A method for selecting points of a survey control network based on real three-dimensional

Abstract

The present application relates to the field of engineering surveying, in particular to a kind of measurement control network selection point method based on real scene three dimensions, comprising the following steps: S1: scene data acquisition and preprocessing, the image data of engineering area is collected using unmanned aerial vehicle oblique photography technology, through oblique model commercial production software, after image aerial triangulation, dense point cloud generation and triangulation network construction and texture mapping, high-precision oblique photography model is generated, based on geological exploration profile or drilling data, three-dimensional geological model is constructed using Kriging interpolation method.The present application constructs integrated real scene three dimensions by fusing above-ground oblique photography model and underground geological model, integrates four core algorithm modules of visual analysis, rock stratum analysis, network type analysis and safety analysis, realizes the visualization and intelligent selection of surveying control point, significantly improves the efficiency and scientificity of point selection, and reduces the cost of engineering construction.

Description

Technical Field

[0001] This invention relates to the field of engineering surveying technology, specifically to a method for selecting points in a measurement control network based on real-world 3D scenes. Background Technology

[0002] Control network selection refers to the process of scientifically determining the location of control points based on project requirements and field conditions before engineering surveying or topographic mapping. Control points are reference points with precise coordinates or elevations, and are divided into horizontal control points (determining location) and vertical control points (determining elevation). Point selection must comprehensively consider principles such as visibility, stability, preservation, and ease of measurement. For example, horizontal control points are often selected in areas with higher elevations and firmer soil to ensure visibility and long-term preservation, while vertical control points are laid out along low-slope routes to ensure the accuracy of leveling measurements.

[0003] The selection of points for existing measurement control networks directly affects measurement accuracy and project quality. Traditional point selection methods require multiple on-site surveys, relying on manual experience to judge visibility, rock stratum stability, network rationality, and safety. The above-ground and underground scenes are fragmented, making it impossible to intuitively analyze the rock stratum conditions below the control points. Visibility analysis relies solely on manual estimation, the optimization of observation pier height lacks quantitative basis, network parameters need to be calculated afterward, the point selection rework rate is high, and the determination of safety distances lacks visual support, making it easy to violate specifications. Therefore, these methods do not meet the current needs. To address this, we propose a measurement control network point selection method based on real-world 3D scenes. Summary of the Invention

[0004] The purpose of this invention is to provide a method for selecting points in a measurement control network based on real-world 3D scenes. This addresses the problems mentioned in the background section, such as the fact that the selection of points in existing measurement control networks directly affects measurement accuracy and engineering quality; traditional point selection methods require multiple on-site surveys; reliance on manual experience to judge visibility, rock stratum stability, network rationality, and safety; fragmented above-ground and underground scenes; inability to intuitively analyze rock stratum conditions below control points; visibility analysis relying solely on manual estimation; lack of quantitative basis for optimizing observation pier height; network parameters requiring post-hoc calculation; high rework rate in point selection; lack of visual support for safety distance determination; and susceptibility to violations of specifications.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for selecting points in a measurement control network based on real-world 3D scenes, comprising the following steps:

[0006] S1: Scene data acquisition and preprocessing. Image data of the engineering area is collected using UAV oblique photography technology. Through commercial oblique model production software, high-precision oblique photography model is generated by image aerial triangulation, dense point cloud generation, triangulation construction and texture mapping. Based on geological exploration profile or borehole data, three-dimensional geological model is constructed using Kriging interpolation.

[0007] S2: Model fusion, which unifies the spatial reference by transforming the inclined model and the geological model in the independent engineering coordinate system through a seven-parameter coordinate transformation, and uses the octree spatial indexing algorithm to splice the data of the inclined model and the geological model in .obj format according to their spatial location to generate an integrated model;

[0008] S3: The line-of-sight analysis algorithm uses a two-point line-of-sight determination sub-algorithm to generate line-of-sight rays and detect obstructions. Then, it uses the optimal height solution sub-algorithm of the observation pier to calculate the optimal height of the standardized observation pier model. Finally, it uses the GNSS height cutoff angle obstruction analysis sub-algorithm to detect the intersection with ground features and calculate the visible airspace ratio.

