Transmission line icing thickness prediction method and device based on monocular depth estimation

Abstract

The invention provides a power transmission line icing thickness prediction method and device based on monocular depth estimation, and the method comprises the steps: carrying out the region division of a target power transmission line icing image based on a pre-trained icing thickness prediction model, and obtaining a power transmission line icing sub-image containing a target region; respectively determining a pre-icing monocular depth image and a post-icing monocular depth image corresponding to the power transmission line icing sub-image; pixel-level subtraction processing is carried out on the pre-icing monocular depth image and the post-icing monocular depth image to obtain a pre-icing and post-icing difference depth map; performing matrix conversion on the difference depth map before and after icing to obtain an icing difference matrix; based on a preset proportional relation calculation formula, calculating the icing difference matrix and real line parameters to obtain a pixel-real thickness proportional relation; and based on a preset real thickness positioning algorithm and the pixel-real thickness proportional relation, performing icing thickness prediction on the power transmission line in the to-be-predicted region to obtain an icing thickness prediction result.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of icing monitoring for transmission lines, and involve but are not limited to a method and device for predicting the icing thickness of transmission lines based on monocular depth estimation. Background Art

[0002] Icing on transmission lines is highly random and irresistible, which may cause mechanical and electrical failures such as tower pole toppling, fitting damage, and conductor burnout. Even worse, it may lead to large-scale power outages and then trigger major power accidents such as grid collapse, seriously threatening the safe and stable operation of transmission lines and power systems. Traditional icing measurement methods (such as manual inspection, weighing method, laser ranging, etc.) have problems such as low efficiency, high cost, and difficulty in real-time monitoring. Therefore, how to monitor the icing of transmission lines by means of intelligence is of great significance for improving the intrinsic safety operation of power systems and ensuring residents' electricity use.

[0003] Monocular depth estimation is one of the key technologies in the field of computer vision. Its core task is to reflect three-dimensional depth geometric information through two-dimensional images. The mainstream methods can be divided into two categories: traditional algorithms and deep learning-based methods. Traditional methods mainly rely on manual features (such as texture gradients, occlusion relationships) and geometric priors (such as Markov random fields), and are achieved through stereo matching or structure from motion, but the accuracy is limited and depends on scene assumptions. Deep learning methods have significantly improved performance through end-to-end training and methods such as multi-scale feature fusion, attention mechanism, and self-supervised learning. In the latest research, the Transformer architecture further improves the detail preservation ability through long-range dependence modeling.

[0004] Considering that the current research on measuring the icing thickness of transmission lines mainly focuses on reflecting the icing thickness on both sides of the transmission line through the difference in the line edges before and after icing, there are certain deficiencies in the research on the icing thickness on the surface of the transmission line. In practical applications, if the icing on the line surface is too thick, it is prone to overloading phenomena and then cause power accidents. Summary of the Invention

[0005] Based on the problems in the related art, the embodiments of the present invention provide a method and device for predicting the icing thickness of transmission lines based on monocular depth estimation.

[0006] The technical solutions of the embodiments of the present invention are implemented as follows:

[0007] The embodiments of the present invention provide a method for predicting the icing thickness of transmission lines based on monocular depth estimation, and the method includes:

[0008] Obtain the target transmission line icing image and the transmission line non-icing image of the area to be predicted;

[0009] Based on a pre-trained icing thickness prediction model, divide the icing image of the target transmission line into regions to obtain a sub-image of the transmission line icing containing the target region;

[0010] By using a preset improved monocular depth estimation algorithm, respectively determine the pre-icing monocular depth image and the post-icing monocular depth image corresponding to the sub-image of the transmission line icing;

[0011] When the depth dimensions of the pre-icing monocular depth image and the post-icing monocular depth image are the same, perform pixel-level subtraction processing on the pre-icing monocular depth image and the post-icing monocular depth image to obtain a depth difference map before and after icing;

[0012] By using a preset image matrix conversion algorithm, perform matrix conversion on the depth difference map before and after icing to obtain an icing difference matrix;

[0013] Based on a preset proportional relationship calculation formula, calculate the icing difference matrix and the real line parameters to obtain a pixel-real thickness proportional relationship; the preset proportional relationship calculation formula is: ; where a represents the pixel-real thickness proportional relationship; R represents the real radius distance of the line; represents the depth value at the rear end of the line center point; represents the depth value at the front end of the line center point;

[0014] Based on a preset real thickness positioning algorithm and the pixel-real thickness proportional relationship, predict the icing thickness of the transmission line in the area to be predicted to obtain an icing thickness prediction result.

[0015] An embodiment of the present invention provides a device for predicting the icing thickness of a transmission line based on monocular depth estimation. The device includes:

[0016] An acquisition module, configured to acquire an icing image of a target transmission line in an area to be predicted and an image of the transmission line without icing;

[0017] A division module, configured to divide the icing image of the target transmission line into regions based on a pre-trained icing thickness prediction model to obtain a sub-image of the transmission line icing containing the target region;

[0018] A determination module, configured to respectively determine a pre-icing monocular depth image and a post-icing monocular depth image corresponding to the sub-image of the transmission line icing by using a preset improved monocular depth estimation algorithm;

[0019] A processing module, configured to perform pixel-level subtraction processing on the pre-icing monocular depth image and the post-icing monocular depth image when the depth dimensions of the pre-icing monocular depth image and the post-icing monocular depth image are the same to obtain a depth difference map before and after icing;

[0020] A conversion module, configured to perform matrix conversion on the difference depth map before and after icing through a preset image matrix conversion algorithm to obtain an icing difference matrix;

[0021] A calculation module, configured to calculate the pixel-real thickness ratio based on a preset ratio relationship calculation formula for the icing difference matrix and the real line parameters; the preset ratio relationship calculation formula is: ; where a represents the pixel-real thickness ratio; R represents the real radius distance of the line; represents the depth value at the back end of the line center point; represents the depth value at the front end of the line center point;

[0022] A prediction module, configured to predict the icing thickness of the transmission line in the area to be predicted based on a preset real thickness positioning algorithm and the pixel-real thickness ratio to obtain an icing thickness prediction result.

