Joint detection method and device for insulator defects, electronic equipment and medium

Abstract

The application relates to a combined detection method and device for defects of an insulator, electronic equipment and a medium. The method comprises the following steps: acquiring an ultraviolet image of an insulator to be detected, preprocessing the ultraviolet image to obtain an ultraviolet preprocessed image containing a discharge light spot area; acquiring a visible light image of the insulator to be detected, preprocessing the visible light image and extracting feature values to obtain a visible light preprocessed image and visible light image feature values; constructing an ultraviolet discharge spectrum of the insulator based on the ultraviolet preprocessed image, and determining a defect discharge level according to the ultraviolet discharge spectrum; fusing the ultraviolet discharge spectrum and the visible light image feature values, the ultraviolet preprocessed image and the visible light preprocessed image to obtain a fusion image and position information of the insulator defects; and adopting a pre-trained insulator defect detection deep learning model to perform defect recognition on the fusion image to obtain an insulator defect level. The application improves the detection accuracy of deteriorated insulators.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system detection and monitoring, and particularly relates to a combined detection method and device for insulator defects, an electronic device and a medium. BACKGROUND

[0002] In a power system, the safe and stable operation of a transmission line is crucial for ensuring social and economic activities and daily life of residents. As a key component in a transmission line, an insulator mainly functions to support a conductor and prevent current from returning to the ground, ensuring the safety and reliability of power transmission. However, during long-term operation, an insulator is affected by various external factors, such as strong electric fields, ultraviolet radiation, temperature changes, humidity, and pollutant deposition, which may cause defects such as surface contamination, aging, cracking, and corrosion of the insulator, thereby affecting its insulating performance and increasing the risk of failure of the transmission line.

[0003] Currently, detection of insulator defects usually adopts visual detection, electrical testing, ultrasonic detection, laser scanning, or single ultraviolet thermal imaging technology, but these technologies generally have problems such as high cost of detection equipment, need for special equipment and personnel, strict detection conditions, poor real-time performance, and low accuracy of single ultraviolet thermal imaging. SUMMARY

[0004] To overcome or partially overcome the deficiencies of the prior art, the embodiments of the present application provide a combined detection method and device for insulator defects, an electronic device and a medium.

[0005] In a first aspect, the present application provides a detection method for insulator defects, comprising:

[0006] obtaining an ultraviolet image of an insulator to be detected, pre-processing the ultraviolet image to obtain an ultraviolet pre-processed image containing a discharge spot area;

[0007] obtaining a visible light image of the insulator to be detected, pre-processing the visible light image and extracting feature values to obtain a visible light pre-processed image and visible light image feature values;

[0008] based on the ultraviolet pre-processed image, constructing an ultraviolet discharge spectrum of the insulator, and determining a defect discharge level according to the ultraviolet discharge spectrum;

[0009] based on the ultraviolet discharge spectrum and the visible light image feature values, fusing the ultraviolet pre-processed image and the visible light pre-processed image to obtain a fused image and position information of the insulator defects;

[0010] using a pre-trained insulator defect detection deep learning model to perform defect recognition on the fused image to obtain an insulator defect level.

[0011] In a second aspect, the application provides a device for joint detection of defects of an insulator, the device comprising:

[0012] a first obtaining unit configured to obtain an ultraviolet image of an insulator to be detected, and to obtain an ultraviolet preprocessed image containing a discharge light spot region by preprocessing the ultraviolet image;

[0013] a second obtaining unit configured to obtain a visible light image of the insulator to be detected, and to obtain a visible light preprocessed image and visible light image feature values by preprocessing and extracting feature values of the visible light image;

[0014] a spectrum constructing unit configured to construct an ultraviolet discharge spectrum of the insulator based on the ultraviolet preprocessed image, and to determine a defect discharge level according to the ultraviolet discharge spectrum;

[0015] a fusion unit configured to fuse the ultraviolet preprocessed image and the visible light preprocessed image based on the ultraviolet discharge spectrum and the visible light image feature values, to obtain a fused image and position information of defects of the insulator;

[0016] a prediction unit configured to perform defect recognition on the fused image by using a pre-trained deep learning model for detection of defects of the insulator, to obtain a defect level of the insulator.

[0017] In a third aspect, an electronic device is provided, comprising: a processor; and a memory arranged to store computer executable instructions that, when executed, cause the processor to perform any of the above joint detection methods of defects of an insulator.

[0018] In a fourth aspect, a computer readable storage medium is provided, which stores one or more programs, which, when executed by an electronic device comprising a plurality of applications, cause the electronic device to perform any of the above joint detection methods of defects of an insulator.

[0019] The above at least one technical solution adopted by the embodiments of the application can achieve the following beneficial effects:

[0020] This application acquires an ultraviolet (UV) image of the insulator to be tested, preprocesses the UV image to obtain a UV preprocessed image containing the discharge spot region, acquires a visible light image of the insulator to be tested, preprocesses the visible light image and extracts feature values ​​to obtain a visible light preprocessed image and visible light image feature values, then constructs the UV discharge spectrum of the insulator based on the UV preprocessed image, and determines the defect discharge level according to the UV discharge spectrum, and fuses the UV preprocessed image and the visible light preprocessed image based on the UV discharge spectrum and the visible light image feature values ​​to obtain a fused image and the location information of the insulator defect, and finally uses a pre-trained deep learning model for insulator defect detection to identify defects in the fused image to obtain the insulator defect level. The defect discharge level and the insulator defect level together constitute the detection result. Compared with existing technologies, this invention uses ultraviolet and visible light images to jointly identify the defect level of insulators. By analyzing the discharge characteristics of the equipment, it improves the accuracy of detecting insulator defects, greatly enhances the effectiveness of insulator detection, and reduces the limitations of using a single device. It not only improves the detection accuracy of deteriorated insulators, but also has significant implications for saving labor. Attached Figure Description

[0021] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0022] Figure 1 A flowchart illustrating a joint detection method for insulator defects according to an embodiment of this application is shown;

[0023] Figure 2 An ultraviolet image acquired according to an embodiment of this application is shown;

[0024] Figure 3 This illustration shows an image after converting an ultraviolet image into a binary image according to an embodiment of this application;

[0025] Figure 4 A schematic diagram of the structure of a feature pyramid network for gradient pyramid decomposition according to an embodiment of this application is shown.

