An unmanned aerial vehicle tilt photography data-driven mountainous area power transmission line risk detection method, device and medium

Abstract

The present application relates to the technical field of image recognition, and particularly relates to a risk detection method, equipment and medium for a mountainous power transmission line driven by unmanned aerial vehicle tilt photography data. The method comprises the following steps: acquiring image data collected by a tilt photography camera and generating a dense point cloud model; constructing a current state digital ground surface model and a real scene three-dimensional texture model based on the dense point cloud model, and extracting a tower foundation area by using a spatial semantic segmentation method combined with image recognition; performing change detection on the current digital ground surface model and a historical digital ground surface model, and performing auxiliary observation on a deformation area by combining the current and historical real scene three-dimensional texture models to determine a suspected landslide deformation area; and performing risk level evaluation on the suspected landslide deformation area in combination with the tower foundation area. The method starts from four aspects of data acquisition, three-dimensional modeling, time series analysis and intelligent recognition, overcomes the limitations of traditional methods in engineering applicability, processing precision and dynamic response, and provides support for power transmission line disaster monitoring.

Description

A method, equipment, and medium for risk detection of power transmission lines in mountainous areas driven by UAV oblique photogrammetry data. Technical Field

[0001] This invention relates to the field of image recognition technology, and in particular to a method, equipment and medium for risk detection of power transmission lines in mountainous areas driven by UAV oblique photography data. Background Technology

[0002] In current power transmission line operation and maintenance practices, the foundations of power poles in mountainous areas are highly susceptible to geological hazards such as landslides, collapses, and debris flows due to the complex geological environment. These hazards are characterized by their high degree of concealment, suddenness, and complex chain of events. Traditional risk identification and monitoring methods are insufficient to meet the current needs for efficient, accurate, and wide-area coverage. Existing technologies mainly rely on manual inspections, fixed-point monitoring instruments, or low-frequency remote sensing image analysis. These methods generally suffer from high costs, long cycles, delayed data updates, and insufficient spatial resolution. They are unable to reflect the development trend of landslide hazards in real time, nor can they achieve fine-grained dynamic risk monitoring of power pole foundations in mountainous power transmission line corridors.

[0003] In recent years, the development of oblique photogrammetry technology has provided new solutions for 3D modeling and terrain reconstruction. Related research and engineering practices have shown that oblique photogrammetry can quickly acquire high-precision surface images and, combined with 3D reconstruction algorithms, achieve structured modeling of target areas. However, existing technologies are mostly focused on general scenarios such as urban building modeling and disaster reconstruction of exposed mountain slopes, failing to address the specific object of in-depth modeling and intelligent identification of power pole foundations in complex mountainous environments. While some existing patents have achieved 3D visualization of landslide areas, they still have significant shortcomings in local deformation analysis, risk factor extraction, and deformation evolution path identification for power pole foundations, lacking the ability to systematically model and judge the interaction mechanism between power facilities and the surrounding geological environment.

[0004] Furthermore, some technical solutions attempt to combine oblique photography results with digital elevation models (DEMs) or digital surface models (DSMs) for preliminary landslide boundary identification and volume estimation. However, these methods mostly rely on static, single-temporal data, making it impossible to achieve dynamic monitoring of landslide hazards over long time series. In mountainous environments, landslide induction mechanisms are often influenced by rainfall, surface water activity, and geological structures, exhibiting characteristics of continuous evolution and periodic accumulation. Single-temporal modeling cannot reflect the complete process of landslides from incubation and development to instability, and it is also difficult to meet the timeliness requirements of power operation and maintenance for risk prediction and early warning.

[0005] In terms of data processing efficiency, the current common oblique photogrammetry process is mainly based on manual and semi-automatic modeling. It involves a large amount of data and a complex processing flow. It lacks structured risk indicator extraction tools and automated analysis processes for power scenarios, making it difficult to efficiently connect with power grid operation and maintenance platforms. This further limits its application and promotion in the safety supervision of large-scale power facilities.

[0006] Therefore, there is an urgent need for a technical solution that can take into account high-precision modeling, multi-temporal analysis, target structure identification, and dynamic risk assessment capabilities. Summary of the Invention

[0007] In view of the aforementioned existing problems, the present invention is proposed.

[0008] Therefore, this invention provides a method, equipment, and medium for risk detection of power transmission lines in mountainous areas driven by UAV oblique photography data, which solves the problems of the inability to effectively integrate multi-source data, the lack of a mechanism for three-dimensional structural modeling and accurate extraction of landslide elements applicable to tower foundation areas, and the inability to dynamically monitor and update changes.

