Method and device for identifying power equipment nameplate and storage medium

Abstract

The application discloses a kind of identification method, device and storage medium of electric equipment nameplate.The method comprises: obtaining at least one original image containing target electric power equipment nameplate;The original image is input into image denoising model, and the target to be identified image is determined based on model output result, wherein the image denoising model is generated model trained based on the generation of adversarial network between sample image containing electric power equipment nameplate and the expected denoising image corresponding to the sample image;The target nameplate region in the target to be identified image is determined based on edge detection method and contour search method, and the equipment nameplate data in the target nameplate region is extracted.The identification efficiency and accuracy of equipment nameplate data are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image recognition, and in particular to a power equipment nameplate recognition method and device and storage medium. BACKGROUND

[0002] Power equipment nameplate recognition plays a crucial role in modern power systems. Nameplates usually contain key information about the equipment, which is essential for equipment maintenance, management, and fault diagnosis. With the development of intelligent recognition technology, more and more intelligent recognition systems are being applied to power equipment nameplate recognition.

[0003] Traditional manual recognition methods are not only inefficient but also prone to errors. Existing automated recognition systems still need to improve their accuracy in complex environments. Existing recognition methods perform poorly in conditions with large changes in lighting, especially in high noise levels, which can severely affect recognition results. SUMMARY

[0004] The present application provides a power equipment nameplate recognition method, device and storage medium to solve the problem of low recognition accuracy of equipment nameplate data.

[0005] According to an aspect of the present application, a power equipment nameplate recognition method is provided, which includes:

[0006] Obtaining at least one original image containing a target power equipment nameplate;

[0007] Inputting the original image into an image denoising model to determine a target to-be-recognized image based on the model output result, wherein the image denoising model is a generative model trained based on a generative adversarial network generated from a pair of sample images containing power equipment nameplates and corresponding expected denoising images;

[0008] Determining a target nameplate region in the target to-be-recognized image based on edge detection and contour finding, and extracting equipment nameplate data in the target nameplate region.

[0009] According to another aspect of the present application, a power equipment nameplate recognition device is provided, which includes:

[0010] An original image acquisition module for acquiring at least one original image containing a target power equipment nameplate;

[0011] A to-be-recognized image determination module for inputting the original image into an image denoising model to determine a target to-be-recognized image based on the model output result, wherein the image denoising model is a generative model trained based on a generative adversarial network generated from a pair of sample images containing power equipment nameplates and corresponding expected denoising images;

[0012] The device nameplate data extraction module is configured to determine a target nameplate region in the target image to be recognized based on an edge detection method and a contour search method, and extract device nameplate data in the target nameplate region.

[0013] According to another aspect of the present application, an electronic device is provided, which comprises:

[0014] at least one processor; and

[0015] a memory in communication with the at least one processor; wherein

[0016] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the method for identifying a power device nameplate according to any one of the embodiments of the present application.

[0017] According to another aspect of the present application, a computer readable storage medium is provided, which stores computer instructions for enabling a processor to perform the method for identifying a power device nameplate according to any one of the embodiments of the present application.

[0018] The technical solution of the embodiments of the present application comprises the following steps: obtaining at least one original image containing a target power device nameplate; inputting the original image into an image denoising model, and determining a target image to be recognized based on a model output result, wherein the image denoising model is a generative model trained based on a generative adversarial network generated from a sample image containing a power device nameplate and a corresponding expected denoising image; improving the image quality of the target image to be recognized, improving the data recognition accuracy based on the target image to be recognized, determining a target nameplate region in the target image to be recognized based on an edge detection method and a contour search method, and extracting device nameplate data in the target nameplate region. The device nameplate data in the target nameplate region is accurately extracted. The problem of low recognition accuracy of device nameplate data is solved, and the beneficial effects of improving the recognition efficiency and accuracy of device nameplate data are achieved.

[0019] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments description. Obviously, the drawings in the following description only show some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative effort based on these drawings.

[0021] Figure 1 is a flow chart of a power equipment nameplate identification method according to the first embodiment of the present application;

[0022] Figure 2 is a flow chart of a power equipment nameplate identification method according to the second embodiment of the present application;

[0023] Figure 3 is a structural schematic diagram of a power equipment nameplate identification device according to the third embodiment of the present application;

[0024] Figure 4 is a structural schematic diagram of an electronic device implementing the power equipment nameplate identification method according to the fourth embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to make the person skilled in the art better understand the present application, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort should be within the scope of the present application.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0027] Embodiment one

[0028] Figure 1A flowchart of a power equipment nameplate recognition method is provided for the first embodiment of the present application. The first embodiment can be applied to the recognition of power equipment nameplates. The method can be executed by a power equipment nameplate recognizer device, which can be implemented in the form of hardware and / or software, and can be configured in an electronic device. As shown in FIG. 10, the method comprises the following steps. Figure 1

[0029] S110, obtaining at least one original image containing a target power equipment nameplate.

