Unmanned aerial vehicle surveying and mapping method and system based on three-dimensional point cloud data analysis

Abstract

The invention discloses an unmanned aerial vehicle surveying and mapping method and system based on three-dimensional point cloud data analysis, and relates to the technical field of unmanned aerial vehicle surveying and mapping. Determining the altitude of a key position node, and constructing an initial three-dimensional model; analyzing the laser data, generating a ground point cloud model, and screening an adjustment strategy with a high initial matching degree through proportional adjustment, rotation and translation optimization; and comparing the ground image with the surveying and mapping block features, determining a secondary matching degree, selecting an optimal point cloud model, and outputting data in combination with surveying and mapping marks. According to the method, GIS positioning, node modeling, point cloud optimization and image calibration are creatively fused, the complex terrain surveying and mapping precision is improved, and the method is suitable for the fields of engineering planning and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of UAV mapping, and in particular to a UAV mapping method and system based on three-dimensional point cloud data analysis. Background Art

[0002] With the rapid development of UAV technology, UAV mapping has been widely used in fields such as geographic information collection, terrain modeling, and urban planning. Traditional UAV mapping methods mainly rely on optical images or single lidar data to generate terrain models through two-dimensional image stitching or sparse point clouds. However, in complex terrain environments (such as mountainous areas, hilly areas, or urban fringe areas), due to factors such as large terrain undulations, complex edge features, and vegetation occlusion, traditional methods often face problems such as insufficient mapping accuracy, difficulty in identifying key feature points, and low model matching efficiency.

[0003] In recent years, three-dimensional point cloud technology has attracted much attention because it can directly reflect the spatial structure characteristics of the earth's surface. Combining lidar-based point cloud data with a UAV platform can obtain high-density three-dimensional coordinate information, providing the possibility for constructing a refined terrain model. However, there are still several limitations in the actual application of existing technologies. First, it is difficult to effectively align single point cloud data with a geographic information system (GIS) or other reference models, resulting in a large model deviation. Second, in the identification of key position nodes, traditional methods mostly rely on manual annotation or simple threshold segmentation, which is difficult to adapt to terrain diversity and affects the accuracy of the initial model. Summary of the Invention

[0004] The purpose of the present invention is to provide a UAV mapping method and system that can more accurately map the ground.

[0005] The present invention discloses a UAV mapping method based on three-dimensional point cloud data analysis, including: Step S100, using a GIS information system to obtain a geographic image at the same height as the UAV, and demarcating a mapping block that needs to be mapped on the geographic image; Step S200, determining the altitude of key position nodes on the mapping block, and constructing an initial mapping three-dimensional model based on the altitude of the key position nodes; Step S300, analyzing the laser detection data of the UAV, constructing a ground three-dimensional point cloud model, and performing several initial model adjustments on the ground three-dimensional point cloud model. Each initial model adjustment includes scale adjustment, rotation, and translation. After each initial model adjustment, calculate the initial matching degree between the ground three-dimensional point cloud model and the initial mapping three-dimensional model, and screen and save the initial model adjustment strategies with an initial matching degree greater than or equal to a preset value; Step S400: Compare and analyze the image features of the ground images captured by the drone and the surveying block, and based on the analysis results, determine the secondary matching degree between the ground three-dimensional point cloud model and the initial surveying three-dimensional model. Based on the secondary matching degree, select the ground three-dimensional point cloud model corresponding to the optimal initial model adjustment strategy, and determine the output of the three-dimensional point cloud model surveying data based on the surveying marks on the surveying block.

[0006] In some embodiments of the present invention, the method for determining the key position nodes on the surveying block includes: Step S201: Perform edge detection on the surveying block, and combine with DEM data to determine the geographical height beside each edge. Step S202: Determine the edge continuous length and edge tortuous characteristics of each edge, and determine the geographical height difference characteristics on both sides of the edge. Step S203: Based on the edge continuous length, edge tortuous characteristics, and geographical height difference characteristics on both sides of the edge, determine the edge criticality of the edge, and identify the edge with edge criticality meeting the preset standard as the key edge. Analyze the geographical height change rate of the key edge, identify the edge section with the geographical height change rate greater than or equal to the preset value as the key edge section, set several key position nodes at preset intervals for the key edge section, identify the edge section outside the key edge section as the non-key edge section, and select the center point of the non-key edge section as the key position node.

