a feature similarity matrix, and obtain a target element set with the largest sum of similarity values ​​in the feature similarity matrix according to the Hungarian algorithm; A third processing module is used to calculate key points of the infrared photon image and the visible photon image corresponding to each element in the target element set to obtain a key point set corresponding to the target element set; An optimization module, used for optimizing the key point set to obtain a target key point set, and constructing a mapping matrix of the target key point set; A registration and fusion module, used to register the target infrared light image and the target visible light image according to the mapping matrix, and fuse the registered target infrared light image and the target visible light image to obtain a target detection image; The detection module is used to detect the target detection image according to the trained defect detection model and output the defect detection result of the target detection image.

[0008] In a third aspect, an embodiment of the present application provides an electronic device, comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the method described in the first aspect.

[0009] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0010] In the embodiment of the present application, by combining the information of infrared light and visible light images, the state of the power equipment can be more comprehensively reflected, thereby improving the accuracy of defect detection; the entire detection process is highly automated, reducing the need for manual intervention and improving detection efficiency. At the same time, defect detection using a trained model realizes intelligent detection; through regular detection and analysis, potential defects of power equipment can be discovered in a timely manner, providing early warning for equipment maintenance and replacement, and reducing the probability of failure; compared with traditional manual detection methods, this method can significantly reduce detection costs while improving detection efficiency and quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 This is a flowchart of a method for identifying defects in electric power equipment provided in some embodiments of the present application. DETAILED DESCRIPTION

[0012] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0013] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here. In addition, the "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally represents that the objects associated with each other are in an "or" relationship.

[0014] In conjunction with the accompanying drawings, a method, system, device and storage medium for identifying defects in electric power equipment provided by an embodiment of the present application are described in detail below through specific embodiments and their application scenarios.

[0015] Example A method for identifying defects in electric power equipment comprises the following steps: S100: Acquire a target infrared light image and a target visible light image of a target power device, and perform image processing on the target infrared light image and the target visible light image to obtain an infrared photon image set and a visible light sub-image set; It should be understood that, first, images of the target power equipment under infrared light and visible light are acquired through specific equipment, and the infrared light image and the visible light image have the same spatial and temporal dimensions; the acquired images are processed to generate an infrared photon image set and a visible light sub-image set, and each image in these sets represents a part or feature of the original image.

[0016] Furthermore, in S100, a target infrared light image and a target visible light image of the target power equipment are obtained, and image processing is performed on the target infrared light image and the target visible light image to obtain an infrared photon image set and a visible light photon image set; specifically: S110: Using a dual-light coaxial camera, simultaneously capture a target infrared image and a target visible light image of the same position of the target power equipment; S120: inputting the target infrared light image and the target visible light image into the trained power equipment recognition model for detection, and identifying the position of the power equipment in the target infrared light image and the target visible light image; S130: Segment the target infrared light image and the target visible light image according to the position of the power equipment to obtain a corresponding infrared photon image set and a corresponding visible light sub-image set.

[0017] It should be understood that, based on the above-mentioned feature definitions, a dual-light coaxial camera is used, which can simultaneously capture infrared images and visible light images of the target power equipment at the same time and at the same location. This design ensures the consistency of the two images in space and time, and facilitates subsequent image processing and alignment; the captured target infrared image and target visible light image are respectively input into pre-trained power equipment recognition models, which can be based on deep learning or traditional image processing technology and can accurately identify the power equipment position in the image; the model processes the input image and outputs the position information of the power equipment in the image, such as a bounding box, center point coordinates, etc.; the target infrared image and target visible light image are segmented according to the position information output by the power equipment recognition model; the segmentation process involves steps such as image cropping and region extraction to obtain a set of infrared photon images and a set of visible light sub-images containing only the power equipment part; each image in these sets represents a specific part or feature of the power equipment under infrared and visible light.

[0018] It should be noted that the use of the dual-light coaxial camera ensures the spatial consistency of the infrared image and the visible light image, providing an accurate basis for subsequent image registration and fusion; through the trained power equipment recognition model, the position of the power equipment in the image can be quickly and accurately identified, reducing the need for manual labeling and improving processing efficiency; image segmentation is performed according to the identified position of the power equipment, and the characteristic information of the power equipment under different lighting conditions can be flexibly extracted, which facilitates subsequent analysis and processing; through the precise segmentation and processing of infrared images and visible light images, the characteristic information of the power equipment can be more accurately extracted, thereby improving the accuracy of defect detection.