[0026] Figure 5 A schematic diagram of the structure of a combined detection device for insulator defects according to an embodiment of this application is shown;

[0027] Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0028] In order to make the purposes, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described clearly and completely below in combination with specific embodiments of the present application and corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work belong to the scope of protection of the present application.

[0029] The technical solutions provided by the embodiments of the present application will be described in detail below in combination with the drawings.

[0030] Currently, the detection of insulator defects usually adopts visual detection, electrical test, ultrasonic detection, laser scanning or single ultraviolet thermal imaging technology, etc., but these technologies generally have problems such as high cost of detection equipment, need for special equipment and personnel, strict detection conditions, poor real-time performance, low accuracy of single ultraviolet thermal imaging, etc. In view of the above problems, the present application is proposed. The main idea of the present application is to fuse the ultraviolet image and the visible light image, and to detect and identify the insulator defects based on the fused data by using a machine learning model, without the need for strict conditions and professional equipment, so as to realize effective identification of insulator defects, and to have higher accuracy compared with the existing single ultraviolet thermal imaging technology, etc.

[0031] Figure 1 A flowchart of a joint detection method of insulator defects according to an embodiment of the present application is shown, from which Figure 1 It can be seen that the present embodiment includes steps S110-S150:

[0032] Step S110, an ultraviolet image of an insulator to be detected is acquired, and the ultraviolet image is preprocessed to obtain an ultraviolet preprocessed image containing a discharge light spot region.

[0033] The application of ultraviolet imaging detection technology in the power system mainly aims at visualizing the discharge phenomenon of high-voltage electrical equipment. Through ultraviolet imaging, the ultraviolet light generated during discharge can be captured, thereby forming visible light spots on the image. The size, brightness and distribution of these light spots are important basis for evaluating the discharge intensity and the state of the equipment.

[0034] In actual scenarios, it is detected whether the meteorological environment meets the industry standard before starting, and if not, the detection is ended. The industry standard followed in some embodiments of the present application is the meteorological environment requirement of “DL / T 664-2008 Application Specification for Ultraviolet Diagnosis of Live Equipment” and “DL / T 345-2010 Application Guide for Ultraviolet Diagnosis Technology of Live Equipment”.

[0035] In some embodiments of the present application, a UAV or a fixed camera can be used to collect the UV image of the insulator of the power transmission line. The UAV or the camera should be equipped with a corresponding UV sensor to ensure that clear image data can be captured, as shown in Figure 2 Figure 2 A collected UV image according to an embodiment of the present application is shown.

[0036] After the UV image is obtained, it is pre-processed, including image grayscale processing, binary image conversion processing, and background region cutting processing, to obtain a discharge light spot region UV image.

[0037] In some embodiments of the present application, after obtaining the UV image data information of the insulator to be detected, it is pre-processed. In some embodiments, pre-processing the UV image data information includes: performing grayscale processing on the UV image to obtain a UV grayscale image; performing binary conversion processing on the UV grayscale image to obtain a UV binary image; performing background region cutting processing on the UV binary image to obtain an image containing a discharge light spot region; using a threshold segmentation or edge detection algorithm to determine the boundary of the discharge light spot to obtain a UV pre-processed image.

[0038] In a feasible embodiment, the collected UV image is subjected to grayscale processing, and for each pixel point in the image, formula 1 is used for grayscale processing:

[0039] Y = 0.299R + 0.587G + 0.114B Formula (1);

[0040] In formula 1, R, G, and B are the values of the red, green, and blue color channels, respectively.

[0041] Then, a threshold value can be automatically determined using, but not limited to, the OTSU method (Otsu's method is an automatic threshold selection technique for image processing), to convert the grayscale image into a binary image, as shown in formula 2, and the converted result is shown in Figure 3

[0042]

[0043] Wherein, G(x, y) is the grayscale value of the original grayscale image at (x, y), and B(x, y) is the pixel value of the image after binary conversion at (x, y).

[0044] Then, the background region is cut, the non-target region in the binary image is removed, and only the temperature anomaly region is retained to obtain an image containing a discharge light spot region.

[0045] Finally, a threshold segmentation or edge detection algorithm is used to determine the boundary of the discharge light spot to obtain a UV pre-processed image. ​​

[0046] In the image containing the discharge light spot area, the discharge light spot area is identified, and the image collected by the ultraviolet imager is a video stream. In order to quantitatively realize the calculation of the light spot area, three parts of background difference, noise suppression and feature extraction are mainly included. The principal component analysis method is suitable for background image extraction, and the electrical equipment is a still image and the discharge light spot is a moving image.

[0047] Specifically, threshold segmentation or edge detection algorithm can be used to determine the boundary and area of the discharge light spot, and the specific process is briefly described as follows.

[0048] Firstly, frame difference method calculation is carried out based on background difference principle, which is specifically shown in formula 3:

[0049] ΔB t (x, y) = |B(x, y, t) - B(x, y, t-1)| Formula (3);

[0050] In formula 3, B(x, y, t) represents the gray value of the pixel point at t time;△B t (x, y) represents the gray value difference at t time.

[0051] Secondly, threshold judgment is carried out, which is specifically shown in formula 4:

[0052]

[0053] In formula 4, ε is the threshold value of background difference.