[0023] In some embodiments, the acquisition module is further configured to collect an initial icing image of the transmission line in the area to be predicted in real time through an image acquisition device; perform noise reduction processing on the initial icing image of the transmission line to obtain a noise-reduced icing image of the transmission line; perform alignment processing on the noise-reduced icing image of the transmission line to obtain an aligned icing image of the transmission line; perform grayscale processing on the aligned icing image of the transmission line to obtain the target icing image of the transmission line.

[0024] In some embodiments, before partitioning the target icing image of the transmission line into sub-images including the target area based on a pre-trained icing thickness prediction model, the method further includes: a training module, configured to acquire a sample icing image of the transmission line; label the icing target area in the sample icing image of the transmission line to obtain a labeled sample icing image of the transmission line; set the training learning rate, batch size, and number of iterations, and train an initial YOLOv10 object detection model to obtain a trained YOLOv10 object detection model; predict the sample icing image of the transmission line based on the trained YOLOv10 object detection model to obtain a sample icing thickness prediction result; calculate the error based on the sample icing thickness prediction result and the labeled sample icing image of the transmission line to obtain the optimal weight.

[0025] In some embodiments, the partitioning module is further configured to perform boundary detection on the target transmission line icing image based on the optimal weights determined by the pre-trained icing thickness prediction model to obtain predicted bounding box information; and perform region cropping on the target transmission line icing image based on the predicted bounding box information to obtain the transmission line icing sub-image.

[0026] In some embodiments, the determination module is further configured to perform normalization processing on the images of the transmission line icing sub-image and the transmission line non-icing image at the same angle but different times respectively, and correspondingly obtain the processed transmission line icing sub-image and the processed transmission line non-icing image; based on the preset improved monocular depth estimation algorithm, perform monocular depth estimation on the processed transmission line icing sub-image and the processed transmission line non-icing image respectively, and correspondingly obtain the first depth estimation value of each pixel in the transmission line icing sub-image and the second depth estimation value of each pixel in the transmission line non-icing image; perform format conversion on the first depth estimation value and the second depth estimation value to respectively obtain the pre-icing monocular depth image and the post-icing monocular depth image.

[0027] In some embodiments, the processing module is further configured to subtract the pixel value of each coordinate in the post-icing monocular depth image from the pixel value of each coordinate in the pre-icing monocular depth image to obtain the pixel value at the corresponding coordinate ; the calculation formula is as follows: where

[0028] An embodiment of the present invention provides a transmission line icing thickness prediction device based on monocular depth estimation, including: a memory for storing executable instructions; a processor for implementing the above-mentioned transmission line icing thickness prediction method based on monocular depth estimation when executing the executable instructions stored in the memory.

[0029] An embodiment of the present invention provides a computer-readable storage medium storing executable instructions for causing a processor to implement the above-mentioned transmission line icing thickness prediction method based on monocular depth estimation when executing the executable instructions.

[0030] One or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages:

[0031] The present invention constructs a method for monitoring the icing thickness on the surface of a transmission line based on monocular depth estimation, achieving accurate measurement of the icing thickness on the surface of the transmission line, and providing a research basis and algorithm tool for calculating the icing weight and evaluating the severity of the transmission line; the present invention proposes an improved diffusion monocular depth estimation model, adding a spatial attention mechanism on the basis of the monocular depth estimation algorithm, enabling the model to focus on the area of information to be detected in the picture, helping the model to more clearly capture the position and thickness information of the icing on the surface of the transmission line, and thus more accurately estimating the depth value; in addition, adding the spatial attention mechanism can also play a role in suppressing noise, by weighting different channels in the image, enhancing the weight of the channels dominated by useful information, reducing the weight of the channels dominated by noise information, enabling it to have the ability to accurately identify noise and non-noise, thereby improving the quality of the original image, providing a more reliable input for the depth estimation model, and comprehensively improving the detection effect of the depth estimation algorithm in the night scene; the present invention proposes an image-matrix conversion model, which can extract the depth value represented by each pixel value in the depth map and convert it into a two-dimensional depth numerical matrix for subsequent numerical calculations. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 FIG. is a schematic structural diagram of a transmission line icing thickness prediction system based on monocular depth estimation provided by the present invention;

[0033] Figure 2 FIG. is a schematic flow diagram of a method for predicting the icing thickness of a transmission line based on monocular depth estimation provided by the present invention;

[0034] Figure 3 FIG. is a schematic flow diagram of an intelligent measurement method for the icing thickness of a transmission line based on monocular depth estimation and real thickness positioning provided by the present invention;

[0035] Figure 4 FIG. is a schematic composition structure diagram of a transmission line icing thickness prediction device based on monocular depth estimation provided by an embodiment of the present invention;

[0036] Figure 5 FIG. is a schematic composition structure diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be construed as limiting the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0038] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict. Unless otherwise defined, all technical and scientific terms used in the embodiments of the present invention have the same meaning as commonly understood by those skilled in the technical field to which the embodiments of the present invention belong. The terms used in the embodiments of the present invention are only for the purpose of describing the embodiments of the present invention and are not intended to limit the present invention.

[0039] In the related art, the traditional icing measurement method has the following disadvantages: 1) Considering that the traditional icing measurement method has problems such as low efficiency, high cost, and difficulty in real-time monitoring; the monocular vision measurement method has attracted attention due to its simple equipment and easy deployment, but the existing technology has defects such as large depth estimation error and difficulty in accurately positioning the real icing boundary.

[0040] 2) Considering the complexity of the icing site environment, the on-site icing images captured by the image acquisition device are easily affected by extreme environments. In these images, the night scene occupies a certain proportion. In the night scene, the overall brightness of the image is low, the difference between the foreground and the background is small, some key texture information is missing, the noise is relatively obvious, and the information to be detected is easily ignored. For such conditions, the detection effect of the monocular depth estimation algorithm (Marigold) is not good, and the relative depth of the icing on the transmission line surface cannot be effectively extracted, and some areas of the detection result are integrated with the background, which is not conducive to the subsequent real thickness positioning operation.

[0041] The following describes an exemplary application of the transmission line icing thickness prediction device based on monocular depth estimation according to an embodiment of the present invention. The transmission line icing thickness prediction device based on monocular depth estimation provided by the embodiment of the present invention can be implemented as a terminal or a server. In one implementation, the transmission line icing thickness prediction device based on monocular depth estimation provided by the embodiment of the present invention can be implemented as various types of terminals such as laptops, tablets, desktop computers, and mobile devices; in another implementation, the transmission line icing thickness prediction device based on monocular depth estimation provided by the embodiment of the present invention can also be implemented as a server. Among them, the server can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs, Content Delivery Networks), and big data and artificial intelligence platforms. The terminal and the server can be directly or indirectly connected through wired or wireless communication methods, which are not limited in the embodiment of the present invention. Next, an exemplary application when the transmission line icing thickness prediction device based on monocular depth estimation is a server will be described.