[0019] S200: Obtain a feature similarity matrix between the infrared photon image set and the visible photon image set, and obtain a target element set having the largest sum of similarity values ​​in the feature similarity matrix according to the Hungarian algorithm; It should be understood that feature extraction methods (such as SIFT, SURF, etc.) are used to calculate the features of images in the infrared photon image set and the visible photon image set, and a feature similarity matrix is ​​constructed, which reflects the degree of similarity between images in different image sets; the Hungarian algorithm is applied to find the target element set with the largest sum of similarity values ​​in the feature similarity matrix. The Hungarian algorithm is a combinatorial optimization algorithm, which is often used to solve the maximum matching problem of bipartite graphs. In this embodiment, it is used to find the best feature matching pair set.

[0020] Furthermore, in S200, a feature similarity matrix between the infrared photon image set and the visible photon image set is obtained, and a target element set having the largest sum of similarity values ​​in the feature similarity matrix is ​​obtained according to the Hungarian algorithm; specifically: S210: performing feature encoding processing on the infrared photon image set and the visible light sub-image set through an improved residual network to obtain an infrared light feature set and a visible light feature set corresponding to the infrared photon image set and the visible light sub-image set respectively; S220: constructing a feature similarity matrix between the infrared light feature set and the visible light feature set, wherein an element in the feature similarity matrix represents a similarity between an infrared light feature in the infrared light feature set and a visible light feature in the visible light feature set; S230: Calculate the similarity value of each element in the feature similarity matrix by using the cosine similarity formula; S240: Obtain a target element set having the largest sum of similarity values ​​in the feature similarity matrix according to the Hungarian algorithm.

[0021] It should be understood that, based on the above-mentioned feature definition, an improved residual network (ResNet-50 network with the fully connected layer removed) is used to perform feature encoding on the infrared photon image set and the visible light photon image set; this network structure can effectively extract deep features in the image by stacking multiple residual blocks; the infrared photon image set and the visible light photon image set are respectively input into the improved residual network, and after layer-by-layer processing of the network, the corresponding infrared light feature set and visible light feature set are finally obtained; these feature sets contain key information of the power equipment in the image for subsequent feature matching; the elements in the matrix represent the similarity between a feature in the infrared light feature set and a feature in the visible light feature set; the similarity can be calculated in various ways, but is usually measured based on the distance or angle between feature vectors; this embodiment uses the cosine similarity formula to calculate the similarity value of each element in the feature similarity matrix; cosine similarity is a commonly used similarity measurement method, which evaluates their similarity by calculating the cosine value of the angle between two feature vectors; the range of the cosine similarity value is [-1, 1], where 1 means that the two eigenvectors are exactly the same, -1 means that the two eigenvectors are completely opposite, and 0 means that the two eigenvectors are unrelated; the feature similarity matrix is ​​used as input, and the Hungarian algorithm is applied to find the target element set with the largest sum of similarity values; the Hungarian algorithm is a classic algorithm for solving the maximum matching problem of bipartite graphs, which can find the target element set with the largest total matching weight (i.e., the sum of similarity values); in this process, the Hungarian algorithm traverses the feature similarity matrix, and finally finds the optimal matching solution by continuously adjusting and matching element pairs.

[0022] It should be noted that the improved residual network can efficiently extract deep features in the image, providing a reliable basis for subsequent feature matching; the cosine similarity formula can accurately evaluate the similarity between two feature vectors, providing an effective means for constructing a feature similarity matrix; the Hungarian algorithm can find the set of target elements with the largest sum of similarity values ​​in the feature similarity matrix, ensuring the accuracy and reliability of feature matching; through automated feature extraction, similarity calculation and matching processes, the efficiency of defect detection is significantly improved, reducing the need for manual intervention.

[0023] S300: Calculating key points of infrared photon images and visible photon images corresponding to each element in the target element set to obtain a key point set corresponding to the target element set; It should be understood that for each element in the target element set (i.e., matching pair), the key points in its corresponding infrared photon image and visible light sub-image are calculated. The key points are usually feature points in the image that are unique and stable, such as corner points, edges, etc.

