A three-dimensional model construction method and system for topographic surveying

Abstract

The application provides a three-dimensional model construction method and system for topographic survey, and belongs to the technical field of model construction. The method comprises the following steps: segmenting satellite images of a target survey area based on the complexity of the terrain to obtain complex terrain areas and open terrain areas; obtaining laser scanning data of the complex terrain areas and low-altitude image data of the open terrain areas; fusing the laser scanning data and the low-altitude image data to obtain a target data set corresponding to the target survey area; constructing a regular grid model corresponding to the target survey area based on the target data set; constructing an irregular triangular grid model corresponding to the target survey area based on the laser scanning data; and fusing the regular grid model and the irregular triangular grid model to obtain a three-dimensional model of the target survey area. The three-dimensional model construction method and system for topographic survey can improve the modeling accuracy.

Description

Technical Field

[0001] This application belongs to the field of model building technology, and more specifically, relates to a method and system for building three-dimensional models for terrain surveying. Background Technology

[0002] In the field of topographic surveying, 3D modeling technology is widely used in engineering construction, resource exploration, environmental monitoring, and other scenarios. Its core requirement is to accurately reconstruct topographic features through digital means. Traditional topographic modeling methods typically employ a single data acquisition approach, such as using satellite remote sensing data or UAV aerial photography data for the entire area, which makes it difficult to account for the varying complexity of different terrains. The resulting 3D models are prone to losing details, and the overall modeling accuracy fails to meet the requirements of high-precision engineering projects. Summary of the Invention

[0003] The purpose of this application is to provide a method and system for constructing three-dimensional models for terrain surveying, so as to improve modeling accuracy.

[0004] A first aspect of this application provides a method for constructing a three-dimensional model for terrain surveying, comprising:

[0005] Satellite images of the target survey area are segmented based on the complexity of the terrain to obtain complex terrain areas and open terrain areas;

[0006] Acquire laser scanning data of complex terrain areas and low-altitude image data of open terrain areas; the laser scanning data is based on data collected by a ground-based 3D laser scanner, and the low-altitude image data is collected by a drone equipped with a camera;

[0007] By fusing laser scanning data and low-altitude image data, a target dataset corresponding to the target survey area is obtained.

[0008] A regular grid model corresponding to the target survey area is constructed based on the target dataset; an irregular triangular grid model corresponding to the target survey area is constructed based on laser scanning data.

[0009] By fusing the regular grid model and the irregular triangular grid model, a three-dimensional model of the target survey area is obtained.

[0010] A second aspect of this application provides a three-dimensional model construction system for terrain surveying, comprising:

[0011] The image segmentation module is used to segment satellite images of the target survey area based on the complexity of the terrain, resulting in complex terrain areas and open terrain areas;

[0012] The data acquisition module is used to acquire laser scanning data of complex terrain areas and low-altitude image data of open terrain areas; the laser scanning data is based on data collected by a ground-based 3D laser scanner, and the low-altitude image data is collected by a drone equipped with a camera;

[0013] The data fusion module is used to fuse laser scanning data and low-altitude image data to obtain the target dataset corresponding to the target survey area.

[0014] The model building module is used to construct a regular grid model corresponding to the target survey area based on the target dataset; and to construct an irregular triangular grid model corresponding to the target survey area based on laser scanning data.

[0015] The model fusion module is used to fuse regular grid models and irregular triangular grid models to obtain a three-dimensional model of the target survey area.

[0016] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for constructing a three-dimensional model for terrain surveying.

[0017] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method for constructing a three-dimensional model for terrain surveying.

[0018] The beneficial effects of the 3D model construction method and system for terrain surveying provided in this application are as follows: First, this application first divides the region based on the complexity of the terrain and matches the acquisition technology. For complex terrain, high-density point cloud data is obtained by ground 3D laser scanning to ensure accurate capture of detailed features such as steep mountains and dense building clusters. For open terrain, low-altitude aerial photography by UAVs is used, which reduces the data acquisition cost of large flat areas due to the high efficiency of coverage. Second, multi-source data errors are eliminated through data fusion to form a unified target dataset, laying a foundation for data consistency for subsequent modeling. Furthermore, regular grid models and irregular triangular grid models are constructed in a targeted manner. The former efficiently presents the macro trend of open terrain, while the latter accurately fits the details of complex terrain. Finally, the advantages of both are integrated through model fusion. Therefore, this application, through the approach of layered acquisition and accurate modeling, not only ensures the modeling accuracy of complex areas but also improves the modeling efficiency of open areas. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating a method for constructing a three-dimensional model for terrain surveying, provided in an embodiment of this application;

[0021] Figure 2 A structural block diagram of a three-dimensional model construction system for terrain surveying provided in one embodiment of this application;

[0022] Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0023] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0025] Please refer to Figure 1 , Figure 1 This application provides a flowchart illustrating a method for constructing a three-dimensional model for terrain surveying, which can be executed by an electronic device. The method may include:

[0026] S101: Based on the complexity of the terrain, satellite images of the target survey area are segmented to obtain complex terrain areas and open terrain areas.

[0027] In this embodiment, topographic surveying refers to the process of measuring, collecting, and analyzing the topography and geomorphic features (such as elevation, slope, aspect, and topographic relief) of the Earth's surface using technical means. This is used to obtain spatial information about the topography and provide data support for engineering construction, resource surveys, and environmental monitoring. Three-dimensional model construction refers to the process of using the collected topographic data and computer technology to construct a digital model that intuitively reflects the three-dimensional spatial morphology of the topography. The three-dimensional model can visually present the topographic relief, spatial distribution, and other features of the topography.

[0028] The target survey area refers to the specific geographical area where topographic surveying and 3D model construction are required. Its scope can be determined according to actual needs (such as engineering areas, research areas, etc.).

[0029] Satellite images are images of the Earth's surface taken by remote sensing equipment carried by artificial satellites. They contain information such as the macroscopic topography and distribution of land features in a region and can serve as basic data for initially classifying the complexity of the terrain.