[0054] Thirdly, noise suppression is mainly based on median filtering, threshold segmentation and morphological method to realize discharge light spot contour extraction, and morphological processing mainly includes opening and closing operation, that is, erosion and expansion operation, which are specifically shown in formula 5 and formula 6 respectively:

[0055] AΘB = {x | (B+x) ∈ A} Formula (5);

[0056]

[0057] Finally, the number of pixels in all boundary areas is counted, so as to calculate the light spot area. If the point with discharge area gray value of 255 is defined as 1, then the light spot area is the total number S of 1 in the A area, which is shown in formula 7:

[0058] S = ∑ (x，y)∈A 1 Formula (7).

[0059] Then the eigenvalue of the ultraviolet image is extracted, so as to construct the ultraviolet discharge spectrum of the insulator. Specifically, in some embodiments of the present application, the constructing the ultraviolet discharge spectrum of the insulator based on the ultraviolet pretreatment image comprises: based on the ultraviolet pretreatment image, performing discharge analysis on the discharge light spot region, extracting discharge spectrum information related to discharge activity; analyzing the spectral features in the discharge spectrum information, and selecting representative eigenvalues of the insulator discharge state to constitute the ultraviolet discharge spectrum of the insulator.

[0060] The ultraviolet pretreatment image is analyzed, the spectrum information related to the discharge activity is extracted, and the spectral features generated by the discharge are analyzed, and the eigenvalues representing the discharge state of the insulator are selected, so as to constitute the ultraviolet discharge spectrum of the insulator.

[0061] In step S120, the visible light image of the insulator to be detected is obtained, the visible light image is preprocessed and eigenvalue extraction is performed, and the visible light pretreatment image and the visible light image eigenvalue are obtained.

[0062] In some embodiments of the present application, the visible light image of the insulator of the power transmission line can be collected by using a drone or a fixed camera. The drone or the camera should be equipped with a corresponding visible light sensor to ensure that clear image data can be captured.

[0063] After obtaining the visible light image of the insulator to be detected, the visible light image is preprocessed, including but not limited to color temperature correction, exposure correction, image gray scale processing, enhancement and denoising processing, etc. Further, the insulator disc surface region is extracted from the gray scale image by using the seed region growing method. Specifically, in some embodiments of the present application, the visible light pretreatment image is obtained by preprocessing the visible light image, comprising: before obtaining the visible light image, sequentially correcting the color temperature of the camera with a white card and correcting the exposure of the camera with a gray card; after obtaining the visible light image, performing gray scale processing on the visible light image to obtain a visible light gray scale image; performing enhancement and denoising processing on the visible light gray scale image; and extracting the insulator disc surface region by using the seed region growing method to obtain the visible light pretreatment image containing the disc surface region.

[0064] Specifically, the camera is first color temperature corrected, such as using a white card to correct the color temperature of the camera, to eliminate the influence of ambient light color temperature; then a gray card is used to correct the exposure of the camera, to eliminate the influence of ambient light illumination. More specifically, the white card correction can be performed according to the following method: a white card picture is taken by the camera to be corrected, the brightness value of the white card should be close to the brightest part of the scene, the brightness value of the white card image is calculated and compared with the ideal brightness value, and the exposure setting of the camera is adjusted according to the difference, so that the brightness value of the white card is close to the ideal value. The gray card correction can be performed according to the following method: a gray card picture is taken by the camera to be corrected, the gray value of the gray card should be close to the average gray value of the scene, the gray value of the gray card image is calculated and compared with the ideal gray value, and the exposure setting of the camera is adjusted according to the difference, so that the gray value of the gray card is close to the ideal value.

[0065] Then the corrected camera is used to collect the visible light image, and after collection, the image is enhanced and denoised, such as using histogram equalization to enhance the contrast of the image, and using median filtering to remove image noise, to improve the image quality and reduce environmental influence.

[0066] Then the disc surface area of the insulator is extracted from the image, to prepare for subsequent feature extraction and analysis. Specifically, the seed region growing method can be used to extract the disc surface area of the insulator, to obtain an image containing the disc surface area, which is referred to as a visible light preprocessed image.

[0067] According to the specific, an initial seed point is selected, which is located in the expected insulator disc surface area. For each pixel in the seed region, check its 8 neighborhood pixels, and add the pixels adjacent to the seed point and meeting the condition to the seed region according to the gray value. For each neighborhood pixel, check whether it meets the growth criterion, and if the neighborhood pixel meets the growth criterion, add it to the seed region list.

[0068] Repeat the above process until there are no new pixels meeting the condition, wherein the gray value criterion is shown in equation 8:

[0069] |I(p)-I(q)|≤T Equation (8);

[0070] Where I(p) is the gray value of the seed point, I(q) is the gray value of the neighborhood pixel, and T is a preset threshold value.

[0071] Then the visible light pretreatment image is subjected to eigenvalue extraction, and in some embodiments, the visible light pretreatment image is subjected to eigenvalue extraction including: extracting the U component mean value and the V component mean value of the disc surface region in the visible light pretreatment image in the YUV color space; comparing the absolute value of the difference between the U component mean value of the disc surface color of the 0-level and IV-level contaminated insulator, the absolute value of the difference between the V component mean value of the disc surface color of the 0-level and IV-level contaminated insulator, and selecting the component corresponding to the larger absolute value as the visible light image eigenvalue. That is, the U and V component mean values of the YUV color space are extracted from the visible light image, and the component with larger difference is selected as the eigenvalue of the visible light image.

[0072] Specifically, the following formula can be used to convert the color space. For a pixel point, the pixel value in the grayscale image is Y, and the expression of Y is as shown in formula 9, and the conversion process is as shown in formula 10-12:

[0073] Y = 0.299R + 0.587G + 0.114B formula (9);

[0074] U = B-Y, V = R-Y formula (10);

[0075]

[0076]

[0077] Wherein, U(x, y) and V(x, y) are the component values of U and V in the YUV color space, R is the insulator disc surface region, and A is the area of the region.