[0024] Furthermore, in S300, key point calculation is performed on the infrared photon image and the visible photon image corresponding to each element in the target element set to obtain a key point set corresponding to the target element set; specifically: S310: using a SIFT feature matching algorithm to respectively calculate a subset of key points between the infrared photon image and the visible photon image corresponding to each element in the target element set; S320: Merge the key point subsets corresponding to each calculated element to form a key point set.

[0025] It should be understood that based on the above feature definition, SIFT (Scale-Invariant Feature SIFT (Scale Invariant Feature Transform) is a computer vision algorithm used to extract local features from images. These features are invariant to the scale, rotation and illumination changes of the image, so they are very suitable for image matching and recognition. For each element in the target element set, that is, each pair of matching infrared photon images and visible photon images, the SIFT algorithm is applied to detect and describe feature points. Scale-invariant feature points are detected in the image, which are usually significant features such as corner points and edge points in the image. A 128-dimensional feature vector is generated for each detected feature point, which describes the local image information around the feature point. Through the SIFT algorithm, a subset of key points between each pair of infrared photon images and visible photon images can be obtained. These key point subsets contain key feature points in the image and their corresponding feature vectors. The key point subsets corresponding to all elements in the target element set are merged to obtain a key point set containing all key feature points. This key point set contains all key feature points between the infrared photon image and the visible photon image, which can be used for subsequent image registration, fusion and defect detection tasks.

[0026] It should be noted that the feature points extracted by the SIFT algorithm have a certain invariance to the scale, rotation and illumination changes of the image, and therefore have high stability and robustness in practical applications; the feature points and feature vectors extracted by the SIFT algorithm can achieve accurate matching between infrared photon images and visible photon images, providing a reliable foundation for subsequent tasks; although the computational complexity of the SIFT algorithm is relatively high, with the support of modern computer hardware, it can still complete the key point extraction and matching tasks of large-scale images within a reasonable time.

[0027] S400: Optimizing the key point set to obtain a target key point set, and constructing a mapping matrix of the target key point set; It is important to understand that the calculated key point set is optimized to obtain a more accurate target key point set; the target key point set is used to construct a mapping matrix, which describes the correspondence between the key points of the infrared image and the visible light image, and is the basis for subsequent image registration and fusion.

[0028] Furthermore, in S400, the key point set is optimized to obtain a target key point set, and a mapping matrix of the target key point set is constructed; specifically: S410: Screening the key point set according to a preset ratio threshold to obtain a key point set that meets the requirements; S420: Perform outlier filtering on the key point set that meets the requirements according to the RANSAC algorithm to obtain a target key point set; S430: Perform homography transformation on the target key point set according to the Homography function to obtain a mapping matrix.

[0029] It should be understood that based on the above-mentioned feature definition, according to the preset ratio threshold (such as the number of key points, distribution density, etc.), the key point set is screened to obtain the key point set that meets the screening requirements; for example, an upper or lower limit of the number of key points can be set, or the key points can be screened according to the distribution density of the key points to ensure that the screened key point set is neither too sparse nor too dense; then a group of sample points are randomly selected from the key point set that meets the screening requirements as inliers through the RANSAC algorithm (random sampling consensus algorithm) for fitting the model; a model (such as a straight line, plane, etc.) is fitted using the inliers; the model is tested with the unselected key points according to certain rules (such as a distance threshold) Add the key points that meet the model to the inliers set; repeat the above steps multiple times (set the number of iterations), and each iteration may generate a new model; select the optimal model based on the number of inliers or the accuracy of the model; finally, those key points that do not meet the optimal model are regarded as outliers and filtered out, and the remaining ones are the target key point set; based on the target key point set, use the Homography function to perform homography transformation; homography transformation is a linear transformation between two-dimensional planes that can describe the correspondence between two images; by calculating the transformation relationship between matching key point pairs, a mapping matrix can be obtained, which can map the key points in one image to the corresponding positions in another image.

[0030] It should be noted that the mapping matrix is ​​constructed by using the Homography function to perform homography transformation; this mapping matrix can accurately describe the correspondence between the two images, thereby achieving accurate mapping of key points. This is crucial for applications such as image registration, image fusion, and target tracking; although the computational complexity of the RANSAC algorithm is high, by reasonably setting the number of iterations and the error threshold, the computational efficiency can be improved while ensuring accuracy. In addition, through the screening process, the amount of data that the RANSAC algorithm needs to process is reduced, further improving the overall computational efficiency.

