Multi-source image fusion reconstruction method and system for power equipment

Abstract

The invention discloses a multi-source image fusion reconstruction method and system for power equipment, and relates to the technical field of image enhancement, and the method comprises the steps: obtaining an infrared image set and a visible light image set of target power equipment; performing preprocessing based on the infrared image set and the visible light image set to correspondingly obtain a first image set and a second image set; respectively inputting the first image set and the second image set into a generator for aggregation reconstruction, and then evaluating through a discriminator to correspondingly obtain a third image set and a fourth image set; feature point extraction is carried out based on the third image set and the fourth image set, image registration is carried out based on the extracted feature points and a matching function, and a third image and a fourth image which establish a matching relation are obtained; and based on the third image and the fourth image which establish the matching relationship, fusing through a fusion function to obtain a fused image. And the reconstruction precision, modeling flexibility and reconstruction efficiency of multi-source image fusion are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image enhancement, and more particularly to a method and system for multi-source image fusion and reconstruction of power equipment. Background Art

[0002] At present, with the rapid advancement of the construction of smart cities and safe cities, video surveillance systems, as the core information collection means, digital images have become important information carriers in the fields of security monitoring, satellite remote sensing, and daily life. However, existing imaging technologies are limited by factors such as sensor performance and environmental interference, and generally have problems such as low image resolution and blurred details. Especially in complex scenarios (such as bad weather, target occlusion, and camouflage interference), it is difficult for a single image to achieve accurate object detection and recognition. Although multi-source image fusion technology can generate a more reliable comprehensive description by integrating multi-dimensional information, existing methods still face severe challenges in terms of accuracy, efficiency, and model adaptability.

[0003] In the field of power equipment operation and maintenance, this problem is particularly prominent. Although the current mainstream infrared thermal imaging technology can reflect the temperature distribution of equipment, the generated two-dimensional infrared images generally have defects such as missing texture details and blurred edges, resulting in difficulties in accurately locating temperature anomaly areas and directly affecting the efficiency of equipment fault diagnosis and the reliability of operation and maintenance decisions. Traditional image reconstruction methods mostly rely on single-sensor data, and it is difficult to effectively fuse multi-modal information such as visible light, infrared, and multi-spectral. Moreover, there are the following limitations at the algorithm level: (1) The fusion method based on traditional feature extraction has insufficient robustness to complex noise and low-quality inputs; (2) Deep learning models generally face problems such as high computational complexity and poor deployment flexibility; (3) The existing neural network architectures have limited ability to mine spatio-temporal correlation features of multi-source images, resulting in difficulty in achieving both reconstruction accuracy and efficiency.

[0004] Therefore, how to improve the reconstruction accuracy, modeling flexibility, and reconstruction efficiency of multi-source image fusion is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a method and system for multi-source image fusion and reconstruction of power equipment, which improves the reconstruction accuracy, modeling flexibility, and reconstruction efficiency of multi-source image fusion.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] A method for multi-source image fusion and reconstruction of power equipment, comprising:

[0008] Obtaining an infrared image set and a visible light image set of a target power equipment;

[0009] Preprocess based on the infrared image set and the visible light image set to obtain a first image set and a second image set correspondingly;

[0010] Input the first image set and the second image set into a generator for aggregation and reconstruction respectively, and then evaluate through a discriminator to obtain a third image set and a fourth image set correspondingly;

[0011] Extract feature points from the third image set and the fourth image set respectively, and perform image registration based on the extracted feature points and a matching function to obtain a third image and a fourth image with a matching relationship established;

[0012] Fuse the third image and the fourth image with a matching relationship established through a fusion function to obtain a fused image.

[0013] Preferably, the preprocessing specifically includes:

[0014] Perform format conversion on the images in the infrared image set and the visible light image set respectively to obtain multi-source images in a unified format, and perform downsampling, edge cropping, and storage processing on the multi-source images in the unified format in sequence to obtain the first image set and the second image set correspondingly.

[0015] Preferably, the aggregation and reconstruction specifically includes:

[0016] Input the first image set and the second image set into the generator respectively, and obtain a first reconstructed image and a second reconstructed image based on the generator function correspondingly.