[0030] Complex terrain areas are those with large topographic relief and complex landform features, such as mountains, canyons, hills, rocky areas, and densely built-up areas. These types of areas require more precise detailed data to accurately represent their terrain features.

[0031] Open terrain areas refer to areas with small topographic relief, relatively flat landforms, and large areas, such as plains, plateaus, large areas of farmland, and open spaces. Macro-topographic information can be obtained for these areas through efficient data collection methods.

[0032] In this embodiment, the satellite image of the target survey area is first segmented based on the complexity of the terrain, dividing the area into complex terrain areas and open terrain areas. The purpose is to select the most suitable data acquisition method according to the characteristics of different terrains. Complex terrain (such as mountains, canyons, building complexes, etc.) requires higher precision detailed data, while open terrain (such as plains, large open areas, etc.) can acquire data in a more efficient way, thereby optimizing acquisition costs and efficiency while ensuring accuracy.

[0033] S102: Acquire laser scanning data of complex terrain areas and low-altitude image data of open terrain areas; the laser scanning data is based on data collected by a ground-based 3D laser scanner, and the low-altitude image data is collected by a drone equipped with a camera.

[0034] In this embodiment, the laser scanning data is point cloud data generated by emitting a laser beam and receiving the reflected signal through a ground-based 3D laser scanner, and then calculating the laser propagation distance. Each point in the point cloud contains 3D coordinate (X, Y, Z) information, which can accurately reflect the detailed features of the terrain surface and has the characteristics of high density and high precision.

[0035] A terrestrial 3D laser scanner is a device that uses laser pulses to measure distance and quickly acquire 3D coordinate information of a target surface. It can collect large amounts of point cloud data in a short time.

[0036] Low-altitude image data is ground image data taken at low altitudes (usually within a few hundred meters of the ground) by cameras carried on low-altitude flight platforms such as drones. It contains information on the texture, color, and spatial distribution of the terrain. The elevation and texture models of the terrain can be generated through image matching, stitching, and 3D reconstruction techniques.

[0037] A camera-equipped drone is an unmanned aerial vehicle that is equipped with optical cameras (such as RGB cameras or multispectral cameras) and flies at low altitudes to take pictures by means of preset routes or manual control.

[0038] In this embodiment, a terrestrial 3D laser scanner is used to collect laser scanning data for complex terrain areas. Laser scanning technology can accurately capture detailed features of the terrain (such as undulations, rock textures, and terrain outlines under vegetation cover) through high-density point cloud data, making it suitable for high-precision data acquisition in complex environments.

[0039] For open terrain areas, drones equipped with cameras are used to collect low-altitude image data. Drone aerial photography is characterized by its wide coverage, flexible operation, and high data acquisition efficiency, making it suitable for collecting terrain data in large open areas. By stitching and processing the images, terrain information for the area can be generated.

[0040] S103: Perform data fusion on laser scanning data and low-altitude image data to obtain the target dataset corresponding to the target survey area.

[0041] In this embodiment, data fusion refers to the process of integrating, processing, and optimizing information from different data sources (such as laser scanning data and low-altitude image data) to eliminate redundancy and conflicts between data and generate a more complete, accurate, and consistent dataset, with the aim of improving data quality and usability.

[0042] The target dataset refers to the unified dataset obtained after data fusion, which is used for subsequent 3D model construction and contains all the necessary topographic spatial information within the target survey area.

[0043] In this embodiment, laser scanning data (details of complex terrain) and low-altitude image data (macroscopic information of open terrain) are fused to eliminate errors between different data sources (such as differences in coordinate systems and scales), integrating the two types of data into a unified target dataset. This ensures the consistency and integrity of the data during subsequent modeling, laying the foundation for constructing a 3D model of the entire region.

[0044] S104: Construct a regular grid model corresponding to the target survey area based on the target dataset; construct an irregular triangular grid model corresponding to the target survey area based on laser scanning data.

[0045] In this embodiment, the regular grid model is a digital model that represents the terrain surface with regularly arranged grid points (such as square grids). Each grid point corresponds to an elevation value, and the elevation changes of the grid points reflect the terrain undulations.

[0046] The Triangulated Irregular Network (TIN) model is a model that uses irregularly distributed discrete points (point cloud data) as vertices, forming a triangular network by connecting adjacent points to fit the terrain surface. The shape and size of the triangles adaptively adjust according to the complexity of the terrain, accurately representing detailed changes in terrain (such as steep cliffs and valleys), making it suitable for modeling complex terrain.

[0047] In this embodiment, a regular grid model (such as a digital elevation model) is constructed based on the target dataset. The regular grid model represents terrain elevation using uniformly distributed grid points; it is simple in structure, computationally efficient, and suitable for representing the overall trend and macroscopic features of open terrain. An irregular triangular grid model is constructed based on laser scanning data. The TIN model fits the terrain surface using irregularly distributed triangle vertices (point cloud data), better adapting to the undulations and detailed changes in complex terrain, and accurately representing the local features of the terrain.

[0048] S105: The regular grid model and the irregular triangular grid model are fused to obtain a three-dimensional model of the target survey area.

[0049] In this embodiment, a regular grid model (macroscopic features of open terrain) and an irregular triangular grid model (detailed features of complex terrain) are fused together. By combining the advantages of the two models, a high-precision 3D model that covers the target survey area and has both macroscopic trends and microscopic details is finally obtained.

[0050] For example, a highway project requires 3D terrain modeling of the transition zone between mountainous and plain areas to provide data support for route design and construction planning. The implementation process is as follows:

[0051] First, satellite remote sensing images of the area were acquired. Then, terrain analysis algorithms were used to calculate indicators such as slope and undulation, dividing the area into a complex terrain zone (mountainous section, approximately 20 square kilometers) and an open terrain zone (plain section, approximately 50 square kilometers). The complex terrain zone includes canyons, steep slopes, and rock formations, while the open terrain zone consists of river valleys, plains, and farmland.