[0009] S4: Rock strata analysis visualization algorithm, based on candidate points. A cylindrical excavation body is generated, and the excavation body is trimmed from the geological model using the Boolean subtraction algorithm. The distribution of the underlying rock strata is visualized, and then parameters such as rock strata type, thickness, integrity coefficient, and uniaxial compressive strength within the excavation area are extracted.

[0010] S5: Network analysis algorithm, which uses a sub-algorithm to extract network parameters from adjacent points in the control network. and Perform spatial side length and three points , and Make an angle That is, angle The calculation is performed, and then the allowable range of side length [L_min, L_max] and allowable range of angle [θ_min, θ_max] for different levels of control network are preset through the network compliance judgment sub-algorithm. The deviation rate between the actual parameters and the threshold is calculated. If either δ_L>5% or δ_θ>1°, a prompt is made to adjust the position of the point.

[0011] S6: Safety distance measurement sub-algorithm. The safety distance measurement sub-algorithm calculates the spatial distance from candidate point P(x, y, z) to road L_road, high-voltage line L_line, and water surface L_water, respectively. Then, the compliance judgment sub-algorithm compares the measurement specifications and presets a safety distance threshold. If any distance is less than the safety distance threshold, the risk level is marked as low, medium, or high, and the point is prompted to be adjusted.

[0012] S7: Point Selection Optimization and Output. For candidate points that pass all analyses, a point selection report is generated, including but not limited to point coordinates, observation pier height, rock strata parameters, network parameters, and safety distance information. For points that do not pass, adjustment schemes are automatically recommended, including but not limited to offsetting coordinates and increasing the observation pier height. The line-of-sight analysis algorithm, rock strata analysis visualization algorithm, network analysis algorithm, and safety distance measurement sub-algorithm are iteratively executed until all constraints are met.

[0013] Preferably, the UAV image data resolution in S1 is greater than 5cm, the Kriging interpolation method is used to generate continuous rock layer interface elevation surfaces for discrete borehole data, and the topology checking algorithm is used to convert geological models in .Revit and .max formats to .obj format, and sequentially detect and repair missing, texture loss and topology errors on the model surfaces of the geological model.

[0014] Preferably, the octree spatial indexing algorithm is used to quickly retrieve spatially overlapping and adjacent model elements through an octree for geometric stitching. The core of the octree spatial indexing algorithm consists of spatial neighborhood retrieval and boundary fusion. The octree spatial indexing algorithm includes the following steps:

[0015] A1: For spatial neighborhood retrieval targets, specifically, to facilitate rapid matching of overlapping areas, find the areas where the tilted model and the geological model overlap and are adjacent in space, determine the splicing boundary, and then traverse all leaf nodes of the geological model, obtain the geological model vertices within the nodes and record them as... For each The leaf nodes of the k-neighborhood of the octree are retrieved, and the vertices of the tilted model are extracted and denoted as . Finally, calculate and Based on the spatial distance, point pairs with a distance less than the splicing threshold are selected to determine the splicing boundary point set B, i.e. ;

[0016] A2: Contour alignment, specifically for splicing boundaries, aligns the rock strata contour of the geological model with the terrain contour of the inclined model. Extracts the boundary contours of the two models from the boundary point set B, that is, extracts the edge lines of the surfaces through Delaunay triangulation and edge detection algorithms. For contours that are not completely overlapping, interpolates along the engineering direction and normal direction to make the vertices of the geological model contour correspond one-to-one with the vertices of the inclined model contour. Using the surface of the inclined model as a reference, adjusts the Z coordinate elevation of the boundary vertices of the geological model to match the surface elevation of the inclined model.

[0017] A3: Face stitching and topology reconstruction. Specifically, based on the aligned boundaries, the face topology of the two models is reconstructed to form a continuous stitched model. Redundant faces of the two models within the stitching region are deleted, such as faces of the geological model extending beyond the surface of the inclined model, or faces of the inclined model penetrating the geological model. Then, adjacent faces in the stitching region are retrieved using an octree, and transition faces, i.e., triangular faces, are constructed using the aligned boundary vertices to connect the boundaries of the inclined model and the geological model. After stitching, it is necessary to verify that the model has no spatial penetration, such as the inclined model penetrating underground geological bodies or the geological model penetrating the surface. Finally, collision detection and optimization are performed using an octree to generate a stitched model in .obj format.

