Unmanned aerial vehicle naked eye 3D two-three dimensional digital mapping method

Abstract

The application provides a UAV naked-eye 3D two-three-dimensional digital mapping method, proposes three-dimensional contour mapping based on point clouds, multi-storey building digital mapping based on real scene models, and digital orthographic image-based digital mapping of ground objects and landforms, realizes rapid mapping of the terrain, ground objects and landforms in the construction area, and the mapping accuracy is better than 5cm, meets the large-scale digital mapping requirements, improves the engineering project construction deployment efficiency, and is suitable for three-dimensional terrain mapping of all construction areas such as railways, highways and municipal engineering. It belongs to the field of surveying and mapping.

Description

Technical Field

[0001] This invention relates to a naked-eye 3D two-dimensional and three-dimensional digital mapping method for unmanned aerial vehicles (UAVs), which is particularly suitable for fully digital mapping of complex terrain, features, and landforms, and belongs to the field of surveying and mapping. Background Technology

[0002] Topographic maps play a crucial role in engineering construction. During the design phase, they are used to determine the specific locations of various projects and corresponding structures, helping designers find suitable locations, lay out facilities, measure distances and elevations, and determine the orientation and slope of the project. During the construction phase, topographic maps serve as fundamental maps, ensuring that construction activities proceed according to the design and avoiding engineering accidents or delays caused by misinterpretations of the terrain. Because there is a certain time lag between the design and construction phases, and due to the influence of the natural environment, the terrain may differ from the design phase after construction begins. Therefore, to ensure construction safety, a new survey of the construction site is necessary.

[0003] Traditional topographic mapping primarily relies on RTK (Real-Time Kinematic) surveying, which suffers from significant fieldwork, low mapping efficiency, and susceptibility to weather conditions. In recent years, with the rapid popularization of drone technology, the industry's main approaches are: First, manually measuring ground points on a real-world model, reconstructing a triangulation network from these points, and then creating 3D contour lines. This method is labor-intensive, requires highly skilled cartographers, and is not easily adopted. Second, loading digital orthophotos (DOM) into CAD software allows for mapping of roads, rivers, farmland, and building areas. However, for multi-story buildings, it cannot accurately measure the area of ​​multi-story buildings, making it unsuitable for direct land acquisition and demolition mapping and management. Furthermore, direct CAD mapping is inefficient, requiring multiple drawing of common edges for farmland, which can easily lead to mapping errors. Summary of the Invention

[0004] This invention provides a naked-eye 3D two-dimensional and three-dimensional digital mapping method for UAVs to improve the efficiency of topographic mapping and ensure that the mapping accuracy meets construction requirements.

[0005] To address the aforementioned issues, a novel naked-eye 3D two- and three-dimensional digital mapping method using unmanned aerial vehicles (UAVs) is proposed, specifically including:

[0006] 1) Unmanned Aerial Vehicle (UAV) Data Acquisition and Processing

[0007] Image control points are set up around the mapping, the coordinates of each image control point are measured using a positioning system, UAV images are collected, and aerial triangulation is performed twice using aerial surveying software to generate LAS format point cloud data, OSGB format 3D reality model and TIFF format DOM image.

[0008] 2) Naked-eye 3D mapping of 3D contour lines based on point cloud data

[0009] The acquired point cloud data were sequentially classified using mathematical morphological filtering, slope filtering, and cloth simulation filtering. The ground point cloud data was then converted into TIN triangulation data using the Delaunay triangulation method. DXF format 3D contour data was generated based on the TIN triangulation data.

[0010] 3) Naked-eye 3D mapping of terrain features based on digital orthophotos

[0011] Import DOM images into ArcGIS to create vectorized layers for, but not limited to, woodlands, roads, farmland, waterways, buildings, and greenhouses. Use topological analysis to establish the relationship between vector graphics and raster graphics, forming an operation flow of raster vector—raster layer—vectorization—raster—vectorization. Use Python scripts to count land use attributes, area, and perimeter, and generate an engineering quantity table.