[0052] To address the complex terrain in mountainous areas, a terrestrial 3D laser scanner was used to collect point cloud data along the survey route. With a scanning interval of 5 meters, laser point cloud data with a density of 200 points / m² was obtained, accurately recording details such as cliff height and gully depth. Simultaneously, in open plain areas, a drone equipped with an RGB camera was used to capture 1200 low-altitude images at a flight altitude of 100 meters and an 80% overlap flight path, generating a DSM (Digital Surface Model) for the region.

[0053] In the data processing stage, the laser point cloud and UAV image data were first unified to the 2000 National Geodetic Coordinate System. Errors were eliminated through coordinate transformation and registration, and the data were merged to form a target dataset covering 70 square kilometers. Based on this dataset, a regular grid DEM with a resolution of 10 meters was constructed to present the overall topographic trend of the plain area. A TIN model was constructed using the laser point cloud data, and a triangular network was used to fit the complex terrain surface of the mountainous area.

[0054] Finally, through model fusion algorithms, the efficient representation of regular grids is preserved in plain areas, while detailed features of TINs are replaced in mountainous areas, generating a 3D model with both macroscopic accuracy and microscopic precision. The model elevation error is controlled within 0.5 meters, providing reliable terrain data support for highway alignment and slope protection design.

[0055] As can be seen from the above, this embodiment first segments the region based on the complexity of the terrain and matches the acquisition technology accordingly. For complex terrain, terrestrial 3D laser scanning is used to acquire high-density point cloud data, ensuring accurate capture of detailed features such as steep mountains and dense building clusters. For open terrain, low-altitude aerial photography by drones is used, leveraging the advantage of efficient coverage to reduce the data acquisition cost for large, flat areas. Secondly, data fusion eliminates errors from multi-source data, forming a unified target dataset and laying a foundation for data consistency in subsequent modeling. Furthermore, regular grid models and irregular triangular grid models are specifically constructed. The former efficiently presents the macroscopic trend of open terrain, while the latter accurately fits the details of complex terrain. Finally, the advantages of both are integrated through model fusion. This layered acquisition and precise modeling approach ensures the modeling accuracy of complex areas while improving the modeling efficiency of open areas.

[0056] In one embodiment of this application, satellite images of the target survey area are segmented based on the complexity of the terrain to obtain complex terrain areas and open terrain areas, including:

[0057] Feature extraction is performed on satellite images of the target survey area to obtain feature parameters related to the complexity of the terrain, including elevation gradient features, surface texture features, and land cover distribution features at different spatial resolutions;

[0058] The elevation gradient features, surface texture features, and land cover distribution features are input into a preset terrain complexity assessment model to obtain the terrain complexity score for each pixel in the satellite image.

[0059] The terrain complexity score of each pixel is binarized to obtain complex terrain regions and open terrain regions.

[0060] In this embodiment, multiple features are extracted from satellite images, focusing on three key parameters directly related to the complexity of the terrain: elevation gradient features (reflecting the intensity of terrain undulations; gradient differences at different spatial resolutions can reflect macroscopic and microscopic terrain changes), surface texture features (reflecting the degree of surface fragmentation through indicators such as texture roughness and directional consistency; for example, the texture of rocky areas is more chaotic than that of plains), and land cover distribution features (identifying the spatial distribution density of buildings, vegetation, water bodies, etc.; densely distributed areas usually correspond to higher complexity).

[0061] After feature extraction, the extracted multidimensional features are input into a preset terrain complexity assessment model. The model generates a quantified terrain complexity score for each pixel of the satellite image through algorithm calculation, realizing the transformation from image features to complexity values, making terrain complexity comparable.

[0062] Finally, the continuous scoring results are transformed into discrete region divisions through binarization: a scoring threshold is set, pixels above the threshold are classified as complex terrain regions, and pixels below the threshold are classified as open terrain regions, thereby completing region segmentation based on pixel-level accuracy and providing accurate region division basis for subsequent targeted data collection.

[0063] For example, in a topographic survey project at the boundary between a mountainous area and a plain, multi-feature extraction and model evaluation were used to achieve regional segmentation. First, 2-meter resolution satellite images of the area were acquired, and three types of features were extracted using remote sensing processing tools: the elevation gradient values ​​at 10-meter and 50-meter resolutions were calculated to obtain micro and macro data on topographic relief; surface texture features were extracted using a gray-level co-occurrence matrix to quantify texture roughness and directional consistency; and buildings, forests, water bodies, etc., were labeled through image recognition, and the distribution density of ground features was statistically analyzed.

[0064] Three types of feature parameters were input into a pre-trained terrain complexity assessment model (built based on the random forest algorithm). The model scored each pixel in the image and output a complexity score from 0 to 100. A threshold of 60 was set, and the scoring results were binarized: pixel regions with scores ≥60 (such as canyons and steep slopes / woodlands) were classified as complex terrain areas, covering approximately 18 square kilometers; pixel regions with scores <60 (such as river valleys / plains and open farmland) were classified as open terrain areas, covering approximately 42 square kilometers. The segmentation results were verified through field sampling, achieving a region classification accuracy of 92%, providing a precise basis for the subsequent data collection plan.

[0065] As can be seen from the above, this embodiment extracts multi-dimensional terrain features, combines them with a preset evaluation model to generate pixel-level complexity scores, and then accurately divides the region through binarization. It comprehensively reflects terrain differences through elevation gradients, textures, and land cover features, and achieves objective classification through quantitative scoring and threshold division, improving the accuracy and efficiency of region segmentation. This provides a precise basis for subsequent targeted data collection, avoiding resource waste and insufficient accuracy.

[0066] In one embodiment of this application, a method for constructing a terrain complexity assessment model includes:

[0067] Collect satellite image samples of multiple regions with known terrain types, label each sample as either a complex terrain sample or an open terrain sample, and label each sample with its corresponding terrain type label;

[0068] Feature extraction is performed on each satellite image sample to obtain the sample's elevation gradient features, surface texture features, and ground feature distribution features, and to construct a sample feature set;

[0069] The sample feature set is divided into a training set and a test set. The training set is input into the initial machine learning model for training. The loss function is optimized by adjusting the model parameters to obtain a preliminary terrain complexity assessment model.

