Transformer substation outdoor image defogging method and device, storage medium and electronic equipment

Abstract

The invention provides a transformer substation outdoor image defogging method and device, a storage medium and electronic equipment, and the method comprises the steps: constructing a training set based on transformer substation outdoor paired images which comprise transformer substation outdoor clear historical images and transformer substation outdoor historical foggy images; constructing a defogging model based on a transformer substation outdoor image defogging network and the training set; and inputting a current outdoor fog-containing image of the transformer substation into the defogging model to obtain an outdoor defogged image of the transformer substation. According to the embodiment of the invention, richer feature information can be extracted, the redundancy of the features is reduced, the definition of the reconstructed outdoor image of the transformer substation is improved, local fog residues are eliminated, and the image quality is improved. And clear images are provided for operators to remotely check the operation condition of power transformation equipment and a subsequent intelligent substation inspection system based on machine vision.

Description

Technical Field

[0001] The present disclosure relates to the technical field of image processing, and in particular to a method, device, storage medium and electronic device for defogging outdoor images of a substation. Background Art

[0002] With the extensive application of information and communication technology in power systems, substations have also changed from "manned" to "unmanned". Operators can use camera monitoring equipment to view the operation of substation equipment in real time and conduct intelligent inspections. However, there are many hills, lakes and rivers in the south, the air is more humid, and foggy weather often occurs, which causes the images taken by substation cameras to be blurred and have low contrast, which seriously affects the judgment of the operation of substation equipment and poses a threat to the safe and stable operation of the power grid. Therefore, defogging the outdoor images of the substation, improving the quality of the outdoor images of the substation in foggy weather, and providing operators with high-quality outdoor images of the substation are of great significance to ensure the safe and stable operation of the power grid.

[0003] The defogging methods for outdoor images of substations mainly include defogging methods based on physical models, defogging methods based on wavelet transform, and defogging methods based on deep learning.

[0004] Among them, the substation outdoor image defogging method based on the physical model needs to first estimate the model parameters according to the physical model formed by the foggy image, and then realize the image defogging according to the estimated parameters; the substation outdoor image defogging method based on wavelet transform needs to construct a suitable wavelet basis function, and then perform wavelet decomposition on the foggy substation outdoor image, process the high and low frequency information of the image in the wavelet domain, and then use the inverse wavelet transform to reconstruct the defogged substation outdoor image; the substation outdoor image defogging method based on deep learning mainly constructs a deep convolutional neural network, and then uses a large number of fog-free images and corresponding foggy images to train the model, so that it can automatically learn the fog features and fog-free image features in the image without human intervention, thereby realizing the substation outdoor image defogging.

[0005] The substation outdoor image defogging method based on deep learning here is more robust than other methods and more suitable for practical applications. However, the clarity of the substation outdoor defogging image obtained by the existing deep learning method still needs to be improved, and there is a phenomenon of local fog residue, which affects the subsequent viewing of operators and the performance of the subsequent substation intelligent inspection system based on machine vision. In addition, the existing substation outdoor image defogging model based on deep learning does not fully extract features and has many redundant features, resulting in insufficient clarity of the reconstructed substation outdoor image and the phenomenon of local fog residue, which affects the image quality. Summary of the invention

[0006] The purpose of the embodiments of the present disclosure is to provide a method, device, storage medium and electronic device for defogging outdoor images of a substation to solve the above-mentioned problems existing in the prior art.

[0007] In order to solve the above technical problems, an embodiment of the present disclosure provides a method for defogging an outdoor image of a substation, comprising:

[0008] Constructing a training set based on paired images of the outdoor substation, wherein the paired images of the outdoor substation include clear historical images of the outdoor substation and historical foggy images of the outdoor substation;

[0009] Building a defogging model based on the substation outdoor image defogging network and the training set;

[0010] The current foggy image outside the substation is input into the defogging model to obtain a defogging image outside the substation.