[0009] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0010] In a first aspect, the present invention provides a method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photography data, comprising:

[0011] Acquire image data from an oblique photogrammetry camera;

[0012] The image data is used to generate a dense point cloud model through the structured beam method and a multi-view stereo matching algorithm.

[0013] Construct a digital surface model and a realistic 3D texture model of the current state based on a dense point cloud model;

[0014] Based on the current state of the digital surface model and the real-scene 3D texture model, the pole foundation area is extracted using spatial semantic segmentation combined with image recognition.

[0015] Change detection is performed on the current digital surface model and the historical digital surface model. The deformation area is then observed by combining the current and historical real-world 3D texture models to identify suspected landslide deformation areas.

[0016] By combining the tower foundation area, the suspected landslide deformation area is classified and assessed to obtain the risk level of the suspected landslide deformation area, and then visualized in conjunction with the transmission line topology map to realize risk labeling and early warning.

[0017] As a preferred embodiment of the UAV oblique photography data-driven risk detection method for power transmission lines in mountainous areas according to the present invention, the image data is used to generate a dense point cloud model through structured bundle method and multi-view stereo matching algorithm, including:

[0018] The structured beam method is used to extract and match feature points from image data and solve camera parameters to generate sparse point clouds and restore the spatial geometric relationship of image data.

[0019] Based on sparse point clouds and spatial geometric relationships, a multi-view stereo matching algorithm is introduced to densify sparse point clouds, resulting in a dense point cloud model with spatial resolution.

[0020] As a preferred embodiment of the UAV oblique photogrammetry data-driven risk detection method for power transmission lines in mountainous areas described in this invention, the method includes: constructing a digital surface model and a realistic 3D texture model based on a dense point cloud model, comprising:

[0021] The dense point cloud model is regularized into a grid, and elevation information is extracted to form a complete surface elevation dataset, generating a digital surface model of the current state.

[0022] A realistic 3D texture model is generated by overlaying the texture information of the original image onto the geometric skeleton of the point cloud in the dense point cloud model.

[0023] As a preferred embodiment of the UAV oblique photography data-driven risk detection method for power transmission lines in mountainous areas described in this invention, the method includes: extracting the tower foundation area based on a digital surface model and a real-world 3D texture model in the current state, using spatial semantic segmentation combined with image recognition, including:

[0024] The tower area was initially determined through spatial semantic segmentation and represented as follows:

[0025] ,

[0026] In the formula, Indicates a condition; Indicates the condition otherwise; For position Elevation value; The average elevation of the target area; For position Normalized Difference Vegetation Index (NDVI) value; The average NDVI value of the target area; To constrain the preset threshold of elevation, A preset threshold to constrain the range of NDVI differences; The binarized result of the recognition; Indicates the "and" condition;

[0027] Based on the preliminary determination of the tower area, a tower target extraction mechanism is formed by combining geometric shape, reflection characteristics, height threshold and image recognition algorithm to extract the tower base area.

[0028] As a preferred embodiment of the UAV oblique photogrammetry data-driven risk detection method for power transmission lines in mountainous areas described in this invention, the method includes: detecting changes between the current digital terrain model and historical digital terrain models, including:

[0029] A surface deformation detection algorithm based on point cloud difference is used to detect changes between the current digital land model and historical digital land models, as shown below:

[0030] ,

[0031] In the formula, Indicates at coordinate point The elevation change between the two digital surface models is shown in meters. Indicates time Time, coordinates The surface elevation value at that location; Indicates time Time, coordinates The surface elevation value.

[0032] As a preferred embodiment of the UAV oblique photography data-driven risk detection method for power transmission lines in mountainous areas described in this invention, the method includes: combining current and historical real-world 3D texture models to assist in the observation of deformation areas and determine suspected landslide deformation areas, including:

[0033] Visual observation of areas with elevation changes is achieved by using current and historical real-world 3D texture models to help determine whether changes exist in the corresponding areas;

[0034] If the auxiliary interpretation shows changes, and the absolute value of the elevation change value exceeds the first preset threshold within a continuous time period, while the slope change rate is greater than the second preset threshold, then the corresponding area is determined to be a suspected landslide deformation area.