[0030] The power equipment nameplate can be understood as a signboard on which the power equipment associated data is marked.

[0031] Specifically, at least one original image containing a target power equipment nameplate is obtained through a camera. The original image can include images of the target power equipment nameplate taken at different angles or different brightness.

[0032] Optionally, the step of obtaining at least one original image containing a target power equipment nameplate comprises: obtaining nameplate shooting images of the target power equipment nameplate based on multiple groups of shooting parameters, and performing fusion processing on the multiple nameplate shooting images to obtain the original image containing the target power equipment nameplate.

[0033] Specifically, the nameplate images are taken from multiple angles or different light source intensities, and the multiple nameplate shooting images are fused to reduce the influence of reflections and shadows. A multi-light source imaging system is used to take nameplate images from multiple angles or different light source intensities. A preset image fusion algorithm is used to fuse the multiple nameplate shooting images. The gray value (or color value) of the corresponding pixel points of different images is weighted and averaged to obtain the original image containing the target power equipment nameplate. Alternatively, the image is decomposed into different scales and different resolutions, and the low-frequency energy information and high-frequency detail information are fused respectively to obtain the original image containing the target power equipment nameplate. The preset image fusion algorithm includes a weighted average image fusion algorithm or a multi-resolution decomposition fusion algorithm.

[0034] ​Preferably, a multi-modal fusion framework can also be designed to combine visual information with sensor data to enhance the understanding and positioning ability for non-standard nameplate styles. Through a visual sensor (e.g., a camera): obtain two-dimensional image data. Through a laser radar: obtain three-dimensional point cloud data; register the two-dimensional image and the three-dimensional point cloud data to establish a correspondence between them. Use three-dimensional point cloud data to assist in the analysis of two-dimensional images; combine visual features and point cloud features for feature-level fusion, and based on the fused multi-modal data, perform subsequent nameplate data recognition (e.g., character recognition, pattern recognition) and matching (e.g., template matching, feature matching). Improve the robustness and accuracy of the system.

[0035] Optionally, after obtaining at least one original image of the target power equipment nameplate, the method further comprises: pre-processing the original image, and updating the original nameplate image based on the pre-processed original image; wherein the pre-processing comprises at least one of gray scale processing, gradient transformation processing, smoothing filtering processing, and thresholding processing.

[0036] Specifically, an original image containing a target power equipment nameplate is read. At least one or a combination of several of gray scale processing, gradient transformation processing, smoothing filtering processing, and thresholding processing is selected as needed to pre-process the original nameplate image. The pre-processed image is updated as a new original nameplate image.

[0037] S120, inputting the original image into an image denoising model to determine a target to-be-recognized image based on a model output result.

[0038] The image denoising model is a generative model trained based on a generative adversarial network generated from a sample image containing a power equipment nameplate and a corresponding expected denoised image. The target to-be-recognized image can be understood as an image of to-be-recognized nameplate data.

[0039] Specifically, the original image is input into a pre-trained generative model, and the original image is denoised based on the generative model to output a denoised original image. The denoised original image is taken as the target to-be-recognized image.

[0040] Optionally, the inputting of the original nameplate image into the image denoising model to obtain a target to-be-recognized nameplate image comprises: filtering the original image to obtain a first denoised image; inputting the first denoised image into the image denoising model to obtain a second denoised image; and determining a target to-be-recognized image based on the first denoised image and the second denoised image.

[0041] The first denoised image can be understood as an image that has been initially filtered by a filtering algorithm. The second denoised image can be understood as a model output image.

[0042] Specifically, first, a preset filtering algorithm (such as median filtering or bilateral filtering) is used to preliminarily denoise the image to obtain a first denoised image. The first denoised image is input into the image denoising model to obtain a second denoised image. Finally, the target to-be-recognized image is determined based on the first denoised image and the second denoised image.

[0043] Optionally, the determining of the target to-be-recognized image based on the first denoised image and the second denoised image comprises: performing fusion processing on the first denoised image and the second denoised image to obtain an initial to-be-recognized image; and performing filling processing on an image gap in the initial to-be-recognized image to obtain the target to-be-recognized image.

[0044] Specifically, the first denoised image and the second denoised image are fused to obtain an initial to-be-recognized image. The initial to-be-recognized image is processed by a preset morphological algorithm to obtain the target to-be-recognized image. The preset morphological algorithm includes erosion, dilation, opening and closing.