[0011] In some embodiments, the construction of a defogging model based on the substation outdoor image defogging network and the training set includes:

[0012] Extract auxiliary features based on the auxiliary feature extraction network and extract main features based on the backbone feature extraction network;

[0013] Constructing the substation outdoor image defogging network based on the auxiliary features and the main features;

[0014] The training set is used to train the substation outdoor image defogging network to obtain the defogging model.

[0015] In some embodiments, the auxiliary feature extraction network includes 9 convolution sub-modules, some of which are composed of a 3×3 convolution module, a spectral normalization module, and a LeakyReLU function in sequence, and other convolution sub-modules are composed of a 3×3 convolution module, a batch normalization module, and a LeakyReLU function in sequence.

[0016] In some embodiments, the backbone feature extraction network includes 10 convolution sub-modules and 6 multi-channel attention modules, each of the convolution sub-modules is composed of a 3×3 convolution, batch normalization, and a LeakyReLU function in sequence, and each of the multi-channel attention modules is composed of 2 1×1 convolutions, 1 convolution group, and 1 adaptive average pooling in sequence.

[0017] In some embodiments, it also includes:

[0018] Building a test set based on the outdoor paired images of the substation;

[0019] The defogging model is tested using the test set to obtain the defogging model that meets predetermined requirements.

[0020] In some embodiments, if the predetermined requirements are not met, the parameters of the substation outdoor image defogging network are adjusted and retrained.

[0021] In some embodiments, the outdoor historical foggy image of the substation is obtained by adding fog of different concentrations to the outdoor clear historical image of the substation.

[0022] Another aspect of the present disclosure provides a substation outdoor image defogging device, comprising:

[0023] A data set construction module is used to construct a training set based on paired images of the outdoor substation, wherein the paired images of the outdoor substation include clear historical images of the outdoor substation and historical foggy images of the outdoor substation;

[0024] A model building module, used to build a defogging model based on the substation outdoor image defogging network and the training set;

[0025] The image acquisition module is used to input the current foggy image outside the substation into the defogging model to obtain the defogging image outside the substation.

[0026] The present disclosure also provides a storage medium storing a computer program, wherein the computer program implements the steps of the above method when executed by a processor.

[0027] The present disclosure also provides an electronic device, which includes at least a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps of the above method when executing the computer program in the memory.

[0028] The disclosed embodiments can extract richer feature information and reduce feature redundancy, thereby improving the clarity of the reconstructed substation outdoor image, eliminating local fog residue, and improving image quality, so as to provide operators with remote viewing of the operation status of substation equipment and subsequent machine vision-based substation intelligent inspection systems with clearer images.

[0029] The device, storage medium and electronic device disclosed in the present invention have all the beneficial effects of the above-mentioned method, which will not be described in detail here. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0031] Figure 1Schematic diagram of the steps of a method for defogging an outdoor image of a substation in one embodiment;

[0032] Figure 2 Schematic diagram of the steps of a method for defogging an outdoor image of a substation in one embodiment;

[0033] Figure 3 Schematic diagram of the structure of the auxiliary feature extraction network in the method for defogging an outdoor image of a substation in one embodiment;

[0034] Figure 4 Schematic diagram of the structure of the backbone feature extraction network in the method for defogging an outdoor image of a substation in one embodiment;

[0035] Figure 5 Schematic diagram of the structure of a multi-channel attention module in a backbone feature extraction network in a method for defogging an outdoor image of a substation in one embodiment;

[0036] Figure 6 Schematic diagram of the process of a method for defogging an outdoor image of a substation in one embodiment. DETAILED DESCRIPTION

[0037] Various aspects and features of the present disclosure are described herein with reference to the drawings.

[0038] It should be understood that various modifications may be made to the embodiments of the present application. Therefore, the above description should not be considered as limiting, but only as an example of an embodiment. Other modifications within the scope and spirit of the present disclosure will occur to those skilled in the art.

[0039] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the present disclosure and, together with the general description of the present disclosure given above and the detailed description of the embodiments given below, serve to explain the principles of the present disclosure.

