Power transmission line reconstruction method combined with multiple parallel line lasers

Abstract

The invention discloses a power transmission line reconstruction method combined with multi-parallel-line laser, and belongs to the field of power transmission line modeling, and the method comprises the steps: determining a target power transmission line region, and carrying out the three-dimensional scanning of a multi-parallel-line laser scanner, and obtaining the point cloud data of the region; shooting a plurality of groups of binocular stereo images by using a binocular camera to obtain regional image data; establishing a registration coordinate system, and registering the point cloud and the image data to obtain an initial three-dimensional model of the power transmission line; constructing a key component set, and carrying out geometric recognition and segmentation on the models to obtain a plurality of key component models; and optimizing the plurality of key component models to obtain a plurality of optimized component models, and performing model optimization on the initial three-dimensional model of the power transmission line to obtain an accurate three-dimensional model of the power transmission line. The technical problems of low precision and poor efficiency of three-dimensional model reconstruction of the power transmission line in the prior art are solved, and the technical effects of improving the three-dimensional model reconstruction precision of the power transmission line and realizing efficient and accurate three-dimensional reconstruction of the power transmission line are achieved.

Description

Technical Field

[0001] The invention relates to the field of power transmission line modeling, and in particular to a power transmission line reconstruction method combining multi-parallel line lasers. Background Art

[0002] Transmission lines are an important part of the power system, and their safe and stable operation directly affects the reliability of the power grid. In order to ensure the safe operation and maintenance of transmission lines, it is necessary to reconstruct the transmission lines in three dimensions to obtain their accurate three-dimensional models. However, traditional three-dimensional reconstruction methods for transmission lines mainly rely on manual measurement or a single data source for modeling, such as using a total station to measure the coordinates of key points for modeling, or only using a single image data for reconstruction. Among them, manual measurement is time-consuming and labor-intensive, and it is difficult to obtain detailed structural information of the transmission line; the single data source does not have enough information, and it is difficult to accurately express the geometric characteristics and spatial position relationship of the transmission line; at the same time, there is a lack of high-precision modeling of key components, which leads to a large difference between the reconstructed model and the actual transmission line. Therefore, the reconstruction accuracy of the three-dimensional model of the transmission line in the existing technology is low and the efficiency is poor. Summary of the invention

[0003] The present application provides a method for reconstructing a power transmission line by combining multiple parallel line lasers, aiming to solve the technical problems of low accuracy and poor efficiency in reconstructing a three-dimensional model of a power transmission line in the prior art.

[0004] The power transmission line reconstruction method combined with multi-parallel line laser disclosed in the present application includes: determining a target power transmission line area, performing three-dimensional scanning on the target power transmission line area using a multi-parallel line laser scanner, and obtaining regional point cloud data, wherein the multi-parallel line laser scanner includes at least two mutually parallel laser beams; using a binocular camera, taking multiple groups of binocular stereo images of the target power transmission line area, and obtaining regional image data; establishing a registration coordinate system, registering the regional point cloud data with the regional image data, unifying them into the registration coordinate system, and obtaining an initial three-dimensional model of the power transmission line; constructing a key component set of the power transmission line, geometrically identifying and segmenting the initial three-dimensional model of the power transmission line based on the key component set, and obtaining multiple key component models of the target power transmission line area, wherein each key component model has a category label and spatial position information; extracting multiple groups of binocular stereo images, respectively optimizing multiple key component models, obtaining multiple optimized component models, and optimizing the initial three-dimensional model of the power transmission line in combination with the category label and the spatial position information, and obtaining an accurate three-dimensional model of the power transmission line.

[0005] One or more technical solutions provided in this application have at least the following technical effects or advantages: Because the target transmission line area is determined, a multi-parallel line laser scanner is used to perform a three-dimensional scan of the target transmission line area to obtain regional point cloud data, wherein the multi-parallel line laser scanner includes at least two mutually parallel laser beams. By using multi-parallel line laser scanning, the three-dimensional point cloud data of the transmission line area is quickly and efficiently obtained, providing rich spatial information for subsequent modeling; a binocular camera is used to take multiple groups of binocular stereo images of the target transmission line area to obtain regional image data, capture the texture, color and other appearance information of the target area, and form a complementary relationship with the point cloud data to jointly provide a data basis for three-dimensional reconstruction; a registration coordinate system is established to register the regional point cloud data with the regional image data, and unify them into the registration coordinate system to obtain an initial transmission line three-dimensional model; through data registration, data from different sources are unified into the same coordinate frame to achieve spatial alignment of point cloud and image data, laying a foundation for subsequent processing; a key component set of the transmission line is constructed, and the initial transmission line three-dimensional model is constructed based on the key component set. Geometric recognition and segmentation are performed to obtain multiple key component models of the target transmission line area, wherein each key component model has a category label and spatial position information. By introducing a key component set, the model segmentation is guided to accurately identify and extract key components of the transmission line, such as towers, insulators, conductors, etc.; multiple groups of binocular stereo images are extracted, and multiple key component models are optimized respectively to obtain multiple optimized component models, and the initial transmission line three-dimensional model is optimized in combination with the category label and spatial position information to obtain an accurate transmission line three-dimensional model. By optimizing the key component model using binocular images and integrating the optimized component model into the initial three-dimensional model, the accuracy of the key components and the overall transmission line model is effectively improved, and a technical solution for obtaining a more realistic and accurate three-dimensional reconstruction result is obtained, which solves the technical problems of low accuracy and poor efficiency of transmission line three-dimensional model reconstruction in the prior art, and achieves the technical effect of improving the reconstruction accuracy of the transmission line three-dimensional model and realizing efficient and accurate transmission line three-dimensional reconstruction.