[0018] Preferably, the two-point visibility determination sub-algorithm consists of a two-point visibility determination sub-algorithm, an optimal height calculation sub-algorithm for the observation pier, and a GNSS height cutoff angle occlusion analysis sub-algorithm. The two-point visibility determination sub-algorithm includes the following steps:

[0019] B1: Candidate control points are set using the two-point visibility determination sub-algorithm. , A spatial ray L is generated, and then features around the ray are quickly filtered based on the spatial index R-tree. A ray triangular facet intersection detection algorithm is used to calculate the intersection set. ,like If the line of sight is completely clear, then the maximum occlusion height is determined; otherwise, the maximum occlusion height is extracted. ;

[0020] B2: Solving the minimum required height using the sub-algorithm of the optimal height of the observation pier. Perform calculations using total cost Optimization function selects the optimal height Load the standardized observation pier model and adjust its height to the optimal height. Re-execute the two-point visibility determination sub-algorithm and iteratively optimize until the occlusion is eliminated;

[0021] B3: Convert the control point coordinates to the station-centered ENU coordinate system. Using the control points as the center, generate a hemispherical grid with an elevation cutoff angle α in the range of 5°-15° and a resolution of 1°×1°. Generate spatial rays for each grid, detect the intersection with ground features, and calculate the visible space occupancy. If If so, the height of the observation pier needs to be adjusted or a new location needs to be selected.

[0022] Preferably, the space ray L is:

[0023] ;

[0024] ;

[0025] ;

[0026] ;

[0027] ;

[0028] ;

[0029] in, The height threshold of the line of sight at the location of the obstruction. =0.1-0.5m is the safety margin, C0 is the fixed cost, and C is the cost per unit height.

[0030] Preferably, the cylindrical excavation body is a cylinder with P as its center, a radius r = 15m, and a depth H = 0-30m.

[0031] Preferably, the ;

[0032] .

[0033] Preferably, the ;

[0034] Where (x0, y0, z0) are the coordinates of the feature points of the road L_road, the high-voltage line L_line, and the water surface L_water.

[0035] Compared with the prior art, the beneficial effects of the present invention are:

[0036] This invention achieves visual fusion of above-ground and underground scenes through integrated modeling, solving the problem of traditional site selection that focuses on above-ground and neglects underground. It uses multi-dimensional algorithms to quantitatively determine visibility, rock strata, network structure, and safety, replacing manual experience-based judgment. The cost function design for observation pier height and rock penetration depth reduces engineering construction costs, reduces the number of on-site surveys, improves site selection efficiency, and reduces rework rate. Attached Figure Description

[0037] Figure 1 This is a flowchart of the measurement control network point selection method of the present invention. Detailed Implementation

[0038] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0039] Please see Figure 1 The present invention provides an embodiment of a method for selecting points in a measurement control network based on a real-world 3D scene, comprising the following steps:

[0040] S1: Scene data acquisition and preprocessing. Image data of the engineering area is collected using UAV oblique photography technology. Through commercial oblique model production software, high-precision oblique photography model is generated by image aerial triangulation, dense point cloud generation, triangulation construction and texture mapping. Based on geological exploration profile or borehole data, three-dimensional geological model is constructed using Kriging interpolation.

[0041] S2: Model fusion, which unifies the spatial reference by transforming the inclined model and the geological model in the independent engineering coordinate system through a seven-parameter coordinate transformation, and uses the octree spatial indexing algorithm to splice the data of the inclined model and the geological model in .obj format according to their spatial location to generate an integrated model;

[0042] S3: The line-of-sight analysis algorithm uses a two-point line-of-sight determination sub-algorithm to generate line-of-sight rays and detect obstructions. Then, it uses the optimal height solution sub-algorithm of the observation pier to calculate the optimal height of the standardized observation pier model. Finally, it uses the GNSS height cutoff angle obstruction analysis sub-algorithm to detect the intersection with ground features and calculate the visible airspace ratio.