[0031] S500: registering the target infrared image and the target visible light image through a mapping matrix, and fusing the registered target infrared image and the target visible light image to obtain a target detection image; It should be understood that the target infrared image and the target visible light image are registered through a mapping matrix so that they are aligned in space; the registered images are fused to generate a target detection image.

[0032] Furthermore, in S500, the target infrared image and the target visible light image are registered by a mapping matrix, and the registered target infrared image and the target visible light image are fused to obtain a target detection image; specifically: S510: Acquire matching positions of a target visible light image and a target infrared light image according to a mapping matrix; S520: transforming the matching position of the target visible light image according to the affine transformation function to obtain a registered visible light image that matches the target infrared light image; S530: Perform feature fusion on the registered visible light image and the target infrared light image according to the feature fusion network to obtain a target detection image.

[0033] Specifically, the mapping matrix in S510 is expressed by the following formula: , in, represents the mapping matrix, Represents the Homography function, represents the i-th infrared photon image, represents the jth visible light sub-image; The registered visible light image in S520 is expressed by the following formula: , in, represents the registered visible light image, represents a target visible light image; The target detection image in S530 is represented by the following formula: , in, represents the target detection image, represents the feature fusion network, Represents the target infrared image.

[0034] It should be understood that, based on the above-mentioned feature definition, the key point positions in the target infrared image are mapped to the target visible light image using a mapping matrix, so as to find the matching position on the visible light image corresponding to the infrared image; this ensures the spatial consistency of the infrared image and the visible light image, and lays the foundation for subsequent image fusion; after determining the matching position, the target visible light image is transformed using an affine transformation function so that it is completely aligned with the target infrared image; the affine transformation is a linear transformation that can maintain the linearity and parallelism of the image and is suitable for image registration tasks; by adjusting the parameters of the affine transformation, the feature points in the visible light image can be made to coincide with the corresponding feature points in the infrared image, so as to obtain the registered visible light image; after the image registration is completed, the feature fusion network is used to extract and fuse the features of the registered visible light image and the target infrared image; the feature fusion network is usually a deep learning model that can automatically learn the complementary information between the two images and fuse them into a unified feature representation; this fused feature representation contains all the important information of the two images and can be used for subsequent target detection, recognition or classification tasks; in this embodiment, the feature fusion network outputs the target detection image.

[0035] It should be noted that through the mapping matrix and affine transformation, the infrared image and the visible light image can be accurately registered to ensure the spatial consistency of the two images; this helps to reduce the error caused by image misalignment and improve the accuracy of subsequent target detection; the feature fusion network can make full use of the complementary information of infrared and visible light images to improve the accuracy and robustness of target detection. Infrared images are sensitive to temperature and can capture the thermal radiation information of objects; while visible light images provide rich texture and color information. By fusing these two types of information, the performance of target detection can be significantly improved.

[0036] S600: inputting the target detection image into the trained defect detection model for detection, and outputting the defect detection result of the target detection image; It is important to understand that the target detection image is fed into a trained defect detection model. The model can be a classifier or regressor based on deep learning or other machine learning methods that can identify defects or anomalies in the image; the model outputs the defect detection result of the target detection image.

[0037] According to a method for identifying defects in electric power equipment of this embodiment, the beneficial effects are as follows: By combining the information of infrared and visible light images, the status of power equipment can be more comprehensively reflected, thereby improving the accuracy of defect detection; the entire detection process is highly automated, reducing the need for manual intervention and improving detection efficiency. At the same time, defect detection using trained models has achieved intelligent detection; through regular detection and analysis, potential defects of power equipment can be discovered in a timely manner, providing early warning for equipment maintenance and replacement, and reducing the probability of failure; compared with traditional manual detection methods, this method can significantly reduce detection costs while improving detection efficiency and quality.

[0038] It should be noted that the execution subject of the method for identifying a defect of electric equipment provided in the embodiment of the present application may be a system for identifying a defect of electric equipment, or a control module in the system for identifying a defect of electric equipment for executing and loading a method for identifying a defect of electric equipment. In the embodiment of the present application, an example of a method for identifying a defect of electric equipment provided in the embodiment of the present application is taken to illustrate the method for identifying a defect of electric equipment provided in the embodiment of the present application by taking a system for identifying a defect of electric equipment for executing and loading a method for identifying a defect of electric equipment.