[0012] 4) Naked-eye 3D mapping of multi-story buildings based on real-scene models

[0013] Import the OSGB format 3D reality model into CASS 3D software, browse the 3D model and link it with the 2D image, draw closed polygons along the building's exterior wall, and view the results on the 3D model at the same time; convert the mapping results to DXF format and import them into ArcGIS, and use data connection design and field creation to statistically analyze the area of ​​multi-story buildings.

[0014] 5) Merging and accuracy verification of topographic maps

[0015] The vectorized files generated in the above steps are merged in GIS and directly merged with DOM images for output; the accuracy of naked-eye 3D digital mapping is evaluated by the accuracy of UAV image control points.

[0016] In the aforementioned method, drone images are acquired through a combination of fully automatic and manual flight modes;

[0017] The UAV data processing employs professional aerial surveying software. First, original UAV imagery is added, and an aerial triangulation algorithm is established to convert the photos into aerial triangulation point cloud data. Second, the coordinates of ground control points (GCPs) are added, and the GCP coordinates are assigned to the aerial triangulation point cloud data using the point-point adjustment method. Through aerial triangulation optimization and the GCP adjustment report, the error of the aerial survey results is checked. For GCPs with a positional error greater than 5cm, re-pointing is required. The UAV data is then converted to the construction coordinate system to ensure the accuracy of UAV digital mapping. Finally, a second aerial triangulation algorithm is used to establish high-precision dense point cloud data, a 3D reality model, and DOM imagery.

[0018] In the aforementioned method, UAV imagery is acquired through a combination of fully automatic and manual flight modes. UAV data processing utilizes professional aerial surveying software. First, the original UAV imagery is added, and an aerial triangulation algorithm is established to convert the photos into aerial triangulation point cloud data. Second, the coordinates of ground control points (GCPs) are added, and the GCP coordinates are assigned to the aerial triangulation point cloud data using the point-point adjustment method. Through aerial triangulation optimization and the GCP adjustment report, the errors in the aerial survey results are checked. For GCPs with positional errors greater than 5cm, re-pointing is required. The UAV data is then converted to a construction coordinate system to ensure the accuracy of UAV digital mapping. Finally, a second aerial triangulation algorithm is used to establish high-precision dense point cloud data, a 3D reality model, and DOM imagery.

[0019] In the aforementioned method, point cloud filtering and classification are performed. First, mathematical morphology filtering is used to perform overall filtering and classification on the mapping point cloud data to obtain coarsely classified point clouds, which quickly separates large areas of ground and non-ground points. Second, slope filtering is used to relocate areas with large holes in the point cloud data, usually steep mountains or steep slopes. By optimizing the slope classification parameters, the classification results of the ground point cloud are re-optimized. Finally, material simulation filtering is used to solve the problem of misclassification or abnormal classification of concrete surfaces.

[0020] In the aforementioned method, a TIN triangulation network is established, and the classified point cloud data is saved as a new point cloud data file. Using the Delaunay triangulation method, the maximum side length of the triangulation network does not exceed 50m, and the display resolution is X=1m, Y=1m. The ground point cloud data is converted into an irregular triangulation network, namely TIN triangulation network data, and the triangulation network file is saved. Three-dimensional contour lines are established. According to the construction requirements, a topographic map scale of 1 / 500 is used, with a corresponding main contour interval of 2.5m and a contour interval of 0.5m. The smoothing type of the contour lines is selected as B-spline region, and the smoothing degree is 70% to 85%. Three-dimensional contour line data is established based on the TIN triangulation network, and the data format file is dxf.

[0021] In the aforementioned method, vectorization parameters are established, DOM images are imported into ArcGIS, coordinate references are set, and vectorization parameters are created using the shapefile tool. Vectorization layer parameters for woodland, roads, farmland, waterways, buildings, and greenhouses are established separately, and different colors are used to distinguish various land features through layer style settings.

[0022] Raster layer vectorization utilizes ArcGIS's topology analysis function to perform topology analysis on raster data. Based on the topology analysis results, the relationship between vector graphics and raster graphics is established, forming an operation flow of raster vector—raster layer—vectorization—raster—vectorization. For areas with non-overlapping boundaries, vectorization is performed directly in the corresponding layers using the area analysis tool.