[0078] In the contaminated insulator, the U component generally refers to a certain spectral characteristic exhibited by the disc surface color of the insulator under contaminated conditions. For 0-level and IV-level contaminated insulators, the color representation of the U component is usually different. 0-level contamination: relatively clean, usually showing a lighter color, the value of the U component may be higher, close to white or transparent state, indicating that the surface of the insulator is less contaminated. IV-level contamination: this level of contamination is heavy, the disc surface color may turn into a darker gray or black, the value of the U component is lower, indicating that the contamination is covered more, and the insulating performance may be affected. The same applies to the V component.

[0079] After obtaining the U component and the V component, the absolute value of the difference between the U component mean value of the disc surface color of the 0-level and IV-level contaminated insulator, and the absolute value of the difference between the V component mean value of the disc surface color of the 0-level and IV-level contaminated insulator are compared, and the component corresponding to the larger one is selected as the visible light image eigenvalue.

[0080] It should be noted that both the ultraviolet image and the visible light image are images containing the insulator to be detected, and the field of view of the two can be the same or different, preferably the same; the size of the two can be the same or different, preferably the same.

[0081] Step S130: Based on the ultraviolet preprocessed image, construct the ultraviolet discharge spectrum of the insulator and determine the defect discharge level.

[0082] Based on the results of ultraviolet discharge spectroscopy processing, the severity of defect discharge is determined, and fuzzy logic and other methods are used to make fuzzy judgments on the discharge time of the insulator in order to determine its defect discharge level.

[0083] Specifically, in some embodiments, determining the defect discharge level based on the ultraviolet discharge spectrum includes: defining fuzzy sets and membership functions to convert precise discharge time data into fuzzy values; and using a fuzzy logic inference algorithm to infer the defect discharge level based on the fuzzy values.

[0084] Fuzzy sets and membership functions are defined to convert precise discharge time data into fuzzy values, and then a fuzzy inference system is used for judgment, as shown in Equation 13:

[0085] Defect level = f(discharge time) Equation (13);

[0086] Where f is a fuzzy logic function.

[0087] The membership function is shown in Equation 14:

[0088]

[0089] Where t1 and t2 are the threshold values ​​for discharge time.

[0090] Step S140: Based on the ultraviolet discharge spectrum and the visible light image feature values, the ultraviolet preprocessed image and the visible light preprocessed image are fused to obtain a fused image and the location information of the insulator defect;

[0091] By fusing the preprocessed ultraviolet and visible light images, the location information of insulator defects can be obtained during the fusion process.

[0092] Specifically, in some embodiments, fusing the ultraviolet preprocessed image and the visible light preprocessed image based on the ultraviolet discharge spectrum and the visible light image feature values ​​to obtain a fused image includes: determining the location information of the insulator defect in the ultraviolet preprocessed image based on the feature values ​​in the ultraviolet discharge spectrum, and performing preliminary localization of the insulator defect in the visible light preprocessed image based on the location information and the visible light feature values; identifying feature points using the gradient pyramid decomposition method, and removing feature points with the same statistical feature vector using a gray-scale statistical local feature matching method; performing feature point matching on the ultraviolet preprocessed image and the visible light preprocessed image after preliminary localization, and performing interpolation calculation on the visible light preprocessed image based on the matching result to process the visible light preprocessed image into an image corresponding to the location in the ultraviolet preprocessed image; fusing the ultraviolet preprocessed image and the visible light preprocessed image based on NSCT transform, and outputting the location information of the insulator defect; and sharpening and enhancing the fused image to obtain the fused image.

[0093] In other words, the feature values ​​extracted in the aforementioned steps can be used for preliminary localization, and then the gradient pyramid decomposition method can be used for precise fusion. In this process, the location information of the insulator defect can be obtained.

[0094] First, the location information of the insulator defect in the ultraviolet preprocessed image is determined based on the ultraviolet image feature values. Then, based on the location information and visible light feature values, the insulator defect is initially located in the visible light preprocessed image. Generally, ultraviolet images have relatively low resolution, while visible light images have higher resolution and are clearer. The location of the insulator defect can be roughly identified based on the ultraviolet image feature values. The insulator defect is located using the ultraviolet image feature values ​​to determine its location information in the ultraviolet preprocessed image. This location information is then matched with the visible light feature values ​​to initially locate the insulator defect in the visible light preprocessed image. This location result is preliminary and relatively coarse.

[0095] Then, the gradient pyramid decomposition method is used for more refined fusion. The gradient pyramid decomposition method is a commonly used image fusion technique that can effectively combine information from multiple images to create a more visually impactful result. Figure 4 A schematic diagram of the feature pyramid network structure of the gradient pyramid decomposition method according to an embodiment of this application is shown. Figure 4As can be seen, the feature pyramid network used to extract feature points presents an overall pyramid shape. This network can enhance key features in the image. It contains multiple sub-networks, and after processing, multiple feature points of the ultraviolet preprocessed image and the visible light preprocessed image can be obtained respectively.

[0096] A local feature matching method based on gray-scale statistics is used to remove feature points with the same statistical feature vectors, making subsequent feature point matching more accurate.

[0097] After deduplication, feature point matching is performed on the ultraviolet preprocessed image and the visible light preprocessed image. Based on the matching results, interpolation calculation is performed on the visible light preprocessed image to process the visible light preprocessed image into an image corresponding to the position of the ultraviolet preprocessed image.

[0098] Finally, the ultraviolet and visible light images are fused based on NSCT transform to improve the spatial and frequency resolution of the image. Further post-processing operations such as sharpening and enhancement can be performed on the fused image to improve its visual quality and usability. Based on this initial fusion, the fused region of the insulator string is detected, and a second fusion is performed. The results are then saved, enabling rapid location of the insulator string.