[0039] A power equipment defect identification system, comprising: An acquisition module, used for acquiring a target infrared image and a target visible light image of a target power device; A first processing module is used to perform image processing on the target infrared light image and the target visible light image to obtain a corresponding infrared photon image set and a visible light sub-image set; The second processing module is used to process the infrared photon image set and the visible photon image set to obtain a feature similarity matrix, and obtain a target element set with the largest sum of similarity values ​​in the feature similarity matrix according to the Hungarian algorithm; The third processing module is used to calculate key points of the infrared photon image and the visible photon image corresponding to each element in the target element set to obtain a key point set corresponding to the target element set; An optimization module, used for optimizing the key point set to obtain a target key point set, and constructing a mapping matrix of the target key point set; A registration and fusion module is used to register the target infrared image and the target visible light image according to the mapping matrix, and fuse the registered target infrared image and the target visible light image to obtain a target detection image; The detection module is used to detect the target detection image according to the trained defect detection model and output the defect detection result of the target detection image.

[0040] A power equipment defect identification system in the embodiment of the present application may be a device, or a component, integrated circuit or chip in a terminal. The device may be a mobile electronic device or a non-mobile electronic device. Exemplarily, the mobile electronic device may be a mobile phone, a tablet computer, a laptop computer, a PDA, an ultra-mobile personal computer (UMPC), a netbook or a personal digital assistant (PDA), etc., and the non-mobile electronic device may be a server, a network attached storage (NAS), a personal computer (PC), etc., which is not specifically limited in the embodiment of the present application.

[0041] A power equipment defect identification system in an embodiment of the present application may be a device having an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0042] The electric power equipment defect identification system provided in the embodiment of the present application can realize Figure 1 The various processes of the method embodiment of a method for identifying defects in electric power equipment are implemented, and to avoid repetition, they are not described here.

[0043] According to a power equipment defect identification system of this embodiment, its beneficial effects are as follows: Through image processing, the quality of infrared images and visible light images is improved, providing more accurate data for subsequent analysis; by calculating the feature similarity matrix and applying the Hungarian algorithm, feature matching between infrared photon images and visible light sub-images is achieved, providing a basis for subsequent key point calculation and mapping matrix construction; through key point calculation, optimization processing and mapping matrix construction, the accuracy and stability of image registration are improved; image fusion makes full use of the information of infrared images and visible light images, improves the accuracy and information content of target detection images, and thus enhances the ability of defect detection; through automated detection, the cost and risk of manual detection are reduced and the detection efficiency is improved.

[0044] Optionally, an embodiment of the present application also provides an electronic device, including a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, each process of the above-mentioned embodiment of the method for identifying defects in electric equipment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0045] An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, each process of the above-mentioned embodiment of the method for identifying defects in electric equipment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0046] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0047] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0048] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0049] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

Claims

1. A method for identifying defects in electric power equipment, characterized in that: The method comprises: Acquire a target infrared light image and a target visible light image of a target electric power device, and perform image processing on the target infrared light image and the target visible light image to obtain an infrared photon image set and a visible light photon image set; Obtaining a feature similarity matrix between the infrared photon image set and the visible photon image set, and obtaining a target element set having the largest sum of similarity values ​​in the feature similarity matrix according to the Hungarian algorithm; Performing key point calculation on the infrared photon image and the visible photon image corresponding to each element in the target element set to obtain a key point set corresponding to the target element set; Optimizing the key point set to obtain a target key point set, and constructing a mapping matrix of the target key point set; Registering the target infrared light image and the target visible light image through the mapping matrix, and fusing the registered target infrared light image and the target visible light image to obtain a target detection image; The target detection image is input into a trained defect detection model for detection, and the defect detection result of the target detection image is output.

2. A method for identifying defects in electric power equipment according to claim 1, characterized in that: The step of acquiring a target infrared light image and a target visible light image of the target power equipment, and performing image processing on the target infrared light image and the target visible light image to obtain an infrared photon image set and a visible light photon image set is as follows: Simultaneously photographing the target infrared light image and the target visible light image at the same position of the target power equipment by a dual-light coaxial camera; Inputting the target infrared light image and the target visible light image into the trained power equipment recognition model for detection, respectively, to identify the positions of the power equipment in the target infrared light image and the target visible light image; The target infrared light image and the target visible light image are segmented according to the position of the electric power equipment to obtain the corresponding infrared photon image set and the visible light sub-image set.