[0042] See Figure 1 , Figure 1 is a schematic structural diagram of a transmission line icing thickness prediction system 10 based on monocular depth estimation provided by an embodiment of the present invention. The embodiment of the present invention can provide a transmission line icing thickness prediction platform based on monocular depth estimation, and the transmission line icing thickness prediction platform based on monocular depth estimation can be implemented as a transmission line icing thickness prediction application based on monocular depth estimation. The transmission line icing thickness prediction system 10 provided by the embodiment of the present invention includes a terminal 110, a network 120, and a server 130. Among them, the server 130 is a server of a transmission line icing thickness prediction application based on monocular depth estimation. The server 130 can constitute the transmission line icing thickness prediction device based on monocular depth estimation according to an embodiment of the present invention. The terminal 110 is connected to the server 130 through the network 120, and the network 120 can be a wide area network, a local area network, or a combination of the two.

[0043] In some embodiments, please refer to Figure 1, when predicting the ice thickness of a transmission line, the terminal 110 sends the initiated ice thickness prediction task of the transmission line to the server 130 through the network 120. In response to the ice thickness prediction task of the transmission line sent by the terminal 110, the server 130 acquires the target transmission line ice-covered image and the transmission line non-ice-covered image of the area to be predicted; based on the pre-trained ice thickness prediction model, divides the target transmission line ice-covered image into regions to obtain the transmission line ice-covered sub-images containing the target regions; determines the pre-ice single-eye depth image and the post-ice single-eye depth image corresponding to the transmission line ice-covered sub-images respectively through a preset improved monocular depth estimation algorithm; when the depth dimensions of the pre-ice single-eye depth image and the post-ice single-eye depth image are consistent, performs pixel-level subtraction processing on the pre-ice single-eye depth image and the post-ice single-eye depth image to obtain the pre- and post-ice difference depth map; performs matrix conversion on the pre- and post-ice difference depth map through a preset image matrix conversion algorithm to obtain the ice-covered difference matrix; calculates the pixel-real thickness ratio relationship based on the preset ratio relationship calculation formula and the real line parameters; predicts the ice thickness of the transmission line in the area to be predicted based on the preset real thickness positioning algorithm and the pixel-real thickness ratio relationship to obtain the ice thickness prediction result. After obtaining the ice thickness prediction result, the server 130 sends the ice thickness prediction result to the terminal 110 through the network 120.

[0044] An embodiment of the present invention provides a method for predicting the ice thickness of a transmission line based on monocular depth estimation. Refer to Figure 2 , Figure 2 is a schematic flowchart of a method for predicting the ice thickness of a transmission line based on monocular depth estimation provided by an embodiment of the present invention, and will be described in conjunction with Figure 2 the steps shown.

[0045] Step S210, acquire the target transmission line ice-covered image and the transmission line non-ice-covered image of the area to be predicted.

[0046] In some embodiments, the area to be predicted refers to a specific geographical range selected based on the operation and maintenance requirements of the transmission line, where the ice thickness of the transmission line needs to be predicted. This area can be delimited according to factors such as the distribution of the transmission line, topography, and meteorological conditions. It can be either the area where a specific length of the transmission line and its surrounding environment are located, or a specific area where multiple transmission lines intersect or pass through. For example, in mountainous areas, high-altitude areas, or specific climate zones where ice disasters are likely to occur, the area containing multiple target transmission lines can be delimited as the area to be predicted. This area is the basic spatial range for subsequent data collection, analysis, and model training of ice thickness prediction, and the determination of its boundary directly affects the accuracy and effectiveness of the prediction result.

[0047] In some embodiments, the icing image of the target transmission line is the image data obtained by an image acquisition device when icing occurs on the transmission line, by taking pictures or collecting the target transmission line within a specific area to be predicted. This image contains the intuitive visual information of the icing on the surface of the target transmission line, such as the shape of the ice (e.g., tubular, sheet-like, dendritic, etc.), thickness distribution, coverage range and other characteristics. The image can be a two-dimensional planar image or a three-dimensional image constructed by multi-view acquisition. By processing and analyzing the icing image of the target transmission line, feature information related to the icing thickness can be extracted, providing key data support for icing thickness prediction.

[0048] In some embodiments, the non-icing image of the transmission line is the image data obtained for the same target transmission line using the same or similar image acquisition device and acquisition conditions when no icing phenomenon occurs on the target transmission line. This image records the appearance characteristics of the transmission line in normal operation, including information such as the structural form of the line, surface texture, color, etc. The non-icing image of the transmission line is used as a reference benchmark for comparative analysis with the icing image of the target transmission line. By comparing the differences in the appearance of the transmission line between the two, the icing area can be more accurately identified, the icing boundary can be determined, and it can assist in calculating the icing thickness. At the same time, the non-icing image can also be used to train the image recognition model to help the model learn the characteristics of the transmission line in the normal state, so as to more effectively detect and analyze the icing situation.

[0049] Step S220: Based on the pre-trained icing thickness prediction model, divide the icing image of the target transmission line into regions to obtain an icing sub-image of the transmission line containing the target region.

[0050] In some embodiments, the pre-trained icing thickness prediction model can be a computational model constructed based on a machine learning algorithm; it can also be a computational model constructed based on a deep learning algorithm. Of course, it can also be a computational model constructed based on other data-driven algorithms, and the embodiments of the present invention do not make specific limitations on this.

