Power transmission and transformation equipment inspection infrared image super-resolution reconstruction method and system

Abstract

The application discloses a kind of power transmission and transformation equipment inspection infrared image super-resolution reconstruction method and system, method includes: obtaining at least one power transmission and transformation equipment inspection infrared image, at least one power transmission and transformation equipment inspection infrared image is preprocessed, obtain at least one low-resolution infrared image;Based on dynamic area perception residual block and variability subpixel convolution constructs super-resolution reconstruction network, and at least one low-resolution infrared image is input into super-resolution reconstruction network as training set and is trained, obtains target super-resolution reconstruction model;Real-time power transmission and transformation equipment inspection infrared image obtained is input into target super-resolution reconstruction model, and target super-resolution reconstruction model is output to obtain reconstruction image.Can low quality, fuzzy infrared image is promoted as high-resolution, clear infrared image, to enhance image details, so that target detection technology can be more accurately identified and analyzed power transmission and transformation equipment heating state.

Description

Technical Field

[0001] This invention belongs to the field of infrared image processing technology, and in particular relates to a method and system for super-resolution reconstruction of infrared images of power transmission and transformation equipment inspection. Background Technology

[0002] Power transmission and transformation equipment is a crucial component of power transmission, used to reliably and economically deliver high-voltage, high-power electrical energy from the generation end to load centers at various levels, and to perform voltage transformation, power distribution, and protection control at substation nodes at various levels. During this process, any loose conductive joints, insulation deterioration, or poor heat dissipation will first manifest as abnormal temperatures.

[0003] Traditional power transmission and transformation equipment inspection mainly relies on manual inspections and periodic maintenance. Manual inspections require experienced technicians to observe with their naked eyes and use manual tools. They are slow to detect subtle changes in equipment temperature, often only noticing when thermal defects have already significantly worsened. This method is not only inefficient but also easily affected by subjective factors, making it difficult to detect potential faults in a timely manner. While periodic maintenance can prevent problems to some extent, it lacks real-time capability and still has a significant lag, making it difficult to meet the stringent safety and high reliability requirements of modern power systems. Summary of the Invention

[0004] This invention provides a method and system for super-resolution reconstruction of infrared images during the inspection of power transmission and transformation equipment, which solves the technical problem that low resolution and poor clarity of infrared images affect the effectiveness of infrared technology in detecting faults in power transmission and transformation equipment.

[0005] In a first aspect, the present invention provides a method for super-resolution reconstruction of infrared images from power transmission and transformation equipment inspections, comprising:

[0006] Acquire an infrared image of at least one power transmission and transformation equipment during inspection, and preprocess the infrared image of the at least one power transmission and transformation equipment to obtain at least one low-resolution infrared image;

[0007] A super-resolution reconstruction network is constructed based on dynamic region-aware residual blocks and variable sub-pixel convolution, and the at least one low-resolution infrared image is used as a training set to be input into the super-resolution reconstruction network for training, thereby obtaining the target super-resolution reconstruction model.

[0008] The acquired real-time infrared images of power transmission and transformation equipment inspection are input into the target super-resolution reconstruction model, and the target super-resolution reconstruction model outputs the reconstructed image.

[0009] Secondly, the present invention provides a super-resolution reconstruction system for infrared images of power transmission and transformation equipment inspection, comprising:

[0010] The acquisition module is configured to acquire at least one infrared image of a power transmission and transformation equipment inspection, and to preprocess the infrared image of the at least one power transmission and transformation equipment inspection to obtain at least one low-resolution infrared image.

[0011] The construction module is configured to build a super-resolution reconstruction network based on dynamic region-aware residual blocks and variable sub-pixel convolution, and input the at least one low-resolution infrared image as a training set into the super-resolution reconstruction network for training to obtain the target super-resolution reconstruction model.

[0012] The output module is configured to input the acquired real-time infrared images of power transmission and transformation equipment inspection into the target super-resolution reconstruction model, and the target super-resolution reconstruction model outputs the reconstructed image.

[0013] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the infrared image super-resolution reconstruction method for power transmission and transformation equipment inspection according to any embodiment of the present invention.

[0014] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the method for super-resolution reconstruction of infrared images of power transmission and transformation equipment inspection according to any embodiment of the present invention.