[0006] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 A schematic flow chart of a power transmission line reconstruction method combining multiple parallel laser lines is provided for the embodiment of the present application; Figure 2 A flow chart of obtaining an accurate three-dimensional model of a power transmission line in a power transmission line reconstruction method combining multi-parallel line lasers is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0008] The overall idea of ​​the technical solution provided by this application is as follows: The embodiment of the present application provides a method for reconstructing a transmission line in combination with a multi-parallel line laser. First, by determining the target transmission line area, a multi-parallel line laser scanner and a binocular camera are used to obtain regional point cloud data and regional image data respectively; then, a registration coordinate system is established to register the two types of data to obtain an initial three-dimensional model of the transmission line. On this basis, a set of key components of the transmission line is constructed, and the initial model is geometrically identified and segmented based on the key component set to obtain multiple key component models with category labels and spatial position information. Afterwards, multiple groups of binocular stereo images are extracted to optimize the key component models, and the initial transmission line three-dimensional model is optimized in combination with the category labels and spatial position information, so as to obtain an accurate three-dimensional model of the transmission line, which overcomes the technical problems of low accuracy and poor efficiency of the reconstruction of the three-dimensional model of the transmission line in the prior art, and achieves the technical effect of improving the reconstruction accuracy of the three-dimensional model of the transmission line and realizing efficient and accurate three-dimensional reconstruction of the transmission line.

[0009] like Figure 1 As shown, the embodiment of the present application provides a method for power transmission line reconstruction combined with multiple parallel line lasers, the method comprising: S1: Determine a target transmission line area, and use a multi-parallel line laser scanner to perform a three-dimensional scan on the target transmission line area to obtain regional point cloud data, wherein the multi-parallel line laser scanner includes at least two mutually parallel laser beams.

[0010] Specifically, first, determine the target transmission line area that needs to be reconstructed in three dimensions. The target transmission line area contains a complete section of the transmission line, and also includes key components such as poles, conductors, insulators, hardware, and the terrain environment around the line. After determining the scope of the target transmission line area, use a multi-parallel line laser scanner to perform a full-scale three-dimensional scan of the area. A multi-parallel line laser scanner is a laser scanning device that includes at least two mutually parallel laser beams. These laser beams will form multiple parallel laser lines on the surface of the target transmission line area during the scanning process. When the laser line is projected onto the target surface, it will be reflected back by the target surface. The scanner calculates a large amount of three-dimensional coordinate point data on each laser line by receiving the reflected laser signal. Since the spacing between the parallel laser beams is very small, the three-dimensional point cloud data obtained by scanning has high density and accuracy.

[0011] During the scanning process, the multi-parallel line laser scanner changes the scanning angle and position to perform multi-angle and multi-directional scanning of the target transmission line area, covering as many parts of the target transmission line area as possible and reducing occlusion and blind spots. The scanning process can be completed by manual control or automatic control. After the scanning is completed, a large-scale three-dimensional point cloud data set covering the entire target transmission line area is obtained, namely, regional point cloud data.

[0012] Through the all-round 3D scanning of multi-parallel line laser scanners, the 3D point cloud data of the transmission line area is acquired efficiently and accurately, laying a solid data foundation for the accurate 3D reconstruction of the transmission line.

[0013] S2: Using a binocular camera, capture multiple groups of binocular stereo images of the target power transmission line area to obtain regional image data.

[0014] Specifically, a binocular camera is a stereo camera with two lenses, and there is a certain distance between the two lenses, simulating the binocular vision characteristics of the human eye. By simultaneously shooting the same target area from different perspectives with two lenses, a pair of stereo images with parallax information can be obtained. Using this parallax information, the depth distance of the object in the image can be calculated, thereby realizing the acquisition of three-dimensional information. Specifically, the target power transmission line area is photographed multiple times and at multiple angles using a binocular camera. Each shot obtains a set of binocular stereo images, that is, a pair of left and right images with parallax. By changing the shooting position and angle of the binocular camera, stereo images of different parts of the target power transmission line area are obtained to cover the entire target power transmission line area. During the shooting process, the lens of the binocular camera is always kept aimed at the target power transmission line area to ensure that a clear and accurate stereo image is obtained. At the same time, in order to improve shooting efficiency and data quality, the shooting parameters of the binocular camera, such as focal length, exposure time, etc., are adjusted by an automatic control system. Through multiple shooting, multiple sets of binocular stereo image data of the target transmission line area, namely regional image data, are obtained, which contains rich surface texture and color information of the target transmission line area. Combined with the obtained regional point cloud data, more detailed and realistic three-dimensional model reconstruction can be achieved.