[0070] The preliminary terrain complexity assessment model is validated using a test set. The classification accuracy of the model is calculated. If the accuracy does not reach the preset threshold, retraining is performed.

[0071] When the model's classification accuracy reaches a preset threshold, the final terrain complexity assessment model is output.

[0072] In this embodiment, satellite image samples of known terrain types are collected and labeled to provide a "standard answer" for model training. The samples need to cover areas with different terrain features to ensure that the model has generalization ability, including complex terrain samples such as mountains and canyons, as well as open terrain samples such as plains and farmland.

[0073] Three core features related to terrain complexity (elevation gradient, surface texture, and land cover distribution) are extracted to construct a feature set, transforming image information into quantifiable data that the model can recognize. These features reflect terrain complexity from different dimensions and together form the basis for the model's judgment.

[0074] Machine learning algorithms (such as random forests and neural networks) are used to learn from the training set, and the model's ability to determine terrain type is optimized by continuously adjusting parameters. The optimization process of the loss function enables the model to gradually master the mapping relationship between features and terrain type.

[0075] The model's classification accuracy is validated using a test set. If the accuracy is not met, iterative optimization is performed to ensure the model maintains stable evaluation capabilities even on unknown samples. The final model can convert input terrain feature parameters into a quantified complexity score, providing an accurate basis for terrain region segmentation.

[0076] As can be seen from the above, this embodiment constructs a dataset through labeled samples and multi-feature extraction, and improves the model accuracy through training, optimization, and validation iterations. Supervised learning enables the model to accurately capture the correlation between terrain features and complexity, and the output evaluation model can objectively quantify terrain complexity, avoiding the subjectivity of manual segmentation. This provides reliable model support for accurate terrain region segmentation and improves the efficiency of terrain surveying.

[0077] In one embodiment of this application, laser scanning data and low-altitude image data are fused to obtain a target dataset corresponding to the target survey area, including:

[0078] Based on the laser scanning data, data points with reflection intensity lower than a preset threshold are removed to obtain the target laser scanning data;

[0079] The coordinates of the low-altitude image data are normalized to obtain the target low-altitude image data;

[0080] Establish a spatial correlation mapping between target laser scanning data and target low-altitude image data;

[0081] Based on spatial correlation mapping, the target laser scanning data and the target low-altitude image data are fused to obtain the target dataset corresponding to the target survey area.

[0082] In this embodiment, the laser scanning data may contain noise points with extremely low reflection intensity (such as impurities in the air or anomalies caused by measurement errors). By setting a preset threshold for reflection intensity, data points below this threshold are discarded, and valid terrain and feature data with reflection intensity meeting the requirements are retained to obtain the target laser scanning data. This step can reduce noise interference and ensure the accuracy of the laser data.

[0083] Low-altitude image data (such as drone aerial images) may be acquired based on different coordinate systems, resulting in scale and positional biases. By performing coordinate normalization processing (such as uniformly converting to the UTM coordinate system or the local projected coordinate system), the coordinate differences between different images are eliminated, aligning all low-altitude image data under the same spatial reference system, thus obtaining the target low-altitude image data and laying the foundation for subsequent spatial correlation.

[0084] In this embodiment, based on the preprocessed target laser scanning data and target low-altitude image data, a spatial positional correspondence between the two is established through a spatial matching algorithm (such as matching common ground feature points, coordinate reference points, or corresponding points). This spatial correlation mapping clarifies the spatial positional association between each data point in the laser point cloud and the corresponding pixel in the low-altitude image, thus solving the spatial alignment problem between the two types of data.

[0085] Based on the established spatial correlation mapping, target laser scanning data (providing high-precision 3D terrain structure and distance information) is fused with target low-altitude image data (providing rich texture, color, and ground feature details). During the fusion process, the spatial information of the two types of data complements each other, ultimately generating a target dataset that simultaneously contains 3D structure and texture details, providing comprehensive data support for subsequent terrain analysis and ground feature identification.

[0086] For example, in a mining area topographic survey project, laser scanning data and low-altitude image data are fused. First, the laser scanning data is processed, and the reflection intensity threshold is set to 30 (0-255 gray value). Low reflection points caused by dust, birds, etc. in the air (about 5% of the data) are removed to obtain the target laser point cloud including the mine slope and stockpile.

[0087] The coordinates of 200 low-altitude images collected by the UAV were normalized and uniformly converted to the Beijing 54 coordinate system to eliminate coordinate deviations caused by flight batches, generating target low-altitude image data. The SIFT algorithm was used to identify common feature points (such as mine towers and road bends) in both types of data, establishing a spatial correlation mapping with matching errors controlled within 0.3 meters.

[0088] Data fusion based on mapping relationships: The depth information of the image is enhanced by the 3D coordinates of the laser point cloud, and the visual features of the point cloud are enriched by the texture and color of the image. Finally, a target dataset containing a 12 square kilometer mining area is generated, which retains the 0.1-meter elevation accuracy of the laser point cloud and has the true color texture of the image, providing a high-quality data foundation for subsequent 3D modeling.

[0089] As can be seen from the above, this embodiment improves data accuracy by removing laser noise points, eliminates image coordinate deviations through coordinate normalization, and achieves precise alignment through spatial correlation mapping. The fused data combines the high-precision 3D structure of laser point clouds with the rich texture details of low-altitude images, providing a complete and reliable target dataset for subsequent modeling, effectively improving terrain data quality and fusion efficiency.

[0090] In one embodiment of this application, a regular grid model and an irregular triangular grid model are fused to obtain a three-dimensional model of the target survey area, including:

[0091] Calculate the elevation deviation between each grid point in the regular grid model and the adjacent triangle vertices in the irregular triangular grid model;

[0092] Fusion rules are constructed based on elevation deviation values, and elevation data of the target model are selected as target data based on the fusion rules; the target model is a regular grid model or an irregular triangular grid model.