[0015] This application discloses a method and system for super-resolution reconstruction of infrared images used in the inspection of power transmission and transformation equipment. Through super-resolution reconstruction technology, low-quality, blurry infrared images can be upgraded to high-resolution, clear infrared images, thereby enhancing image details and enabling target detection technology to more accurately identify and analyze the heating status of power transmission and transformation equipment. This method effectively solves the problem of insufficient infrared image quality in traditional inspections, improves detection accuracy and reliability, and avoids power outages and safety accidents caused by power transmission and transformation equipment failures. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart of a method for super-resolution reconstruction of infrared images of power transmission and transformation equipment according to an embodiment of the present invention;

[0018] Figure 2 This is a structural block diagram of an infrared image super-resolution reconstruction system for power transmission and transformation equipment inspection, provided in an embodiment of the present invention.

[0019] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 The flowchart illustrates a method for super-resolution reconstruction of infrared images of power transmission and transformation equipment inspection according to this application.

[0022] like Figure 1 As shown, the method for super-resolution reconstruction of infrared images during power transmission and transformation equipment inspection specifically includes the following steps:

[0023] Step S101: Obtain an infrared image of at least one power transmission and transformation equipment inspection, and preprocess the infrared image of the at least one power transmission and transformation equipment inspection to obtain at least one low-resolution infrared image.

[0024] In this step, infrared images of power transmission and transformation equipment will be used for inspection. The image is divided into multiple regions. For each region, a different convolution kernel is applied for blurring. The blurred regions are then merged back into the first low-resolution image.

[0025] Infrared images of power transmission and transformation equipment inspection Perform Fourier transform on the infrared images of power transmission and transformation equipment inspection. Transform from the time domain to the frequency domain, and examine the infrared images of power transmission and transformation equipment in the frequency domain. Downsampling is performed, and the frequency domain image is converted back to the time domain using inverse Fourier transform to obtain a second low-resolution image;

[0026] The first low-resolution image and the second low-resolution image are fused together to generate an infrared image for the inspection of the power transmission and transformation equipment. The corresponding low-resolution image, wherein the expression for the low-resolution image is:

[0027] ,

[0028] In the formula, Infrared images for inspecting power transmission and transformation equipment The corresponding low-resolution image, The contribution of the first low-resolution image obtained through region adaptive blurring operation. Let the convolution kernel for the i-th region be of size . or The size of the convolution kernel is adjusted according to the complexity of the region. For the i-th region, This indicates that for the i-th region Apply the i-th convolution kernel Perform two-dimensional convolution operations. For the number of regions, This is a two-dimensional inverse Fourier transform, used to convert frequency domain information back to the time domain to generate low-resolution images. For filters, For the nonlinear transformation function of sound suppression, This is the second low-resolution image obtained through Fourier transform and downsampling.

[0029] Step S102: Construct a super-resolution reconstruction network based on dynamic region-aware residual blocks and variable sub-pixel convolution, and input the at least one low-resolution infrared image as a training set into the super-resolution reconstruction network for training to obtain the target super-resolution reconstruction model.

[0030] In this step, the super-resolution reconstruction network includes: an input layer; a dynamic region-aware residual block connected to the input layer; a deformable pixel-compressed convolutional layer connected to the dynamic region-aware residual block; an output layer connected to the deformable pixel-compressed convolutional layer; and a discriminator connected to the output layer.

[0031] Specifically, the expression for the loss function of the discriminator in the target super-resolution reconstruction model is as follows:

[0032] ,

[0033] ,

[0034] ,

[0035] ,

[0036] ,

[0037] In the formula, Let the loss function of the discriminator be , The feature similarity loss function is... As the weight for deformation loss, For texture loss weights, The total number of feature layers. For structural loss weights, For the loss of deformation smoothness, Multi-scale texture consistency loss measures the texture of generated and real images at different scales. Spatial consistency loss is used to measure the spatial consistency between the generated image and the real image. To generate the image in the first The feature map of the layer represents the first layer. Feature information extracted from layers, For the real image in the first The feature map of the layer represents the high-level feature information of the real image. For a custom feature correlation loss function, it represents the sum of the features of the generated image and the real image at the 1st... The features extracted from the layers are compared. Given the square of the L2 norm, calculate the distance and difference between feature maps. For the first The channel offset in pixels The spatial gradient at a given location is used to adjust the convolution sampling position of each pixel. To use the Sobel operator to generate the image at the 1st... Texture information at the layer scale To use the Sobel operator on real images In the Texture information at the layer scale To generate the image in the first Texture information at the layer scale This is a scale-level index, representing different image spatial scales. For the real image in the first Texture information at the layer scale Images generated for super-resolution, For the original high-resolution image, To calculate the pixel position of the generated image gradient at, To calculate the pixel location of the real image gradient at, For each pixel position Summation, To generate an image and real images At pixel position The difference at that location represents a pixel-level error.