[0017] Preferably, the discriminator evaluation specifically includes:

[0018] Input the first reconstructed image and the second reconstructed image into the discriminator respectively, and obtain a first discriminant value and a second discriminant value based on the discriminator function correspondingly;

[0019] Judge whether the first discriminant value is greater than a second threshold;

[0020] If so, use the first reconstructed image as the third image to obtain the third image set;

[0021] Otherwise, input the first reconstructed image into the generator for training until the discriminant value of the output image is greater than the second threshold, and use the current output image as the third image to obtain the third image set;

[0022] Judge whether the second discriminant value is greater than the second threshold;

[0023] If so, use the second reconstructed image as the fourth image to obtain the fourth image set;

[0024] Otherwise, input the second reconstructed image into the generator for training until the discrimination value of the output image is greater than the second threshold, and use the current output image as the fourth image to obtain the fourth image set.

[0025] Preferably, the generator function Specifically:

[0026]

[0027] Wherein, represents the real image expectation, D Ra represents the relative average discriminator, X r represents the real image, X f represents the fake image, represents the fake image expectation.

[0028] Preferably, the discriminator function Specifically:

[0029]

[0030] Preferably,

[0031]

[0032] Wherein, σ represents the sigmoid activation function, and C represents the output.

[0033] Preferably, the matching function E(u, v) is specifically:

[0034]

[0035] Wherein, w i represents the window weight matrix, (x, y) represents the pixel coordinate position corresponding to window i; I(x + u, y + v) represents the image gray value at the pixel coordinate position (x + u, y + v), and I(x, y) represents the image gray value at the pixel coordinate position (x, y).

[0036] Preferably, the fusion function is specifically:

[0037] F ij (n) = S ij

[0038]

[0039] Wherein, F ij (n) represents the feedback input of the neuron at position (ij) in the nth iteration, S ij represents the input image at position (ij), L ij(n) represents the neuron link input at position (ij) in the nth iteration, V L represents the amplitude gain of the link input, W' ij represents the link weight matrix, M×N represents the link domain range, Y ij (n - 1) represents the neuron spike output in the (n - 1)th iteration, U ij (n) represents the neuron internal activity term in the nth iteration, β represents the neuron internal activity link coefficient, Y ij (n) represents the neuron spike output in the nth iteration, E ij (n - 1) represents the dynamic threshold in the (n - 1)th iteration, E ij (n) represents the dynamic threshold in the nth iteration, V E represents the amplitude coefficient of the dynamic threshold function, a f represents the time decay coefficient of the feedback input.

[0040] A multi-source image fusion and reconstruction system for power equipment, comprising: an image acquisition module, an image processing module, a reconstruction evaluation module, an image registration module, and an image fusion module;

[0041] The image acquisition module is used to acquire an infrared image set and a visible light image set of the target power equipment;

[0042] The image processing module is used to perform preprocessing based on the infrared image set and the visible light image set, and correspondingly obtain a first image set and a second image set;

[0043] The reconstruction evaluation module is used to input the first image set and the second image set into a generator for aggregated reconstruction and then evaluate through a discriminator, and correspondingly obtain a third image set and a fourth image set;

[0044] The image registration module is used to extract feature points based on the third image set and the fourth image set respectively and perform image registration based on the extracted feature points and a matching function to obtain a third image and a fourth image with a matching relationship established;

[0045] The image fusion module is used to fuse the third image and the fourth image with a matching relationship established through a fusion function to obtain a fused image.

[0046] Through the above technical solutions, compared with the prior art, the present invention discloses a multi-source image fusion and reconstruction method and system for power equipment, having the following beneficial effects:

[0047] 1. Improve image quality: By aggregating and reconstructing the infrared image set and the visible light image set, the details in the image are enhanced by the generator. It can effectively combine two different types of image data, thereby improving the quality of the final fused image.

[0048] 2. Enhance the fusion effect: Use the fusion function to fuse the registered third image and fourth image, which can integrate the advantages of two different modality images, such as the thermal imaging ability of the infrared image and the high-resolution characteristics of the visible light image, thereby generating a more comprehensive and information-rich fused image. It is particularly important for the condition monitoring and fault diagnosis of power equipment, and can provide more accurate temperature distribution and structural detail information.

[0049] 3. Compared with general super-resolution reconstruction, the reconstruction process of the present invention contains a screening process, which effectively reduces the occupation of computer computing power by invalid images and reduces the consumption of human resources in subsequent work.