[0093] By fusing the target data, a three-dimensional model of the target survey area is obtained.

[0094] In this embodiment, the elevation deviation value is calculated to establish a quantitative evaluation basis. For each grid point in the regular grid model, its corresponding adjacent triangle vertices in the irregular triangular grid model (usually the 3-4 closest vertices) are located. The elevation difference between the two models is calculated through coordinate matching to obtain the elevation deviation value of each grid point. This step transforms the difference between the two models into a quantifiable numerical value, providing data support for subsequent fusion decisions.

[0095] Secondly, intelligent selection is achieved by constructing fusion rules based on deviation values. Judgment criteria are set according to the magnitude of the elevation deviation value: when the deviation value is less than a preset threshold (e.g., 0.5 meters), it indicates that the elevation data of the two models in this area are highly consistent, and the regular grid model data can be selected (balancing efficiency and accuracy); when the deviation value is greater than or equal to the threshold, it indicates that there are significant differences in detail in complex terrain areas, and the irregular triangular grid model data is prioritized (preserving high-precision details). The core of the fusion rules is to allow each area to automatically adapt to a better data source. Finally, model fusion is completed based on the target data. Elevation data for the entire area is filtered and integrated according to the fusion rules. In open terrain areas, the efficient representation of the regular grid is retained, while in complex terrain areas, it is replaced with the accurate data of the irregular triangular grid. Simultaneously, an edge smoothing algorithm is used to eliminate transition errors at the splicing points of different models, ultimately generating a 3D model that combines macroscopic trends and microscopic details, ensuring both overall modeling efficiency and improving accuracy in complex areas.

[0096] For example, in a terrain modeling project along a mountain highway, it is necessary to integrate a 5-meter resolution regular grid DEM with a TIN model generated from laser point clouds. The specific steps are as follows:

[0097] First, the elevation deviation between each grid point in the DEM and the three nearest triangle vertices in the TIN is calculated. In gentle intermontane basins, the elevation difference between the grid point and the TIN vertices is mostly between 0.2 and 0.3 meters; while in sharp bends and steep slopes, the deviation generally reaches 0.8 to 1.5 meters, and even exceeds 2 meters in some rocky slopes.

[0098] A fusion threshold of 0.5 meters is set: for basin areas with a deviation of less than 0.5 meters, DEM data is retained to ensure modeling efficiency; for steep slopes and cliff areas with a deviation of 0.5 meters or more, TIN data is used to preserve terrain details.

[0099] During the fusion process, a weighted average algorithm was used for the transition zone (approximately 15 meters wide) between the two models: points closer to the DEM area were assigned a higher DEM weight, and points closer to the TIN area were assigned a higher TIN weight, ensuring a smooth and continuous change in edge elevation. In the final generated 3D model, open areas retain the uniform grid characteristics of the DEM, while complex road sections preserve the precise details of the TIN. Field verification showed that the overall elevation accuracy of the model is within 0.3 meters, meeting both the macro-topographic analysis requirements for highway alignment and providing accurate micro-topographic data for slope protection design.

[0100] As can be seen from the above, this embodiment constructs fusion rules by calculating elevation deviation values, intelligently selecting better elevation data. When the deviation is small, a rule-based grid is selected to ensure efficiency; when the deviation is large, an irregular triangular grid is used to preserve details. After smoothing processing to eliminate splicing errors, the final model has both macroscopic trends and microscopic accuracy, improving the modeling quality.

[0101] In one embodiment of this application, the fusion rule includes:

[0102] If the elevation deviation is less than the preset deviation value, the elevation data of the regular grid model will be used as the target data.

[0103] In response to an elevation deviation value greater than or equal to a preset deviation value, the elevation data of the irregular triangular grid model is used as the target data.

[0104] In this embodiment, a preset deviation value is used as the judgment standard. The preset deviation value can be set according to the survey accuracy requirements and reflects the acceptable error range of the terrain data. When the elevation deviation between the grid points of the regular grid model and the adjacent vertices of the irregular triangular grid model is less than the preset threshold, it indicates that the elevation data of the two types of models are highly consistent in this area, and the terrain features are relatively gentle (such as open plains). At this time, the regular grid model data is selected as the target data because of its regular structure, high computational efficiency, and ability to efficiently present the macroscopic terrain trend while avoiding data redundancy.

[0105] When the elevation deviation is greater than or equal to a preset threshold, it indicates that the terrain in the area is complex (such as mountains or canyons). Regular grid models, due to their fixed resolution, struggle to accurately capture details, while irregular triangular grid models, through adaptive triangular networks, can better fit the terrain undulations. In this case, selecting irregular triangular grid model data can preserve high-precision details of the complex terrain, ensuring the accuracy of modeling key areas.

[0106] As can be seen from the above, this embodiment dynamically selects the optimal elevation data by setting a preset deviation threshold: when the deviation is small, a regular grid is used to maintain efficiency, and when the deviation is large, an irregular triangular grid is used to maintain detail. This leverages both the efficiency of the former and the accuracy of the latter, achieving a balance between efficiency and precision, and improving the overall quality and applicability of the 3D model.

[0107] In one embodiment of this application, the method further includes:

[0108] Obtain measured elevation data of a preset number of ground control points within the target survey area; the measured elevation data of the ground control points are the elevation values ​​of specific points collected on-site within the target survey area;

[0109] Calculate the deviation between the elevation values ​​of the corresponding ground control points in the 3D model and the measured elevation data;

[0110] If the deviation value is greater than the preset accuracy threshold, the area corresponding to the deviation value is located, and the elevation data of the target model in the area is corrected based on the measured elevation data of the ground control points to obtain the adjusted regular grid model or irregular triangular grid model.

[0111] The target model is either a regular grid model or an irregular triangular grid model.