[0023] Extract the quantities of work, select the vectorized layer data, and use data design to establish fields for land use attributes, units, quantities, areas, and perimeters. Perform statistical analysis on the vectorized layer to obtain a table of quantities of work.

[0024] In the aforementioned method, digital mapping of multi-story buildings involves importing a 3D reality model in OSGB format using CASS 3D software, setting building layers, and linking the 3D model with 2D images. A multi-building surface drawing method is used to draw closed polygons along the building's exterior walls, while simultaneously viewing the results on the 3D model. For quantity calculation, the multi-story building mapping file is converted to DXF format and imported into ArcGIS. Through data connection design and field creation, area statistical analysis of the multi-story building is achieved.

[0025] In the aforementioned method, vectorized files in shp, dxf, and dwg formats are directly merged using GIS. The merged files are then used directly for access road alignment and earthwork calculation, added directly to the Aowei software for viewing, and merged directly with high-resolution DOM images for output by construction personnel.

[0026] Compared with existing technologies, this invention comprehensively adopts methods such as naked-eye 3D mapping of 3D contour lines based on point cloud data, naked-eye 3D mapping of terrain features based on digital orthophotos, and naked-eye 3D mapping of multi-story buildings based on real-scene models. It achieves rapid vector mapping of terrain, features, and landforms within the construction area, with mapping accuracy better than 5cm, meeting the requirements of large-scale digital mapping. This improves the efficiency of engineering project deployment, enhances the accuracy and precision of digital mapping, reduces the workload of surveyors in the field, and increases the efficiency of both field and office work. It enables efficient modeling in complex terrain areas, has strong digital interactivity, and improves the level of digital management in engineering projects. It forms a technical closed loop of "high-precision data acquisition - intelligent processing and analysis - engineering application adaptation," not only breaking through the multiple bottlenecks of traditional UAV mapping in terms of accuracy, efficiency, and scene adaptability, but also constructing a digital mapping technology standard suitable for modern engineering construction. It has significant industry demonstration effects and industrialization promotion value, and is applicable to 3D topographic mapping of all construction areas such as railways, highways, and municipal works. Attached Figure Description

[0027] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention.

[0029] Example

[0030] See attached document Figure 1 This embodiment improves the mapping method, specifically by proposing a naked-eye 3D two-dimensional and three-dimensional digital mapping method for UAVs, including:

[0031] (1) Data acquisition and processing of unmanned aerial vehicles (UAVs);

[0032] To improve the accuracy of UAV mapping, after the design unit handed over the control point coordinates, based on previous UAV mapping experience, a method was proposed to be used for every 1km... 2 The density standard of setting up 3 ground control points improves the control accuracy by 2-3 times compared to traditional UAV mapping (usually 1 point every 2-4 km²). It ensures that the coordinate transformation error is ≤5cm from the source. Ground control points are set up around the mapping area, and the coordinates of each control point are measured using the Beidou positioning system. The coordinate reference adopts the national CGCS2000 coordinate system, and the elevation reference is the 1985 Yellow Sea Height System. A unified framework of millimeter-level geographic reference is constructed, which solves the positioning deviation problem in areas where traditional GNSS signals are blocked. It achieves seamless docking between the construction coordinate system and the national reference coordinate system and avoids the cumulative error of coordinate transformation in the later stage.

[0033] For UAV data acquisition, a combination of fully automatic and manual flight is adopted. Based on digital mapping experience, the parameters can be set as follows: the flight altitude should be controlled within 80m to 100m, the overlap rate of UAV aerial photos should be 65% to 85%, and the UAV lens should be 8° to 20°. After setting the flight parameters, high-resolution UAV aerial photo data for mapping is established.

[0034] The UAV data processing utilizes professional aerial surveying software and introduces a novel "double aerial triangulation" process. First, original UAV imagery is added, and an aerial triangulation algorithm is established to convert the photos into aerial triangulation point cloud data. Second, the coordinates of ground control points (GCPs) are added, and the coordinates are assigned to the GCPs using the point-pointing method. Through aerial triangulation optimization and GCP adjustment reports, the errors in the aerial survey results are checked. For GCPs with positional errors exceeding 5cm, re-pointing is required. A second calculation generates a high-density point cloud and model, transforming the UAV data into a construction coordinate system, ensuring the accuracy of UAV digital mapping and facilitating use by construction personnel. Finally, a second aerial triangulation algorithm is used to establish high-precision dense point cloud data (LAS format), a 3D reality model (OSGB format), and DOM imagery (TIFF format).