3. A method for identifying defects in electric power equipment according to claim 1, characterized in that: The step of obtaining a feature similarity matrix between the infrared photon image set and the visible photon image set, and obtaining a target element set having the largest sum of similarity values ​​in the feature similarity matrix according to the Hungarian algorithm is as follows: Performing feature encoding processing on the infrared photon image set and the visible light sub-image set through an improved residual network to obtain an infrared light feature set and a visible light feature set corresponding to the infrared photon image set and the visible light sub-image set respectively; Constructing a feature similarity matrix between the infrared light feature set and the visible light feature set, wherein an element in the feature similarity matrix represents a similarity between an infrared light feature in the infrared light feature set and a visible light feature in the visible light feature set; Calculate the similarity value of each element in the feature similarity matrix by using the cosine similarity formula; According to the Hungarian algorithm, a target element set having the largest sum of similarity values ​​in the feature similarity matrix is ​​obtained.

4. A method for identifying defects in electric power equipment according to claim 1, characterized in that: The key point calculation is performed on the infrared photon image and the visible photon image corresponding to each element in the target element set to obtain a key point set corresponding to the target element set; specifically: Using the SIFT feature matching algorithm to respectively calculate a subset of key points between the infrared photon image and the visible photon image corresponding to each element in the target element set; The key point subsets corresponding to each calculated element are merged to form the key point set.

5. A method for identifying defects in electric power equipment according to claim 1, characterized in that: The step of optimizing the key point set to obtain a target key point set and constructing a mapping matrix of the target key point set is as follows: The key point set is screened according to a preset ratio threshold to obtain a key point set that meets the requirements; Perform outlier filtering on the key point set that meets the requirements according to the RANSAC algorithm to obtain the target key point set; The target key point set is subjected to homography transformation according to the Homography function to obtain a mapping matrix.

6. A method for identifying defects in electric power equipment according to claim 1, characterized in that: The target infrared light image and the target visible light image are registered by the mapping matrix, and the registered target infrared light image and the target visible light image are fused to obtain the target detection image; specifically: Acquire a matching position between the target visible light image and the target infrared light image according to a mapping matrix; transforming the matching position of the target visible light image according to an affine transformation function to obtain a registered visible light image that matches the target infrared light image; The registered visible light image and the target infrared light image are feature fused according to a feature fusion network to obtain a target detection image.

7. A method for identifying defects in electric power equipment according to claim 6, characterized in that: The mapping matrix is ​​expressed by the following formula: ， in, represents the mapping matrix, Represents the Homography function, represents the i-th infrared photon image, represents the jth visible light sub-image; The registered visible light image is expressed by the following formula: ， in, represents the registered visible light image, represents a target visible light image; The target detection image is represented by the following formula: ， in, represents the target detection image, represents the feature fusion network, Represents the target infrared image.

8. A power equipment defect identification system, implementing a power equipment defect identification method as claimed in any one of claims 1 to 7, characterized in that: The system comprises: An acquisition module, used for acquiring a target infrared image and a target visible light image of a target power device; A first processing module, used for performing image processing on the target infrared light image and the target visible light image to obtain a corresponding infrared photon image set and a visible light sub-image set; A second processing module is used to process the infrared photon image set and the visible photon image set to obtain a feature similarity matrix, and obtain a target element set with the largest sum of similarity values ​​in the feature similarity matrix according to the Hungarian algorithm; A third processing module is used to calculate key points of the infrared photon image and the visible photon image corresponding to each element in the target element set to obtain a key point set corresponding to the target element set; An optimization module, used for optimizing the key point set to obtain a target key point set, and constructing a mapping matrix of the target key point set; A registration and fusion module, used to register the target infrared light image and the target visible light image according to the mapping matrix, and fuse the registered target infrared light image and the target visible light image to obtain a target detection image; The detection module is used to detect the target detection image according to the trained defect detection model and output the defect detection result of the target detection image.

9. An electronic device, characterized in that: It includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein when the program or instruction is executed by the processor, the steps of a method for identifying defects in electric power equipment as described in any one of claims 1 to 7 are implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, the steps of the method for identifying defects of electric power equipment as described in any one of claims 1-7 are implemented.

Images