[0112] In this embodiment, a predetermined number of ground control points are selected within the target survey area (these points should uniformly cover complex and open terrain, with denser coverage in key areas). Precise elevation values ​​of these points are collected in the field using high-precision measurement techniques (such as total stations and GNSS RTK) to form measured elevation data. This data serves as a baseline, providing an objective reference standard for model accuracy evaluation and ensuring the reliability of the verification results.

[0113] The calculated elevation values ​​(elevation data from regular grid models or irregular triangular grid models) of the corresponding ground control points in the 3D model are compared with the measured elevation data. Coordinate matching is used to locate the corresponding position of each control point in the model, and the elevation difference between the two is calculated to obtain the deviation value. The model error is then converted into a quantifiable numerical value, clarifying the magnitude and distribution of the error.

[0114] Set a preset accuracy threshold (e.g., 0.3 meters as required by engineering). When the calculated deviation value exceeds this accuracy threshold, the model accuracy of the corresponding area is deemed substandard. Locate the specific geographical area corresponding to the deviation value using spatial coordinates to determine the range that needs correction (e.g., a steep slope or canyon area), focusing on the source of error and avoiding indiscriminate correction that would waste efficiency.

[0115] For areas with positioning deviations, the elevation data of the region is adjusted based on the original model type used (regular grid model or irregular triangular grid model) and the measured elevation data of ground control points. If it's a regular grid model, the elevation values ​​of the corresponding grid points are corrected; if it's an irregular triangular grid model, the elevation parameters of the triangle vertices are adjusted. The correction process maintains smooth transitions at the model edges, ultimately resulting in an adjusted model with acceptable accuracy, achieving precise elimination of local errors.

[0116] For example, in a water conservancy project topographic modeling project, measured data from ground control points were used to optimize the accuracy of the 3D model. Within a 50-square-kilometer survey area, 30 ground control points were evenly distributed, with the density increased to 15 in complex terrain areas (reservoir slopes, canyon sections) and 15 in open terrain areas (reservoir plains). Measured elevation data were collected using GNSSRTK technology, achieving an accuracy of ±0.05 meters.

[0117] The generated 3D model was compared with the control point data to calculate the deviation: the deviation of control points in the plain area was mostly between 0.2 and 0.3 meters, meeting the preset accuracy threshold of 0.3 meters; however, the deviation of three control points on the steep slope section west of the reservoir reached 0.5 to 0.8 meters, exceeding the threshold. The 2-square-kilometer deviation area was located by coordinate positioning. This area was originally constructed using an irregular triangular grid model (TIN).

[0118] During the correction, the measured elevations of three control points were used as a benchmark. The original laser point cloud data for the area was extracted to reconstruct the local TIN model, and the elevation parameters of the triangle vertices were adjusted: the cliff vertices were corrected from the model-calculated 125.6 meters to the measured 126.3 meters, and the slope base vertices were corrected from 110.2 meters to 110.8 meters. A weighted smoothing algorithm was used to transition the corrected area to the surrounding model, eliminating abrupt elevation changes. After correction and review, the deviation of the control points in the area was reduced to within 0.2 meters, and the overall model accuracy was improved to within the 0.3-meter threshold, providing reliable topographic data for the reservoir dam design.

[0119] As can be seen from the above, this embodiment overcomes the limitation of the model relying solely on the original collected data by introducing actual measured data as an absolute reference. Through quantitative evaluation and targeted correction, it significantly improves the accuracy and reliability of the 3D model in key areas, ensuring that the model meets the requirements of engineering applications.

[0120] In one embodiment of this application, the elevation data of the target model in the area is corrected based on the measured elevation data of ground control points to obtain an adjusted regular grid model or an irregular triangular grid model, including:

[0121] Based on the region corresponding to the deviation value, the original laser scanning data of the region is extracted to construct a local regular grid model, and the original low-altitude image data of the region is extracted to construct a local irregular triangular grid model.

[0122] Using the measured data of ground control points as constraints, the elevation data corresponding to the regular grid model is adjusted based on the local regular grid model to obtain the adjusted regular grid model.

[0123] The elevation data corresponding to the irregular triangular grid model is adjusted based on the local irregular triangular grid model to obtain the adjusted irregular triangular grid model.

[0124] In this embodiment, for the region corresponding to the deviation value, laser scanning data and low-altitude image data for that region are extracted from the original data, and local regular grid models and local irregular triangular grid models are constructed respectively. This avoids indiscriminate processing of the entire region's data, and provides basic data support for subsequent corrections through localized refined modeling, ensuring the targeted nature of the corrections.

[0125] Using measured elevation data from ground control points as a hard constraint, the elevation data of the deviation areas in the original regular grid model are adjusted. With the local regular grid model as a reference, corresponding grid points in the original model are located through coordinate matching. Based on the deviation between the measured data and the model's calculated values, the elevation values ​​of the grid points are corrected. During the correction process, the regularity of the grid structure is maintained, ensuring a continuous elevation transition between the adjusted local area and the surrounding grid, ultimately resulting in the adjusted regular grid model.

[0126] Based on a locally irregular triangular grid model, the elevation data of the deviation areas in the original irregular triangular grid model are adjusted. Using the measured data of ground control points as a benchmark, the vertices of the triangles within the deviation areas are located. By adjusting the three-dimensional coordinate parameters of the vertices (with a focus on optimizing the Z-axis elevation value), the terrain fitting accuracy of the local model is made to closely match the measured data. At the same time, the topological structure of the triangular network is kept stable to avoid cracks appearing on the model surface due to vertex adjustments, ultimately generating an adjusted irregular triangular grid model.

[0127] For example, based on the aforementioned water conservancy project topographic modeling project, targeted corrections were made to the steep slope deviation area on the west side of the reservoir. Laser scan point cloud (density 200 points / ㎡) and UAV low-altitude image (resolution 0.1 meters) of this 2 square kilometer area were extracted from the original data, and a local 5-meter resolution regular grid model and a local TIN model were reconstructed.