[0007] In some embodiments of the present invention, the method for determining the edge tortuous characteristics includes: Step S2031: Uniformly set several first edge detection points on the edge, randomly select several first edge detection points, denoted as random detection points, and randomly combine the random detection points in pairs to obtain several random edge detection point groups. Connect the random edge detection points in each random edge detection point group to obtain several random edge detection lines. Step S2032: Randomly combine the random edge detection lines, determine the included angle between the random edge detection lines in the combination, and calculate the average value of all the included angles, denoted as the average included angle. Identify the average included angle as the edge tortuous parameter.

[0008] In some embodiments of the present invention, the method for determining the geographical height difference characteristics on both sides of the edge includes: Step S2033: Uniformly set several second edge detection points on the edge, and set height detection points on both sides of each second edge detection point, and the connection line between the height detection points is perpendicular to the edge. Step S2034: Calculate the height difference between relative height detection points, screen out the height differences greater than or equal to a preset value, denote them as the height differences to be concerned, identify the second edge detection points corresponding to the height differences to be concerned as the edge detection points to be concerned, calculate the proportion of the number of edge detection points to be concerned corresponding to the edge detection points to be concerned among all the second edge detection points, and identify the proportion as the geographical height difference parameter.

[0009] In some embodiments of the present invention, the method for determining the edge criticality of an edge based on the edge continuous length, edge tortuosity feature, and geographical height difference feature on both sides of the edge includes: Step S2035: Perform parametric definition conversion on the edge tortuosity feature to obtain an edge tortuosity parameter, and perform parametric definition conversion on the geographical height difference feature on both sides of the edge to obtain a geographical height difference parameter; Step S2036: Determine the edge critical degree of the edge based on the edge continuous length, edge tortuosity parameter, and edge height difference parameter of the edge.

[0010] In some embodiments of the present invention, the method for constructing an initial surveying and mapping three-dimensional model and a ground three-dimensional point cloud model includes: Step S204: Construct a three-dimensional coordinate system, and map the key position nodes into the three-dimensional coordinate system based on the belonging positions and altitudes of the key position nodes to form an initial surveying and mapping three-dimensional model; Step S205: Analyze the laser detection data, and construct a laser point cloud in the three-dimensional coordinate system to form a ground three-dimensional point cloud model.

[0011] In some embodiments of the present invention, the method for determining the initial matching degree between the ground three-dimensional point cloud model and the initial surveying and mapping three-dimensional model includes: Step S301: Overlap the initial surveying and mapping three-dimensional model and the ground three-dimensional point cloud model, determine the laser mapping point closest to each key position node in the ground three-dimensional point cloud model, denote the combination of the two as a model mapping point group, and calculate the inter-point distance; Step S302: Determine the initial matching degree based on the positional relationship between the key position nodes and the inter-point distances of the model mapping point groups corresponding to each key position node.

[0012] In some embodiments of the present invention, the method for determining the secondary matching degree between the ground three-dimensional point cloud model and the initial surveying and mapping three-dimensional model includes: Step S401: Grayscale the ground image and the surveying and mapping block respectively to obtain a ground image grayscale map and a surveying and mapping block grayscale map, and overlap the ground image grayscale map and the surveying and mapping block grayscale map; Step S402: Randomly select a number of first grayscale map blocks from the ground image grayscale map, and determine the first average grayscale value and the first average grayscale change rate of the first grayscale map block; Step S403: Perform dynamic block scanning on the surveyed block grayscale map, and determine the second average grayscale value and the second average grayscale change rate within the dynamic block during each dynamic block stay analysis. If the second average grayscale value of the dynamic block is the same as the first average grayscale value and the second average grayscale change rate is the same as the first average grayscale change rate, then the dynamic block at this time is identified as the second grayscale block, calculate the block distance between the first grayscale block and the second grayscale block, and retain the second grayscale blocks with a block distance less than or equal to the preset value; Step 404: Statistically analyze the block distance between the corresponding first grayscale block and the second grayscale block, and determine the proximity parameter between the first grayscale block and the second grayscale block based on the preset block distance interval to which the block distance belongs. Statistically analyze the sum of the proximity parameters corresponding to all block distances, which is recorded as the secondary matching degree.