[0015] By using a binocular camera to collect multiple sets of stereo images of the target transmission line area, regional image data is obtained, which complements the point cloud data and provides appearance and texture information support for the subsequent fine reconstruction of the three-dimensional model of the transmission line.

[0016] S3: Establishing a registration coordinate system, registering the regional point cloud data with the regional image data, unifying them into the registration coordinate system, and obtaining an initial three-dimensional model of the power transmission line.

[0017] Specifically, after obtaining the regional point cloud data and regional image data of the target transmission line area, these two different types of data are registered and fused to form an initial transmission line three-dimensional model.

[0018] First, a unified registration coordinate system is established. The registration coordinate system is a virtual three-dimensional spatial reference system used to unify data from different sources and formats into the same coordinate framework. Among them, the establishment of the registration coordinate system requires comprehensive consideration of the spatial distribution characteristics of the interval point cloud data and the interval image data, and the selection of appropriate coordinate origins, coordinate axis directions, and scale factors. For example, based on the coordinate system of the point cloud data, a corresponding registration coordinate system is established. Then, the regional point cloud data and the regional image data are registered with the registration coordinate system respectively, and the data in different coordinate systems are unified into the registration coordinate system through coordinate transformation. For the regional point cloud data, its coordinates are converted to the registration coordinate system through rigid body transformations such as translation and rotation; for the image data, the internal and external parameters of the camera are considered, and the image pixel coordinates are corresponded to the three-dimensional coordinates in the registration coordinate system. Through registration, the regional point cloud data and the regional image data are spatially unified and aligned. The registered data have consistent scale, direction, and position relationship in the registration coordinate system, laying the foundation for subsequent data fusion processing.

[0019] After that, the registered regional point cloud data and regional image data are fused to generate the initial three-dimensional model of the transmission line. The fusion process comprehensively utilizes the characteristics of the two types of data, extracts effective information, and optimizes the combination. Specifically, the regional point cloud data provides accurate three-dimensional geometric information for reconstructing the main structure and spatial position relationship of the transmission line; the regional image data provides rich texture and color information for improving the visual realism of the model. By matching the regional point cloud data with the regional image data and mapping the texture to the three-dimensional geometric model, a three-dimensional model of the transmission line with both geometric structure and real texture is obtained, that is, the initial three-dimensional model of the transmission line.

[0020] By establishing a registration coordinate system, the regional point cloud data and the regional image data are registered and fused to generate the initial transmission line three-dimensional model, thus achieving the organic unification and integrated utilization of different types of data, and laying the foundation for the subsequent refined reconstruction of the transmission line three-dimensional model.

[0021] S4: construct a key component set of the transmission line, perform geometric identification and segmentation on the initial transmission line three-dimensional model based on the key component set, and obtain multiple key component models of the target transmission line area, wherein each key component model has a category label and spatial location information.

[0022] Specifically, first, according to the structural characteristics and engineering requirements of the transmission line, a set of key components of the transmission line is constructed. The key component set includes important components of the transmission line, such as poles, conductors, insulators, hardware, etc. Among them, each type of key component has a clear definition and description in the set, including its geometric shape, size range, spatial relationship and other characteristic information. Then, with the key component set as a reference, the initial transmission line three-dimensional model is geometrically identified and segmented. Geometric identification refers to detecting and locating the geometric structure matching the key component from the initial transmission line three-dimensional model and determining its category label; segmentation refers to separating the identified key components from the entire initial transmission line three-dimensional model to obtain an independent key component model. Among them, the segmentation process ensures the integrity and consistency of the segmentation results based on the topological relationship and boundary conditions between the components.

[0023] After geometric recognition and segmentation, a series of 3D models of key components, i.e., multiple key component models, are obtained from the initial 3D model of the transmission line. Each key component model has a clear category label, indicating the type of component it belongs to, such as a tower model, an insulator model, etc. At the same time, during the recognition and segmentation process, the spatial position information of each key component model in the original initial 3D model of the transmission line is recorded, including its coordinates, direction, scale, etc.

[0024] By constructing a set of key components and performing geometric identification and segmentation on the initial three-dimensional model of the transmission line based on the key component set, a series of key component models with category labels and spatial position information are obtained, realizing the decomposition and extraction from the overall model to local components, providing a basis for subsequent fine modeling and optimization adjustment.

[0025] S5: extract multiple groups of binocular stereo images, optimize multiple key component models respectively, obtain multiple optimized component models, and optimize the initial transmission line three-dimensional model in combination with the category label and the spatial position information to obtain an accurate transmission line three-dimensional model.