[0045] Illustratively, filling the image gap of the initial to-be-recognized image by using the composite morphological algorithm comprises: first performing a closing operation, then performing morphological operations in the order of erosion and then dilation, and finally making the fitting degree of the processing result of the morphological filtering to the target in the original image optimal. The initial to-be-recognized image is closed to fill the small gaps in the object. The isolated points and small structures in the initial to-be-recognized image are further removed by erosion operation to make the image smoother. Subsequently, the boundaries of the objects in the initial to-be-recognized image are restored by dilation operation, and new small holes or gaps that may be generated by the erosion operation are filled. By adjusting the size and shape of the structure element and the number of erosion and dilation operations, the fitting degree of the processing result of the morphological filtering to the target in the original image can be optimized, and finally the fitting degree of the processing result of the morphological filtering to the target in the original image is optimal.

[0046] Optionally, before inputting the original image into the image denoising model, further comprising: obtaining a plurality of training samples, wherein the training samples include a plurality of sample original images with noise and expected denoised images corresponding to the sample original images; inputting the sample original images into a generative model of a pre-established generative adversarial network to obtain model denoised images corresponding to the sample loss images; determining a first image denoising loss between the model denoised images and the expected denoised images corresponding to the model denoised images through a preset loss function; determining a second image denoising loss between the model denoised images and the expected denoised images corresponding to the model denoised images through a discriminative model of the generative adversarial network; and adjusting model parameters of the generative model based on the first image denoising loss and the second image denoising loss to obtain the image denoising model.

[0047] Specifically, a plurality of pairs of original images with noise and corresponding expected denoised images are collected. These image pairs constitute a training data set for training the generative adversarial network. The sample original images with noise are input into the generative model. The generative model generates a denoised image, i.e., a model denoised image, from the input image with noise. A preset loss function (such as mean square error, peak signal-to-noise ratio, etc.) is used to calculate the difference between the model denoised image and the expected denoised image, obtaining a first image denoising loss. The discriminative model discriminates between the model denoised image and the expected denoised image and outputs a discrimination result, i.e., a second image denoising loss. Based on the first image denoising loss and the second image denoising loss, an optimization algorithm is used to adjust the parameters of the generative model to reduce the two loss values. At the same time, the parameters of the discriminative model will also be adjusted according to the discrimination result to improve its discrimination ability. The above training process is repeated until the generative model can generate a model denoised image that is close enough to the expected denoised image, and the discriminative model cannot distinguish between the model denoised image and the real image. It is considered that the model has converged, the model training is completed, and the image denoising model is obtained.

[0048] S130, determining a target nameplate region in the target to-be-recognized image based on an edge detection method and a contour finding method, and extracting device nameplate data in the target nameplate region.

[0049] The target nameplate region can be understood as a rectangular region of the target nameplate in the target to-be-recognized image. The device nameplate data can be understood as power equipment associated data recorded on the target nameplate.

[0050] Specifically, a region with a probability greater than a preset probability threshold in the target to-be-recognized image is determined as the target nameplate region based on the edge detection method and the contour finding method, and characters in the nameplate region are recognized through an optical character recognition algorithm to extract the device nameplate data in the target nameplate region.

[0051] Optionally, the nameplate data includes at least one of a device model, a device serial number, a rated voltage, a rated current, a rated frequency, and a manufacturing date.

[0052] The technical scheme of the embodiment of the application comprises the following steps: obtaining at least one original image containing a target power device nameplate; inputting the original image into an image denoising model, and determining a target to-be-recognized image based on a model output result, wherein the image denoising model is a generated model trained based on a generative adversarial network generated from a sample image containing a power device nameplate and a corresponding expected denoised image; improving the image quality of the target to-be-recognized image, improving the data recognition accuracy based on the target to-be-recognized image, and finally determining a target nameplate region in the target to-be-recognized image based on an edge detection method and a contour searching method, and extracting device nameplate data in the target nameplate region. The problem of low recognition accuracy of device nameplate data is solved, and the beneficial effects of improved recognition efficiency and accuracy of device nameplate data are achieved.

[0053] Embodiment two

[0054] Figure 2 A flowchart of a power device nameplate recognition method provided by the second embodiment of the application is shown in the figure. The embodiment is a further optimization of how to determine a target nameplate region in a target to-be-recognized image based on an edge detection method and a contour searching method in the above-mentioned embodiment. Optionally, the determination of the target nameplate region in the target to-be-recognized image based on the edge detection method and the contour searching method comprises: determining edge data in the target to-be-recognized image based on the edge detection method, and determining contour data in the target to-be-recognized image based on a contour searching algorithm; determining a candidate nameplate region based on the edge data and the contour data, determining a plurality of corner points in the candidate nameplate region based on a corner point detection algorithm, and performing perspective projection transformation on the plurality of corner points to obtain a target nameplate region.