[0013] In some embodiments of the present invention, a drone surveying and mapping system based on three-dimensional point cloud data analysis is also disclosed, including: The first module is used to obtain the geographical image at the same altitude of the drone by using the GIS information system, and demarcate the surveyed block that needs to be surveyed on the geographical image; The second module is used to determine the altitude of the key position nodes on the surveyed block, and construct an initial surveyed three-dimensional model based on the altitude of the key position nodes; The third module is used to analyze the laser detection data of the drone, construct a ground three-dimensional point cloud model, and perform several initial model adjustments on the ground three-dimensional point cloud model. Each initial model adjustment includes scale adjustment, rotation, and translation. After each initial model adjustment, calculate the initial matching degree between the ground three-dimensional point cloud model and the initial surveyed three-dimensional model, and screen and save the initial model adjustment strategies with an initial matching degree greater than or equal to the preset value; The fourth module is used to perform image feature comparison and analysis on the ground image and the surveyed block captured by the drone, and determine the secondary matching degree between the ground three-dimensional point cloud model and the initial surveyed three-dimensional model based on the analysis result. Based on the secondary matching degree, select the ground three-dimensional point cloud model corresponding to the optimal initial model adjustment strategy, and determine the output of the three-dimensional point cloud model surveying and mapping data based on the surveying marks on the surveyed block.

[0014] The present invention discloses a UAV mapping method and system based on three-dimensional point cloud data analysis, which relates to the technical field of UAV mapping. It uses GIS to obtain geographical images at the same height as the UAV and delimits mapping blocks; determines the altitude of key position nodes and constructs an initial three-dimensional model; analyzes laser data to generate a ground point cloud model, optimizes it through scale adjustment, rotation, and translation, and screens adjustment strategies with a high initial matching degree; determines the secondary matching degree by comparing the ground image with the characteristics of the mapping block, selects the optimal point cloud model, and outputs data in combination with mapping marks. The present invention innovatively integrates GIS positioning, node modeling, point cloud optimization, and image calibration, improves the mapping accuracy of complex terrains, and is applicable to fields such as engineering planning.

[0015] The technical solution of the present invention will be further described in detail below through the accompanying drawings and embodiments. Description of the Drawings

[0016] Figure 1 It is a method step diagram of the UAV mapping method based on three-dimensional point cloud data analysis disclosed in the embodiment of the present invention. Detailed Embodiments

[0017] The technical solution of the present invention will be further described below through the accompanying drawings and embodiments.

[0018] Hereinafter, the technical solution of the present invention will be clearly and completely described in conjunction with the accompanying drawings and specific embodiments. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention and should not be construed as limiting the protection scope of the present invention. Those skilled in the art can make some non-essential improvements and adjustments based on the content of the present invention described below. In the present invention, unless otherwise clearly specified and limited, the technical terms used in the present invention should have the ordinary meaning understood by those skilled in the art of the present invention.

[0019] Embodiment:

[0020] The present invention discloses a UAV mapping method based on three-dimensional point cloud data analysis. Refer to Figure 1 , including: Step S100, using a GIS information system, obtaining a geographical image at the same height as the UAV, and delimiting a mapping block that needs to be mapped on the geographical image.