[0040] These and other characteristics of the present disclosure will become apparent from the following description of a preferred form of embodiment given as a non-limiting example, with reference to the accompanying drawings.

[0041] It should also be understood that, although the present disclosure has been described with reference to some specific examples, those skilled in the art will be able to realize many other equivalent forms of the present disclosure that have the features described in the claims and are therefore within the scope of protection defined thereby.

[0042] The above and other aspects, features and advantages of the present disclosure will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings.

[0043] Specific embodiments of the present disclosure are described hereinafter with reference to the accompanying drawings; however, it should be understood that the embodiments claimed are merely examples of the present disclosure, which may be implemented in a variety of ways. Well-known and / or repeated functions and structures are not described in detail to avoid obscuring the present disclosure with unnecessary or redundant details. Therefore, the specific structural and functional details claimed herein are not intended to be limiting, but merely serve as a basis and representative basis for the claims to teach those skilled in the art to substantially any combination.

[0044] The first embodiment of the present disclosure provides a method for defogging outdoor images of a substation based on deep learning, which is particularly implemented based on a U-Net network, wherein the U-Net network belongs to a fully convolutional neural network, such as Figure 1 As shown, the substation outdoor image defogging method includes:

[0045] S101, constructing a training set based on paired images of an outdoor substation, wherein the paired images of an outdoor substation include clear historical images of an outdoor substation and historical foggy images of an outdoor substation.

[0046] In this step, a training set is constructed based on paired images of the substation outdoor, wherein the paired images of the substation outdoor include clear historical images of the substation outdoor and historical foggy images of the substation outdoor. The step S101 is mainly used for the construction of a data set. Specifically, clear images under different backgrounds are selected from existing outdoor images of the substation to obtain clear historical images of the substation outdoor, and different concentrations of fog are added to the clear historical images of the substation outdoor to obtain different historical foggy images of the substation outdoor, so that the clear historical images of the substation outdoor and the historical foggy images of the substation outdoor after fogging form paired images of the substation outdoor, and then a training set is constructed through the obtained paired images of the substation outdoor.

[0047] S102, building a defogging model based on the substation outdoor image defogging network and the training set.

[0048] After the training set is constructed based on the paired images of the substation outdoor through the above step S101, in this step, a defogging model is constructed based on the substation outdoor image defogging network and the training set. In this step, it is necessary to first construct the substation outdoor image defogging network, which is mainly implemented based on the U-Net network as a fully convolutional neural network. After the substation outdoor image defogging network is constructed, iterative training is continuously performed through the training set to construct the final defogging model.

[0049] Furthermore, the defogging model is constructed based on the substation outdoor image defogging network and the training set, such as Figure 2 As shown, including:

[0050] S201, extracting auxiliary features based on an auxiliary feature extraction network and extracting main features based on a backbone feature extraction network.

[0051] In this step, auxiliary features are extracted based on the auxiliary feature extraction network and main features are extracted based on the backbone feature extraction network.

[0052] Specifically, Figures 3 to 6 As shown, the auxiliary feature extraction network includes a first convolution submodule 1, a second convolution submodule 2, a third convolution submodule 3, a fourth convolution submodule 4, a fifth convolution submodule 5, a sixth convolution submodule 6, a seventh convolution submodule 7, an eighth convolution submodule 8 and a ninth convolution submodule 9, wherein the first convolution submodule 1, the fourth convolution module 4 and the seventh convolution module 7 are each composed of a 3×3 convolution module, a spectral normalization module and a LeakyReLU function in sequence, and the other convolution submodules are each composed of a 3×3 convolution module, a batch normalization module and a LeakyReLU function in sequence.