[0099] Image fusion can be represented by Equation 15:

[0100] F = α * I 紫外 (x, y) + (1-α) * I 可见光 (x, y) Formula (15);

[0101] Where α is the weighting coefficient.

[0102] In some embodiments, the above-mentioned fusion of the ultraviolet preprocessed image and the visible light preprocessed image based on nonlinear principal component analysis (NSCT) includes: performing wavelet transform on the ultraviolet preprocessed image and the visible light preprocessed image respectively using frequency domain analysis methods to obtain first wavelet coefficients and second wavelet coefficients respectively; in the wavelet transform domain, performing layer-by-layer and scale-by-scale comparison and selection on the first wavelet coefficients and the second wavelet coefficients to retain wavelet coefficients containing more defect information; and reconstructing the fused image using the retained wavelet coefficients in the inverse wavelet transform.

[0103] Step S150: Using a pre-trained deep learning model for insulator defect detection, defect identification is performed on the fused image to obtain the insulator defect level.

[0104] For the identification of insulator defect levels, this application adopts a deep learning-based deep learning model for insulator defect detection. Specifically, the deep learning model for insulator defect detection is constructed and pre-trained according to the following method: The deep learning model for insulator defect detection includes a pyramid network, which comprises multiple residual blocks and a pyramid pooling block. Each residual block contains multiple convolutional layers and a batch normalization layer. The pyramid pooling block comprises multiple pooling layers of different sizes, each followed by a convolutional layer. At least one fully connected layer is also connected to the top of the pyramid network. A training sample set is constructed, which includes at least a training set containing multiple pairs of one-to-one correspondences of ultraviolet and visible light images, labeled with insulator defect levels. The training sample set is input into the deep learning model for insulator defect detection. Image features of the training samples are extracted and fused through the pyramid network. The fused diverse features are mapped to a preset defect level through the fully connected layer to obtain the insulator defect level.

[0105] Specifically, a large amount of ultraviolet and visible light image data of insulators is collected and labeled with their defect levels as training data. This training data can be divided into two parts: a training set and a validation set. Deep learning algorithms are used, leveraging a feature pyramid network to learn and classify image features, training a machine learning model to identify the defect levels of insulators. In this application, the pyramid network typically consists of multiple residual blocks and a pyramid pooling module. Each residual block contains several convolutional layers and batch normalization layers for feature extraction; the pyramid pooling module is used to capture contextual information at different scales.

[0106] Taking a pair of ultraviolet and visible light images as an example, they are input into the deep learning model for insulator defect detection. The pyramid network is used to extract the image features of the training samples and fuse them. Then, one or more fully connected layers are used to map the fused features onto the prediction of the defect level. The output layer uses the softmax function to perform multi-classification, thereby obtaining the defect level of the insulator.

[0107] The model is trained on the training set, and its performance is monitored using the validation set. Hyperparameters are then adjusted to improve the model's accuracy, resulting in a deep learning model for insulator defect detection.

[0108] During the prediction process, a trained deep learning model for insulator defect detection is applied to analyze the fused image data, identify insulator defects, classify and predict the quality of the insulators, and classify the defects into levels based on the degree of deviation from the core wire and the characteristics of the abnormal area using a machine learning model.

[0109] In one embodiment, this application designs a pyramid pooling module containing four scale pooling layers, with each scale pooling layer followed by a 1x1 convolutional layer for channel dimensionality reduction. The four scale pooling layers are: Scale 1: 2×2 pooling; Scale 2: 4×4 pooling; Scale 3: 6×6 pooling; Scale 4: 8×8 pooling.

[0110] During training, feature maps of different scales are obtained by extraction. These feature maps of different scales are upsampled and stitched together to form a fused feature map. Then, bilinear interpolation is used for upsampling, and feature maps of different scales are stitched together to increase the diversity of features and obtain a fused image.

[0111] At the top of the pyramid network, one or more fully connected layers are used to map the features of the fused image onto the prediction of the defect level, and the output layer uses a softmax function for multi-class classification. In some embodiments of this application, the model can be trained using the cross-entropy loss function, utilizing the ADMA optimizer with a learning rate set to 0.001. The model is trained on the training set, and its performance is monitored using a validation set. Hyperparameters are tuned to improve the model's accuracy.

[0112] In the prediction process, a new image is input, and a trained model is used to predict the new image, outputting the probability distribution of each defect. The defect level is determined based on the highest probability, the defect location is marked on the image, and different colors are used to label each defect region according to its level, generating a label to display its predicted level. In some embodiments of this application, the setting and determination of the defect level can be achieved as follows:

[0113]

[0114] Where θ1, θ2, θ3, and θ4 are preset threshold values.

[0115] Regarding the output of the results, in some embodiments of this application, the prediction results can be output in a visual manner, including the marking of defect locations, the indication of defect levels, etc., to provide decision support for the maintenance and repair of power systems.

[0116] The final detection result can be formed based on the aforementioned defect discharge level and the defect level predicted by the machine learning model. Together, they provide decision support for the maintenance and repair of the power system.

[0117] The following are some specific application examples to better illustrate this application:

[0118] Depend on Figure 1As shown in the method, this application acquires an ultraviolet image of the insulator to be tested, preprocesses the ultraviolet image to obtain an ultraviolet preprocessed image containing the discharge spot region; acquires a visible light image of the insulator to be tested, preprocesses the visible light image and extracts feature values ​​to obtain a visible light preprocessed image and visible light image feature values; then, based on the ultraviolet preprocessed image, constructs the ultraviolet discharge spectrum of the insulator, and determines the defect discharge level according to the ultraviolet discharge spectrum; based on the ultraviolet discharge spectrum and the visible light image feature values, the ultraviolet preprocessed image and the visible light preprocessed image are fused to obtain a fused image and the location information of the insulator defect; finally, a pre-trained deep learning model for insulator defect detection is used to identify defects in the fused image to obtain the insulator defect level. The defect discharge level and the insulator defect level together constitute the detection result. Compared with existing technologies, this invention uses ultraviolet and visible light images to jointly identify the defect level of insulators. By analyzing the discharge characteristics of the equipment, it improves the accuracy of detecting insulator defects, greatly enhances the effectiveness of insulator detection, and reduces the limitations of using a single device. It not only improves the detection accuracy of deteriorated insulators, but also has significant implications for saving labor.