[0026] Specifically, the rich texture information provided by high-resolution binocular stereo images is used to visually enhance and optimize the geometric details of the initial transmission line 3D model. First, local image blocks corresponding to different parts of the transmission line are extracted from multiple sets of binocular stereo images as the data source for texture mapping. Then, these image blocks are mapped to the surfaces of multiple key component models, and multiple key component models are given realistic surface textures through methods such as UV mapping. At the same time, according to the detailed information of the image, the geometric structures of multiple key component models are fine-tuned and optimized to make them more consistent with the detailed features of the real scene. After texture mapping and geometric optimization of the key component models, the optimized component models are reassembled into the initial transmission line 3D model, and the overall model is adjusted and optimized, including geometric correction of the model, texture smoothing, lighting rendering and other processing, to improve the visual quality and realism of the entire transmission line 3D model, thereby obtaining an accurate transmission line 3D model.

[0027] By utilizing the texture and detail information of binocular stereo images, the initial three-dimensional model of the transmission line is refined and optimized, achieving visual enhancement and geometric detail expression of the three-dimensional model, and obtaining an accurate and realistic three-dimensional model of the transmission line.

[0028] Furthermore, the embodiment of the present application also includes: Acquire spatial data information of the target transmission line area, and generate a spatial reference coordinate system based on the spatial data information; generate a laser scanning route and an image acquisition route based on the spatial reference coordinate system; according to the laser scanning route, control the multi-parallel line laser scanner to perform a three-dimensional scan of the target transmission line area to acquire regional point cloud data; according to the image acquisition route, control the binocular camera to perform image acquisition on the target transmission line area to acquire regional image data.

[0029] In a feasible implementation, by pre-acquiring spatial data information of the target transmission line area and planning the laser scanning route and image acquisition route accordingly, the three-dimensional data of the transmission line can be acquired more efficiently and comprehensively.

[0030] First, obtain the spatial data information of the target transmission line area. The spatial data information includes the geographical location, topography, surrounding environment and other data of the target transmission line area, which are obtained through field surveys, satellite remote sensing, geographic information systems, etc. Secondly, based on the spatial data information, establish a three-dimensional spatial reference coordinate system in the target transmission line area. The spatial reference coordinate system takes a certain characteristic point (such as the center of the tower base) of the target transmission line area as the origin, and takes the east, north, and vertical upward directions as the coordinate axes to establish a three-dimensional coordinate system in the actual environment.

[0031] Then, based on the spatial reference coordinate system, the laser scanning route and image acquisition route are generated. Among them, the laser scanning route refers to the motion trajectory and scanning direction of the multi-parallel line laser scanner, which determines the scanning coverage and scanning density of the multi-parallel line laser scanner in the target transmission line area; the image acquisition route refers to the shooting position and shooting angle of the binocular camera, which determines the distribution and overlap of the binocular image. By comprehensively considering the spatial reference coordinate system, the spatial distribution of the transmission line, the terrain conditions and other factors, through computer-aided design, path planning and other technologies, the laser scanning route and image acquisition route are obtained to improve the efficiency and quality of data acquisition and reduce blind spots and redundancy. After that, the multi-parallel line laser scanner is controlled to move and scan in the target transmission line area according to the planned laser scanning route. During the scanning process, the position and posture of the multi-parallel line laser scanner are recorded in real time in the spatial reference coordinate system, corresponding to the point cloud data obtained by scanning. Through multiple scans, complete regional point cloud data is obtained to reflect the three-dimensional structure of the transmission line and its surrounding environment. At the same time, the binocular camera is controlled to shoot at different positions and angles in the target transmission line area according to the planned image acquisition route. Each shot obtains a set of binocular stereo images, including left and right perspective images. Through multiple shots and image stitching, full coverage of regional image data is obtained to reflect the appearance and texture of the transmission line and its surrounding environment.

[0032] By acquiring complete regional point cloud data and regional image data of the target transmission line area, a high-quality data foundation is provided for subsequent 3D reconstruction. A spatial reference coordinate system is pre-built and the data collection route is planned accordingly to improve the pertinence and reliability of data collection and reduce the randomness of manual operation, thereby obtaining 3D data of the transmission line more efficiently and accurately.

[0033] Furthermore, the embodiment of the present application also includes: An interactive model building platform is provided, wherein a registration coordinate system is established in the model building platform according to the spatial reference coordinate system; based on the mapping relationship between the spatial reference coordinate system and the registration coordinate system, a first fitting route is generated according to the laser scanning route, and a second fitting route is generated according to the image acquisition route; according to the first fitting route, the regional point cloud data is fitted into the registration coordinate system to obtain a first transmission line three-dimensional model; according to the second fitting route, the regional image data is fitted into the first transmission line three-dimensional model to obtain an initial transmission line three-dimensional model.

[0034] In a preferred embodiment, by constructing a registration coordinate system in the model building platform and fitting the data into the registration coordinate system, data registration and preliminary construction of the three-dimensional model can be completed more accurately and efficiently.