[0053] The input of the auxiliary feature extraction network here is the outdoor historical foggy image F1 of the substation. The outdoor historical foggy image F1 of the substation is passed through the first convolution submodule 1 to obtain a feature map F2. The feature map F2 is passed through the second convolution submodule 2 to obtain a feature map F3. The feature map F3 is passed through the third convolution submodule 3 to obtain a feature map F4. The feature map F4 is element-wise multiplied with the feature map F2 to obtain a feature map F5. The feature map F5 is element-wise added with the outdoor historical foggy image F1 of the substation to obtain a feature map F6. The number of channels of the feature maps F2 to F6 is 3. The feature map F6 is convolved with 1×1 to obtain a feature map F7. The feature map obtained after 1×1 convolution of the image F1 is concatenated with the feature map F7 to obtain a feature map F14. The feature map F7 is passed through the fourth convolution submodule 4 to obtain a feature map F8. The feature map F8 is passed through the fifth convolution submodule 5 to obtain a feature map F9. The feature map F9 is passed through the The sixth convolution submodule 6 obtains a feature map F10, and the feature map F10 and the feature map F7 are element-wise multiplied to obtain a feature map F11, and the feature map F11 and the feature map F7 are element-wise added to obtain a feature map F12. The feature map F12 is subjected to a 1×1 convolution to obtain a feature map F13, and the feature map F13 and the feature map F14 are concatenated to obtain a feature map F21; the feature map F13 is subjected to the seventh convolution submodule 7 to obtain a feature map F15, and the feature map F15 is subjected to the eighth convolution submodule 8 to obtain a feature map F16, and the feature map F16 is subjected to the ninth convolution submodule 9 to obtain a feature map F17, and the feature map F17 and the feature map F15 are element-wise multiplied to obtain a feature map F18, and the feature map F18 and the image F13 are element-wise added to obtain a feature map F19, and the feature map F19 is subjected to a 1×1 convolution to obtain a feature map F20, and the feature map F20 and the feature map F21 are element-wise concatenated to obtain the output feature map.

[0054] The backbone feature extraction network here includes a tenth convolution submodule 10, an eleventh convolution submodule 11, a twelfth convolution submodule 12, a thirteenth convolution submodule 13, a fourteenth convolution submodule 14, a fifteenth convolution submodule 15, a sixteenth convolution submodule 16, a seventeenth convolution submodule 17, an eighteenth convolution submodule 18 and a nineteenth convolution submodule 19. All of the above convolution submodules are composed of a 3×3 convolution, batch normalization and LeakyReLU function in sequence.

[0055] The backbone feature extraction network further includes a first multi-channel attention module 21, a second multi-channel attention module 22, a third multi-channel attention module 23, a fourth multi-channel attention module 24, a fifth multi-channel attention module 25 and a sixth multi-channel attention module 26. All of the above multi-channel attention modules have the same network structure.

[0056] Specifically, each multi-channel attention module is composed of two 1×1 convolutions, a convolution group and an adaptive average pooling in sequence. Assuming that the number of channels of the input feature map of the multi-channel attention module is C, the number of channels of the output feature map of the first 1×1 convolution is C / 2, and the number of channels of the output feature map of the second 1×1 convolution is C / 4; the convolution group consists of four branches, the first branch is used to transfer the input feature map of the convolution group, the second to the fourth branches are respectively composed of 3×3 convolutions with an output channel number of C / 8, 3×3 convolutions with an output channel number of C / 16 and 3×3 convolutions with an output channel number of C / 16. The output feature maps of the second to third branches are concatenated to obtain a feature map with a channel number of C / 4. The feature map is adjusted by a 1×1 convolution and a Sigmoid function to obtain the weight of the feature map in the channel dimension. The obtained weight is element-wise multiplied with the input feature of the convolution group to obtain the feature map D. The feature map D is subjected to adaptive average pooling to obtain the output feature map of the multi-channel attention module.