[0050] 4. Optimize feature extraction and matching: Perform feature point extraction on the third image set and the fourth image set, and perform precise image registration based on these feature points, which helps to ensure the accurate spatial alignment of the images from two sources, thereby reducing the registration error caused by perspective differences or different sensor characteristics.

[0051] 5. The edge information of the reconstructed infrared image, i.e., the fused image, is clear, the texture information increases, and the amount of data that can be mined increases, which is helpful for subsequent work such as image-based condition judgment, fault detection, and operation and maintenance decision-making generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0053] Figure 1 It is a flowchart of a multi-source image fusion and reconstruction method for power equipment provided by the present invention.

[0054] Figure 2 It is a schematic structural diagram of a multi-source image fusion and reconstruction system for power equipment provided by the present invention.

[0055] Figure 3 It is a structural block diagram of a computer device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0057] Embodiment 1

[0058] As Figure 1 shown, an embodiment of the present invention discloses a multi-source image fusion and reconstruction method for power equipment, including:

[0059] Obtaining an infrared image set and a visible light image set of a target power equipment;

[0060] Performing preprocessing based on the infrared image set and the visible light image set to respectively obtain a first image set and a second image set;

[0061] Respectively inputting the first image set and the second image set into a generator for aggregation and reconstruction, and then evaluating through a discriminator to respectively obtain a third image set and a fourth image set;

[0062] Respectively extracting feature points from the third image set and the fourth image set, and performing image registration based on the extracted feature points and a matching function to obtain the third image set and the fourth image set with a matching relationship established;

[0063] Fusing the third image set and the fourth image set with a matching relationship established through a fusion function to obtain a fused image set.

[0064] Embodiment 2

[0065] An embodiment of the present invention discloses a multi-source image fusion and reconstruction method for power equipment, including:

[0066] Obtaining an infrared image set and a visible light image set of a target power equipment.

[0067] Preferably, in this embodiment, a multi-source camera is used to capture infrared images and visible light images of the target power equipment at similar angles, and the corresponding infrared image set and visible light image set are obtained as a multi-source original image set.

[0068] Preferably, the difference in the similar angles in this embodiment does not exceed 15°, the shooting area covers at least the expected detection position, and a weather with weak light such as a cloudy day is selected for shooting to reduce the influence of light on the accuracy of the original images.

[0069] Preferably, the number of images in the infrared image set and the visible light image set is 10 to 15.

[0070] Preprocessing is performed based on the infrared image set and the visible light image set, and the first image set and the second image set are correspondingly obtained.

[0071] Preferably, the preprocessing specifically includes:

[0072] Format conversion is respectively performed on the images in the infrared image set and the visible light image set to obtain multi-source images in a unified format. The multi-source images in the unified format are successively subjected to downsampling, edge cropping, and storage processing, and the first image set and the second image set are correspondingly obtained.

[0073] Preferably, in this embodiment, the image format is unified into YCBCR images.

[0074] Preferably, the first image set and the second image set jointly form a multi-source low-resolution image set.

[0075] Based on the first image set and the second image set, they are respectively input into the generator for aggregation and reconstruction, and then evaluated by the discriminator, and the third image set and the fourth image set are correspondingly obtained.

[0076] Preferably, the aggregation and reconstruction specifically includes:

[0077] The first image set and the second image set are respectively input into the generator, and the first reconstructed image and the second reconstructed image are correspondingly obtained based on the generator function.

[0078] Preferably, in this embodiment, an ESRCAN generator and an ESRCAN discriminator are adopted, and the ESRCAN generator and the ESRCAN discriminator are cascaded to form an ESRCAN convolutional network.

[0079] Preferably, the ESRCAN generator takes the deep residual network as the core, and its core function is the generator function as:

[0080]

[0081] Among them, represents the real image expectation, D Ra represents the relative average discriminator, and the relative average discriminator indicates that X Ra (X f , X r ) value is close to 1 to indicate that X r is more real than X f , X r represents the real image, and X f represents the fake image, represents the fake image expectation.