[0055] As Figure 2 shown, the method comprises:

[0056] S210, obtaining at least one original image containing a target power device nameplate.

[0057] S220, inputting the original image into an image denoising model, and determining a target to-be-recognized image based on a model output result.

[0058] S230, determining edge data in the target to-be-recognized image based on an edge detection method, and determining contour data in the target to-be-recognized image based on a contour searching algorithm.

[0059] Specifically, an edge detection algorithm is used to detect the edges of the plaque region in the target image to be recognized to obtain edge data. A contour finding algorithm is used to identify the contour of the plaque region. According to the size, shape and other characteristics of the contour, candidate plaque regions are screened out.

[0060] S240, based on the edge data and the contour data, determine a candidate plaque region, determine a plurality of corner points in the candidate plaque region based on a corner point detection algorithm, perform perspective projection transformation on the plurality of corner points to obtain a target plaque region, and extract equipment plaque data in the target plaque region.

[0061] Specifically, in the candidate plaque region, a plurality of corner points in the candidate plaque region are determined by a corner point detection algorithm. Four corner points are selected from the detected plurality of corner points. The four corner points should represent the four vertices of the candidate plaque region. A perspective transformation matrix is calculated using the four corner points and their corresponding target positions. Based on the perspective transformation matrix, perspective projection transformation is performed on the candidate plaque region to obtain the target plaque region. The corner point can be a point with a sharp change in brightness in the image, for example, a corner or intersection of an object boundary.

[0062] Preferably, an adaptive lighting adjustment technique can be introduced to dynamically adjust the exposure and contrast of the collected original image, improving the recognition robustness; specifically including: analyzing the brightness distribution of the original image, calculating the global and local brightness statistical information; according to the brightness statistical information, dynamically adjusting the exposure and contrast of the image, to ensure that the plaque region in the original image has good visibility. A light invariance feature extractor based on deep learning is used, which is trained by a large number of plaque images under different lighting conditions, so as to learn the key features that are not affected by lighting changes. For example, the overall brightness average and standard deviation of the original image are calculated to evaluate the global brightness level and contrast of the original image. The original image is divided into multiple grids, and the brightness of each grid is counted to identify the highlight and shadow areas in the original image. According to the global brightness average, if the image is too dark, increase the exposure; if the image is too bright, reduce the exposure. According to the global brightness standard deviation, if the contrast is low, a contrast enhancement algorithm is applied. The adaptive contrast enhancement technique is used to improve the contrast of local areas. A large number of sample plaque images containing plaques under different lighting conditions are collected to ensure that the data set covers various lighting scenarios, such as strong light, weak light, backlight, shadow, etc. A convolutional neural network is trained based on the sample plaque images, so that it can learn the light invariance features of the plaque under different lighting conditions. The image after adaptive lighting adjustment is input into the trained deep learning model, and the model outputs the light invariance features of the plaque.

[0063] The technical scheme of the embodiment of the present application determines the edge data in the target image to be recognized based on an edge detection method, and determines the contour data in the target image to be recognized based on a contour search algorithm; the positioning accuracy of the nameplate region is improved, the candidate nameplate region is determined based on the edge data and the contour data, a plurality of corner points in the candidate nameplate region are determined based on a corner detection algorithm, and perspective projection transformation is performed on the plurality of corner points to obtain a target nameplate region. This is helpful to simplify the subsequent image recognition process and improve the data recognition efficiency.

[0064] As an optional example of the embodiment of the present application, the identification method of the power equipment nameplate of the embodiment specifically includes the following steps:

[0065] Step S1: Obtain at least one original image containing a target power equipment nameplate.

[0066] Step S2: Preprocess the original image, and update the original nameplate image based on the preprocessed original image; wherein the preprocessing includes at least one of gray scale processing, gradient transformation processing, smoothing filtering processing and thresholding processing.

[0067] Step S3: Perform adaptive denoising processing on the original nameplate image in step S2 to eliminate isolated noise points in the region image to obtain a target image to be recognized.

[0068] The adaptive denoising processing is specifically to use the generator part in the generative adversarial network to generate a noise-free image, and the discriminator is used to evaluate the denoising effect and continuously optimize the denoising process.