[0051] In actual model training, by learning and training a large amount of historical data, a mapping relationship between the relevant parameters of the transmission line and the ice thickness is established, so as to realize the prediction of the ice thickness of the transmission line. Before the model is put into actual application, it will use multi-source data such as the ice-covered images, non-ice-covered images, meteorological data (such as temperature, humidity, wind speed, air pressure, etc.), and physical parameters of the transmission line (such as line material, diameter, tension, etc.) collected historically as training samples, and use optimization algorithms such as gradient descent and backpropagation to iteratively adjust the model parameters to minimize the error between the prediction result and the actual ice thickness. After sufficient training, the model can quickly and accurately output the predicted value of the ice thickness of the target transmission line based on the newly input data to be predicted (such as real-time collected image data, meteorological parameters, etc.). The architecture of the model can be a convolutional neural network (CNN), a recurrent neural network (RNN), a long short-term memory network (LSTM), etc., or an ensemble learning model, and its performance and prediction accuracy depend on the quality and quantity of the training data and the degree of model optimization. The embodiments of the present invention do not make specific limitations on this.

[0052] In some embodiments, regional division refers to the process of dividing the area to be predicted into multiple sub-areas according to certain rules and standards according to the requirements of predicting the ice thickness of the transmission line. The division rules can be determined based on factors such as geographical features (such as topography, altitude, longitude and latitude range), meteorological condition distribution (such as temperature gradient, humidity difference, wind direction and wind speed change area), and transmission line topology (such as tower distribution, line orientation, section situation).

[0053] Through regional division, a large-scale and complex area to be predicted can be decomposed into relatively independent sub-areas with similar characteristics, making the subsequent monitoring, analysis and prediction of the ice-covered situation of the transmission line more targeted and refined. The ice-covered situation of the transmission line in each sub-area can adopt the same or similar processing methods, which helps to improve the prediction efficiency and accuracy, and is also convenient for differential management and response to the ice-covered risks in different areas.

[0054] In some embodiments, the ice-covered sub-image of the transmission line refers to the local image data covering the ice-covered information of the transmission line in the target area, which is extracted from the ice-covered image of the target transmission line according to the result of regional division or specific analysis requirements. After the image of the transmission line is collected, since the original image may contain scene information in a large range, in order to analyze the ice-covered condition of a specific target area more accurately, it is necessary to crop and segment the sub-image containing the target area from the original ice-covered image. This sub-image focuses on the transmission line in the target area and clearly presents the details such as the ice-covered form and distribution characteristics on the surface of the line in this area, excluding the interference of other irrelevant areas. For example, when taking the transmission line between a certain section of specific poles and towers as the target area, the sub-image containing only this section of the line and its ice cover is extracted from the original ice-covered image through an image segmentation algorithm. These sub-images can be used as input data for further feature extraction, ice-covered thickness calculation, and input into a pre-trained ice-covered thickness prediction model, providing direct image data support for accurately predicting the ice-covered thickness of the transmission line in the target area.

[0055] Step S230, respectively determine the pre-ice-covered monocular depth image and the post-ice-covered monocular depth image corresponding to the ice-covered sub-image of the transmission line through a preset improved monocular depth estimation algorithm.

[0056] In some embodiments, the preset improved monocular depth estimation algorithm refers to an image depth information estimation method that is optimized and improved for the ice-covered thickness prediction scenario of the transmission line on the basis of the traditional monocular depth estimation algorithm. The preset improved monocular depth estimation algorithm makes targeted improvements to these problems. For example, it introduces prior knowledge about the structure of the transmission line and ice-covered characteristics, optimizes the feature extraction network to better capture depth cues related to ice cover, improves the loss function to enhance the accuracy of depth estimation for the ice-covered area, or adopts a multi-modal data fusion method (such as combining meteorological data, transmission line parameters, etc.) to assist depth estimation.

[0057] In the present invention, the preset improved monocular depth estimation algorithm performs a monocular depth estimation operation on the above dataset, inputs the pre-ice-covered and post-ice-covered images after preprocessing into the loaded Marigold model respectively, and through forward propagation of the model, calculates the depth estimation value of each pixel. Let the input of the model be I (i.e., the ice-covered image of the target transmission line and the non-ice-covered image of the transmission line in the above embodiments), and the output be D (i.e., the pre-ice-covered monocular depth image and the post-ice-covered monocular depth image in the above embodiments). Then the process of forward propagation can be expressed by the formula: ; where f represents the mapping function of the Marigold model.

[0058] In some embodiments, the post-icing monocular depth image refers to the depth image generated by applying a preset improved monocular depth estimation algorithm to the corresponding sub-image of the transmission line after icing. In the icing state, the image acquisition device obtains a sub-image containing the iced transmission line, and the algorithm will consider factors such as the appearance change of the line caused by icing, the shape and thickness of the ice, etc., and perform depth estimation on each pixel point in the image. The post-icing monocular depth image can clearly present the three-dimensional shape of the iced transmission line and the distribution of the ice, reflecting the position change of the transmission line in space after icing and the thickness information of the ice. By comparing with the pre-icing monocular depth image, the undulation change of the line surface caused by icing can be visually observed, and then the distribution of the ice thickness can be accurately calculated, providing key data support for the prediction and evaluation of the ice thickness of the transmission line.

[0059] Step S240, when the depth dimensions of the pre-icing monocular depth image and the post-icing monocular depth image are the same, perform pixel-level subtraction processing on the pre-icing monocular depth image and the post-icing monocular depth image to obtain a depth difference map before and after icing.

[0060] In some embodiments, pixel-level subtraction processing is an operation method used to compare the differences between two images in image analysis. Specifically in the scenario of transmission line icing monitoring, it is to subtract the depth values of the corresponding pixel points in the pre-icing monocular depth image and the post-icing monocular depth image. Since each pixel point represents the depth information at a specific position in the image, through this per-pixel subtraction operation, the depth change at each position before and after icing can be accurately obtained.

[0061] In some embodiments, the depth difference map before and after icing is an image obtained by performing pixel-level subtraction processing on the pre-icing monocular depth image and the post-icing monocular depth image. It visually shows the change of depth information before and after the transmission line is iced. In the difference depth map, the value of each pixel point represents the depth difference at that position before and after icing. If the pixel value in a certain area is positive and large, it means that the depth in this area has increased significantly after icing, that is, the ice thickness is large; on the contrary, if the pixel value is negative or close to zero, it means that the ice thickness in this area is small or there is no obvious icing. By analyzing the difference depth map, the distribution of ice on the transmission line and the difference in ice thickness at different positions can be clearly observed. For example, different depth difference ranges can be represented by color coding, with red indicating areas with a large increase in depth (thicker ice), and blue indicating areas with a small depth change (thinner ice or no ice), so as to visually present the overall distribution trend of ice on the transmission line, providing important visual basis for the evaluation and prediction of the icing condition of the transmission line.