[0119] Figure 5 A schematic diagram of a combined detection device for insulator defects according to an embodiment of this application is shown. Figure 5 It can be seen that the combined insulator defect detection device 500 includes:

[0120] The first acquisition unit 510 is used to acquire an ultraviolet image of the insulator to be tested, and to preprocess the ultraviolet image to obtain an ultraviolet preprocessed image containing the discharge spot area.

[0121] The second acquisition unit 520 is used to acquire a visible light image of the insulator to be detected, preprocess the visible light image and extract feature values ​​to obtain a preprocessed visible light image and visible light image feature values.

[0122] The spectral construction unit 530 is used to construct the ultraviolet discharge spectrum of the insulator based on the ultraviolet preprocessed image, and determine the defect discharge level according to the ultraviolet discharge spectrum;

[0123] The fusion unit 540 is used to fuse the ultraviolet preprocessed image and the visible light preprocessed image based on the ultraviolet discharge spectrum and the visible light image feature values ​​to obtain a fused image and the location information of the insulator defect;

[0124] The prediction unit 550 is used to perform defect identification on the fused image using a pre-trained deep learning model for insulator defect detection, and to obtain the insulator defect level.

[0125] In some embodiments of this application, in the above-described apparatus, the first acquisition unit 510 is used to perform grayscale processing on the ultraviolet image to obtain an ultraviolet grayscale image; perform binarization conversion processing on the ultraviolet grayscale image to obtain an ultraviolet binary image; perform background region segmentation processing on the ultraviolet binary image to obtain an image containing a discharge spot region; and use threshold segmentation or edge detection algorithms to determine the boundary of the discharge spot to obtain an ultraviolet preprocessed image.

[0126] In some embodiments of this application, in the above-described apparatus, the spectral construction unit 530 is used to perform discharge analysis on the discharge spot region based on the ultraviolet preprocessed image, extract discharge spectral information related to discharge activity, analyze the spectral features in the discharge spectral information, and select representative feature values ​​of the insulator discharge state to construct the ultraviolet discharge spectrum of the insulator.

[0127] In some embodiments of this application, in the above-described apparatus, the second acquisition unit 520 is used to sequentially perform color temperature correction on the camera using a white card and exposure correction on the camera using a gray card before acquiring the visible light image; after acquiring the visible light image, the visible light image is processed into grayscale to obtain a visible light grayscale image; the visible light grayscale image is enhanced and denoised; and the insulator disk area is extracted using a seed region growth method to obtain a visible light preprocessed image containing the disk area.

[0128] In some embodiments of this application, in the above-described apparatus, the second acquisition unit 520 is used to extract the mean U component and the mean V component of the disk area in the YUV color space in the visible light preprocessed image; compare the absolute value of the difference between the mean U component of the disk color of Class 0 and Class IV polluted insulators and the absolute value of the difference between the mean V component of the disk color of Class 0 and Class IV polluted insulators, and take the component corresponding to the larger absolute value as the visible light image feature value.

[0129] In some embodiments of this application, in the above-described apparatus, the spectral construction unit 530 is used to define fuzzy sets and membership functions, convert precise discharge time data into fuzzy values, and use a fuzzy logic reasoning algorithm to infer the defect discharge level based on the fuzzy values.

[0130] In some embodiments of this application, in the above-described apparatus, the fusion unit 540 is used to determine the location information of the insulator defect in the ultraviolet preprocessed image based on the feature values ​​in the ultraviolet discharge spectrum, and to perform preliminary localization of the insulator defect in the visible light preprocessed image based on the location information and the feature values ​​of the visible light image; to identify feature points using the gradient pyramid decomposition method, and to remove feature points with the same statistical feature vector using a gray-scale statistical local feature matching method; to perform feature point matching on the ultraviolet preprocessed image and the visible light preprocessed image after preliminary localization, and to perform interpolation calculation on the visible light preprocessed image based on the matching result, so as to process the visible light preprocessed image into an image corresponding to the location of the ultraviolet preprocessed image; to fuse the ultraviolet preprocessed image and the visible light preprocessed image based on nonlinear principal component analysis (NSCT) transform, and to output the location information of the insulator defect; and to sharpen and enhance the fused image to obtain the fused image.

[0131] In some embodiments of this application, in the above-described apparatus, the deep learning model for insulator defect detection is trained using the following method: Constructing a deep learning model for insulator defect detection, the model includes: a pyramid network comprising multiple residual blocks and a pyramid pooling block, each residual block containing multiple convolutional layers and a batch normalization layer, the pyramid pooling block comprising multiple pooling layers of different sizes, each pooling layer followed by a convolutional layer; the top of the pyramid network is also connected to at least one fully connected layer; constructing a training sample set, the training sample set including at least a training set, the training set comprising multiple pairs of one-to-one corresponding ultraviolet images and visible light images, the ultraviolet images and the visible light images being labeled with insulator defect levels; inputting the training sample set into the deep learning model for insulator defect detection, extracting and fusing image features from the training samples through the pyramid network, and mapping the fused diverse features to a preset defect level through the fully connected layer to obtain the insulator defect level.

[0132] It is understood that the above-mentioned joint detection device for insulator defects can realize all the steps of the joint detection method for insulator defects provided in the foregoing embodiments. The relevant explanations of the joint detection method for insulator defects are applicable to the joint detection device for insulator defects, and will not be repeated here.