[0035] As a preferred embodiment of the UAV oblique photography data-driven risk detection method for power transmission lines in mountainous areas described in this invention, the method involves: combining the tower foundation area to conduct a graded assessment of suspected landslide deformation areas, thereby obtaining the risk level of the suspected landslide deformation areas, including:

[0036] Based on the geological, slope, soil moisture, and relative location indicators of the suspected landslide deformation area, a risk assessment calculation and risk score are performed. Specifically, it is expressed as follows:

[0037] ,

[0038] In the formula, The slope value represents the degree of inclination of the ground surface in the suspected landslide deformation area; This represents the change in surface elevation between two different periods; The horizontal distance between the suspected landslide deformation area and the foundation area of ​​the nearest power transmission tower; Soil moisture content; , , and These are the weighting coefficients for the corresponding indicators.

[0039] As a preferred embodiment of the UAV oblique photogrammetry data-driven risk detection method for power transmission lines in mountainous areas according to the present invention, wherein: acquiring the image data collected by the oblique photogrammetry camera includes:

[0040] The measurement camera's flight altitude is set between 80 and 120 meters, with a ground resolution greater than 5 cm / pixel.

[0041] The horizontal overlap rate of the images should be set to no less than 70%, and the vertical overlap rate should be set to no less than 80%.

[0042] The flight tilt angle is set between 35° and 45°.

[0043] In a second aspect, the present invention provides an electronic device, comprising:

[0044] Memory and processor;

[0045] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of a method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photography data.

[0046] Thirdly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the method for risk detection of mountain power transmission lines driven by UAV oblique photography data.

[0047] Compared with existing technologies, the beneficial effects of this invention are as follows: Utilizing a drone platform equipped with multi-lens oblique photography enables comprehensive, high-density, and rapid data acquisition of power transmission lines in mountainous areas; Through structured beamforming and multi-view stereo reconstruction techniques, a high-precision digital surface model (DSM) and dense point cloud model are constructed, achieving refined reconstruction of the tower foundations and surrounding landforms; semantic rules and intelligent image algorithms are combined to automatically extract and segment tower targets, providing spatial semantic support for landslide hazard identification; through periodic re-flight and temporal oblique image comparison, a multi-temporal 3D model is established; DSM differential and slope change are used to jointly identify landslide evolution characteristics, enabling dynamic modeling and trend early warning of the entire landslide process from incubation to disaster; the oblique photography modeling results are fused with multi-source remote sensing factors to quantify the level of landslide hazards and improve the accuracy of risk assessment. Attached Figure Description

[0048] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 is a schematic diagram of the overall process of a method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photography data according to an embodiment of the present invention. Detailed Implementation

[0050] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0051] Example 1, referring to Figure 1, is an embodiment of the present invention, providing a method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photogrammetry data, including:

[0052] S100: Acquire image data from the oblique photogrammetry camera;

[0053] S200: The image data is used to generate a dense point cloud model through the structured beam method and a multi-view stereo matching algorithm;

[0054] S300: Constructs a digital surface model and a realistic 3D texture model of the current state based on a dense point cloud model;

[0055] S400: Based on the current state of the digital surface model and the real-world 3D texture model, the pole foundation area is extracted using spatial semantic segmentation combined with image recognition.

[0056] S500: Detects changes between the current digital surface model and the historical digital surface model, and combines the current and historical real-world 3D texture models to conduct auxiliary observation of the deformation area and identify suspected landslide deformation areas;

[0057] S600: Combined with the tower foundation area, the suspected landslide deformation area is classified and assessed to obtain the risk level of the suspected landslide deformation area, and visualized in conjunction with the transmission line topology map to realize risk labeling and early warning.

[0058] It should be noted that existing manual inspection or fixed-point deployment methods are subject to many limitations in mountainous areas, resulting in low data acquisition efficiency and difficulty in achieving high-precision, high-coverage data acquisition in complex terrain environments. At the same time, most solutions only target landslide identification on exposed mountain slopes, ignoring the structural characteristics and disaster sensitivity of engineering targets such as power transmission towers in three-dimensional space. Furthermore, most solutions are based on single-temporal images, lacking the ability to characterize the temporal evolution process of landslides, and rely heavily on single images or elevation data, making it difficult to comprehensively assess the intensity of regional geological activity.

[0059] To address the aforementioned main issues, steps S100-S600 utilize a drone equipped with an oblique photography camera to achieve full coverage and low-altitude, high-resolution image acquisition along the mountainous power transmission lines, overcoming the limitations of traditional methods in terms of acquisition means. Combining 3D reconstruction algorithms, spatial topology analysis, and fine segmentation of the target area, structured modeling of the tower foundation area and accurate identification of landslide hazard areas are achieved. Multi-temporal oblique photography data overlay analysis and deformation feature extraction are introduced to achieve dynamic updates and trend predictions of landslide risk. Overall, by integrating multiple remote sensing features such as oblique imagery, DSM, slope field, and Normalized Difference Vegetation Index (NDVI), multi-dimensional coupling of landslide disaster factors and intelligent risk classification are achieved, improving the accuracy and robustness of detection.