[0057] Here, the outdoor historical foggy image F1 of the substation is used as the input image, and the feature map T1 is obtained after the tenth convolution submodule 10. The feature map obtained after downsampling of the feature map T1 is input into the eleventh convolution submodule 11 to obtain the feature map T2. The feature map obtained after downsampling of the feature map T2 is input into the twelfth convolution submodule 12 to obtain the feature map T3. The feature map obtained after downsampling of the feature map T3 is input into the thirteenth convolution submodule 13 to obtain the feature map T4. The feature map obtained after downsampling of the feature map T4 is input into the fourteenth convolution submodule 14 to obtain the feature map T5. The feature map T5 is input into the fifth multi-channel attention module 25 to obtain the feature map T6; the feature map T6 is input into the fifteenth convolution submodule 15 to obtain the feature map T7. The feature map T7 is upsampled and input into the fourth multi-channel attention module 24 to obtain the feature map T4. The feature map T8 is spliced ​​to obtain a feature map T8; the feature map T8 is input into the sixteenth convolution submodule 16 to obtain a feature map T9; the feature map T9 is upsampled and then input into the third multi-channel attention module 23 to obtain a feature map T10; the feature map T10 is input into the seventeenth convolution submodule 17 to obtain a feature map T11; the feature map T11 is upsampled and then input into the second multi-channel attention module 2 to obtain a feature map T2 to obtain a feature map T12; the feature map T12 is input into the eighteenth convolution submodule 18 to obtain a feature map T13; the feature map T13 is upsampled and then input into the first multi-channel attention module 21 to obtain a feature map T14; the feature map T14 is input into the convolution submodule 19 to obtain a feature map T15. At the same time, the feature map T7 is input into the sixth multi-channel attention module 26 to obtain the feature map T16, and the feature map T15 and the feature map T16 are element-wise added to obtain the output feature map of the backbone feature extraction network.

[0058] Based on the auxiliary feature extraction network and the backbone feature extraction network, it is possible to extract richer feature information and reduce feature redundancy.

[0059] S202: constructing the substation outdoor image defogging network based on the auxiliary features and the main features.

[0060] After extracting auxiliary features based on the auxiliary feature extraction network and extracting main features based on the backbone feature extraction network in the above step S201, in this step, the substation outdoor image defogging network is constructed based on the auxiliary features and the main features.

[0061] S203, training the substation outdoor image defogging network by using the training set to obtain the defogging model.

[0062] After constructing the substation outdoor image defogging network based on the auxiliary feature extraction network and the backbone feature extraction network through the above step S202, in this step, the substation outdoor image defogging network is trained using the training set to obtain the defogging model.

[0063] Furthermore, it also includes:

[0064] Building a test set based on the outdoor paired images of the substation;

[0065] The defogging model is tested using the test set to obtain the defogging model that meets predetermined requirements.

[0066] Specifically, a pair of outdoor substation images is formed by combining the outdoor clear historical image of the substation with the outdoor historical foggy image of the substation after fogging, and a test set is constructed by the outdoor substation paired images. The test set here is used to test the constructed defogging model. The purpose of the test here is to obtain the defogging model that meets the predetermined requirements. This is achieved in an iterative manner. For example, a predetermined number of iterations can be set, and the defogging model is tested by iterative operations within the predetermined number of iterations.

[0067] In addition, if the predetermined requirements are not met, the parameters of the substation outdoor image defogging network are adjusted and retrained. Specifically, if the defogging model that meets the predetermined requirements cannot be obtained within a predetermined number of iterations, the parameters in the substation outdoor image defogging network are adjusted, and the substation outdoor image defogging network after the parameter adjustment is retrained according to the above step S203 until the defogging model that meets the predetermined requirements is obtained.

[0068] S103, inputting the current foggy image outside the substation into the defogging model to obtain a defogging image outside the substation.

[0069] After building a defogging model based on the substation outdoor image defogging network and the training set through the above step S102, in this step, the current foggy image outside the substation is input into the defogging model to obtain the defogging image outside the substation. Specifically, in this step, the defogging model is used for the actual image defogging operation, the currently acquired foggy image is input into the defogging model, and the current defogging image outside the substation is output. In this way, the foggy outdoor image of the substation is passed through the defogging model of the full convolutional network to obtain the defogging outdoor image of the substation, thereby improving the clarity of the reconstructed outdoor image of the substation, eliminating local fog residue and improving the image quality.