[0062] In the present invention, when performing pixel subtraction processing on the pre-icing monocular depth image and the post-icing monocular depth image, it is necessary to normalize the pre-icing and post-icing monocular depth images at the same angle at different times so that the sizes and formats of these images are consistent.

[0063] Step S250: Through a preset image matrix conversion algorithm, perform matrix conversion on the pre- and post-icing differential depth map to obtain an icing difference matrix.

[0064] In some embodiments, the preset image matrix conversion algorithm is an algorithm for converting image data into a specific matrix form. In the context of transmission line icing monitoring, this algorithm operates on the pre- and post-icing differential depth map. Its purpose is to convert the depth difference information represented in the form of pixel points in the differential depth map into a matrix form that is convenient for computer processing and analysis.

[0065] In some embodiments, the icing difference matrix is the result obtained by performing matrix conversion on the pre- and post-icing differential depth map through a preset image matrix conversion algorithm. It is a two-dimensional matrix, and each element in it corresponds to the converted value of the depth difference information at a specific position in the pre- and post-icing differential depth map.

[0066] In some embodiments, the rows and columns of the icing difference matrix can respectively represent the positions or area divisions of the transmission line in different directions, and the values of the matrix elements represent the icing difference degrees at the corresponding positions. For example, a certain row of the matrix may correspond to a certain horizontal position of a section of the transmission line, and a certain column may correspond to different height intervals in the vertical direction. Then the element value at the intersection of this row and this column in the matrix represents the depth difference situation before and after icing at this specific position. Through the icing difference matrix, various mathematical operations and analyses can be conveniently carried out, such as calculating the average thickness of the ice coating, determining the positions where the maximum and minimum values of the ice coating thickness are located, and analyzing the distribution trend of the ice coating on the transmission line.

[0067] Step S260: Based on a preset proportional relationship calculation formula, calculate the icing difference matrix and the actual line parameters to obtain a pixel-actual thickness proportional relationship.

[0068] Among them, the preset proportional relationship calculation formula is: ; in the formula, a represents the pixel-actual thickness proportional relationship; R represents the actual radius distance of the line; represents the depth value at the back end of the line center point; represents the depth value at the front end of the line center point.

[0069] Step S270: Based on a preset actual thickness positioning algorithm and the pixel-actual thickness proportional relationship, predict the ice coating thickness of the transmission line in the to-be-predicted area to obtain an ice coating thickness prediction result.

[0070] In some embodiments, the preset true thickness positioning algorithm is an algorithm for determining the position and corresponding relationship of the true thickness of ice coating on a transmission line in an image.

[0071] In some embodiments, the pixel-true thickness proportional relationship refers to the corresponding proportional relationship established between the value represented by a pixel point in an image and the actual true thickness of ice coating in the analysis of an ice-coated transmission line image.

[0072] The present invention constructs a method for monitoring the ice coating thickness on the surface of a transmission line based on monocular depth estimation, realizing the accurate measurement of the ice coating thickness on the surface of the transmission line, and providing a research basis and algorithm tool for calculating the ice coating weight and evaluating the severity of the transmission line; the present invention proposes an improved diffusion monocular depth estimation model, adding a spatial attention mechanism on the basis of the monocular depth estimation algorithm, enabling the model to focus on the area of information to be detected in the picture, helping the model to more clearly capture the position and thickness information of the ice coating on the surface of the transmission line, so as to more accurately estimate the depth value; in addition, adding the spatial attention mechanism can also play a role in suppressing noise, by weighting different channels in the image, enhancing the weight of the channel dominated by useful information and reducing the weight of the channel dominated by noise information, enabling it to have the ability to accurately identify noise and non-noise, thereby improving the quality of the original image, providing a more reliable input for the depth estimation model, and comprehensively improving the detection effect of the depth estimation algorithm in the night scene; the present invention proposes an image-matrix conversion model, which can extract the depth value represented by each pixel value in the depth map and convert it into a two-dimensional depth numerical matrix for subsequent numerical calculations.

[0073] In some embodiments, the above step S210 can be implemented through the following steps S211 to S214:

[0074] Step S211, using an image acquisition device to collect the initial ice-coated transmission line image of the area to be predicted in real time.

[0075] In some embodiments, the image acquisition device can be a device with an image information acquisition function such as a high-definition camera, an imaging device carried by a drone, a satellite remote sensing device, etc.

[0076] Step S212, performing noise reduction processing on the initial ice-coated transmission line image to obtain a noise-reduced ice-coated transmission line image.

[0077] In some embodiments, the purpose of noise reduction processing is to remove or reduce the noise in the image, restore the true information of the image, improve the quality of the image, and provide a more reliable data basis for subsequent operations such as monocular depth estimation and ice coating thickness calculation. In the embodiments of the present invention, the noise reduction methods include but are not limited to mean filtering, median filtering, Gaussian filtering, and wavelet noise reduction.

[0078] Step S213: Align the denoised ice-covered transmission line image to obtain an aligned ice-covered transmission line image.

[0079] In some embodiments, the alignment process is to eliminate the differences caused by factors such as shooting angle, position, and lighting conditions, adjust the images before and after icing to have the same coordinate system or spatial scale, so that the corresponding points in the images can be accurately matched, providing a data basis for subsequent analysis of the ice thickness.

[0080] Step S214: Grayscale the aligned ice-covered transmission line image to obtain the target ice-covered transmission line image.

[0081] In some embodiments, after grayscaling the image, the grayscaling process can eliminate the uncertainty brought by color information, making the analysis of the image focus more on features related to icing such as brightness and texture, and improving the accuracy of ice detection and thickness calculation.

[0082] In some embodiments, before performing the above step S220, the method may include being implemented through steps S111 to S115:

[0083] Step S111: Obtain a sample ice-covered transmission line image.

[0084] Step S112: Label the ice-covered target area in the sample ice-covered transmission line image to obtain a labeled sample ice-covered transmission line image.

[0085] Step S113: Set the training learning rate, batch size, and number of iterations, and train the initial YOLOv10 object detection model to obtain a trained YOLOv10 object detection model.

[0086] Step S114: Based on the trained YOLOv10 object detection model, predict the sample ice-covered transmission line image to obtain a sample ice thickness prediction result.