[0035] First, in the model building platform, a registration coordinate system is established according to the spatial reference coordinate system. The model building platform is a virtual three-dimensional environment for digital modeling and simulation. In the platform, a virtual coordinate system corresponding to the spatial reference coordinate system, namely the registration coordinate system, is constructed. The registration coordinate system is consistent with the spatial reference coordinate system in terms of direction, scale, origin position, etc., and is a digital mapping of the real environment coordinate system. By constructing the registration coordinate system in the model building platform, a unified spatial reference framework is provided for subsequent data registration and model construction. Then, based on the mapping relationship between the spatial reference coordinate system and the registration coordinate system, the fitting routes of the laser scanning route and the image acquisition route are generated respectively. Since the laser scanning route and the image acquisition route are planned in the spatial reference coordinate system of the real environment, and the data registration and model construction are carried out in the registration coordinate system of the virtual environment, the laser scanning route and the image acquisition route are mapped to the registration coordinate system to obtain the corresponding fitting routes. The laser scanning route is mapped to the registration coordinate system to obtain the first fitting route; the image acquisition route is mapped to the registration coordinate system to obtain the second fitting route. By generating the first fitting route and the second fitting route, a mapping relationship is established between the data collection route of the real environment and the registration coordinate system of the virtual environment, laying a foundation for subsequent data registration.

[0036] Subsequently, according to the first fitting route, the coordinates of each three-dimensional point in the regional point cloud data are converted to the registration coordinate system to obtain the registered point cloud data, and the registered point cloud data is triangulated, surface reconstructed, etc. to generate the first transmission line three-dimensional model, which reflects the geometric structure of the transmission line. Afterwards, according to the second fitting route, each binocular stereo image in the regional image data is registered with the first transmission line three-dimensional model to determine the mapping position and direction of the image on the model surface. Next, the registered binocular stereo image is used as a texture and mapped to the surface of the first transmission line three-dimensional model to obtain the initial transmission line three-dimensional model, which not only contains the geometric structure of the transmission line, but also contains the real surface texture information.

[0037] By constructing a registration coordinate system corresponding to the real environment in the model building platform and fitting the regional point cloud data and regional image data to the coordinate system, the initial three-dimensional model of the transmission line is obtained, which integrates the high-precision geometric information of the point cloud data and the high-quality texture information of the image data, laying the foundation for subsequent model optimization and refined processing.

[0038] Furthermore, the embodiment of the present application also includes: Based on the key component set, a component geometry segmentor is constructed, wherein the component geometry segmentor includes a component geometry recognition branch and a component geometry segmentation branch; geometric recognition is performed on the initial three-dimensional transmission line model through the component geometry recognition branch to obtain key component recognition results, wherein the key component recognition results include category labels; according to the key component recognition results, model segmentation is performed on the initial three-dimensional transmission line model through the component geometry segmentation branch to obtain multiple key component models, and spatial position information of each key component model in the initial three-dimensional transmission line model is recorded during model segmentation.

[0039] In a feasible implementation, first, a component geometry segmenter for geometric identification and segmentation is constructed based on a pre-constructed set of key components of the transmission line. The key component set includes important components in the transmission line, such as towers, conductors, insulators, hardware, etc. Each component has its own characteristic descriptions such as geometric shape, size range, spatial relationship, etc. Utilizing these characteristic information, a component geometry segmenter is trained through machine learning and other technologies. The component geometry segmenter consists of two functional branches, namely the component geometry identification branch and the component geometry segmentation branch. Among them, the component geometry identification branch is responsible for identifying each key component from the three-dimensional model of the transmission line and attaching category labels; the component geometry segmentation branch is responsible for segmenting the identified key components from the entire three-dimensional model to obtain independent component models.

[0040] Subsequently, the initial transmission line 3D model is input into the component geometry recognition branch to detect and locate the key components in the initial transmission line 3D model, and a category label is associated with each detected component object to indicate which category the component belongs to, such as a tower, a conductor, an insulator, etc., to obtain the key component recognition result. The key component recognition result is represented by a series of category labels of key components, describing which key components are included in the initial transmission line 3D model. Next, according to the key components in the key component recognition result, the segmentation boundaries of each key component are determined in the initial transmission line 3D model to form segmentation lines. Afterwards, according to these segmentation lines, these key components are copied and extracted one by one from the initial transmission line 3D model to obtain multiple key component models. At the same time, in the process of extracting each key component model, its spatial position information in the original initial transmission line 3D model is recorded, including parameters such as coordinates, direction, and scale.

[0041] By constructing a component geometry segmentor and using it to perform geometric identification and segmentation on the initial transmission line 3D model, multiple key component models are obtained. Each model has a clear category label and spatial location information, providing a basis for 3D reconstruction of the transmission line.

[0042] Furthermore, the embodiment of the present application also includes: An interactive power transmission line database is provided, wherein a three-dimensional power transmission line sample set is constructed based on the power transmission line database, including a plurality of three-dimensional power transmission line samples; key component identification and segmentation are performed on the plurality of three-dimensional power transmission line samples based on the key component set to obtain a three-dimensional power transmission line segmentation sample set; and the component geometry identification branch is trained and obtained based on the three-dimensional power transmission line sample set and the three-dimensional power transmission line segmentation sample set.