[0038] It should be noted that the dynamic region-aware residual block contains a region classifier, a convolutional kernel generator, and a basic convolutional layer connected in sequence.

[0039] The region classifier is used to calculate the probability distribution of each channel, expressed as:

[0040] ,

[0041] , ,

[0042] In the formula, The output of the region classifier represents the pixel. Category The probability, This indicates summing over 64 feature channels. For the first The channel weights are responsible for generating classification scores based on the channel features. For from the first The first layer Feature map transmitted from the channel. For bias terms, Corresponding to pixels The region mask, representing the category. The probability distribution, Let be the classification score of the r-th class at position p;

[0043] The convolution kernel generator performs global average pooling on each channel of the image and passes the global information to the fully connected layer to generate the corresponding dynamic convolution kernel.

[0044] The deformable pixel convolutional layer contains an offset convolutional layer, a convolutional layer, and a pixel rearrangement layer connected in sequence;

[0045] The specific execution process of the deformable pixel convolutional layer includes:

[0046] The offset is generated through an offset convolutional layer, expressed as follows:

[0047] , ,

[0048] In the formula, For in pixels Place, No. The channel offset, generated by the offset convolutional layer, indicates the degree of offset of each pixel during sampling. The weights are the kernel weights for offset prediction, representing the parameters of the convolution operation that generates the offsets. For the input feature map, For convolution operations, the input feature map is... Convolutional kernel weights with offset prediction Perform element-wise weighted summation to generate the output feature for each position. Channel index representing offset The value ranges from 1 to 18, representing 18 offset channels.

[0049] Step S103: The acquired real-time infrared image of the power transmission and transformation equipment inspection is input into the target super-resolution reconstruction model, and the target super-resolution reconstruction model outputs the reconstructed image.

[0050] In summary, the method of this application, through super-resolution reconstruction technology, can upscale low-quality, blurry infrared images to high-resolution, clear infrared images, thereby enhancing image details and enabling target detection technology to more accurately identify and analyze the heating status of power transmission and transformation equipment. This method effectively solves the problem of insufficient infrared image quality in traditional inspections, improves detection accuracy and reliability, and avoids power outages and safety accidents caused by power transmission and transformation equipment failures.

[0051] Please see Figure 2 The diagram shows the structural block diagram of the infrared image super-resolution reconstruction system for power transmission and transformation equipment inspection according to this application.

[0052] like Figure 2 As shown, the infrared image super-resolution reconstruction system 200 for power transmission and transformation equipment inspection includes an acquisition module 210, a construction module 220, and an output module 230.

[0053] The acquisition module 210 is configured to acquire at least one infrared image of a power transmission and transformation equipment inspection, preprocess the infrared image of the at least one power transmission and transformation equipment inspection, and obtain at least one low-resolution infrared image; the construction module 220 is configured to construct a super-resolution reconstruction network based on dynamic region-aware residual blocks and variable sub-pixel convolution, and input the at least one low-resolution infrared image as a training set into the super-resolution reconstruction network for training to obtain a target super-resolution reconstruction model; the output module 230 is configured to input the acquired real-time infrared image of the power transmission and transformation equipment inspection into the target super-resolution reconstruction model, and the target super-resolution reconstruction model outputs a reconstructed image.

[0054] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0055] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the super-resolution reconstruction method for infrared images of power transmission and transformation equipment inspection in any of the above method embodiments.

[0056] In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows:

[0057] Acquire an infrared image of at least one power transmission and transformation equipment during inspection, and preprocess the infrared image of the at least one power transmission and transformation equipment to obtain at least one low-resolution infrared image;

[0058] A super-resolution reconstruction network is constructed based on dynamic region-aware residual blocks and variable sub-pixel convolution, and the at least one low-resolution infrared image is used as a training set to be input into the super-resolution reconstruction network for training, thereby obtaining the target super-resolution reconstruction model.

[0059] The acquired real-time infrared images of power transmission and transformation equipment inspection are input into the target super-resolution reconstruction model, and the target super-resolution reconstruction model outputs the reconstructed image.

[0060] Computer-readable storage media may include a stored program area and a stored data area, wherein the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of the infrared image super-resolution reconstruction system for power transmission and transformation equipment inspection. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely configured relative to a processor, which can be connected to the infrared image super-resolution reconstruction system for power transmission and transformation equipment inspection via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0061] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby implementing the infrared image super-resolution reconstruction method for power transmission and transformation equipment inspection described in the above embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the infrared image super-resolution reconstruction system for power transmission and transformation equipment inspection. The output device 340 may include a display screen or other display device.