[0087] Step S115: Calculate the error based on the sample ice thickness prediction result and the labeled sample ice-covered transmission line image to obtain the optimal weight.

[0088] In some embodiments, the above step S220 can be implemented through steps S221 to S222:

[0089] Step S221: Based on the optimal weight determined by the pre-trained ice thickness prediction model, perform boundary detection on the target ice-covered transmission line image to obtain predicted bounding box information.

[0090] Step S222: Based on the predicted bounding box information, crop the target transmission line icing image to obtain the transmission line icing sub-image.

[0091] In some embodiments, the above step S230 can be implemented through steps S231 to S233:

[0092] Step S231: Normalize the images of the same angle at different times in the transmission line icing sub-image and the transmission line non-icing image respectively, and correspondingly obtain the processed transmission line icing sub-image and the processed transmission line non-icing image.

[0093] Step S232: Based on the preset improved monocular depth estimation algorithm, perform monocular depth estimation on the processed transmission line icing sub-image and the processed transmission line non-icing image respectively, and correspondingly obtain the first depth estimation value of each pixel in the transmission line icing sub-image and the second depth estimation value of each pixel in the transmission line non-icing image.

[0094] Step S233: Convert the formats of the first depth estimation value and the second depth estimation value to obtain the pre-icing monocular depth image and the post-icing monocular depth image respectively.

[0095] In some embodiments, the above step S240 can be implemented through the following content:

[0096] Subtract the pixel value of each coordinate in the post-icing monocular depth image from the pixel value of each coordinate in the pre-icing monocular depth image to obtain the pixel value at the corresponding coordinate ; The calculation formula is as follows: ; In the formula, represents the coordinate at the i-th row and the j-th column.

[0097] Next, an exemplary application of the embodiments of the present invention in an actual application scenario will be described.

[0098] The present invention proposes an intelligent measurement method for the icing thickness of transmission lines based on monocular depth estimation and true thickness positioning. The key to this invention lies in constructing a method that is applicable to the calculation of the icing thickness on both sides of the line and can also identify the icing thickness on the surface of the transmission line. First, this method captures the icing images on the line surface through image acquisition devices such as drones and fixed cameras. Then, it uses an object detection algorithm (YOLOv10) to identify and crop the icing area of the line in the original image. Then, it performs monocular depth estimation operations on the cropped image through an improved monocular depth estimation algorithm (Marigold) to obtain the monocular depth images before and after icing respectively. Then, it performs pixel-level subtraction on the two to obtain the differential depth map before and after icing, and combines it with the image-matrix conversion module to obtain the icing difference matrix. Finally, through the above-obtained matrix and the true line parameters, the pixel-actual thickness ratio relationship a can be obtained through relevant formulas, and combined with the true thickness positioning algorithm, the icing thickness result on the surface of the transmission line in the image can be obtained.

[0099] Figure 3 It is a schematic flow diagram of an intelligent measurement method for the icing thickness of transmission lines based on monocular depth estimation and true thickness positioning. First, through an object detection algorithm, preprocessing operations such as region recognition and cropping are performed on the input original line icing image. Then, the processed image is evaluated through a discriminant function. Then, after being processed by a monocular depth estimation algorithm, the depth map of the target image is obtained. Then, pixel-level subtraction is performed on it. Combining with the image-matrix conversion module, the icing difference matrix before and after icing can be obtained. Finally, combined with the true thickness positioning algorithm, the icing thickness value can be obtained, realizing the intelligent measurement of the icing thickness on the surface of the transmission line.

[0100] The present invention proposes an intelligent measurement method for the icing thickness of transmission lines based on monocular depth estimation and true thickness positioning. See Figure 3 , and the specific principle steps are as follows:

[0101] Step 1: Data collection and preprocessing: Collect the icing data of the transmission line obtained by the image acquisition device or drone inspection, and perform preprocessing such as noise reduction, data alignment, and image grayscale conversion on the above data.

[0102] Step 2: Target Region Identification and Cropping: First, annotate the collected images to mark the target regions of line icing in the images. Then, train the single-stage object detection algorithm YOLOv10 and set parameters such as the learning rate, batch size, and number of training epochs for the training. Next, use the prepared dataset to train the set YOLOv10 model. During the training process, the model will continuously adjust its own parameters to minimize the error between the prediction results and the annotation results, and finally output an optimal weight. Then, the model will use this weight to perform object detection on the input image, output the predicted bounding box information, and finally, according to the predicted bounding box information, crop the original image, delete the background interference part in the image, and intelligently select the target region from the candidate regions to obtain a sub-image containing the target region.

[0103] Among them, the discriminant function is expressed as in formula (1):

[0104] (1);

[0105] In the formula, Clarity represents the clarity of the target; Contrast represents the contrast of the target; Distinction represents the distinguishability between the target and the background; α, β, and γ represent weight coefficients used to adjust the importance of each factor.

[0106] Step 3: Obtaining the Depth Map: First, collect the image dataset after cropping in Step 2. This dataset contains pre-icing and post-icing line images at the same angle and different times, and perform normalization processing on this dataset to make the sizes and formats of these images consistent. Then, use the improved monocular depth estimation algorithm (Marigold) to perform monocular depth estimation operations on the above dataset. Input the pre-icing and post-icing images after preprocessing into the loaded Marigold model respectively, and through the forward propagation of the model, calculate the depth estimation value of each pixel. Let the input of the model be I and the output be D, then the forward propagation process can be expressed by formula (2):

[0107] (2);

[0108] Among them, f represents the mapping function of the Marigold model.

[0109] Convert the calculated depth values into image formats to obtain the pre-icing and post-icing monocular depth images respectively.

[0110] Step 4: Obtaining the Differential Depth Map: First, perform size scaling processing on the monocular depth images obtained in Step 3 to ensure that the sizes of the two depth maps are the same, that is, the width W and height H are the same. Then, for the pre-icing monocular depth map and the post-icing monocular depth map Perform pixel-by-pixel subtraction operation to obtain the differential depth map The pixel value at this coordinate Can be calculated by formula (3):

[0111] (3);

[0112] In the formula, Represents the coordinates at the i-th row and j-th column;

[0113] Step 5: Image-matrix conversion: The differential depth map before and after icing obtained in Step 4 is processed by the image-matrix conversion module to obtain the icing difference matrix. Among them, the converted two-dimensional depth matrix D, the matrix mathematical representation is as formula (4):

[0114] (4);

[0115] Among them, d(x,y) represents the depth value at the image coordinate (x,y).