[0043] In a preferred embodiment, first, interact with the transmission line database, obtain multiple transmission line three-dimensional model samples therefrom, and construct a transmission line three-dimensional sample set. The transmission line database is a special database that stores a large amount of transmission line three-dimensional model data, which comes from different transmission line projects and covers various types and specifications of transmission line designs. By querying and retrieving the transmission line database, a batch of representative transmission line three-dimensional model samples are obtained, which are organized into a transmission line three-dimensional sample set for subsequent recognition training. Subsequently, professional labelers identify each key component in each transmission line three-dimensional sample according to the definition and feature description of each key component provided by the key component set, and attach corresponding category labels to them. Then, based on the recognition results, professional labelers draw the segmentation boundaries of each key component on the transmission line three-dimensional sample, segment it from the overall model, and obtain an independent component model. By manually identifying and segmenting all transmission line three-dimensional samples, a transmission line three-dimensional segmentation sample set is obtained, which includes the identified and segmented key component models and the corresponding category labels. Afterwards, the 3D sample set of the transmission line is used as training data, and the 3D segmentation sample set of the transmission line is used as annotation data, which are input into the machine learning model for training. By continuously adjusting and optimizing its internal parameters, it can accurately identify the key components in the 3D samples of the transmission line and give the correct category labels. Through multiple iterations, the model is continuously fine-tuned and verified with new sample data until its performance reaches the preset accuracy, and the component geometry recognition branch is obtained.

[0044] By using the transmission line database to construct a three-dimensional sample set of transmission lines, and identifying and segmenting key components, a three-dimensional segmentation sample set of transmission lines is obtained. These two sample sets are used for training and optimization to obtain a component geometry recognition model to provide support for component geometry recognition.

[0045] Further, such as Figure 2 As shown, the embodiment of the present application also includes: According to the spatial position information, multiple key component models are positioned in the initial three-dimensional model of the transmission line; multiple local image blocks corresponding to the multiple key component models are extracted from the multiple groups of binocular stereo images; for each key component model, texture mapping is performed using the corresponding local image block to obtain a key component model with realistic texture as an optimized component model; according to the category label and the spatial position information, the multiple optimized component models are integrated into the initial three-dimensional model of the transmission line to obtain an accurate three-dimensional model of the transmission line.

[0046] In a feasible implementation, first, the spatial position information of each key component model in the registration coordinate system is extracted, including coordinates, direction, scale, etc. The spatial position parameters are used to accurately locate the corresponding positions of the key component models on the initial three-dimensional model of the transmission line, providing a basis for subsequent model optimization. Then, based on the spatial position information of each key component model, its corresponding imaging area is determined in multiple sets of binocular stereo images, and the area is extracted to form a local image block. The local image block represents the image of the key component in the high-resolution image and contains detailed texture information on the surface of the key component. By extracting local image blocks for all key component models, multiple high-resolution local image blocks corresponding to the key component models are obtained.

[0047] Afterwards, the extracted local image blocks are used to perform texture mapping on each key component model to obtain an optimized component model with realistic texture. Texture mapping is to map the pixel information on the local image block to the surface of the key component model so that it presents a realistic texture effect. Specifically, the surface of each key component model is divided into a series of small triangular facets, and the pixels on the local image block are mapped to these triangular facets one by one according to the mapping relationship to form a continuous texture map. Through texture mapping, the surface of the key component model is endowed with realistic texture information, and its visual effect is greatly improved to become an optimized component model. Then, according to the category label and spatial position information of each optimized component model, it is matched with the corresponding position in the initial transmission line 3D model, and the optimized component model is copied to the initial transmission line 3D model to obtain an accurate transmission line 3D model. The model is highly consistent with the real transmission line in terms of geometric structure, and is close to the real scene in terms of visual effect, supporting various high-precision visualization applications and analysis tasks.

[0048] By using binocular stereo images to optimize the texture of the initial transmission line 3D model, an accurate and realistic 3D model of the transmission line is obtained, which improves the visual quality and realism of the 3D model.

[0049] Furthermore, the embodiment of the present application also includes: Distortion correction and geometric correction are performed on the local image block to obtain a corrected image block; texture mapping is performed on the surface of the key component model using the corrected image block to obtain an optimized component model; wherein, UV mapping is established between the mesh surface of the key component model and the corrected image block.

[0050] In a preferred embodiment, distortion correction and geometric correction are performed on local image blocks to obtain high-quality corrected image blocks, and the corrected image blocks are used to perform UV mapping on the surface of key component models to obtain optimized component models with realistic textures, thereby solving the distortion and geometric distortion problems in the original image blocks and improving the accuracy and realism of texture mapping.