[0035] Compared to single aerial triangulation calculations, the accuracy of horizontal / vertical estimation is improved by 40%, realizing the transformation from "post-verification" to "process control" and ensuring that the data results directly meet the requirements of 1 / 500 large-scale mapping.

[0036] (2) Naked-eye 3D mapping of 3D contour lines based on point cloud data

[0037] Point cloud filtering and classification processing. Since point cloud data contains a large number of non-ground points, such as buildings, vegetation, power facilities, and vehicles, these points affect the accuracy of topographic mapping, necessitating point cloud filtering and classification processing. First, a mathematical morphology filtering classification method is used to perform overall filtering and classification of the mapping point cloud data, obtaining a coarsely classified point cloud that quickly separates large areas of ground from non-ground points, significantly improving processing efficiency compared to a single algorithm. Second, a slope-based filtering classification method is used to re-select areas with large holes in the point cloud data, typically steep mountains or embankments. By optimizing the slope classification parameters, the classification results of the ground point cloud are re-optimized, eliminating inaccurate point cloud filtering classification results in mountainous areas, where some mountainous areas exhibit numerous holes that fail to accurately represent the original ground features. For mountain / embankment areas with slopes >25°, dynamically adjusting the slope classification threshold can eliminate most hole artifacts. Finally, a cloth simulation filtering classification method is used to eliminate problems such as indistinguishable concrete pavements and concrete roof slabs within building complexes, or classification anomalies. By using the above three-level progressive point cloud filtering and classification algorithm, accurate classification of ground point cloud data is achieved, breaking through the adaptability bottleneck of traditional single filtering algorithms in complex scenarios. The accuracy of ground point cloud classification is significantly improved, laying the foundation for high-precision contour line generation.

[0038] Create a TIN triangulation. Save the classified point cloud data as a new point cloud data file. Using the Delaunay triangulation method, the maximum side length of the triangulation should not exceed 50m, and the display resolution is X=1m, Y=1m. Convert the ground point cloud data into an irregular triangulation, i.e., TIN triangulation data, and save the triangulation file.

[0039] Establish three-dimensional contour lines. The contour line construction method using a triangulation network is employed. Based on construction requirements, the highest precision topographic map scale used in engineering projects is typically 1 / 500, corresponding to a principal contour interval of 2.5m and a contour interval of 0.5m. The contour line smoothing type is selected as B-spline region, with a smoothing degree of 70%–85%. Three-dimensional contour line data is established based on the TIN triangulation network, and the data format file is dxf.

[0040] (3) Naked-eye 3D mapping of land features and topography based on digital orthophotos

[0041] Establish vectorization parameters. Import DOM imagery into ArcGIS, set coordinate references, and use the shapefile tool to create new vectorization parameters. Vectorization layer parameters were created for forest land, roads, farmland, waterways, buildings, greenhouses, etc. Different colors were used to distinguish different types of land features through layer style settings.

[0042] Raster layer vectorization. ArcGIS's topology analysis function is used to perform topological analysis on raster data. Based on the analysis results, the relationship between vector graphics and raster graphics is established, forming an operation flow of raster vectorization—raster layer—vectorization—raster—vectorization. For areas with non-overlapping boundaries, such as greenhouses, buildings, waterways, and woodlands, vectorization is performed directly in the corresponding layers using the area analysis tool. For road vectorization, there are intersection relationships between roads, which need to be addressed through topological intersection analysis. For farmland vectorization, there are many common edges. If vectorization were done using CAD, all boundaries of each plot would need to be drawn. To improve vectorization efficiency, topological adjacency analysis is used to directly extract the common edges between plots, enabling rapid land vectorization.

[0043] Extract quantities. Select vectorized layer data, and use data design to create fields for land use attributes, units, quantities, areas, and perimeters. All data types are single precision, and the calculation rules are performed using Python. Perform statistical analysis on the vectorized layer to obtain a table of quantities.