[0060] Example 2, referring to Figure 1, is an embodiment of the present invention. Based on the above embodiment, a method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photography data is provided.

[0061] In this embodiment of the invention, step S100 involves acquiring image data collected by an oblique photogrammetry camera.

[0062] For example, multiple fixed-wing or multi-rotor UAV platforms can be deployed in mountainous areas where power transmission lines are laid, equipped with oblique photogrammetry cameras (a five-lens system, including one vertical lens and four oblique lenses), enabling multi-view, high-overlap image acquisition capabilities. To ensure flight path accuracy and data synchronization, the UAVs are also equipped with high-precision positioning modules (GNSS / RTK) and inertial measurement units (IMUs) to acquire attitude parameters and spatial coordinates. Image and navigation data undergo preliminary fusion processing via edge computing devices and are transmitted to ground receiving and processing stations in real time or with delay via communication modules (such as 4G / 5G or data transmission links) to perform low-altitude multi-view image acquisition tasks according to the planned flight path.

[0063] The image data includes the latest measured image data as well as historical image data.

[0064] Furthermore, in order to improve model accuracy and reconstruction stability, in this embodiment of the application, step S100 also includes the following specific flight parameter settings:

[0065] The measurement camera's flight altitude is set between 80 and 120 meters, with a ground resolution greater than 5 cm / pixel.

[0066] The horizontal overlap rate of the images should be set to no less than 70%, and the vertical overlap rate should be set to no less than 80%.

[0067] The flight tilt angle is set between 35° and 45° to optimize the stereoscopic vision reconstruction effect;

[0068] Based on the above implementation method, a real-time dynamic carrier phase differential (RTK / PPK) module can also be used for high-precision position synchronization, ensuring that the accuracy of the route control point is better than 5cm.

[0069] In this embodiment of the application, the image data in step S200 generates a dense point cloud model using the structured bundle method (SfM) and a multi-view stereo matching algorithm, including the following steps A1-A2:

[0070] A1: The structured beam method is used to extract and match feature points from image data and solve camera parameters to generate sparse point clouds and restore the spatial geometric relationship of image data.

[0071] Specifically, in step A1, the structured bundle method extracts key points or feature points from multiple images. These points are usually easily identifiable parts of the image, such as corners and edges. These feature points are then matched across different images to find corresponding point pairs. Using the matched point pairs, the camera's internal parameters, such as focal length and principal point, as well as its external parameters, such as position and orientation, are calculated. Then, using triangulation, the three-dimensional coordinates of the feature points are calculated based on the matched points and camera parameters, forming a sparse point cloud. At this point, the geometric relationships between the images are established, including the camera's position and orientation, as well as the approximate structure of the point cloud.

[0072] A2: Based on sparse point clouds and spatial geometric relationships, a multi-view stereo matching algorithm (MVS) is introduced to densify sparse point clouds and obtain a dense point cloud model with spatial resolution.

[0073] Specifically, step A2 uses the MVS algorithm to estimate the depth of each pixel using multiple images and known camera geometry. For each pixel or image region, its depth value is calculated by comparing the corresponding regions in different images. This process needs to consider illumination consistency, texture information, and geometric constraints to ensure the accuracy of matching, thereby generating more points to fill the gaps between sparse point clouds and obtain a dense point cloud model.

[0074] It should be noted that obtaining a dense point cloud model can provide more detailed three-dimensional surface information and has a higher spatial resolution. As a core intermediate result, it can be further used to construct two different types of subsequent three-dimensional data results.

[0075] In this embodiment of the application, step S300, which constructs a digital surface model and a realistic 3D texture model based on a dense point cloud model, includes the following steps B1-B2:

[0076] B1: Regularize the dense point cloud model into a grid and extract elevation information to form a complete surface elevation dataset, generating a digital surface model of the current state;

[0077] Specifically, a dense point cloud model contains a large number of three-dimensional spatial points, each with X, Y, and Z coordinates, which together depict the undulations of the ground and the outlines of objects on the ground.

[0078] Regularization into a grid involves defining a regular grid on the ground plane (XY plane); this grid consists of countless square cells (i.e., pixels) of the same size. The size of each grid cell determines the final resolution of the digital surface model (DSM), for example, 0.1 meters per pixel. Subsequently, each cell in the regular grid is assigned an elevation value. Once all grid cells have been assigned elevation values, a complete digital surface model stored in the form of a regular grid is formed.