[0043] S4: Rock strata analysis visualization algorithm, based on candidate points. A cylindrical excavation body is generated, and the excavation body is trimmed from the geological model using the Boolean subtraction algorithm. The distribution of the underlying rock strata is visualized, and then parameters such as rock strata type, thickness, integrity coefficient, and uniaxial compressive strength within the excavation area are extracted.

[0044] S5: Network analysis algorithm, which uses a sub-algorithm to extract network parameters from adjacent points in the control network. and Perform spatial side length and three points , and Make an angle That is, angle The calculation is performed, and then the allowable range of side length [L_min, L_max] and allowable range of angle [θ_min, θ_max] for different levels of control network are preset through the network compliance judgment sub-algorithm. The deviation rate between the actual parameters and the threshold is calculated. If either δ_L>5% or δ_θ>1°, a prompt is made to adjust the position of the point.

[0045] S6: Safety distance measurement sub-algorithm. The safety distance measurement sub-algorithm calculates the spatial distance from candidate point P(x, y, z) to road L_road, high-voltage line L_line, and water surface L_water, respectively. Then, the compliance judgment sub-algorithm compares the measurement specifications and presets a safety distance threshold. If any distance is less than the safety distance threshold, the risk level is marked as low, medium, or high, and the point is prompted to be adjusted.

[0046] S7: Point Selection Optimization and Output. For candidate points that pass all analyses, a point selection report is generated, including but not limited to point coordinates, observation pier height, rock strata parameters, network parameters, and safety distance information. For points that do not pass, adjustment schemes are automatically recommended, including but not limited to offsetting coordinates and increasing the observation pier height. The line-of-sight analysis algorithm, rock strata analysis visualization algorithm, network analysis algorithm, and safety distance measurement sub-algorithm are iteratively executed until all constraints are met.

[0047] Among them, the image data of the UAV in S1 has a resolution greater than 5cm, and the Kriging interpolation method is used to generate a continuous rock interface elevation surface for discrete borehole data.

[0048] The topology checking algorithm is used to convert geological models in .Revit and .max formats to .obj format, and sequentially detects and repairs missing features, lost textures, and topological errors on the model faces of the geological models.

[0049] Furthermore, the octree spatial indexing algorithm is used to quickly retrieve spatially overlapping and adjacent model elements through an octree for geometric stitching. The core of the octree spatial indexing algorithm consists of spatial neighborhood retrieval and boundary fusion. The octree spatial indexing algorithm includes the following steps:

[0050] A1: For spatial neighborhood retrieval targets, specifically, to facilitate rapid matching of overlapping areas, find the areas where the tilted model and the geological model overlap and are adjacent in space, determine the splicing boundary, and then traverse all leaf nodes of the geological model, obtain the geological model vertices within the nodes and record them as... For each The leaf nodes of the k-neighborhood of the octree are retrieved, and the vertices of the tilted model are extracted and denoted as . Finally, calculate and Based on the spatial distance, point pairs with a distance less than the splicing threshold are selected to determine the splicing boundary point set B, i.e. ;

[0051] A2: Contour alignment, specifically for splicing boundaries, aligns the rock strata contour of the geological model with the terrain contour of the inclined model. Extracts the boundary contours of the two models from the boundary point set B, that is, extracts the edge lines of the surfaces through Delaunay triangulation and edge detection algorithms. For contours that are not completely overlapping, interpolates along the engineering direction and normal direction to make the vertices of the geological model contour correspond one-to-one with the vertices of the inclined model contour. Using the surface of the inclined model as a reference, adjusts the Z coordinate elevation of the boundary vertices of the geological model to match the surface elevation of the inclined model.

[0052] A3: Face stitching and topology reconstruction. Specifically, based on the aligned boundaries, the face topology of the two models is reconstructed to form a continuous stitched model. Redundant faces of the two models within the stitching region are deleted, such as faces of the geological model extending beyond the surface of the inclined model, or faces of the inclined model penetrating the geological model. Then, adjacent faces in the stitching region are retrieved using an octree, and transition faces, i.e., triangular faces, are constructed using the aligned boundary vertices to connect the boundaries of the inclined model and the geological model. After stitching, it is necessary to verify that the model has no spatial penetration, such as the inclined model penetrating underground geological bodies or the geological model penetrating the surface. Finally, collision detection and optimization are performed using an octree to generate a stitched model in .obj format.