[0044] (4) Naked-eye 3D mapping of multi-story buildings based on real-scene models

[0045] Digital mapping of multi-story buildings proposes a collaborative workflow of "3D model positioning + 2D image verification": Using CASS 3D software, an OSGB format 3D real-world model is imported, building layers are set, and the 3D model is linked with the 2D image. A multi-building surface drawing method is used to draw closed polygons along the building's exterior walls, while simultaneously viewing the results on the 3D model. A single-story building is named F1, a two-story building F2, a three-story building F3, a four-story building F4, and so on, with an N-story building named FN. This solves the problem of misjudging the number of building floors in traditional 2D mapping.

[0046] Quantity calculation. By converting the multi-story building survey drawings to DXF format and importing them into ArcGIS, and through data connection design and field creation, the area of ​​multi-story buildings can be statistically analyzed. This solves the problem that CASS cannot directly calculate the area of ​​multi-story buildings, and obtains the quantity of the building's work.

[0047] In building complexes with a floor area ratio greater than 2.5, the mapping efficiency is several times higher than that of total station measurements, the floor identification accuracy reaches 100%, and the area statistical error is less than 1%.

[0048] (5) Merging and accuracy verification of topographic maps.

[0049] Since the coordinate references of the three types of naked-eye 3D digital mapping mentioned above are consistent, vectorized files in formats such as shp, dxf, and dwg can be directly merged through GIS. The merged files can be directly used for access road alignment, earthwork calculation, etc., and can be directly added to Aowei software for viewing. They can be directly merged and output with high-resolution DOM images, making it convenient for construction personnel to use. This breaks the problem of data silos among multiple software in traditional mapping, significantly shortens the mapping cycle, and reduces the input of field personnel.

[0050] For the accuracy of naked-eye 3D digital mapping, the accuracy of UAV ground control points is used for evaluation. The known point verification method is used, which involves measuring three coordinate points on the actual ground, importing them into the digital mapping file, and comparing the differences between the coordinate values ​​before and after mapping. Since the topographic map scale required in construction is 1 / 500, according to the accuracy requirements of 1 / 500 mapping, the check points only need to meet the requirement that the plane error is less than 5cm and the elevation error is less than 5cm.

[0051] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A UAV naked eye 3D two-three-dimensional digital mapping method, characterized in that, Specifically includes: 1) unmanned aerial vehicle data acquisition and processing Around the layout of the mapping control points, the coordinates of each control point are measured by the positioning system, the unmanned aerial vehicle image is collected, and the aerial triangulation is solved twice by using the photogrammetry software to generate the LAS format point cloud data, the OSGB format three-dimensional real scene model and the TIFF format DOM image; 2) 3D contour naked eye 3D mapping based on point cloud data The obtained point cloud data is sequentially classified by mathematical morphology filtering, slope filtering and cloth simulation filtering; the ground point cloud data is converted into TIN triangular network data by Delaunay triangle construction method; and the DXF format three-dimensional contour data is generated based on the TIN triangular network data; 3) naked eye 3D mapping of ground features and landforms based on digital orthographic image Import the DOM image into ArcGIS to establish vectorized layers including but not limited to forest land, road, farmland, river, building and greenhouse, and use topological analysis to establish the correlation between vector graphics and raster graphics to form the operation process of raster-vector-raster layer-vectorization-raster-vectorization; the engineering quantity table is generated by Python script to count the attributes, area and perimeter of the land class; 4) multi-storey building body naked eye 3D mapping based on real scene model Multi-storey building digital mapping, a "three-dimensional model positioning + two-dimensional image verification" linkage operation mode is proposed: import the osgb format three-dimensional real scene model into CASS 3D software, set the building body layer, and through the three-dimensional model browsing and two-dimensional image linkage, the closed polygon is drawn along the building body outer wall through the multi-building body area drawing method, and the three-dimensional model viewing result is obtained. For single-storey building, it is named F1, for two-storey building, it is named F2, for three-storey building, it is named F3, for four-storey building, it is named F4, and so on. N-storey building is named FN, which solves the problem of misjudgment of building storey in traditional two-dimensional mapping; the mapping result is converted into DXF format and imported into ArcGIS, and the multi-storey building area is counted through data connection design and field creation; 5) topographic map merging and precision inspection Merge the vectorized files generated in the above steps in GIS, directly merge with DOM image and output; the naked eye 3D digital mapping precision is evaluated through the precision of unmanned aerial vehicle control points.