[0021] The principle of step S100 is to use Geographic Information System (GIS) to obtain the geographical image corresponding to the flight altitude of the drone, and on this basis, delimit the blocks to be surveyed. The core of this process lies in converting the geographical data in the real world into operable digital information, laying the foundation for subsequent three-dimensional point cloud analysis. The GIS system can generate a two-dimensional geographical image consistent with the drone's flight altitude by integrating satellite images, map data, and elevation information. This image not only contains surface features (such as rivers, mountains, buildings, etc.), but can also correspond to the actual geographical location through the coordinate system. Delimiting the survey blocks means selecting specific areas on the geographical image according to the task requirements, such as a mountainous area or a building complex area that needs to be surveyed for topography. This process is similar to "selecting" the target area on a digital map, ensuring that subsequent survey work focuses on the key areas. For example, assuming that the slope change of a hillside needs to be surveyed, the GIS system can provide satellite images of the area and delimit the scope of the hillside in combination with the task requirements. Through this step, the system can clarify the boundaries and scope of the survey, avoid redundant data collection, and at the same time provide a spatial reference for the identification of subsequent key nodes and model construction. The application of GIS improves the accuracy and efficiency of the survey because it not only provides visual geographical information, but also can preset initial conditions for three-dimensional modeling through data layer superposition (such as elevation data). The essence of this process is to abstract the complex real geographical environment into a computable two-dimensional plane, providing a clear "starting line" for drone surveying.

[0022] In step S200, determine the altitude of the key position nodes on the surveyed blocks, and construct an initial three-dimensional survey model based on the altitude of the key position nodes.

[0023] The principle of step S200 is to construct an initial surveying and mapping 3D model by determining the elevation of key position nodes within the surveying and mapping block, providing a reference benchmark for subsequent point cloud data analysis. The core of this process lies in extracting 3D information from 2D geographical images and using key position nodes as representative points of terrain features. The determination of key position nodes usually depends on the significance of terrain features, such as ridges, valleys, or surface mutations, which can reflect the main height changes in the surveying area. By combining Digital Elevation Model (DEM) data, accurate elevation values can be assigned to each key node. For example, when surveying a hilly area, the mountaintop, the foot of the mountain, and the slope turning points may be selected as key nodes, and the height data of these points (such as 500 meters for the mountaintop and 200 meters for the foot of the mountain) can be obtained from the DEM. Subsequently, based on the elevation and geographical coordinates of these nodes, an initial surveying and mapping model is constructed in a 3D coordinate system. This model is essentially a rough 3D framework, similar to a terrain "skeleton" built with several key support points. Its function is to provide a comparison benchmark for subsequent laser point cloud data, avoiding the deviation that may occur when directly modeling using the raw data collected by the drone. For example, if directly modeling with point cloud data, terrain distortion may occur due to vegetation occlusion or equipment errors, while the initial model presets the overall contour of the terrain through key nodes. The key to this process lies in the representativeness of node selection and the accuracy of elevation data, which together determine whether the initial model can effectively guide subsequent adjustments. Through this step, the surveying and mapping system can establish a preliminary spatial awareness in complex terrains, laying a foundation for refined modeling.

[0024] In step S300, analyze the laser detection data of the drone, construct a ground 3D point cloud model, and perform several initial model adjustments on the ground 3D point cloud model. Each initial model adjustment includes scale adjustment, rotation, and translation. After each initial model adjustment, calculate the initial matching degree between the ground 3D point cloud model and the initial surveying and mapping 3D model, and screen and save the initial model adjustment strategies with an initial matching degree greater than or equal to the preset value.