[0079] It should be noted that a digital surface model is a two-dimensional raster image, but the value of each pixel is not color, but elevation. It can accurately reflect the three-dimensional morphology of the surface and its attachments, and is used for subsequent slope analysis, terrain change detection and landslide deformation identification.

[0080] B2: By overlaying the texture information of the original image onto the point cloud geometric skeleton in the dense point cloud model, a realistic 3D texture model is generated.

[0081] The point cloud geometric skeleton is the three-dimensional coordinates of the corresponding points.

[0082] The original image is the real image acquired in S100, which provides color and texture information.

[0083] Specifically, surface reconstruction algorithms, such as Poisson reconstruction and triangulation, can be used to connect adjacent point clouds in a dense point cloud model to construct a triangular mesh model composed of countless tiny triangles. This mesh model can accurately describe the geometry of the ground surface.

[0084] Furthermore, since the camera parameters (i.e., position and pose) for each photo have been accurately calculated in step A1 above, each 3D point cloud point can now be precisely mapped to the corresponding pixel on each original photo. Based on the camera parameters, the pixel color or texture of the corresponding area on the photo is extracted, and a final optimal texture map is generated through a fusion algorithm, such as weighted averaging. Finally, the generated texture map is wrapped onto the triangular mesh model to obtain a real-world 3D texture model, which is used for structural identification, material / reflection characteristic analysis, and intuitive display of risk results in the tower foundation area.

[0085] It should be noted that in subsequent processing steps, the DSM model mainly undertakes the functions of terrain geometry quantization and deformation calculation, while the real-scene 3D texture model mainly undertakes the functions of target feature extraction and visualization.

[0086] In this embodiment of the application, step S400, based on the current state of the digital surface model and the real-world 3D texture model, uses spatial semantic segmentation combined with image recognition to extract the tower foundation area, including:

[0087] The tower area was initially determined through spatial semantic segmentation and represented as follows:

[0088] ,

[0089] In the formula, Indicates a condition; Indicates the condition otherwise; For position Elevation value; The average elevation of the target area; For position Normalized Difference Vegetation Index (NDVI) value; The average NDVI value of the target area; To constrain the preset threshold of elevation, A preset threshold to constrain the range of NDVI differences; The binarized result of the recognition; It indicates the condition "and".

[0090] It should be noted that the formula logic for identifying the tower foundation area is as follows: if the difference between the elevation value of a certain point and the target average elevation is less than the preset threshold of the constraint elevation... ( (A value of 1 meter can be used), and the difference between its NDVI value and the target average NDVI is also less than the preset threshold constraining the NDVI difference range. ( If the value is 0.1, the point is determined to belong to the target area; otherwise, it is determined not to belong to the target area. This formula is mainly used to identify and extract the initial tower area: a small elevation difference indicates that the point is similar to the tower base in terms of height characteristics, and a small NDVI difference indicates that the point is similar to the tower base in terms of vegetation / non-vegetation characteristics; when both conditions are met simultaneously, the point can be classified as part of the tower base. The tower foundation area segmentation result obtained by this formula is used as a spatial constraint input into the subsequent risk calculation process to limit the calculation range of factors such as slope, elevation difference (ΔZ), and landslide distance, thereby ensuring that the risk assessment targets the tower foundation and its adjacent area, rather than the entire terrain scene.

[0091] Furthermore, based on the initially determined tower area, a tower target extraction mechanism is formed by combining geometric shape, reflection characteristics, and height threshold with image recognition algorithms to extract the tower base area.

[0092] Specifically, the geometric shape can be the aspect ratio of the tower base and the structural texture; the reflective properties can be the reflectivity of the metallic material; and the height threshold can be the structure that is higher than the surrounding ground features.

[0093] It should be noted that in the target extraction process of the tower foundation area, this invention adopts a method of semantic segmentation (i.e., formula constraints) combined with multi-feature fusion. The formula section primarily relies on the difference between elevation and NDVI values ​​to determine whether a target point belongs to the tower area. This section provides clear mathematical criteria, ensuring interpretability in terms of geometric height and vegetation cover characteristics. Meanwhile, geometric morphology (tower base aspect ratio, structural texture), reflectivity (reflectivity of metallic materials), and height threshold (characteristics of being higher than surrounding features) are not directly involved in the formula but serve as supplementary features. They participate in the fusion judgment through rule constraints and image intelligence algorithms. Their role is to: based on the initial results of formula screening, use prior geometric and spectral features to perform secondary correction on candidate areas, avoiding missed or false detections caused by a single threshold formula.