[0116] Step 6: True thickness calculation: Through the above-obtained matrix and the true line parameters, the pixel-actual thickness ratio relationship a can be obtained by formula (5), and then combined with the true thickness positioning algorithm, the icing thickness result on the surface of the transmission line in the image can be obtained. Among them, the true thickness positioning algorithm is represented as shown in formula (6):

[0117] (5);

[0118] Among them, a represents the pixel-true thickness ratio relationship; R represents the true radius distance of the line; Represents the depth value at the back end of the line center point; Represents the depth value at the front end of the line center point.

[0119] (6);

[0120] Among them, T represents the icing thickness at any point on the surface of the transmission line; a represents the pixel-true thickness ratio relationship; ΔD represents the differential depth value represented by this point in the difference matrix.

[0121] Figure 4 Is the schematic diagram of the composition structure of the transmission line icing thickness prediction device provided by the embodiment of the present invention, as Figure 4As shown in the figure, the transmission line icing thickness prediction device 400 based on monocular depth estimation includes: an acquisition module 401, configured to acquire a target transmission line icing image and a transmission line non-icing image of the area to be predicted; a division module 402, configured to perform area division on the target transmission line icing image based on a pre-trained icing thickness prediction model to obtain a transmission line icing sub-image including a target area; a determination module 403, configured to respectively determine a pre-icing monocular depth image and a post-icing monocular depth image corresponding to the transmission line icing sub-image through a preset improved monocular depth estimation algorithm; a processing module 404, configured to perform pixel-level subtraction processing on the pre-icing monocular depth image and the post-icing monocular depth image when the depth dimensions of the pre-icing monocular depth image and the post-icing monocular depth image are consistent to obtain a pre- and post-icing difference depth map; a conversion module 405, configured to perform matrix conversion on the pre- and post-icing difference depth map through a preset image matrix conversion algorithm to obtain an icing difference matrix; a calculation module 406, configured to calculate the icing difference matrix and real line parameters based on a preset proportional relationship calculation formula to obtain a pixel-real thickness proportional relationship; the preset proportional relationship calculation formula is: ; where a represents the pixel-real thickness proportional relationship; R represents the real radius distance of the line; represents the depth value at the back end of the line center point; represents the depth value at the front end of the line center point; a prediction module 407, configured to predict the icing thickness of the transmission line in the area to be predicted based on a preset real thickness positioning algorithm and the pixel-real thickness proportional relationship to obtain an icing thickness prediction result.

[0122] It should be noted that the description of the device in the embodiments of the present invention is similar to the description of the above method embodiments and has similar beneficial effects to the method embodiments, so details will not be repeated. For the technical details not disclosed in the embodiments of this device, please refer to the description of the method embodiments of the present invention for understanding.

[0123] It should be noted that in the embodiments of the present invention, if the above-mentioned method for predicting the icing thickness of a transmission line based on monocular depth estimation is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention, in essence, or the part that contributes to the related art can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a terminal to execute all or part of the methods described in the embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), magnetic disks, or optical discs that can store program codes. In this way, the embodiments of the present invention are not limited to any specific combination of hardware and software.

[0124] Correspondingly, the embodiments of the present invention provide an electronic device. Figure 5 It is a schematic structural diagram of the electronic device provided by the embodiments of the present invention. As Figure 5 shown, the electronic device 500 at least includes: a processor 501 and a computer-readable storage medium 502 configured to store executable instructions, where the processor 501 generally controls the overall operation of the electronic device 500. The computer-readable storage medium 502 is configured to store instructions and applications executable by the processor 501, and can also cache data to be processed or already processed by the processor 501 and each module in the electronic device 500, and can be implemented by flash memory (FLASH) or random access memory (Random Access Memory, RAM).

[0125] The embodiments of the present invention provide a storage medium storing executable instructions, where the executable instructions, when executed by a processor, will cause the processor to execute the methods provided by the embodiments of the present invention. For example, as Figure 2 shown in the method.

[0126] In some embodiments, the storage medium may be a computer-readable storage medium. For example, it can be a ferroelectric memory (FRAM, Ferromagnetic Random Access Memory), read-only memory (ROM, Read Only Memory), programmable read-only memory (PROM, Programmable Read Only Memory), erasable programmable read-only memory (EPROM, Erasable Programmable Read Only Memory), electrically erasable programmable read-only memory (EEPROM, Electrically Erasable Programmable Read Only Memory), flash memory, magnetic surface memory, optical disc, or a compact disk-read only memory (CD-ROM, Compact Disk-Read Only Memory), etc.; it can also be various devices including one or any combination of the above memories.

[0127] In some embodiments, the executable instructions may be in the form of a program, software, software module, script, or code, and can be written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and can be deployed in any form, including being deployed as an independent program or being deployed as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0128] As an example, the executable instructions may or may not correspond to a file in the file system, and may be stored as part of a file that stores other programs or data. For example, they can be stored in one or more scripts in a hypertext markup language (HTML, Hyper Text Markup Language) document, stored in a single file dedicated to the program under discussion, or stored in multiple cooperating files (for example, files that store one or more modules, subroutines, or code portions). As an example, the executable instructions can be deployed to execute on one electronic device, or on multiple electronic devices located at one location, or on multiple electronic devices distributed at multiple locations and interconnected through a communication network.

[0129] As described above, the above are only embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and scope of the present invention are all included in the protection scope of the present invention.

[0130] It should be understood that the "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present invention. Therefore, the "in one embodiment" or "in an embodiment" that appears throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present invention, the magnitude of the sequence numbers of the above processes does not mean the order of execution, and the order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention. The serial numbers of the embodiments of the present invention above are only for description and do not represent the advantages or disadvantages of the embodiments.

[0131] It should be noted that in this text, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such a process, method or device. Without more limitations, the element defined by the statement "including a..." does not exclude the existence of another identical element in the process, method, article or device including the element. In several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed.