[0082] Preferably, the discriminator evaluation specifically includes:

[0083] The first reconstructed image and the second reconstructed image are respectively input into a discriminator, and a first discrimination value and a second discrimination value are respectively obtained based on the discriminator function;

[0084] Determine whether the first discrimination value is greater than a second threshold;

[0085] If so, use the first reconstructed image as a third image to obtain a third image set;

[0086] Otherwise, input the first reconstructed image into a generator for training until the discrimination value of the output image is greater than the second threshold, use the current output image as the third image to obtain a third image set;

[0087] Determine whether the second discrimination value is greater than the second threshold;

[0088] If so, use the second reconstructed image as a fourth image to obtain a fourth image set;

[0089] Otherwise, input the second reconstructed image into a generator for training until the discrimination value of the output image is greater than the second threshold, use the current output image as the fourth image to obtain a fourth image set.

[0090] Preferably, the ESRCAN discriminator takes the Residual-in-Residual Dense Block (RRDB) as the core, and this structure effectively improves the generalization ability of the model, reduces the computational complexity and memory occupation.

[0091] Preferably, the discriminator function Specifically:

[0092]

[0093] Preferably,

[0094]

[0095] wherein, σ represents the sigmoid activation function, and C represents the output.

[0096] Preferably, the third image set is an infrared super-resolution image set, and the fourth image set is a visible light super-resolution image set.

[0097] Feature points are respectively extracted based on the third image set and the fourth image set, and image registration is performed based on the extracted feature points and a matching function to obtain the third image and the fourth image with a matching relationship established.

[0098] Preferably, based on the third image set and the fourth image set, feature points are extracted using the Harris operator as a standard. The third image set and the fourth image set are registered based on the feature points extracted respectively to obtain the third image and the fourth image with a matching relationship established, that is, the multi-source matching super-resolution image.

[0099] Preferably, the Harris operator only uses pixel point cross-differences and filtering, without the need to set thresholds. It has a high degree of detection automation and the number of extracted feature points can be quantified. Setting different numbers of feature points for different targets can effectively reduce memory requirements. In addition, the Harris operator has good stability and is less affected by noise, rotation, and perspective changes.

[0100] Preferably, the matching function E(u, v) is specifically:

[0101]

[0102] where w i represents the window weight matrix, (x, y) represents the pixel coordinate position corresponding to window i; I(x + u, y + v) represents the image gray value at the pixel coordinate position (x + u, y + v), and I(x, y) represents the image gray value at the pixel coordinate position (x, y).

[0103] Preferably, performing feature point extraction on the third image set and the fourth image set and performing precise image registration based on these feature points helps to ensure the accurate spatial alignment of the images from two sources, thereby reducing the registration error caused by perspective differences or different sensor characteristics.

[0104] Based on the third image and the fourth image with a matching relationship established, they are fused through a fusion function to obtain a fused image.

[0105] Preferably, the fusion function is specifically:

[0106] F ij (n) = S ij ;

[0107]

[0108] where F ij (n) represents the feedback input of the neuron at position (i, j) in the nth iteration, S ij represents the input image at position (i, j), L ij (n) represents the neuron link input at position (i, j) in the nth iteration, V L represents the amplitude gain of the link input, W' ij represents the link weight matrix, M × N represents the link domain range, Y ij(n - 1) represents the neuron impulse output in the (n - 1)-th iteration, U ij (n) represents the neuron internal activity term in the n-th iteration, β represents the internal activity link coefficient of the neuron, Y ij (n) represents the neuron impulse output in the n-th iteration, E ij (n - 1) represents the dynamic threshold in the (n - 1)-th iteration, E ij (n) represents the dynamic threshold in the n-th iteration, V E represents the amplitude coefficient of the dynamic threshold function, a f represents the time decay coefficient of the feedback input.

[0109] Preferably, the third image and the fourth image based on the established matching relationship are fused through a fusion function, and the fourth image is used to correct the third image to obtain a fused image.

[0110] Preferably, the edge information of the fused image is clear, the texture information increases, the amount of data that can be mined increases, which is helpful for subsequent work such as image-based state judgment, fault detection, and operation and maintenance decision-making generation.