[0062] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0063] In one implementation, the above-described electronic device is used in a super-resolution reconstruction system for infrared images of power transmission and transformation equipment inspection, serving as a client, and includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0064] Acquire an infrared image of at least one power transmission and transformation equipment during inspection, and preprocess the infrared image of the at least one power transmission and transformation equipment to obtain at least one low-resolution infrared image;

[0065] A super-resolution reconstruction network is constructed based on dynamic region-aware residual blocks and variable sub-pixel convolution, and the at least one low-resolution infrared image is used as a training set to be input into the super-resolution reconstruction network for training, thereby obtaining the target super-resolution reconstruction model.

[0066] The acquired real-time infrared images of power transmission and transformation equipment inspection are input into the target super-resolution reconstruction model, and the target super-resolution reconstruction model outputs the reconstructed image.

[0067] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0068] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for super-resolution reconstruction of infrared images from power transmission and transformation equipment inspection, characterized in that, include: Acquire at least one infrared image of a power transmission and transformation equipment inspection, and preprocess the at least one infrared image of the power transmission and transformation equipment inspection to obtain at least one low-resolution infrared image. The preprocessing of the at least one infrared image of the power transmission and transformation equipment inspection to obtain at least one low-resolution infrared image includes: Infrared images of power transmission and transformation equipment inspection The image is divided into multiple regions. For each region, a different convolution kernel is applied for blurring. The blurred regions are then merged back into the first low-resolution image. Infrared images of power transmission and transformation equipment inspection Perform Fourier transform on the infrared images of power transmission and transformation equipment inspection. Transform from the time domain to the frequency domain, and examine the infrared images of power transmission and transformation equipment in the frequency domain. Downsampling is performed, and the frequency domain image is converted back to the time domain using inverse Fourier transform to obtain a second low-resolution image; The first low-resolution image and the second low-resolution image are fused together to generate an infrared image for the inspection of the power transmission and transformation equipment. The corresponding low-resolution image, wherein the expression for the low-resolution image is: ， In the formula, Infrared images for inspecting power transmission and transformation equipment The corresponding low-resolution image, The contribution of the first low-resolution image obtained through region adaptive blurring operation. Let the convolution kernel for the i-th region be of size . or The size of the convolution kernel is adjusted according to the complexity of the region. For the i-th region, This indicates that for the i-th region Apply the i-th convolution kernel Perform two-dimensional convolution operations. For the number of regions, This is a two-dimensional inverse Fourier transform, used to convert frequency domain information back to the time domain to generate low-resolution images. For filters, For the nonlinear transformation function of sound suppression, This is the second low-resolution image obtained through Fourier transform and downsampling; A super-resolution reconstruction network is constructed based on dynamic region-aware residual blocks and variable sub-pixel convolution, and the at least one low-resolution infrared image is used as a training set to be input into the super-resolution reconstruction network for training, thereby obtaining the target super-resolution reconstruction model. The acquired real-time infrared images of power transmission and transformation equipment inspection are input into the target super-resolution reconstruction model, and the target super-resolution reconstruction model outputs the reconstructed image.

2. The method for super-resolution reconstruction of infrared images for power transmission and transformation equipment inspection according to claim 1, characterized in that, The super-resolution reconstruction network includes: Input layer; Dynamic region-aware residual blocks connected to the input layer; Deformable compressed pixel convolutional layer connected to the dynamic region-aware residual block; The output layer connected to the deformable pixel convolutional layer; A discriminator connected to the output layer.

3. The method for super-resolution reconstruction of infrared images for power transmission and transformation equipment inspection according to claim 2, characterized in that, in, The expression for the loss function of the discriminator in the target super-resolution reconstruction model is as follows: ， ， ， ， ， In the formula, Let the loss function of the discriminator be , The feature similarity loss function is... As the weight for deformation loss, For texture loss weights, The total number of feature layers. For structural loss weights, For the loss of deformation smoothness, Multi-scale texture consistency loss measures the texture of generated and real images at different scales. Spatial consistency loss is used to measure the spatial consistency between the generated image and the real image. To generate the image in the first The feature map of the layer represents the first layer. Feature information extracted from layers, For the real image in the first The feature map of the layer represents the high-level feature information of the real image. For a custom feature correlation loss function, it represents the sum of the features of the generated image and the real image at the 1st... The features extracted from the layers are compared. Given the square of the L2 norm, calculate the distance and difference between feature maps. For the first The channel offset in pixels The spatial gradient at a given location is used to adjust the convolution sampling position of each pixel. To use the Sobel operator to generate the image at the 1st... Texture information at the layer scale To use the Sobel operator on real images In the Texture information at the layer scale To generate the image in the first Texture information at the layer scale This is a scale-level index, representing different image spatial scales. For the real image in the first Texture information at the layer scale Images generated for super-resolution, For the original high-resolution image, To calculate the pixel position of the generated image gradient at, To calculate the pixel location of the real image gradient at, For each pixel position Summation, To generate an image and real images At pixel position The difference at that location represents a pixel-level error.