[0069] Step S4: Fill in the graph gap of the target image to be recognized obtained in step S3 using a preset morphological algorithm to filter out the protruding blocks affecting the target image.

[0070] Step S5: Determine the target nameplate region in the target image to be recognized based on the edge detection method and the contour search method, and extract the equipment nameplate data in the target nameplate region.

[0071] In the embodiment, the preprocessing of the original image in step S2 is implemented in the following manner:

[0072] Step S21: Perform gray scale processing on the original image to obtain a single-channel gray scale image.

[0073] Step S22: Perform gradient transformation processing on the gray scale image obtained in step S21 to obtain a gray scale image capable of recognizing edges; the gradient transformation processing is to use a Scharr operator to construct the gradient amplitude of the gray scale image in the horizontal and vertical directions.

[0074] Step S23: performing a smoothing filtering process on the image obtained in step S22 to obtain a filtered image; the smoothing filtering process is performed by using a low-pass smoothing filter, for example, using a 5*5 smoothing matrix.

[0075] Step S24: performing a thresholding process on the filtered image obtained in step S23 to obtain a region image with clear wheel flares.

[0076] The thresholding process specifically includes the following steps:

[0077] Step S241: determining the threshold value between the foreground and the background of the image by using the maximum inter-class variance method, the proportion of the number of pixel points in the background to the total number of pixel points in the image is set as p0, the proportion of the number of foreground pixel points to the total number of pixel points in the image is recorded as p1, the non-average gray values of the background and the foreground are q0 and q1 respectively, and the average gray value of the image is:

[0078] average = p0 x q0 + p1 x q1;

[0079] Step S242: calculating the variance between the foreground and the background of the image, that is:

[0080] variance0 = p0 x ( q0-average ) 2 + p1 x ( q1-average ) 2 ;

[0081] Step S243: determining the segmentation threshold value when the variance is maximum.

[0082] In this embodiment, in step S3, the adaptive denoising process further includes a hybrid filtering technique, which combines a traditional filtering algorithm and a method based on deep learning to further refine the denoising result; specifically including the following steps:

[0083] Step S31: performing a filtering process on the original image to obtain a first denoised image;

[0084] Specifically, a preset filtering algorithm (such as median filtering or bilateral filtering) is used to preliminarily denoise the image to obtain a first denoised image.

[0085] Step S32: inputting the first denoised image into the image denoising model to obtain a second denoised image;

[0086] Specifically, the preliminarily denoised image is input into a pre-trained deep neural network, which has been trained with a large number of image pairs with and without noise, and can automatically learn and eliminate the best strategy for different types of noise.

[0087] Step S33: determining a target to-be-recognized image based on the first denoised image and the second denoised image.

[0088] Specifically, the results of the initial filtering and the deep learning denoising are fused to obtain the target to-be-recognized image.

[0089] In this embodiment, in step S4, the preset morphological algorithm is a composite morphological algorithm including four morphologies of erosion, dilation, opening and closing; the target to-be-recognized image obtained in step S3 is subjected to a figure gap filling by using the preset morphological algorithm, and the specific mode is to perform a closing operation first, and then perform morphological operations in the order of erosion first and then dilation again, so that the fitting degree of the processing result of the morphological filtering to the original image is best; and the specific mode includes the following steps:

[0090] Step S41: first, a closing operation is performed on the image to fill small holes and broken lines.

[0091] Step S42: then, an erosion operation is performed to remove small interference structures;

[0092] Step S43: then, a dilation operation is performed to restore the eroded part and further smooth the edges;

[0093] Step S44: repeat the above steps until the fitting degree of the image is best.

[0094] In this embodiment, in step S5, the specific method of contour extraction is to find out the largest rectangular frame surrounding the plaque by using a contour finding algorithm, and mark the area where the plaque is located, then use corner point detection to identify 4 edge points of the plaque in the area, and perform perspective projection operation on the 4 edge points, finally convert the irregular quadrilateral into a rectangle; and the specific method includes the following steps:

[0095] Step S51: determining edge data in the target to-be-recognized image based on an edge detection method;

[0096] Step S52: determining contour data in the target to-be-recognized image based on a contour finding algorithm;

[0097] Step S53: determining a candidate plaque area based on the edge data and the contour data;

[0098] Step S54: determining a plurality of corner points in the candidate plaque area based on a corner point detection algorithm;

[0099] Step S55: performing perspective projection transformation on the plurality of corner points to obtain a target plaque area.