[0128] Using the measured elevations of three ground control points (126.3 meters, 118.5 meters, and 110.8 meters) as constraints, the regular grid model was first modified: the corresponding grid points in the original model were located by coordinate matching, and the elevations of 28 grid points with deviations exceeding 0.3 meters were adjusted according to the measured data gradient. For example, the elevation of the grid point at (X:35210, Y:42680) was corrected from 125.6 meters to 126.2 meters to ensure that the elevation difference between grids is ≤0.5 meters and to maintain transition continuity.

[0129] Further adjustments were made to the TIN model: For the 12 key triangle vertices within the location deviation area, the elevation of the cliff vertex (X:35205, Y:42675) was adjusted from 125.8 meters to the measured 126.3 meters, and the elevation of the slope bottom vertex (X:35190, Y:42710) was corrected from 110.2 meters to 110.8 meters. The triangle network topology remained unchanged during the adjustment, and a vertex smoothing algorithm was used for the three edge triangles to eliminate abrupt elevation changes. After the correction, the local model deviation from the measured data was ≤0.2 meters, and the edge connection error between the two models was ≤0.1 meters, meeting the engineering accuracy requirements.

[0130] As can be seen from the above, this embodiment focuses on the deviation area through local model reconstruction, and optimizes the elevation data of regular grids and irregular triangular grids respectively, using measured data from ground control points as constraints. This ensures both targeted correction and maintains the stability and edge continuity of the model structure, accurately eliminating local errors and improving the model's elevation accuracy and engineering applicability.

[0131] A method for constructing a three-dimensional model for terrain surveying, corresponding to the above embodiment. Figure 2 This is a structural block diagram of a three-dimensional model construction system for terrain surveying, provided as an embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2 The three-dimensional model construction system 20 for terrain surveying includes: an image segmentation module 21, a data acquisition module 22, a data fusion module 23, a model construction module 24, and a model fusion module 25.

[0132] Among them, the image segmentation module 21 is used to segment the satellite image of the target survey area based on the terrain complexity to obtain complex terrain areas and open terrain areas;

[0133] The data acquisition module 22 is used to acquire laser scanning data of complex terrain areas and low-altitude image data of open terrain areas; the laser scanning data is based on data collected by a ground-based 3D laser scanner, and the low-altitude image data is collected by a drone equipped with a camera;

[0134] Data fusion module 23 is used to fuse laser scanning data and low-altitude image data to obtain the target dataset corresponding to the target survey area;

[0135] Model building module 24 is used to build a regular grid model corresponding to the target survey area based on the target dataset; and to build an irregular triangular grid model corresponding to the target survey area based on laser scanning data.

[0136] The model fusion module 25 is used to fuse the regular grid model and the irregular triangular grid model to obtain a three-dimensional model of the target survey area.

[0137] In one embodiment of this application, the image segmentation module 21 is specifically used for:

[0138] Feature extraction is performed on satellite images of the target survey area to obtain feature parameters related to the complexity of the terrain, including elevation gradient features, surface texture features, and land cover distribution features at different spatial resolutions;

[0139] The elevation gradient features, surface texture features, and land cover distribution features are input into a preset terrain complexity assessment model to obtain the terrain complexity score for each pixel in the satellite image.

[0140] The terrain complexity score of each pixel is binarized to obtain complex terrain regions and open terrain regions.

[0141] In one embodiment of this application, the image segmentation module 21 is further configured to:

[0142] Based on the laser scanning data, data points with reflection intensity lower than a preset threshold are removed to obtain the target laser scanning data;

[0143] The coordinates of the low-altitude image data are normalized to obtain the target low-altitude image data;

[0144] Establish a spatial correlation mapping between target laser scanning data and target low-altitude image data;

[0145] Based on spatial correlation mapping, the target laser scanning data and the target low-altitude image data are fused to obtain the target dataset corresponding to the target survey area.

[0146] In one embodiment of this application, the model fusion module 25 is specifically used for:

[0147] Calculate the elevation deviation between each grid point in the regular grid model and the adjacent triangle vertices in the irregular triangular grid model;

[0148] Fusion rules are constructed based on elevation deviation values, and elevation data of the target model are selected as target data based on the fusion rules; the target model is a regular grid model or an irregular triangular grid model.

[0149] By fusing the target data, a three-dimensional model of the target survey area is obtained.

[0150] In one embodiment of this application, the fusion rule includes:

[0151] If the elevation deviation is less than the preset deviation value, the elevation data of the regular grid model will be used as the target data.

[0152] In response to an elevation deviation value greater than or equal to a preset deviation value, the elevation data of the irregular triangular grid model is used as the target data.

[0153] In one embodiment of this application, a three-dimensional model construction system 20 for terrain surveying further includes: a model correction module, specifically used for:

[0154] Obtain measured elevation data of a preset number of ground control points within the target survey area; the measured elevation data of the ground control points are the elevation values ​​of specific points collected on-site within the target survey area;

[0155] Calculate the deviation between the elevation values ​​of the corresponding ground control points in the 3D model and the measured elevation data;

[0156] If the deviation value is greater than the preset accuracy threshold, the area corresponding to the deviation value is located, and the elevation data of the target model in the area is corrected based on the measured elevation data of the ground control points to obtain the adjusted regular grid model or irregular triangular grid model.

[0157] The target model is either a regular grid model or an irregular triangular grid model.

[0158] In one embodiment of this application, the model correction module is further configured to:

[0159] Based on the region corresponding to the deviation value, the original laser scanning data of the region is extracted to construct a local regular grid model, and the original low-altitude image data of the region is extracted to construct a local irregular triangular grid model.

[0160] Using the measured data of ground control points as constraints, the elevation data corresponding to the regular grid model is adjusted based on the local regular grid model to obtain the adjusted regular grid model.

[0161] The elevation data corresponding to the irregular triangular grid model is adjusted based on the local irregular triangular grid model to obtain the adjusted irregular triangular grid model.

[0162] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of the modules in the aforementioned system embodiments, for example... Figure 2 The functions of the image segmentation module 21, data acquisition module 22, data fusion module 23, model building module 24, and model fusion module 25 are shown.