[0053] Furthermore, the two-point visibility determination sub-algorithm consists of a two-point visibility determination sub-algorithm, an optimal height calculation sub-algorithm for observation piers, and a GNSS height cutoff angle occlusion analysis sub-algorithm. The two-point visibility determination sub-algorithm includes the following steps:

[0054] B1: Candidate control points are set using the two-point visibility determination sub-algorithm. , A spatial ray L is generated, and then features around the ray are quickly filtered based on the spatial index R-tree. A ray triangular facet intersection detection algorithm is used to calculate the intersection set. ,like If the line of sight is completely clear, then the maximum occlusion height is determined; otherwise, the maximum occlusion height is extracted. ;

[0055] B2: Solving the minimum required height using the sub-algorithm of the optimal height of the observation pier. Perform calculations using total cost Optimization function selects the optimal height Load the standardized observation pier model and adjust its height to the optimal height. Re-execute the two-point visibility determination sub-algorithm and iteratively optimize until the occlusion is eliminated;

[0056] B3: Convert the control point coordinates to the station-centered ENU coordinate system. Using the control points as the center, generate a hemispherical grid with an elevation cutoff angle α in the range of 5°-15° and a resolution of 1°×1°. Generate spatial rays for each grid, detect the intersection with ground features, and calculate the visible space occupancy. If If so, the height of the observation pier needs to be adjusted or a new location needs to be selected.

[0057] ;

[0058] ;

[0059] ;

[0060] ;

[0061] ;

[0062] ;

[0063] in, The height threshold of the line of sight at the location of the obstruction. =0.1-0.5m is the safety margin, C0 is the fixed cost, and C is the cost per unit height.

[0064] The cylindrical excavation body is a cylinder with center P, radius r = 15m and depth H = 0-30m.

[0065] ;

[0066] .

[0067] ;

[0068] Where (x0, y0, z0) are the coordinates of the feature points of the road L_road, the high-voltage line L_line, and the water surface L_water.

[0069] Example:

[0070] The Anhui section of the Gansu-Zhejiang ±800 kV DC transmission project is included in the power development plan and is a key power transmission project under construction. In order to save space corridor resources to the greatest extent and reduce the impact on the economic and social development of the areas along the route, the project plans to build a GIL (Gas Infrared Transmission) tunnel project that crosses the bottom of the Yangtze River in the Anhui section. The total length of the project is about 3.55 kilometers, of which about 1.65 kilometers cross the river.

[0071] Before construction begins, a high-precision control network needs to be established to provide accurate and reliable basic control data for subsequent construction layout and tunnel boring machine (TBM) surveying. The application of precision measurement technology ensures that the position and attitude of the utility tunnel are precisely controlled during its crossing of the Yangtze River, thereby achieving the goal of safely crossing the riverbed.

[0072] Implementation process:

[0073] 1. Data Acquisition: A DJI M300 drone was used to collect oblique photographic data of the engineering area. The resolution of the photographic data was 3cm, resulting in 2600 images. Data from 46 boreholes and geological profiles were also collected. Figure 1 open;

[0074] 2. Model construction: The inclined model is generated by the SfM algorithm, and the geological model is constructed by Kriging interpolation. After being converted to .obj format, the model is based on three common control points. The spatial reference is unified through seven-parameter transformation and integrated to generate an integrated real-scene 3D model.

[0075] 3. Point selection analysis: Eight densified control points were selected in the model, and a line-of-sight analysis was performed. Three of the points were found to be obstructed. The optimal observation pier heights were determined to be 1.2m, 0.8m, and 1.5m, respectively. Rock layer analysis showed that two of the points were located below soft soil. The depth of the observation pier was adjusted to 1.5m to meet the stability requirements.

[0076] Network analysis revealed that the primary control network, deployed according to a Class C GNSS network, did not meet the requirement of an average distance of 5km between adjacent points. After adjusting the control points, the final average distance was 5.6km.