2. The unmanned aerial vehicle naked-eye 3D two-three-dimensional digital mapping method according to claim 1, characterized in that, Unmanned aerial vehicle images are collected through full-automatic and manual combined flight mode; The professional photogrammetry software is used for unmanned aerial vehicle data processing. First, add the original image of the unmanned aerial vehicle, establish the aerial triangulation algorithm, and calculate the aerial triangulation point cloud data; second, add the coordinates of the control points, use the prick point method to assign the coordinates of the control points to the aerial triangulation point cloud data, and through the aerial triangulation optimization method, the aerial triangulation point cloud data is obtained. Through the control point adjustment report, the photogrammetry result error is queried, and for the control point position error greater than 5 cm, the unmanned aerial vehicle data needs to be converted to the construction coordinate system to ensure the precision of unmanned aerial vehicle digital mapping; finally, through the second aerial triangulation algorithm, high-precision dense point cloud data, three-dimensional real scene model and DOM image are established.

3. The unmanned aerial vehicle naked-eye 3D two-three-dimensional digital mapping method according to claim 1, characterized in that: The point cloud is filtered and classified. Firstly, the mathematical morphology filtering classification method is adopted to filter and classify the point cloud data of the surveying and mapping points as a whole, so as to obtain the rough classified point cloud and quickly separate the large-area ground and non-ground points. Secondly, the slope filtering classification method is adopted to reselect the point cloud data of the region with large voids, usually the mountain or cliff region with large slope, and the classification result of the ground point cloud is reoptimized by optimizing the slope classification parameters. Finally, the cloth simulation filtering classification method is adopted to solve the problem of misclassification or abnormal classification of the concrete surface.

4. The unmanned aerial vehicle naked-eye 3D two-three-dimensional digital mapping method according to claim 1, characterized in that: The TIN triangular network is established, the classified point cloud data is saved as a new point cloud data file, the Delaunay triangle construction method is used, the maximum side length of the triangular network is not more than 50 m, the display resolution is X=1 m and Y=1 m, the ground point cloud data is converted into an irregular triangular network, i.e. the TIN triangular network data, and the triangular network file is saved; the three-dimensional contour line is established, according to the construction requirement, the topographic scale is 1 / 500, the main contour interval is 2.5 m, the contour interval is 0.5 m, the smoothing type of the contour line is selected as B-spline area, the smoothing degree is 70% to 85%, the three-dimensional contour line data is established based on the TIN triangular network, and the data format file is dxf.

5. The unmanned aerial vehicle naked-eye 3D two-three-dimensional digital mapping method according to claim 1, characterized in that: The vectorization parameters are established, the DOM image is imported into ArcGIS, the coordinate reference is set, the shapefile tool is used to create vectorization parameters, the forest land, road, farmland, river, building and greenhouse vectorization layer parameters are established respectively, different colors are used to distinguish various ground objects through layer style setting; The raster layer is vectorized, the topological analysis function of ArcGIS is used to perform topological analysis on the raster data, the association between the vector graphics and the raster graphics is established according to the topological analysis result, the operation process of raster-vector-raster layer-vectorization-raster-vectorization is formed, and for the regions without boundary coincidence, the vectorization processing is directly performed in the corresponding layer by using the area domain analysis tool; The engineering quantity is extracted, the vectorization layer data is selected, the land attribute, unit, quantity, area and perimeter fields are established by using data design, the vectorization layer is statistically analyzed, and the engineering quantity table is obtained.

6. The unmanned aerial vehicle naked-eye 3D two-three-dimensional digital mapping method according to claim 1, characterized in that: The vectorization files in shp, dxf and dwg formats are directly merged by using GIS, the merged file is directly used for the convenient road alignment and earthwork calculation, is directly added to the Ovi software for viewing, is directly merged with the high-resolution DOM image for output, and is used by the construction personnel.

Images