[0070] The disclosed embodiments can extract richer feature information and reduce feature redundancy, thereby improving the clarity of the reconstructed substation outdoor image, eliminating local fog residue, and improving image quality, thereby providing operators with a way to remotely view the operation status of substation equipment and subsequent machine vision-based substation intelligent inspection systems with clearer images.

[0071] Based on the same inventive concept, the second embodiment of the present disclosure provides a substation outdoor image defogging device, comprising:

[0072] A data set construction module is used to construct a training set based on paired images of the outdoor substation, wherein the paired images of the outdoor substation include clear historical images of the outdoor substation and historical foggy images of the outdoor substation;

[0073] A model building module, used to build a defogging model based on the substation outdoor image defogging network and the training set;

[0074] The image acquisition module is used to input the current foggy image outside the substation into the defogging model to obtain the defogging image outside the substation.

[0075] Furthermore, the model building module includes:

[0076] A feature extraction unit, used for extracting auxiliary features based on an auxiliary feature extraction network and extracting main features based on a backbone feature extraction network;

[0077] A network construction unit, used for constructing the substation outdoor image defogging network based on the auxiliary features and the main features;

[0078] The model building unit is used to train the substation outdoor image defogging network using the training set to obtain the defogging model.

[0079] Furthermore, the auxiliary feature extraction network includes 9 convolution sub-modules, some of which are composed of a 3×3 convolution module, a spectral normalization module and a LeakyReLU function in sequence, and other convolution sub-modules are composed of a 3×3 convolution module, a batch normalization module and a LeakyReLU function in sequence.

[0080] Furthermore, the backbone feature extraction network includes 10 convolution sub-modules and 6 multi-channel attention modules. The convolution sub-modules are each composed of a 3×3 convolution, a batch normalization, and a LeakyReLU function in sequence. The multi-channel attention modules are each composed of 2 1×1 convolutions, a convolution group, and an adaptive average pooling in sequence.

[0081] Furthermore, the data set construction module is also used to construct a test set based on the paired images outdoors of the substation; the model construction module is also used to test the defogging model using the test set to obtain the defogging model that meets predetermined requirements.

[0082] Furthermore, the model building module is also used to: if the predetermined requirements are not met, adjust the parameters of the substation outdoor image defogging network and retrain it.

[0083] Furthermore, the outdoor historical foggy image of the substation is obtained by adding fog of different concentrations to the outdoor clear historical image of the substation.

[0084] The disclosed embodiments can extract richer feature information and reduce feature redundancy, thereby improving the clarity of the reconstructed substation outdoor image, eliminating local fog residue, and improving image quality, thereby providing operators with a way to remotely view the operation status of substation equipment and subsequent machine vision-based substation intelligent inspection systems with clearer images.

[0085] A third embodiment of the present disclosure provides a storage medium storing a computer program. When the computer program is executed by a processor, the steps of implementing the above method include:

[0086] S11, constructing a training set based on paired images of the outdoor substation, wherein the paired images of the outdoor substation include clear historical images of the outdoor substation and historical foggy images of the outdoor substation;

[0087] S12, constructing a defogging model based on the substation outdoor image defogging network and the training set;

[0088] S13, inputting the current foggy image outside the substation into the defogging model to obtain a defogging image outside the substation.

[0089] The storage medium of the present disclosure stores a computer program, which can implement the methods in different implementation modes of the above-mentioned first embodiment when executed by a processor.

[0090] The disclosed embodiments can extract richer feature information and reduce feature redundancy, thereby improving the clarity of the reconstructed substation outdoor image, eliminating local fog residue, and improving image quality, thereby providing operators with a way to remotely view the operation status of substation equipment and subsequent machine vision-based substation intelligent inspection systems with clearer images.

[0091] A fourth embodiment of the present disclosure provides an electronic device, the electronic device comprising at least a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the above method when executing the computer program on the memory, specifically comprising:

[0092] S21, constructing a training set based on paired images of the outdoor substation, wherein the paired images of the outdoor substation include clear historical images of the outdoor substation and historical foggy images of the outdoor substation;

[0093] S22, constructing a defogging model based on the substation outdoor image defogging network and the training set;

[0094] S23, inputting the current foggy image outside the substation into the defogging model to obtain a defogging image outside the substation.