[0094] In terms of image intelligence algorithms, this invention employs a Convolutional Neural Network (CNN) to spatially classify the texture features of oblique photographic images, extracting the differences between the tower and surrounding vegetation and soil. It also combines a deep learning model (PointNet++) to learn the three-dimensional structural features of dense point clouds, identifying the geometric contours of the tower foundation. Simultaneously, it integrates a Support Vector Machine (SVM) classifier to fuse and discriminate numerical features such as elevation difference, NDVI, and reflectivity. This creates a complementary relationship among the three: CNN emphasizes two-dimensional image texture, PointNet++ emphasizes three-dimensional point cloud structure, and SVM emphasizes numerical feature classification. Through multi-model collaborative decision-making, the robustness and generalization ability of the tower foundation area extraction are further improved.

[0095] The entire extraction mechanism forms a two-layer system that combines "deterministic screening of elevation-NDVI formula" with "multi-model collaborative learning of artificial intelligence," which not only ensures the interpretability of the recognition process but also has adaptability in complex environments, thereby achieving high-precision extraction of tower foundation areas.

[0096] In this embodiment of the application, step S500 involves detecting changes between the current digital land surface model and the historical digital land surface model, including:

[0097] C1: A surface deformation detection algorithm based on point cloud difference is used to detect changes between the current digital land model and historical digital land models, represented as:

[0098] ,

[0099] In the formula, Indicates at coordinate point The elevation change between the two digital land surface models, measured in meters, is a key indicator for detecting land surface deformation (such as landslides and subsidence). Indicates time Time, coordinates The surface elevation value at that location is derived from the current phase of oblique photogrammetry or remote sensing 3D modeling results. Indicates time Time, coordinates The surface elevation value.

[0100] In this embodiment of the application, step S500 involves combining the current and historical real-world 3D texture models to perform auxiliary observation of the deformation area and determine the suspected landslide deformation area, including:

[0101] C2: Visualize and observe areas with elevation changes through current and historical real-world 3D texture models to help determine whether changes exist in the corresponding areas;

[0102] C3: If there is a change in the auxiliary interpretation, and the absolute value of the elevation change value exceeds the first preset threshold within a continuous time period, and the slope change rate is greater than the second preset threshold, then the corresponding area is determined to be a suspected landslide deformation area.

[0103] For example, in C3, the elevation change value over a continuous time period If the absolute value exceeds the first preset threshold (e.g., ≥15cm) and the slope change rate is greater than the second preset threshold (e.g., ≥10°), then the corresponding area is determined to be a suspected landslide deformation area.

[0104] It should be noted that, to achieve dynamic monitoring of the landslide evolution process, this invention employs a multi-temporal aerial survey strategy, periodically (e.g., monthly or quarterly) re-flying to acquire oblique photogrammetric image data. After each re-fly, the image data is used to generate dense point clouds through SfM and MVS algorithms, and further constructed into the latest digital surface model, namely the current state DSM model and the corresponding three-dimensional texture model. The current state DSM is differentially calculated with the historical DSM to quantitatively identify surface deformation characteristics (such as landslide leading edge displacement, basement settlement, etc.); while the three-dimensional texture model is used for intuitive verification and visualization of the deformed area, such as assisting in the judgment of whether cracks, collapses, or exposed surfaces have appeared. The S500 step achieves dynamic monitoring of the landslide evolution process through quantitative detection of DSM and qualitative verification of the texture model.

[0105] In this embodiment of the application, step S600, in conjunction with the tower foundation area, performs a graded assessment of the suspected landslide deformation area to obtain the risk level of the suspected landslide deformation area, including:

[0106] Based on the geological, slope, soil moisture, and relative location indicators of the suspected landslide deformation area, a risk assessment calculation and risk score are performed. Specifically, it is expressed as follows:

[0107] ,

[0108] In the formula, The slope value represents the degree of inclination of the ground surface in the suspected landslide deformation area, expressed in degrees or percentages. It represents the change in surface elevation between two different periods, derived from the differential elevation formula mentioned in C1, and the unit is meters; The horizontal distance between the suspected landslide deformation area and the nearest power transmission tower foundation area is expressed in meters. Soil moisture content; , , and These are the weighting coefficients for the corresponding indicators.