4. The method for super-resolution reconstruction of infrared images of power transmission and transformation equipment according to claim 2, characterized in that, The dynamic region-aware residual block contains a region classifier, a convolutional kernel generator, and a basic convolutional layer connected in sequence. The region classifier is used to calculate the probability distribution of each channel, expressed as: ， ， ， In the formula, The output of the region classifier represents the pixel. Category The probability, This indicates summing over 64 feature channels. For the first The channel weights are responsible for generating classification scores based on the channel features. For from the first The first layer Feature map transmitted from the channel. For bias terms, Corresponding to pixels The region mask, representing the category. The probability distribution, Let be the classification score of the r-th class at position p; The convolution kernel generator performs global average pooling on each channel of the image and passes the global information to the fully connected layer to generate the corresponding dynamic convolution kernel.

5. The method for super-resolution reconstruction of infrared images of power transmission and transformation equipment according to claim 2, characterized in that, The deformable subpixel convolutional layer includes an offset convolutional layer, a convolutional layer, and a pixel rearrangement layer connected in sequence. The specific execution process of the deformable pixel convolutional layer includes: The offset is generated through an offset convolutional layer, expressed as follows: ， ， In the formula, For in pixels Place, No. The channel offset, generated by the offset convolutional layer, indicates the degree of offset of each pixel during sampling. The weights are the kernel weights for offset prediction, representing the parameters of the convolution operation that generates the offsets. For the input feature map, For convolution operations, the input feature map is... Convolutional kernel weights with offset prediction Perform element-wise weighted summation to generate the output feature for each position. Channel index representing offset The value ranges from 1 to 18, representing 18 offset channels.

6. A super-resolution reconstruction system for infrared images of power transmission and transformation equipment inspection, characterized in that, include: The acquisition module is configured to acquire at least one infrared image of a power transmission and transformation equipment inspection, and to preprocess the infrared image of the at least one power transmission and transformation equipment inspection to obtain at least one low-resolution infrared image. The preprocessing of the infrared image of the at least one power transmission and transformation equipment inspection to obtain at least one low-resolution infrared image includes: Infrared images of power transmission and transformation equipment inspection The image is divided into multiple regions. For each region, a different convolution kernel is applied for blurring. The blurred regions are then merged back into the first low-resolution image. Infrared images of power transmission and transformation equipment inspection Perform Fourier transform on the infrared images of power transmission and transformation equipment inspection. Transform from the time domain to the frequency domain, and examine the infrared images of power transmission and transformation equipment in the frequency domain. Downsampling is performed, and the frequency domain image is converted back to the time domain using inverse Fourier transform to obtain a second low-resolution image; The first low-resolution image and the second low-resolution image are fused together to generate an infrared image for the inspection of the power transmission and transformation equipment. The corresponding low-resolution image, wherein the expression for the low-resolution image is: ， In the formula, Infrared images for inspecting power transmission and transformation equipment The corresponding low-resolution image, The contribution of the first low-resolution image obtained through region adaptive blurring operation. Let the convolution kernel for the i-th region be of size . or The size of the convolution kernel is adjusted according to the complexity of the region. For the i-th region, This indicates that for the i-th region Apply the i-th convolution kernel Perform two-dimensional convolution operations. For the number of regions, This is a two-dimensional inverse Fourier transform, used to convert frequency domain information back to the time domain to generate low-resolution images. For filters, For the nonlinear transformation function of sound suppression, This is the second low-resolution image obtained through Fourier transform and downsampling; The construction module is configured to build a super-resolution reconstruction network based on dynamic region-aware residual blocks and variable sub-pixel convolution, and input the at least one low-resolution infrared image as a training set into the super-resolution reconstruction network for training to obtain the target super-resolution reconstruction model. The output module is configured to input the acquired real-time infrared images of power transmission and transformation equipment inspection into the target super-resolution reconstruction model, and the target super-resolution reconstruction model outputs the reconstructed image.

7. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method described in any one of claims 1 to 5.

Images