[0100] Preferably, in this embodiment, an adaptive lighting adjustment technique is also introduced to dynamically adjust exposure and contrast based on the captured raw image, improving recognition robustness; Specifically, the following steps are included:

[0101] Step S61: Analyze the brightness distribution of the original image, and calculate the global and local brightness statistical information;

[0102] Step S62: Dynamically adjust the exposure and contrast of the original image based on the brightness statistical information, to ensure that the nameplate region in the original image has good visibility;

[0103] Step S63: Use a deep learning-based lighting invariant feature extractor, which is trained on a large number of nameplate images under different lighting conditions, to learn key features that are not affected by lighting changes.

[0104] In this embodiment, a multi-light source imaging technique is also included, which obtains nameplate shooting images of the target power equipment nameplate based on multiple sets of shooting parameters, and fuses multiple said nameplate shooting images to obtain an original image containing the target power equipment nameplate.

[0105] For example, nameplate images are taken from multiple angles or different light source intensities, and then these images are fused by an algorithm to reduce the effects of reflections and shadows; Specifically, the following steps are included:

[0106] Step S71: Use a multi-light source imaging system to obtain nameplate shooting images of the target power equipment nameplate based on multiple sets of shooting parameters;

[0107] Step S72: Fuse multiple said nameplate shooting images to obtain an original image containing the target power equipment nameplate.

[0108] Specifically, the original image reduces the effects of reflections and shadows, improving recognition accuracy.

[0109] In this embodiment, a multi-modal fusion framework is also designed to combine visual information with other sensor data (such as laser radar) to enhance the understanding and positioning ability for non-standard nameplate styles; Specifically, the following steps are included:

[0110] Step S81: Obtain three-dimensional point cloud data of the target nameplate through laser radar, and obtain two-dimensional image data of the target nameplate through camera.

[0111] Step S82: Register the two-dimensional image data and the three-dimensional point cloud data to establish a correspondence between the two-dimensional image data and the three-dimensional point cloud data;

[0112] Step S83: Use three-dimensional point cloud data to assist in analyzing two-dimensional image data;

[0113] For example, the three-dimensional geometric information is used to help locate and identify the position and shape of the nameplate.

[0114] Step S84: The result of the multi-modal data fusion is used in the subsequent identification and matching process to improve the robustness and accuracy of the system.

[0115] The technical solution of the embodiment of the application uses a generative adversarial network for adaptive denoising processing. It can effectively eliminate isolated noise points and also handle more complex noise patterns. An adaptive illumination adjustment technique is introduced to dynamically adjust exposure and contrast based on the collected images. This enables the system to maintain good recognition performance under different lighting conditions. By analyzing the brightness distribution of the image and dynamically adjusting the exposure and contrast, it ensures that the nameplate region has good visibility. In addition, the illumination-invariant feature extractor based on deep learning further enhances the robustness, making it immune to changes in lighting conditions. Through the preprocessing, adaptive denoising, morphological processing, and contour extraction steps, the accuracy and robustness of power equipment nameplate recognition are significantly improved, making it particularly suitable for complex environments and diverse nameplate styles.

[0116] Embodiment Three

[0117] Figure 3 A structural schematic diagram of a power equipment nameplate recognition device provided by Embodiment Three of the application. As shown in the figure, the device includes an original image acquisition module 310, a to-be-identified image determination module 320, and a device nameplate data extraction module 330. Figure 3

[0118] The original image acquisition module 310 is used to acquire at least one original image containing a target power equipment nameplate. The to-be-identified image determination module 320 is used to input the original image into an image denoising model and determine a target to-be-identified image based on the model output result. The image denoising model is a generative model trained based on a generative adversarial network using sample images containing power equipment nameplates and corresponding expected denoised images. The device nameplate data extraction module 330 is used to determine a target nameplate region in the target to-be-identified image based on an edge detection method and a contour search method, and extract device nameplate data in the target nameplate region.

[0119] ​The technical scheme of the embodiment of the application comprises the following steps: an original image acquisition module is used to acquire at least one original image containing a target power equipment nameplate; then, a to-be-recognized image determination module is used to input the original image into an image denoising model, and determine a target to-be-recognized image based on a model output result, wherein the image denoising model is a generative model trained based on a generative adversarial network generated from a sample image containing a power equipment nameplate and an expected denoising image corresponding to the sample image; the image quality of the target to-be-recognized image is improved, the data recognition accuracy based on the target to-be-recognized image is improved, finally, a device nameplate data extraction module is used to determine a target nameplate region in the target to-be-recognized image based on an edge detection method and a contour searching method, and extract device nameplate data in the target nameplate region, so that the device nameplate data in the target nameplate region is accurately extracted. The problem of low recognition accuracy of device nameplate data is solved, and the beneficial effects of improving the recognition efficiency and accuracy of device nameplate data are achieved.