[0111] Embodiment 3

[0112] As Figure 2 shown, a multi-source image fusion and reconstruction system for power equipment includes: an image acquisition module, an image processing module, a reconstruction evaluation module, an image registration module, and an image fusion module;

[0113] The image acquisition module is used to acquire an infrared image set and a visible light image set of the target power equipment;

[0114] The image processing module is used to perform preprocessing based on the infrared image set and the visible light image set, and correspondingly obtain a first image set and a second image set;

[0115] The reconstruction evaluation module is used to respectively input the first image set and the second image set into a generator for aggregation and reconstruction and then evaluate through a discriminator, and correspondingly obtain a third image set and a fourth image set;

[0116] The image registration module is used to respectively perform feature point extraction on the third image set and the fourth image set and perform image registration based on the extracted feature points and a matching function to obtain a third image and a fourth image with an established matching relationship;

[0117] The image fusion module is used to fuse the third image and the fourth image with an established matching relationship through a fusion function to obtain a fused image.

[0118] Preferably, in this embodiment, the function implementation of each module corresponds one by one to the above method, and will not be elaborated here one by one.

[0119] Embodiment 4

[0120] Based on the same inventive concept, the present invention further provides a computer device, including a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete mutual communication through the communication bus;

[0121] The memory is used to store a computer program;

[0122] When the processor is used to execute the program stored in the memory, it can implement a multi-source image fusion and reconstruction method for a power device as described in Embodiment 1 or 2.

[0123] As Figure 3 shown, the electronic device may include: a processor 31, a communication interface 32, a memory 33, and a communication bus 34. Among them, the processor 31, the communication interface 32, and the memory 33 complete mutual communication through the communication bus 34. The processor 31 can call the logical instructions in the memory 33 to execute a multi-source image fusion and reconstruction method for a power device as described in Embodiment 1 or 2.

[0124] In addition, when the logical instructions in the above-mentioned memory 33 are implemented in the form of a software functional unit and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in the various embodiments of the present invention. And the aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0125] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a multi-source image fusion and reconstruction method and system for a power device, having the following beneficial effects:

[0126] 1. Improve image quality: By aggregating and reconstructing the infrared image set and the visible light image set, the detail information in the image is enhanced by the generator. It can effectively combine two different types of image data, thereby improving the quality of the final fused image.

[0127] 2. Enhance the fusion effect: Use a fusion function to fuse the registered third image and fourth image, which can integrate the advantages of two different modality images, such as the thermal imaging ability of infrared images and the high-resolution characteristics of visible light images, thereby generating a more comprehensive and information-rich fused image. This is particularly important for the condition monitoring and fault diagnosis of power equipment, as it can provide more accurate temperature distribution and structural detail information.

[0128] 3. Compared with general super-resolution reconstruction, the reconstruction process of the present invention contains a screening process, which effectively reduces the occupation of computer computing power by invalid images and simultaneously reduces the consumption of human resources in subsequent work.

[0129] 4. Optimize feature extraction and matching: Perform feature point extraction on the third image set and the fourth image set, and perform precise image registration based on these feature points, which helps to ensure the accurate spatial alignment of images from two sources, thereby reducing registration errors caused by differences in viewing angles or sensor characteristics.

[0130] 5. The edge information of the reconstructed infrared image, i.e., the fused image, is clear, the texture information increases, and the amount of data that can be mined increases, which helps with subsequent work such as image-based condition judgment, fault detection, and operation and maintenance decision-making generation.

[0131] The various embodiments in this specification are described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. For the same or similar parts among the various embodiments, reference can be made to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and reference can be made to the description in the method part for related parts.

[0132] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for multi-source image fusion and reconstruction of electric power equipment, characterized in that: include: Acquire infrared image sets and visible light image sets of target power equipment; Preprocessing is performed based on the infrared image set and the visible light image set to obtain a first image set and a second image set respectively; Based on the first image set and the second image set respectively inputted into the generator for aggregation reconstruction and then evaluated by the discriminator, a third image set and a fourth image set are obtained correspondingly; Extracting feature points based on the third image set and the fourth image set respectively and performing image registration based on the extracted feature points and a matching function to obtain a third image and a fourth image that establish a matching relationship; The third image and the fourth image based on the established matching relationship are fused through a fusion function to obtain a fused image.

2. The method for multi-source image fusion and reconstruction of electric power equipment according to claim 1, characterized in that: The pre-processing specifically includes: Based on the images in the infrared image set and the visible light image set, format conversion is performed respectively to obtain multi-source images in a unified format, and the multi-source images in the unified format are sequentially down-sampled, edge cropped and stored to obtain the first image set and the second image set accordingly.