[0095] The memory of the electronic device of the present disclosure stores a computer program, and the processor implements the methods of other implementations in the above-mentioned first embodiment when executing the computer program in the memory.

[0096] The disclosed embodiments can extract richer feature information and reduce feature redundancy, thereby improving the clarity of the reconstructed substation outdoor image, eliminating local fog residue, and improving image quality, thereby providing operators with a way to remotely view the operation status of substation equipment and subsequent machine vision-based substation intelligent inspection systems with clearer images.

[0097] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0098] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0099] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0100] In the embodiments provided in the present application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0101] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0102] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0103] If the integrated module is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned method embodiment when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0104] In addition, the features of the embodiments shown in the drawings of the present application or the various embodiments mentioned in this specification are not necessarily understood as independent embodiments. Instead, each feature described in one example of an embodiment can be combined with one or more other desired features from other embodiments to produce other embodiments not described in words or with reference to the drawings.

[0105] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for defogging outdoor images of a substation, characterized in that: include: Constructing a training set based on paired images of the outdoor substation, wherein the paired images of the outdoor substation include clear historical images of the outdoor substation and historical foggy images of the outdoor substation; Building a defogging model based on the substation outdoor image defogging network and the training set; The current foggy image outside the substation is input into the defogging model to obtain a defogging image outside the substation.

2. The method for defogging outdoor images of a substation according to claim 1 is characterized in that: The defogging model is constructed based on the substation outdoor image defogging network and the training set, including: Extract auxiliary features based on the auxiliary feature extraction network and extract main features based on the backbone feature extraction network; Constructing the substation outdoor image defogging network based on the auxiliary features and the main features; The training set is used to train the substation outdoor image defogging network to obtain the defogging model.

3. The method for defogging outdoor images of a substation according to claim 2 is characterized in that: The auxiliary feature extraction network includes 9 convolution sub-modules, some of which are composed of a 3×3 convolution module, a spectral normalization module and a LeakyReLU function in sequence, and other convolution sub-modules are composed of a 3×3 convolution module, a batch normalization module and a LeakyReLU function in sequence.

4. The method for defogging outdoor images of a substation according to claim 2 is characterized in that: The backbone feature extraction network includes 10 convolution sub-modules and 6 multi-channel attention modules. The convolution sub-modules are each composed of a 3×3 convolution, a batch normalization, and a LeakyReLU function in sequence. The multi-channel attention modules are each composed of 2 1×1 convolutions, a convolution group, and an adaptive average pooling in sequence.

5. The method for defogging outdoor images of a substation according to claim 1, characterized in that: Also includes: Building a test set based on the outdoor paired images of the substation; The defogging model is tested using the test set to obtain the defogging model that meets predetermined requirements.

6. The method for defogging outdoor images of a substation according to claim 5, characterized in that: If the predetermined requirements are not met, the parameters of the substation outdoor image defogging network are adjusted and retrained.

7. The method for defogging outdoor images of a substation according to claim 1, characterized in that: The outdoor historical foggy image of the substation is obtained by adding fog of different concentrations to the outdoor clear historical image of the substation.

8. A substation outdoor image defogging device, characterized in that: include: A data set construction module is used to construct a training set based on paired images of the outdoor substation, wherein the paired images of the outdoor substation include clear historical images of the outdoor substation and historical foggy images of the outdoor substation; A model building module, used to build a defogging model based on the substation outdoor image defogging network and the training set; The image acquisition module is used to input the current foggy image outside the substation into the defogging model to obtain the defogging image outside the substation.

9. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. An electronic device, comprising at least a memory and a processor, wherein a computer program is stored in the memory, wherein: The processor implements the steps of the method of any one of claims 1 to 7 when executing the computer program on the memory.

Images