[0133] Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Please refer to it. Figure 6At the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and memory. The memory may include main memory, such as high-speed random-access memory (RAM), or non-volatile memory, such as at least one disk drive. Of course, the electronic device may also include other hardware required for other business operations.

[0134] The processor, network interface, and memory can be interconnected via an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0135] Memory is used to store programs. Specifically, programs may include program code, which includes computer operation instructions. Memory may include main memory and non-volatile memory, and provides instructions and data to the processor.

[0136] The processor reads the corresponding computer program from non-volatile memory into main memory and then executes it, forming a joint detection device for insulator defects at the logical level. The processor executes the program stored in memory and specifically performs the following operations:

[0137] Acquire an ultraviolet image of the insulator to be tested, and preprocess the ultraviolet image to obtain an ultraviolet preprocessed image containing the discharge spot region;

[0138] A visible light image of the insulator to be detected is acquired, and the visible light image is preprocessed and feature values ​​are extracted to obtain a preprocessed visible light image and visible light image feature values.

[0139] Based on the ultraviolet preprocessed image, the ultraviolet discharge spectrum of the insulator is constructed, and the defect discharge level is determined according to the ultraviolet discharge spectrum;

[0140] Based on the ultraviolet discharge spectrum and the visible light image feature values, the ultraviolet preprocessed image and the visible light preprocessed image are fused to obtain a fused image and the location information of the insulator defect;

[0141] A pre-trained deep learning model for insulator defect detection is used to identify defects in the fused image to obtain the insulator defect level.

[0142] The above is as stated in this application. Figure 5 The method executed by the joint detection device for insulator defects disclosed in the illustrated embodiment can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the field, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0143] The electronic device can also perform Figure 5 The method for implementing a joint detection device for insulator defects, and the realization of the joint detection device for insulator defects in... Figure 5 The functions of the embodiments shown are not described again in this application.

[0144] This application also proposes a computer-readable storage medium that stores one or more programs, the programs including instructions that, when executed by an electronic device including multiple applications, enable the electronic device to perform... Figure 5 The method executed by the combined insulator defect detection device in the illustrated embodiment is specifically used to perform:

[0145] Acquire an ultraviolet image of the insulator to be tested, and preprocess the ultraviolet image to obtain an ultraviolet preprocessed image containing the discharge spot region;

[0146] A visible light image of the insulator to be detected is acquired, and the visible light image is preprocessed and feature values ​​are extracted to obtain a preprocessed visible light image and visible light image feature values.

[0147] Based on the ultraviolet preprocessed image, the ultraviolet discharge spectrum of the insulator is constructed, and the defect discharge level is determined according to the ultraviolet discharge spectrum;

[0148] Based on the ultraviolet discharge spectrum and the visible light image feature values, the ultraviolet preprocessed image and the visible light preprocessed image are fused to obtain a fused image and the location information of the insulator defect;

[0149] A pre-trained deep learning model for insulator defect detection is used to identify defects in the fused image to obtain the insulator defect level.

[0150] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0151] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0152] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0153] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0154] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0155] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0156] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0157] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0158] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0159] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A combined detection method for insulator defects, characterized in that, include: Acquire an ultraviolet image of the insulator to be tested, and preprocess the ultraviolet image to obtain an ultraviolet preprocessed image containing the discharge spot region; A visible light image of the insulator to be detected is acquired, and the visible light image is preprocessed and feature values ​​are extracted to obtain a preprocessed visible light image and visible light image feature values. Based on the ultraviolet preprocessed image, the ultraviolet discharge spectrum of the insulator is constructed, and the defect discharge level is determined according to the ultraviolet discharge spectrum; Based on the ultraviolet discharge spectrum and the visible light image feature values, the ultraviolet preprocessed image and the visible light preprocessed image are fused to obtain a fused image and the location information of the insulator defect; A pre-trained deep learning model for insulator defect detection is used to identify defects in the fused image to obtain the insulator defect level. The detection results are generated based on the defect discharge level, insulator defect level, and insulator defect location information; The process of fusing the preprocessed ultraviolet image and the preprocessed visible light image based on the ultraviolet discharge spectrum and the visible light image feature values ​​to obtain a fused image and the location information of the insulator defect includes: Based on the characteristic values ​​in the ultraviolet discharge spectrum, the location information of the insulator defect in the ultraviolet preprocessed image is determined, and based on the location information combined with the characteristic values ​​of the visible light image, the insulator defect is initially located in the visible light preprocessed image. The gradient pyramid decomposition method is used to identify feature points, and the local feature matching method based on gray-scale statistics is used to remove feature points with the same statistical feature vectors, thereby obtaining multiple feature points in the ultraviolet preprocessed image and the visible light preprocessed image respectively. Feature point matching is performed on the ultraviolet preprocessed image and the visible light preprocessed image after initial positioning, and interpolation calculation is performed on the visible light preprocessed image based on the matching result to process the visible light preprocessed image into an image corresponding to the position of the ultraviolet preprocessed image. The ultraviolet preprocessed image and the visible light preprocessed image are fused based on nonlinear principal component analysis (NSCT) transform, and the location information of the insulator defect is output. The merged image is then sharpened and enhanced to obtain the fused image. The fusion of the UV preprocessed image and the visible light preprocessed image based on nonlinear principal component analysis (NSCT) transform includes: Wavelet transforms were performed on the ultraviolet preprocessed image and the visible light preprocessed image using frequency domain analysis to obtain the first wavelet coefficient and the second wavelet coefficient, respectively. In the wavelet transform domain, the first wavelet coefficients and the second wavelet coefficients are compared and selected layer by layer and scale by scale, retaining wavelet coefficients that contain more defect information. The retained wavelet coefficients are reconstructed into the fused image using inverse wavelet transform.