[0077] Safety analysis revealed that one location was 40m from the high-voltage line, and after being offset by 15m, it met the 50m safety distance requirement.

[0078] Generate the final point selection scheme, select 8 control points of the primary control network, and generate a compliance verification report for the coordinates of the 8 control points of the densified control network, the height of the observation piers, and the results of the on-site reconnaissance verification and model analysis. The deviation between the results is ≤5cm.

[0079] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for selecting points in a measurement control network based on a real-world 3D scene, characterized in that, Includes the following steps: S1: Scene data acquisition and preprocessing. Image data of the engineering area is collected using UAV oblique photography technology. Through commercial oblique model production software, high-precision oblique photography model is generated by image aerial triangulation, dense point cloud generation, triangulation construction and texture mapping. Based on geological exploration profile or borehole data, three-dimensional geological model is constructed using Kriging interpolation. S2: Model fusion, which unifies the spatial reference by transforming the inclined model and the geological model in the independent engineering coordinate system through a seven-parameter coordinate transformation, and uses the octree spatial indexing algorithm to splice the data of the inclined model and the geological model in .obj format according to their spatial location to generate an integrated model; S3: The line-of-sight analysis algorithm uses a two-point line-of-sight determination sub-algorithm to generate line-of-sight rays and detect obstructions. Then, it uses the optimal height solution sub-algorithm of the observation pier to calculate the optimal height of the standardized observation pier model. Finally, it uses the GNSS height cutoff angle obstruction analysis sub-algorithm to detect the intersection with ground features and calculate the visible airspace ratio. S4: Rock strata analysis visualization algorithm, based on candidate points. A cylindrical excavation body is generated, and the excavation body is trimmed from the geological model using the Boolean subtraction algorithm. The distribution of the underlying rock strata is visualized, and then parameters such as rock strata type, thickness, integrity coefficient, and uniaxial compressive strength within the excavation area are extracted. S5: Network analysis algorithm, which uses a sub-algorithm to extract network parameters from adjacent points in the control network. and Perform spatial side length and three points , and Make an angle That is, angle The calculation is performed, and then the allowable range of side length [L_min, L_max] and allowable range of angle [θ_min, θ_max] for different levels of control network are preset through the network compliance judgment sub-algorithm. The deviation rate between the actual parameters and the threshold is calculated. If either δ_L>5% or δ_θ>1°, a prompt is made to adjust the position of the point. S6: Safety distance measurement sub-algorithm. The safety distance measurement sub-algorithm calculates the spatial distance from candidate point P(x, y, z) to road L_road, high-voltage line L_line, and water surface L_water, respectively. Then, the compliance judgment sub-algorithm compares the measurement specifications and presets a safety distance threshold. If any distance is less than the safety distance threshold, the risk level is marked as low, medium, or high, and the point is prompted to be adjusted. S7: Point Selection Optimization and Output. For candidate points that pass all analyses, a point selection report is generated, including but not limited to point coordinates, observation pier height, rock strata parameters, network parameters, and safety distance information. For points that do not pass, adjustment schemes are automatically recommended, including but not limited to offsetting coordinates and increasing the observation pier height. The line-of-sight analysis algorithm, rock strata analysis visualization algorithm, network analysis algorithm, and safety distance measurement sub-algorithm are iteratively executed until all constraints are met.

2. The method for selecting points in a measurement control network based on a real-world 3D scene according to claim 1, characterized in that: The UAV image data resolution in S1 is greater than 5cm. The Kriging interpolation method is used to generate continuous rock strata interface elevation surfaces from discrete borehole data. The topology checking algorithm is used to convert geological models in .Revit and .max formats to .obj format, and to sequentially detect and repair missing, texture loss, and topology errors on the model surfaces of the geological models.