[0051] First, the extracted local image blocks are subjected to distortion correction and geometric correction to obtain the corrected image blocks. Specifically, due to factors such as lens distortion and non-ideality of photosensitive elements during the imaging process, there are often certain distortions in the original local image blocks, such as barrel distortion and pincushion distortion. These distortions will cause geometric deformation of objects on the image blocks, affecting the accuracy of texture mapping. To address this problem, the local image blocks are subjected to distortion correction. By establishing a distortion model and estimating the distortion parameters, each pixel in the local image blocks is reprojected to eliminate the distortion effect and obtain an undistorted image block. Then, the undistorted image blocks are subjected to geometric correction. According to the spatial position and posture of the key component model, the image blocks are subjected to geometric transformations such as rotation, scaling, and translation to match the surface shape of the key component model. After distortion correction and geometric correction, a corrected image block that is highly consistent with the key component model is obtained, providing support for accurate texture mapping. After that, the corrected image blocks are used to perform texture mapping on the surface of the key component model to obtain an optimized component model with realistic texture. Texture mapping is to establish a mapping relationship between the corrected image block and the surface of the key component model, and accurately map the pixels on the corrected image block to each triangular facet of the key component model. To this end, the UV mapping technology is used. Specifically, first, the surface of the key component model is meshed and decomposed into a series of regular triangular mesh facets; then, a set of UV coordinates is assigned to each triangular facet to represent the corresponding position of each point on the facet in the corrected image block. By establishing a mapping relationship between the vertex coordinates of the triangular facet and the UV coordinates, the pixels on the corrected image block are mapped to the triangular facets one by one according to the UV coordinates to form a continuous texture map. By UV mapping all triangular facets, the entire surface of the key component model is given realistic texture information, the visual quality is greatly improved, and it becomes an optimized component model.

[0052] Furthermore, the embodiment of the present application also includes: Traverse multiple key component models, obtain a first key component model, and extract spatial position information of the first key component model as first spatial position information; call a first optimized component model corresponding to the first key component model; integrate the first optimized component model into the initial transmission line three-dimensional model according to the first spatial position information; iteratively perform model integration to obtain an accurate transmission line three-dimensional model.

[0053] In a feasible implementation, first, multiple key component models are obtained, and multiple key component models are traversed in a certain order (such as in the lexicographic order of the category label). During the traversal process, each key component model is obtained in turn as the first key component model, and its corresponding spatial position information is extracted as the first spatial position information. After that, the optimized component model corresponding to the first key component model is retrieved and marked as the first optimized component model. Subsequently, in the initial transmission line three-dimensional model, the specific position of the first key component model is found according to the first spatial position information, and the first optimized component model is copied to the corresponding component model in the initial transmission line three-dimensional model. By replacing the model, the low-precision component model in the initial transmission line three-dimensional model is updated with the high-precision, high-quality optimized component model. During the copying process, the first spatial position information is used to transform the coordinates and scale the first optimized component model so that it perfectly matches the corresponding position in the initial transmission line three-dimensional model. By iterative integration, the remaining key component models are replaced with the corresponding optimized component models to obtain an accurate transmission line three-dimensional model.

[0054] In summary, the power transmission line reconstruction method combined with multi-parallel line lasers provided in the embodiments of the present application has the following technical effects: Determine the target transmission line area, use a multi-parallel line laser scanner to perform a three-dimensional scan of the target transmission line area, obtain regional point cloud data, efficiently and accurately collect the spatial geometric information of the transmission line, and provide basic data for subsequent three-dimensional reconstruction. Use a binocular camera to take multiple sets of binocular stereo images of the target transmission line area, obtain regional image data, capture the texture, color and other appearance information of the transmission line area, and enrich the visual effect of the three-dimensional model. Establish a registration coordinate system, register the regional point cloud data with the regional image data, unify them into the registration coordinate system, obtain the initial three-dimensional model of the transmission line, and spatially align data from different sources so that they can complement and fuse each other in subsequent processing to improve the accuracy of three-dimensional reconstruction. Construct a key component set of the transmission line, perform geometric identification and segmentation on the initial transmission line 3D model based on the key component set, and obtain multiple key component models of the target transmission line area, where each key component model has a category label and spatial position information. By constructing a key component set and performing geometric identification and segmentation on the initial model based on the component set, the key components of the transmission line, such as towers, insulators, and conductors, are accurately located and extracted, laying the foundation for subsequent fine modeling. Extract multiple sets of binocular stereo images, optimize multiple key component models respectively, obtain multiple optimized component models, and optimize the initial transmission line 3D model in combination with the category label and spatial position information to obtain an accurate transmission line 3D model. Optimize the key component model by extracting binocular stereo images, and integrate the optimized component model into the initial transmission line 3D model, effectively improving the local and overall modeling accuracy, and obtaining a more realistic and accurate transmission line 3D model.

[0055] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0056] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to multiple elements that can be selected individually or in full. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these changes and variations.

Claims

1. A method for power transmission line reconstruction combining multi-parallel laser lines, characterized in that: include: Determine a target transmission line area, and use a multi-parallel line laser scanner to perform a three-dimensional scan on the target transmission line area to obtain regional point cloud data, wherein the multi-parallel line laser scanner includes at least two mutually parallel laser beams; Using a binocular camera, taking multiple groups of binocular stereo images of the target power transmission line area to obtain regional image data; Establishing a registration coordinate system, registering the regional point cloud data with the regional image data, unifying them into the registration coordinate system, and obtaining an initial three-dimensional model of the transmission line; Constructing a key component set of the transmission line, and geometrically identifying and segmenting the initial three-dimensional transmission line model based on the key component set to obtain multiple key component models of the target transmission line area, wherein each key component model has a category label and spatial location information; Extract multiple groups of binocular stereo images, optimize multiple key component models respectively, obtain multiple optimized component models, and optimize the initial transmission line three-dimensional model in combination with the category label and the spatial position information to obtain an accurate transmission line three-dimensional model.