2. The method according to claim 1, characterized in that, Preprocessing the ultraviolet image includes: The ultraviolet image is processed to obtain an ultraviolet grayscale image; The ultraviolet grayscale image is binarized to obtain an ultraviolet binary image; The ultraviolet binary image is processed by background region segmentation to obtain an image containing the discharge spot region; Threshold segmentation or edge detection algorithms are used to determine the boundary of the discharge spot, resulting in a pre-processed ultraviolet image.

3. The method according to claim 2, characterized in that, The process of constructing the ultraviolet discharge spectrum of the insulator based on the ultraviolet preprocessed image includes: Based on the ultraviolet preprocessed image, discharge analysis is performed on the discharge spot region to extract discharge spectral information related to discharge activity; Analyze the spectral characteristics in the discharge spectrum information, select representative characteristic values ​​of the insulator's discharge state, and construct the ultraviolet discharge spectrum of the insulator.

4. The method according to claim 1, characterized in that, Preprocessing the visible light image includes: Before acquiring the visible light image, the camera is color temperature corrected using a white card and exposure corrected using a gray card. After acquiring the visible light image, the visible light image is processed into grayscale to obtain a visible light grayscale image; The visible light grayscale image is enhanced and denoised. The seed region growth method was used to extract the surface region of the insulator disk, resulting in a visible light preprocessed image containing the disk surface region.

5. The method according to claim 4, characterized in that, Feature extraction of the visible light image is performed according to the following method: Extract the mean values ​​of the U component and V component of the disk area in the YUV color space from the preprocessed visible light image; Compare the absolute values ​​of the difference between the mean values ​​of the U component of the color of the insulator discs of Class 0 and Class IV polluted insulators, and the absolute values ​​of the difference between the mean values ​​of the V component of the color of the insulator discs of Class 0 and Class IV polluted insulators. The component corresponding to the larger absolute value is taken as the visible light image feature value.

6. The method according to claim 1, characterized in that, The step of determining the defect discharge level based on the ultraviolet discharge spectrum includes: Define fuzzy sets and membership functions to convert precise discharge time data into fuzzy values; The defect discharge level is inferred based on the fuzzy value using a fuzzy logic reasoning algorithm.

7. The method according to claim 1, characterized in that, The deep learning model for insulator defect detection was trained using the following method: A deep learning model for insulator defect detection is constructed, comprising: a pyramid network, wherein the pyramid network includes multiple residual blocks and a pyramid pooling block, each residual block contains multiple convolutional layers and a batch normalization layer, and the pyramid pooling block includes multiple pooling layers of different sizes, each pooling layer being followed by a convolutional layer; the top of the pyramid network is also connected to at least one fully connected layer. Construct a training sample set, which includes at least a training set, which includes multiple pairs of one-to-one correspondences of ultraviolet images and visible light images, wherein the ultraviolet images and the visible light images are labeled with insulator defect levels; The training sample set is input into the deep learning model for insulator defect detection. The image features of the training samples are extracted and fused through the pyramid network. The fused diverse features are then mapped to a preset defect level through the fully connected layer to obtain the insulator defect level.

8. A combined detection device for insulator defects, characterized in that, The device includes: The first acquisition unit is used to acquire an ultraviolet image of the insulator to be tested, and to preprocess the ultraviolet image to obtain an ultraviolet preprocessed image containing the discharge spot area; The second acquisition unit is used to acquire a visible light image of the insulator to be detected, preprocess the visible light image and extract feature values ​​to obtain a preprocessed visible light image and visible light image feature values. A spectral construction unit is used to construct the ultraviolet discharge spectrum of the insulator based on the ultraviolet preprocessed image, and to determine the defect discharge level according to the ultraviolet discharge spectrum; The fusion unit is used to fuse the ultraviolet preprocessed image and the visible light preprocessed image based on the ultraviolet discharge spectrum and the visible light image feature values ​​to obtain the fused image and the location information of the insulator defect; The prediction unit is used to perform defect identification on the fused image using a pre-trained deep learning model for insulator defect detection, obtain the insulator defect level, and form a detection result based on the defect discharge level, the insulator defect level, and the location information of the insulator defect. The fusion unit is used to determine the location information of the insulator defect in the ultraviolet preprocessed image based on the feature values ​​in the ultraviolet discharge spectrum, and to perform preliminary localization of the insulator defect in the visible light preprocessed image based on the location information and the feature values ​​of the visible light image. The gradient pyramid decomposition method is used to identify feature points, and the local feature matching method based on gray-scale statistics is used to remove feature points with the same statistical feature vectors, thereby obtaining multiple feature points in the ultraviolet preprocessed image and the visible light preprocessed image respectively. Feature point matching is performed on the ultraviolet preprocessed image and the visible light preprocessed image after initial positioning, and interpolation calculation is performed on the visible light preprocessed image based on the matching result to process the visible light preprocessed image into an image corresponding to the position of the ultraviolet preprocessed image. The ultraviolet preprocessed image and the visible light preprocessed image are fused based on nonlinear principal component analysis (NSCT) transform, and the location information of the insulator defect is output. The merged image is then sharpened and enhanced to obtain the fused image. The fusion unit is also used to perform wavelet transform on the ultraviolet preprocessed image and the visible light preprocessed image respectively using frequency domain analysis method to obtain the first wavelet coefficient and the second wavelet coefficient respectively; In the wavelet transform domain, the first wavelet coefficients and the second wavelet coefficients are compared and selected layer by layer and scale by scale, retaining wavelet coefficients that contain more defect information. The retained wavelet coefficients are reconstructed into the fused image using inverse wavelet transform.

9. A computer-readable storage medium storing one or more programs, characterized in that, When the one or more programs are executed by an electronic device including multiple applications, the electronic device performs the joint detection method for insulator defects according to any one of claims 1 to 7.

Images