3. The method for selecting points in a measurement control network based on a real-world 3D scene according to claim 2, characterized in that: The octree spatial indexing algorithm is used to quickly retrieve spatially overlapping and adjacent model elements and perform geometric stitching. The core of the octree spatial indexing algorithm consists of spatial neighborhood retrieval and boundary fusion. The octree spatial indexing algorithm includes the following steps: A1: For spatial neighborhood retrieval targets, specifically, to facilitate rapid matching of overlapping areas, find the areas where the tilted model and the geological model overlap and are adjacent in space, determine the splicing boundary, and then traverse all leaf nodes of the geological model, obtain the geological model vertices within the nodes and record them as... For each The leaf nodes of the k-neighborhood of the octree are retrieved, and the vertices of the tilted model are extracted and denoted as . Finally, calculate and Based on the spatial distance, point pairs with a distance less than the splicing threshold are selected to determine the splicing boundary point set B, i.e. ; A2: Contour alignment, specifically for splicing boundaries, aligns the rock strata contour of the geological model with the terrain contour of the inclined model. Extracts the boundary contours of the two models from the boundary point set B, that is, extracts the edge lines of the surfaces through Delaunay triangulation and edge detection algorithms. For contours that are not completely overlapping, interpolates along the engineering direction and normal direction to make the vertices of the geological model contour correspond one-to-one with the vertices of the inclined model contour. Using the surface of the inclined model as a reference, adjusts the Z coordinate elevation of the boundary vertices of the geological model to match the surface elevation of the inclined model. A3: Face stitching and topology reconstruction. Specifically, based on the aligned boundaries, the face topology of the two models is reconstructed to form a continuous stitched model. Redundant faces of the two models within the stitching region are deleted, such as faces of the geological model extending beyond the surface of the inclined model, or faces of the inclined model penetrating the geological model. Then, adjacent faces in the stitching region are retrieved using an octree, and transition faces, i.e., triangular faces, are constructed using the aligned boundary vertices to connect the boundaries of the inclined model and the geological model. After stitching, it is necessary to verify that the model has no spatial penetration, such as the inclined model penetrating underground geological bodies or the geological model penetrating the surface. Finally, collision detection and optimization are performed using an octree to generate a stitched model in .obj format.

4. The method for selecting points in a measurement control network based on a real-world 3D scene according to claim 3, characterized in that: The two-point visibility determination sub-algorithm consists of a two-point visibility determination sub-algorithm, an optimal height calculation sub-algorithm for observation piers, and a GNSS height cutoff angle occlusion analysis sub-algorithm. The two-point visibility determination sub-algorithm includes the following steps: B1: Candidate control points are set using the two-point visibility determination sub-algorithm. , A spatial ray L is generated, and then features around the ray are quickly filtered based on the spatial index R-tree. A ray triangular facet intersection detection algorithm is used to calculate the intersection set. ,like If the line of sight is completely clear, then the maximum occlusion height is determined; otherwise, the maximum occlusion height is extracted. ; B2: Solving the minimum required height using the sub-algorithm of the optimal height of the observation pier. Perform calculations using total cost Optimization function selects the optimal height Load the standardized observation pier model and adjust its height to the optimal height. Re-execute the two-point visibility determination sub-algorithm and iteratively optimize until the occlusion is eliminated; B3: Convert the control point coordinates to the station-centered ENU coordinate system. Using the control points as the center, generate a hemispherical grid with an elevation cutoff angle α in the range of 5°-15° and a resolution of 1°×1°. Generate spatial rays for each grid, detect the intersection with ground features, and calculate the visible space occupancy. If If so, the height of the observation pier needs to be adjusted or a new location needs to be selected.

5. The method for selecting points in a measurement control network based on real-world 3D as described in claim 4, characterized in that: The space ray L is: ； ； ； ； ； ； in, The height threshold of the line of sight at the location of the obstruction. =0.1-0.5m is the safety margin, C0 is the fixed cost, and C is the cost per unit height.

6. The method for selecting points in a measurement control network based on a real-world 3D scene according to claim 5, characterized in that: The cylindrical excavation body is a cylinder with P as its center, a radius r = 15m, and a depth H = 0-30m.

7. The method for selecting points in a measurement control network based on a real-world 3D scene according to claim 6, characterized in that: The ; 。 8. The method for selecting points in a measurement control network based on a real-world 3D scene according to claim 7, characterized in that: The ; Where (x0, y0, z0) are the coordinates of the feature points of the road L_road, the high-voltage line L_line, and the water surface L_water.

Images