[0109] Specifically, the parameters in the risk assessment need to be normalized to enable score calculation. Historical data or theoretical value ranges for each parameter are collected to determine the minimum and maximum values ​​of each indicator. For example, slope typically ranges from 0 to 90 degrees, elevation changes can range from a few meters to tens of meters, distance to the pole can range from a few meters to hundreds of meters, and soil moisture content generally ranges from 0% to 100%. Through minimum-maximum normalization, the original values ​​of each indicator are converted into values ​​between 0 and 1. Generally, for slope, elevation change, and soil moisture content, larger values ​​indicate higher risk, so larger normalized values ​​directly represent higher risk. However, the opposite is true for pole distance; closer distances indicate higher risk. Therefore, the normalized pole distance values ​​need to be reversed by subtracting the normalized value from 1, so that smaller original distances correspond to larger risk values. This transforms all indicators into unitless relative values ​​for easy direct comparison.

[0110] It should be noted that the steeper the slope, the higher the likelihood of a landslide. A large absolute value indicates significant terrain deformation, a key precursor to landslides. The power transmission tower foundation area is analyzed using the aforementioned formula. Automatic identification is used; the closer the distance, the greater the likelihood that a landslide poses a threat to the tower. Soil moisture content, the degree of water saturation in surface or near-surface soil, can be obtained through remote sensing inversion (such as microwave remote sensing) or on-site sensors. High soil moisture is often associated with landslide-inducing factors (rainfall, infiltration, etc.). The weighting coefficients can be set based on expert knowledge or historical sample training results; for example, the closer to the tower, the steeper the slope, and the greater the settlement, the higher the score. For example, the risk level can be divided into five levels: low risk, low-to-medium risk, medium risk, medium-to-high risk, and high risk.

[0111] Furthermore, the final scoring results in S600 will be presented in Geographic Information System (GIS) format through a 3D visualization platform. Risk areas will be marked on the 3D model with color codes and linked to the transmission line topology map to achieve "model visibility, hazard perception, and risk controllability." It supports the generation of operation and maintenance auxiliary information such as landslide early warning reports, historical trend maps, and tower foundation safety scoring tables, providing power departments with intuitive and operable risk management basis.

[0112] In summary, this invention utilizes an unmanned aerial vehicle (UAV) platform equipped with a multi-lens oblique photography system, which possesses advantages such as high resolution, multiple perspectives, and flexible low-altitude flight. It enables comprehensive, high-density, and rapid data collection of power transmission lines in mountainous areas without the need for manual intervention. This is particularly suitable for power transmission corridors located at high altitudes, with dense vegetation and poor transportation access, significantly improving the timeliness and spatial resolution of disaster monitoring. Compared to traditional remote sensing or ordinary photogrammetry methods that only provide two-dimensional images or rough elevation data, this invention constructs a high-precision digital surface model (DSM) and dense point cloud model using structured beamforming and multi-view stereo reconstruction technology. This allows for refined reconstruction of the tower foundations and surrounding topographic structures. Semantic rules and artificial intelligence algorithms are combined to automatically extract and segment tower targets. Furthermore, through periodic re-flying and comparison with temporal oblique images, a multi-temporal three-dimensional model is established. By using DSM differential analysis and slope changes to jointly identify landslide evolution characteristics (such as leading edge retreat and foundation settlement), dynamic modeling and trend early warning of the entire landslide process from incubation to disaster are achieved. By fusing oblique photogrammetry modeling results with multi-source remote sensing factors such as temporal deformation, NDVI vegetation index, and soil moisture content, a unified disaster risk factor matrix is ​​constructed to quantify the level of landslide hazards. Simultaneously, a highly automated processing workflow is developed, which can be integrated into the power company's existing operation and maintenance systems or smart grid platforms. This supports batch data processing, visualization, and operation and maintenance decision support, significantly reducing manpower input and improving the response speed and prediction capabilities for disasters such as landslides. Ultimately, it can significantly enhance the operation and maintenance efficiency and disaster resistance capabilities of power grid infrastructure.

[0113] Example 3: The above is an illustrative scheme of a risk detection method for mountain power transmission lines driven by UAV oblique photography data.

[0114] This embodiment also provides an electronic device applicable to a method for risk detection of mountain transmission lines driven by UAV oblique photogrammetry data, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for risk detection of mountain transmission lines driven by UAV oblique photogrammetry data as proposed in the above embodiment.

[0115] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photography data, as proposed in the above embodiments.