[0120] Optionally, the to-be-recognized image determination module comprises:

[0121] a first denoising unit, configured to perform filtering processing on the original image to obtain a first denoising image;

[0122] a second denoising unit, configured to input the first denoising image into the image denoising model to obtain a second denoising image;

[0123] a to-be-recognized image determination unit, configured to determine a target to-be-recognized image based on the first denoising image and the second denoising image.

[0124] Optionally, the to-be-recognized image determination unit comprises:

[0125] a fusion processing subunit, configured to perform fusion processing on the first denoising image and the second denoising image to obtain an initial to-be-recognized image;

[0126] a filling processing subunit, configured to perform filling processing on an image gap in the initial to-be-recognized image to obtain a target to-be-recognized image.

[0127] Optionally, the device further comprises a sample acquisition module, a model training module, a first loss determination module, a second loss determination module, and a model adjustment module.

[0128] The sample acquisition module is configured to acquire a plurality of training samples before the original image is input into the image denoising model, wherein the training samples comprise a plurality of sample original images with noise and expected denoising images corresponding to the sample original images;

[0129] The model training module is configured to input the sample original image into a generative model of a pre-established generative adversarial network to obtain a model denoised image corresponding to the sample loss image.

[0130] The first loss determination module is configured to determine a first image denoising loss between the model denoised image and the expected denoised image corresponding to the model denoised image by using a preset loss function.

[0131] The second loss determination module is configured to determine a second image denoising loss between the model denoised image and the expected denoised image corresponding to the model denoised image by using a discriminative model of the generative adversarial network.

[0132] The model adjustment module is configured to adjust model parameters of the generative model based on the first image denoising loss and the second image denoising loss to obtain the image denoising model.

[0133] Optionally, the equipment nameplate data extraction module comprises:

[0134] The data extraction unit is configured to determine edge data in the target to-be-recognized image based on an edge detection method, and determine contour data in the target to-be-recognized image based on a contour search algorithm.

[0135] The target region determination unit is configured to determine a candidate nameplate region based on the edge data and the contour data, determine a plurality of corner points in the candidate nameplate region based on a corner point detection algorithm, and perform perspective projection transformation on the plurality of corner points to obtain a target nameplate region.

[0136] Optionally, the original image acquisition module is specifically configured to:

[0137] acquire a plurality of nameplate shooting images of the target power equipment nameplate based on a plurality of shooting parameters, and perform fusion processing on the plurality of nameplate shooting images to obtain the original image containing the target power equipment nameplate.

[0138] Optionally, the device further comprises a preprocessing module.

[0139] The preprocessing module is configured to perform preprocessing on the original image after the at least one original image of the target power equipment nameplate is acquired, and update the original nameplate image based on the preprocessed original image; wherein the preprocessing comprises at least one of gray scale processing, gradient transformation processing, smoothing filtering processing, and thresholding processing.

[0140] Optionally, the nameplate data comprises at least one of a device model, a device serial number, a rated voltage, a rated current, a rated frequency, and a manufacturing date.

[0141] The power equipment nameplate recognition device provided by the embodiment of the present application can execute the power equipment nameplate recognition method provided by any embodiment of the present application, has the function modules and beneficial effects corresponding to the execution method.

[0142] Embodiment four

[0143] Figure 4 A structural schematic diagram of an electronic device 10 that can be used to implement embodiments of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (e.g., headsets, glasses, watches, etc.), and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present application described and / or claimed in this document.

[0144] As shown in Figure 4 The electronic device 10 includes at least one processor 11, and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11, wherein the memory stores a computer program that can be executed by the at least one processor. The processor 11 can execute various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0145] A plurality of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, a loudspeaker, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0146] The processor 11 may be any general-purpose and / or specialized processing component with processing and computing capabilities. Examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the method for identifying the nameplate of an electric power device.

[0147] In some embodiments, the method for identifying the nameplate of an electric device may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for identifying the nameplate of an electric device described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the method for identifying the nameplate of an electric device in any other suitable manner (e.g., via firmware).

[0148] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0149] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0150] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer-readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of a program of instructions in a transitory signal, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0151] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0152] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0153] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0154] It should be understood that the various forms of flow shown above can be reordered, added to, or have steps deleted. For example, the steps described in the present application can be performed in parallel, in series, or in a different order, as long as the desired results of the technical solutions of the present application can be achieved, and this is not limited herein.