2. The method for power transmission line reconstruction combined with multi-parallel line lasers according to claim 1, characterized in that: The method comprises: Acquire spatial data information of the target transmission line area, and generate a spatial reference coordinate system based on the spatial data information; Generate a laser scanning route and an image acquisition route based on the spatial reference coordinate system; According to the laser scanning route, controlling the multi-parallel line laser scanner to perform three-dimensional scanning on the target power transmission line area to obtain regional point cloud data; According to the image acquisition route, the binocular camera is controlled to acquire images of the target power transmission line area to obtain regional image data.

3. The method for power transmission line reconstruction combined with multi-parallel line lasers according to claim 2, characterized in that: Establishing a registration coordinate system, registering the regional point cloud data with the regional image data, unifying them into the registration coordinate system, and obtaining an initial three-dimensional model of the transmission line, including: An interactive model building platform is used to establish a registration coordinate system in the model building platform according to the spatial reference coordinate system; Based on the mapping relationship between the spatial reference coordinate system and the registration coordinate system, a first fitting route is generated according to the laser scanning route, and a second fitting route is generated according to the image acquisition route; Fitting the regional point cloud data into the registration coordinate system according to the first fitting route to obtain a first transmission line three-dimensional model; The regional image data is fitted into the first power transmission line three-dimensional model according to the second fitting route to obtain an initial power transmission line three-dimensional model.

4. The method for power transmission line reconstruction combined with multi-parallel line lasers according to claim 1, characterized in that: Construct a set of key components of the transmission line, perform geometric identification and segmentation on the initial three-dimensional model of the transmission line based on the set of key components, and obtain multiple key component models of the target transmission line area, wherein each key component model has a category label and spatial location information, including: Based on the key component set, construct a component geometry segmentor, wherein the component geometry segmentor includes a component geometry recognition branch and a component geometry segmentation branch; Performing geometric recognition on the initial three-dimensional model of the transmission line through the component geometric recognition branch to obtain key component recognition results, wherein the key component recognition results include category labels; According to the key component identification result, the initial transmission line three-dimensional model is segmented through the component geometric segmentation branch to obtain multiple key component models, and the spatial position information of each key component model in the initial transmission line three-dimensional model is recorded during the model segmentation.

5. The method for power transmission line reconstruction combined with multi-parallel laser lines according to claim 4, characterized in that: Also includes: An interactive power transmission line database, based on which a three-dimensional power transmission line sample set is constructed, comprising a plurality of three-dimensional power transmission line samples; Based on the key component set, key component identification and segmentation are performed on the plurality of three-dimensional samples of the transmission lines to obtain a three-dimensional segmentation sample set of the transmission lines; The component geometry recognition branch is acquired through training according to the three-dimensional sample set of the transmission line and the three-dimensional segmentation sample set of the transmission line.

6. The method for power transmission line reconstruction combined with multi-parallel line lasers according to claim 1, characterized in that: Extracting multiple sets of binocular stereo images, optimizing multiple key component models respectively, obtaining multiple optimized component models, and optimizing the initial transmission line three-dimensional model in combination with the category label and the spatial position information to obtain an accurate transmission line three-dimensional model, including: Positioning a plurality of key component models in the initial three-dimensional model of the power transmission line according to the spatial position information; Extracting a plurality of local image blocks corresponding to a plurality of key component models from the plurality of groups of binocular stereo images; For each key component model, texture mapping is performed using the corresponding local image block to obtain a key component model with realistic texture as an optimized component model; According to the category labels and the spatial position information, a plurality of optimized component models are integrated into the initial three-dimensional model of the transmission line to obtain an accurate three-dimensional model of the transmission line.

7. The method for power transmission line reconstruction combined with multi-parallel laser lines according to claim 6, characterized in that: For each key component model, texture mapping is performed using the corresponding local image block to obtain a key component model with realistic texture as the optimized component model, including: Performing distortion correction and geometric correction on the local image block to obtain a corrected image block; Using the corrected image blocks to perform texture mapping on the surface of the key component model to obtain an optimized component model; Among them, UV mapping is established between the mesh patches of the key component model and the corrected image blocks.

8. The method for power transmission line reconstruction combined with multi-parallel laser lines according to claim 6, characterized in that: According to the category label and the spatial position information, multiple optimized component models are integrated into the initial three-dimensional model of the transmission line to obtain an accurate three-dimensional model of the transmission line, including: Traversing multiple key component models, acquiring a first key component model, and extracting spatial position information of the first key component model as first spatial position information; Retrieving a first optimized component model corresponding to the first key component model; Integrating the first optimized component model into the initial transmission line three-dimensional model according to the first spatial position information; The model integration is performed iteratively to obtain an accurate 3D model of the transmission line.

Images