[0132] As described above, it is only the implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for predicting ice thickness of power transmission lines based on monocular depth estimation, characterized in that: The method comprises: Acquire an ice-covered image of a target transmission line and an ice-free image of a transmission line in a region to be predicted; Based on a pre-trained ice thickness prediction model, the target transmission line ice image is divided into regions to obtain a transmission line ice sub-image containing the target region; By presetting an improved monocular depth estimation algorithm, respectively determining a monocular depth image before and after ice-covering corresponding to the ice-covered sub-image of the transmission line; In a case where the depth sizes of the monocular depth image before ice coating and the monocular depth image after ice coating are consistent, performing pixel-level subtraction processing on the monocular depth image before ice coating and the monocular depth image after ice coating to obtain a difference depth map before and after ice coating; Performing matrix conversion on the difference depth map before and after ice coating by using a preset image matrix conversion algorithm to obtain an ice coating difference matrix; Based on a preset proportional relationship calculation formula, the ice cover difference matrix and the real line parameters are calculated to obtain a pixel-real thickness proportional relationship; the preset proportional relationship calculation formula is: ; In the formula, a represents the ratio of pixel to real thickness; R represents the real radius distance of the line; Indicates the depth value of the rear end of the line center point; Indicates the depth value at the front end of the center point of the line; Based on a preset true thickness positioning algorithm and the pixel-true thickness ratio relationship, ice thickness prediction is performed on the power transmission lines in the area to be predicted to obtain an ice thickness prediction result.

2. The method according to claim 1, characterized in that The step of obtaining an ice-covered image of a target transmission line in a region to be predicted includes: The initial transmission line ice coverage image of the area to be predicted is collected in real time by an image collection device; Performing noise reduction processing on the initial transmission line ice-covered image to obtain a noise-reduced transmission line ice-covered image; Performing alignment processing on the denoised transmission line ice-covered image to obtain an aligned transmission line ice-covered image; Grayscale processing is performed on the aligned transmission line ice-covered image to obtain the target transmission line ice-covered image.

3. The method according to claim 1, characterized in that Before dividing the target transmission line ice-covered image into regions based on the pre-trained ice-covered thickness prediction model to obtain a sub-image containing the target region, the method further includes: Acquire ice-covered images of sample power transmission lines; Annotating an ice-covered target area in the sample transmission line ice-covered image to obtain an annotated sample transmission line ice-covered image; Set the training learning rate, batch size, and number of iterations to train the initial YOLOv10 target detection model to obtain the trained YOLOv10 target detection model; Based on the trained YOLOv10 target detection model, the sample transmission line ice-covered image is predicted to obtain a sample ice-covered thickness prediction result; An error calculation is performed based on the sample ice thickness prediction result and the labeled sample transmission line ice image to obtain an optimal weight.

4. The method according to claim 3, characterized in that The ice thickness prediction model based on the pre-trained method is used to divide the target transmission line ice image into regions to obtain a sub-image containing the target region, including: Based on the optimal weight determined by the pre-trained ice thickness prediction model, boundary detection is performed on the ice-covered image of the target transmission line to obtain prediction boundary box information; Based on the predicted bounding box information, the target power transmission line ice-covered image is region cropped to obtain the power transmission line ice-covered sub-image.

5. The method according to claim 1, characterized in that The method of respectively determining the monocular depth image before and after ice-covering corresponding to the ice-covered sub-image of the power transmission line by presetting an improved monocular depth estimation algorithm comprises: Normalizing the images of the iced transmission line sub-image and the un-iced transmission line image at the same angle at different times respectively, and obtaining a processed iced transmission line sub-image and a processed un-iced transmission line image respectively; Based on the preset improved monocular depth estimation algorithm, monocular depth estimation is performed on the processed iced transmission line sub-image and the processed un-iced transmission line image, respectively, to obtain a first depth estimation value for each pixel in the iced transmission line sub-image and a second depth estimation value for each pixel in the un-iced transmission line image; The first depth estimation value and the second depth estimation value are format-converted to obtain the monocular depth image before ice coverage and the monocular depth image after ice coverage, respectively.

6. The method according to claim 1, characterized in that The performing pixel-level subtraction processing on the monocular depth image before ice covering and the monocular depth image after ice covering to obtain a difference depth map before and after ice covering includes: The pixel value of each coordinate in the monocular depth image after ice coverage Subtract the pixel value of each coordinate in the monocular depth image before ice cover , get the pixel value at the corresponding coordinate ; The calculation formula is as follows: ; In the formula, Represents the coordinates at the i-th row and j-th column.

7. A transmission line ice thickness prediction device based on monocular depth estimation, characterized in that: The device comprises: An acquisition module, used for acquiring an ice-covered image of a target transmission line and an ice-free image of a transmission line in a region to be predicted; A partitioning module is used to partition the target transmission line ice-covered image into regions based on a pre-trained ice-covered thickness prediction model, so as to obtain a transmission line ice-covered sub-image containing the target region; A determination module, used to determine the monocular depth image before and after icing corresponding to the icing sub-image of the transmission line respectively by using a preset improved monocular depth estimation algorithm; A processing module, configured to perform pixel-level subtraction processing on the monocular depth image before ice-covering and the monocular depth image after ice-covering, so as to obtain a difference depth map before and after ice-covering, when the depth sizes of the monocular depth image before ice-covering and the monocular depth image after ice-covering are consistent; A conversion module, used to perform matrix conversion on the difference depth image before and after ice coating by using a preset image matrix conversion algorithm to obtain an ice coating difference matrix; A calculation module is used to calculate the ice cover difference matrix and the real line parameters based on a preset proportional relationship calculation formula to obtain a pixel-real thickness proportional relationship; the preset proportional relationship calculation formula is: ; In the formula, a represents the ratio of pixel to real thickness; R represents the real radius distance of the line; Indicates the depth value of the rear end of the line center point; Indicates the depth value at the front end of the center point of the line; The prediction module is used to predict the ice thickness of the power transmission lines in the area to be predicted based on a preset true thickness positioning algorithm and the pixel-true thickness ratio relationship to obtain an ice thickness prediction result.

8. An electronic device, characterized in that: include: A memory for storing executable instructions; A processor, for implementing the method for predicting ice thickness of transmission lines based on monocular depth estimation as described in any one of claims 1 to 6 when executing executable instructions stored in the memory.

9. A computer-readable storage medium storing executable instructions for causing a processor to execute the executable instructions to implement the method for predicting ice thickness of transmission lines based on monocular depth estimation as described in any one of claims 1 to 6.

Images