[0155] The above detailed description does not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method of identifying a nameplate of an electrical equipment, characterized by, The method comprises the following steps: acquiring at least one original image containing a target power equipment nameplate; inputting the original image into an image denoising model to determine a target to-be-recognized image based on a model output result, wherein the image denoising model is a generated model trained based on a generative adversarial network generated from a sample image containing a power equipment nameplate and a corresponding expected denoising image; before inputting the original image into the image denoising model, further comprising: acquiring a plurality of training samples, wherein the training samples comprise a plurality of sample original images with noise and corresponding expected denoising images; inputting the sample original images into a generated model of a pre-established generative adversarial network to obtain model denoising images corresponding to sample loss images; determining a first image denoising loss between the model denoising images and the corresponding expected denoising images by a preset loss function; determining a second image denoising loss between the model denoising images and the corresponding expected denoising images by a discriminative model of the generative adversarial network; adjusting model parameters of the generated model based on the first image denoising loss and the second image denoising loss to obtain the image denoising model; determining a target nameplate region in the target to-be-recognized image based on an edge detection method and a contour searching method, comprising: determining edge data in the target to-be-recognized image based on the edge detection method, and determining contour data in the target to-be-recognized image based on a contour searching algorithm; determining a candidate nameplate region based on the edge data and the contour data, determining a plurality of corner points in the candidate nameplate region based on a corner point detection algorithm, and performing perspective projection transformation on the plurality of corner points to obtain a target nameplate region; extracting equipment nameplate data in the target nameplate region.

2. The method of claim 1, wherein, The inputting of the original nameplate image into the image denoising model to obtain a target to-be-recognized nameplate image comprises: performing filter processing on the original image to obtain a first denoising image; inputting the first denoising image into the image denoising model to obtain a second denoising image; determining a target to-be-recognized image based on the first denoising image and the second denoising image.

3. The method of claim 2, wherein, The determination of a target to-be-recognized image based on the first denoising image and the second denoising image comprises: performing fusion processing on the first denoising image and the second denoising image to obtain an initial to-be-recognized image; performing gap filling processing on the initial to-be-recognized image to obtain a target to-be-recognized image.

4. The method of claim 1, wherein, The acquiring of at least one original image containing a target power equipment nameplate comprises: acquiring nameplate shooting images of the target power equipment nameplate based on a plurality of shooting parameters, and performing fusion processing on the plurality of nameplate shooting images to obtain an original image containing the target power equipment nameplate.

5. The method of claim 1, wherein, After acquiring at least one original image of the target power equipment nameplate, further comprising: The original image is preprocessed, and the original plaque image is updated based on the preprocessed original image; wherein the preprocessing includes at least one of gray processing, gradient transformation processing, smoothing filtering processing and thresholding processing.

6. The method of claim 1, wherein, Wherein, The plaque data includes at least one of device model, device serial number, rated voltage, rated current, rated frequency and manufacturing date.

7. A device for identifying nameplates of electric equipment, characterized in that: Including: An original image acquisition module is configured to acquire at least one original image containing a target power equipment plaque; A to-be-recognized image determination module is configured to input the original image into an image denoising model, and determine a target to-be-recognized image based on a model output result, wherein the image denoising model is a generative model trained based on a generative adversarial network generated from a sample image containing a power equipment plaque and an expected denoising image corresponding to the sample image; A sample acquisition module is configured to acquire a plurality of training samples before the original image is input into the image denoising model, wherein the training samples include a plurality of sample original images with noise and expected denoising images corresponding to the sample original images; A model training module is configured to input the sample original images into a generative model of a pre-established generative adversarial network to obtain model denoising images corresponding to sample loss images; A first loss determination module is configured to determine a first image denoising loss between the model denoising images and the expected denoising images corresponding to the model denoising images through a preset loss function; A second loss determination module is configured to determine a second image denoising loss between the model denoising images and the expected denoising images corresponding to the model denoising images through a discriminative model of the generative adversarial network; A model adjustment module is configured to adjust model parameters of the generative model based on the first image denoising loss and the second image denoising loss to obtain the image denoising model; A device plaque data extraction module is configured to determine a target plaque region in the target to-be-recognized image based on an edge detection method and a contour finding method, and extract device plaque data in the target plaque region; The device plaque data extraction module includes: A data extraction unit is configured to determine edge data in the target to-be-recognized image based on an edge detection method, and determine contour data in the target to-be-recognized image based on a contour finding algorithm; A target region determination unit is configured to determine a candidate plaque region based on the edge data and the contour data, determine a plurality of corner points in the candidate plaque region based on a corner point detection algorithm, and perform perspective projection transformation on the plurality of corner points to obtain a target plaque region.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for causing the processor to execute the power equipment plaque recognition method of any one of claims 1-6 when executed.

Images