[0163] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0164] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0165] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store information such as regular grid models and irregular triangular grid models.

[0166] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation method described in the three-dimensional model construction method for terrain surveying provided in the embodiments of this application, or they can execute the implementation method of the electronic device described in the embodiments of this application, which will not be repeated here.

[0167] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to implement these processes. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or system capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0168] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0169] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0170] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0171] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connections shown or discussed may be indirect coupling or communication connections through some interfaces or units, or they may be electrical, mechanical, or other forms of connection.

[0172] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0173] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0174] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method of constructing a three-dimensional model for topographic surveying, characterized by, The method comprises the following steps: segmenting satellite images of a target survey area based on terrain complexity to obtain complex terrain areas and open terrain areas; obtaining laser scanning data of the complex terrain areas and low-altitude image data of the open terrain areas; the laser scanning data is based on data collected by a ground three-dimensional laser scanner, and the low-altitude image data is collected by a camera-carrying unmanned aerial vehicle; performing data fusion on the laser scanning data and the low-altitude image data to obtain a target data set corresponding to the target survey area; constructing a regular grid model corresponding to the target survey area based on the target data set and constructing an irregular triangular grid model corresponding to the target survey area based on the laser scanning data; calculating elevation deviation values of each grid point in the regular grid model and adjacent triangular vertices in the irregular triangular grid model; constructing a fusion rule based on the elevation deviation values and selecting elevation data of a target model as target data based on the fusion rule; the target model is the regular grid model or the irregular triangular grid model; performing fusion based on the target data to obtain a three-dimensional model of the target survey area; wherein the fusion rule comprises: in response to the elevation deviation value being less than a preset deviation value, using the elevation data of the regular grid model as the target data; in response to the elevation deviation value being greater than or equal to the preset deviation value, using the elevation data of the irregular triangular grid model as the target data.

2. The method of claim 1, wherein The method of segmenting satellite images of a target survey area based on terrain complexity to obtain complex terrain areas and open terrain areas comprises: performing feature extraction on satellite images of a target survey area to obtain feature parameters related to terrain complexity, including elevation gradient features, surface texture features and feature distribution features at different spatial resolutions; inputting the elevation gradient features, the surface texture features and the feature distribution features into a preset terrain complexity evaluation model to obtain a terrain complexity score of each pixel in the satellite images; performing binaryzation processing on the terrain complexity score of each pixel to obtain complex terrain areas and open terrain areas.

3. The method of claim 2, wherein the three-dimensional model is constructed by a method comprising: The method of performing data fusion on the laser scanning data and the low-altitude image data to obtain a target data set corresponding to the target survey area comprises: based on the laser scanning data, removing data points with a reflection intensity lower than a preset threshold to obtain target laser scanning data; performing normalization processing on the coordinates of the low-altitude image data to obtain target low-altitude image data; establishing a spatial correlation mapping between the target laser scanning data and the target low-altitude image data; based on the spatial correlation mapping, fusing the target laser scanning data and the target low-altitude image data to obtain a target data set corresponding to the target survey area.

4. The method of claim 1, wherein The method further comprises: obtaining measured elevation data of a preset number of ground control points in the target survey area; the measured elevation data of the ground control points is an elevation value of a specific point collected in the field in the target survey area; calculating deviation values of elevation values of positions of the ground control points in the three-dimensional model and the measured elevation data. In response to the deviation value being greater than a preset accuracy threshold, the area corresponding to the deviation value is located, and the elevation data of the target model in the area is corrected based on the measured elevation data of the ground control points to obtain an adjusted regular grid model or an irregular triangular grid model. The target model is either the regular grid model or the irregular triangular grid model.

5. The method of claim 4, wherein the three-dimensional model is constructed by a method comprising: The elevation data of the target model in the area is corrected based on the measured elevation data of ground control points to obtain an adjusted regular grid model or irregular triangular grid model, including: ​ Based on the region corresponding to the deviation value, the original laser scanning data of the region is extracted to construct a local regular grid model, and the original low-altitude image data of the region is extracted to construct a local irregular triangular grid model. Using the measured data of ground control points as constraints, the elevation data corresponding to the local regular grid model is adjusted based on the local regular grid model to obtain the adjusted regular grid model. The elevation data corresponding to the irregular triangular grid model is adjusted based on the local irregular triangular grid model to obtain the adjusted irregular triangular grid model.

6. A three-dimensional model construction system for topographic survey, characterized by comprising: a three-dimensional model construction device; a three-dimensional model display device; and a three-dimensional model display control device. include: The image segmentation module is used to segment satellite images of the target survey area based on the complexity of the terrain, resulting in complex terrain areas and open terrain areas; The data acquisition module is used to acquire laser scan data of the complex terrain area and low-altitude image data of the open terrain area; the laser scan data is based on data collected by a ground-based three-dimensional laser scanner, and the low-altitude image data is collected by a drone equipped with a camera; The data fusion module is used to fuse the laser scanning data and the low-altitude image data to obtain the target dataset corresponding to the target survey area; The model building module is used to construct a regular grid model corresponding to the target survey area based on the target dataset; and to construct an irregular triangular grid model corresponding to the target survey area based on the laser scanning data. The model fusion module is used to calculate the elevation deviation between each grid point in the regular grid model and the adjacent triangle vertices in the irregular triangular grid model. Based on the elevation deviation value, a fusion rule is constructed, and based on the fusion rule, the elevation data of the target model is selected as the target data. The target model is either the regular grid model or the irregular triangular grid model; Based on the target data, a three-dimensional model of the target survey area is obtained by fusing the data. The fusion rules include: In response to the elevation deviation value being less than a preset deviation value, the elevation data of the regular grid model is used as the target data; In response to the elevation deviation value being greater than or equal to a preset deviation value, the elevation data of the irregular triangular grid model is used as the target data.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, the computer-readable storage medium comprising: When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 5.

Images