[0116] The electronic device and medium proposed in this embodiment belong to the same inventive concept as the method for risk detection of mountain power transmission lines driven by UAV oblique photography data proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0117] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0118] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photogrammetry data, characterized in that, include: Acquire image data from an oblique photogrammetry camera; generate a dense point cloud model from the image data using a structured beamforming method and a multi-view stereo matching algorithm; construct a digital surface model and a realistic 3D texture model based on the dense point cloud model; and extract the tower foundation area using a spatial semantic segmentation method combined with image recognition based on the current state digital surface model and realistic 3D texture model, including: initially determining the tower area through spatial semantic segmentation, represented as: In the formula, Indicates a condition; Indicates the condition otherwise; For position Elevation value; The average elevation of the target area; For position Normalized Difference Vegetation Index (NDVI) value; The average NDVI value of the target area; To constrain the preset threshold of elevation, A preset threshold to constrain the range of NDVI differences; The binarized result of the recognition; This section describes a process involving the "and" condition. Based on the preliminary assessment of the tower area, a tower target extraction mechanism is established using geometric shape, reflection characteristics, height thresholds, and image recognition algorithms to extract the tower foundation area. Change detection is performed between the current and historical digital surface models, and the deformation area is further analyzed using current and historical real-world 3D texture models to identify suspected landslide deformation areas. Based on the tower foundation area, a graded assessment of suspected landslide deformation areas is conducted to determine their risk level, which is then visualized using the transmission line topology map to achieve risk labeling and early warning. Specifically, the graded assessment of suspected landslide deformation areas, based on the tower foundation area, includes risk assessment calculations and risk scoring based on the geological, slope, soil moisture, and relative location indicators of the suspected landslide deformation area. Specifically, it is expressed as follows: In the formula, The slope value represents the degree of inclination of the ground surface in the suspected landslide deformation area; This represents the change in surface elevation between two different periods; The horizontal distance between the suspected landslide deformation area and the foundation area of ​​the nearest power transmission tower; Soil moisture content; 、 、 and These are the weighting coefficients for the corresponding indicators.

2. The method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photogrammetry data as described in claim 1, characterized in that, The image data is used to generate a dense point cloud model through structured beamforming and multi-view stereo matching algorithms. This includes: extracting and matching feature points and calculating camera parameters from the image data using structured beamforming to generate a sparse point cloud and restore the spatial geometric relationship of the image data; and introducing a multi-view stereo matching algorithm based on the sparse point cloud and spatial geometric relationship to densify the sparse point cloud and obtain a dense point cloud model with spatial resolution.

3. The method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photogrammetry data as described in claim 2, characterized in that, The process involves constructing a digital land surface model and a realistic 3D texture model based on a dense point cloud model. This includes: regularizing the dense point cloud model into a grid and extracting elevation information to form a complete land elevation dataset, thereby generating a digital land surface model in the current state; and generating a realistic 3D texture model by overlaying texture information from the original image onto the point cloud geometric skeleton in the dense point cloud model.

4. The method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photogrammetry data as described in claim 3, characterized in that, Change detection is performed between the current digital land surface model and historical digital land surface models, including: using a point cloud-based difference-based land surface deformation detection algorithm to detect changes between the current and historical digital land surface models, as shown below: In the formula, Indicates at coordinate point The elevation change between the two digital surface models is shown in meters. Indicates time Time, coordinates The surface elevation value at that location; Indicates time Time, coordinates The surface elevation value.

5. The method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photogrammetry data as described in claim 4, characterized in that, The current and historical real-world 3D texture models are used to assist in the observation of deformed areas and identify suspected landslide deformation areas. This includes: visually observing areas with elevation changes using the current and historical real-world 3D texture models to help determine whether changes exist in the corresponding areas. Whether changes exist in the corresponding areas includes whether cracks, collapses, or exposed ground surfaces appear. If the auxiliary judgment shows changes, and the absolute value of the elevation change value exceeds a first preset threshold within a continuous time period, while the slope change rate is greater than a second preset threshold, then the corresponding area is determined to be a suspected landslide deformation area.

6. The method for risk detection of power transmission lines in mountainous areas driven by UAV oblique photogrammetry data as described in claim 5, characterized in that, The acquisition of image data collected by the oblique photogrammetry camera includes: setting the camera's flight altitude between 80 and 120 m, with a ground resolution greater than 5 cm / pixel; setting the lateral overlap rate of the images to be no less than 70%, and the longitudinal overlap rate of the images to be no less than 80%; and setting the flight tilt angle between 35° and 45°.

7. An electronic device, characterized in that, include: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the method for risk detection of mountain power transmission lines driven by UAV oblique photography data as described in any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that, The device stores computer-executable instructions that, when executed by a processor, implement the steps of the method for risk detection of mountain power transmission lines driven by UAV oblique photogrammetry data as described in any one of claims 1 to 6.

Images