3. The method for multi-source image fusion and reconstruction of electric power equipment according to claim 1, characterized in that: The aggregation reconstruction specifically includes: The first image set and the second image set are respectively input to the generator, and a first reconstructed image and a second reconstructed image are correspondingly obtained based on a generator function.

4. The method for multi-source image fusion and reconstruction of electric power equipment according to claim 3, characterized in that: The discriminator evaluation specifically includes: Based on the first reconstructed image and the second reconstructed image, the first discriminant value and the second discriminant value are respectively input into the discriminator, and the first discriminant value and the second discriminant value are correspondingly obtained based on the discriminator function; Determining whether the first discrimination value is greater than a second threshold; If yes, taking the first reconstructed image as the third image to obtain the third image set; otherwise, inputting the first reconstructed image into the generator for training until the discrimination value of the output image is greater than the second threshold, taking the current output image as the third image, and obtaining the third image set; Determining whether the second discrimination value is greater than a second threshold; If yes, taking the second reconstructed image as the fourth image to obtain the fourth image set; Otherwise, the second reconstructed image is input into the generator for training until the discrimination value of the output image is greater than the second threshold, and the current output image is used as the fourth image to obtain the fourth image set.

5. The method for multi-source image fusion and reconstruction of electric power equipment according to claim 4, characterized in that: The generator function Specifically: in, represents the real image expectation, D Ra represents the relative average discriminator, X r represents the real image, X f represents a false image, Represents false image expectation.

6. The method for multi-source image fusion and reconstruction of electric power equipment according to claim 5, characterized in that: The discriminator function Specifically:

7. The method for multi-source image fusion and reconstruction of electric power equipment according to claim 6, characterized in that: Among them, σ represents the sigmoid activation function and C represents the output.

8. The method for multi-source image fusion and reconstruction of electric power equipment according to claim 6, characterized in that: The matching function E(u,v) is specifically: Among them, w i represents the window weight matrix, (x, y) represents the pixel coordinate position corresponding to window i; I(x+u,y+v) represents the image grayscale value of the pixel coordinate position (x+u,y+v), and I(x,y) represents the image grayscale value of the pixel coordinate position (x,y).

9. The method for multi-source image fusion and reconstruction of electric power equipment according to claim 6, characterized in that: The fusion function is specifically: F ij (n)=S ij Among them, F ij (n) represents the feedback input of the neuron at position (ij) in the nth iteration, S ij represents the input image at position (ij), L ij (n) represents the neuron link input at position (ij) in the nth iteration, V L represents the amplitude gain of the link input, W' ij represents the link weight matrix, M×N represents the link domain range, Y ij (n-1) represents the neuron pulse output in the n-1th iteration, U ij (n) represents the internal activity term of the neuron in the nth iteration, β represents the internal activity link coefficient of the neuron, and Y ij (n) represents the neuron pulse output in the nth iteration, E ij (n-1) represents the dynamic threshold in the n-1th iteration, E ij (n) represents the dynamic threshold in the nth iteration, V E represents the amplitude coefficient of the dynamic threshold function, a f Represents the time decay coefficient of the feedback input.

10. A multi-source image fusion and reconstruction system for electric power equipment, applied to a multi-source image fusion and reconstruction method for electric power equipment as claimed in any one of claims 1 to 9, characterized in that: include: Image acquisition module, image processing module, reconstruction evaluation module, image registration module and image fusion module; The image acquisition module is used to acquire an infrared image set and a visible light image set of the target power equipment; The image processing module is used to perform preprocessing based on the infrared image set and the visible light image set to obtain a first image set and a second image set respectively; The reconstruction evaluation module is used to obtain a third image set and a fourth image set based on the first image set and the second image set respectively input into the generator for aggregation reconstruction and then evaluated by the discriminator; The image registration module is used to extract feature points based on the third image set and the fourth image set respectively and perform image registration based on the extracted feature points and a matching function to obtain a third image and a fourth image that establish a matching relationship; The image fusion module is used to fuse the third image and the fourth image that have established a matching relationship through a fusion function to obtain a fused image.

Images