[0025] The principle of step S300 is to construct a ground three-dimensional point cloud model by analyzing the laser detection data of the UAV, and make it align with the initial surveying and mapping model through multiple adjustments, so as to improve the surveying and mapping accuracy. The core of this process lies in converting the discrete point cloud data collected by lidar into a continuous three-dimensional terrain representation, and realizing the fusion with the initial model through iterative optimization. The laser detection data contains a large number of three-dimensional coordinate points, reflecting the true shape of the ground surface, such as ground undulations, vegetation heights or building outlines. The system first maps these points into a three-dimensional coordinate system to form a ground point cloud model. However, since the initial model is only based on key nodes, its details and accuracy are limited. Therefore, it is necessary to adjust the point cloud model. The adjustments include scale scaling (matching the overall scale), rotation (correcting the direction deviation), and translation (aligning the spatial position). After each adjustment, the initial matching degree of the two models is calculated. The matching degree may be evaluated by the distance between points or the overlap rate. For example, if the average distance between the ground points in the point cloud model and the nodes of the initial model is less than 0.3 meters, the matching degree is relatively high. The system filters out the adjustment strategies with a matching degree greater than a preset value (such as 85%) and saves them. For example, when surveying a farmland, the initial model may only reflect the positions of the ridges, while the point cloud model contains details of crops and soil. By adjusting, the crop interference can be eliminated and the real ground can be approximated. The significance of this process is to use the high resolution of the point cloud data to make up for the roughness of the initial model, and at the same time reduce errors through multiple iterative optimizations. For example, if the initial model is deflected by 5 degrees, the rotation adjustment can correct it and improve the coincidence degree of the two models. This step is essentially the preliminary fusion of the point cloud data and the preset framework, providing a reliable intermediate model for subsequent image feature comparison, ensuring that the surveying and mapping results not only conform to the macroscopic terrain but also have microscopic accuracy.

[0026] Step S400: Perform image feature comparison and analysis on the ground images taken by the UAV and the surveying and mapping block. Based on the analysis results, determine the secondary matching degree between the ground three-dimensional point cloud model and the initial surveying and mapping three-dimensional model. Based on the secondary matching degree, select the ground three-dimensional point cloud model corresponding to the optimal initial model adjustment strategy, and determine the output of the three-dimensional point cloud model surveying and mapping data based on the surveying and mapping marks on the surveying and mapping block.

[0027] The principle of step S400 is to perform feature comparison and analysis on the ground images captured by the drone with the surveying and mapping block, determine the secondary matching degree between the ground three-dimensional point cloud model and the initial surveying and mapping model, and based on this, select the ground three-dimensional point cloud model optimized by the optimal initial model adjustment strategy. Finally, surveying and mapping data is output in combination with surveying and mapping marks. The core of this process lies in using image information to perform secondary verification on the multiple adjusted point cloud models generated in step S300 to ensure their alignment accuracy with the actual terrain. The significance of this process is to introduce an additional calibration dimension through image features to make up for possible local deviations in the laser point cloud (such as height errors caused by vegetation occlusion), so as to select the model that best fits the real terrain. For example, in forest areas, image comparison can help distinguish the ground from the tree canopy and optimize the point cloud model. The finally output surveying and mapping data not only reflects the terrain height, but also is accurately aligned with the actual landform features, and is applicable to scenarios such as agricultural planning or disaster assessment.

[0028] In some embodiments of the present invention, the method for determining key position nodes on the surveying and mapping block includes: Step S201, perform edge detection on the surveying and mapping block, and in combination with DEM data, determine the geographical height beside each edge.

[0029] The principle of step S201 is to identify significant terrain boundaries within the surveyed area through edge detection technology, and combine Digital Elevation Model (DEM) data to assign precise geographical heights to both sides of these boundaries, providing basic data for the determination of subsequent key nodes. The core of this process lies in transforming the two-dimensional geographical image of the surveyed area into a structured feature containing height information, using edges as the initial signs of terrain changes. Edge detection usually employs image processing algorithms (such as Canny or Sobel operators), which identify distinct dividing lines of surface features by analyzing pixel grayscale or color gradients, such as ridgelines, river edges, or sudden slope changes. For example, when surveying a hilly area, edge detection may identify a ridgeline, which appears as a significant change in brightness or color on the geographical image. Combining DEM data, the system assigns height values to both sides of this edge. For instance, the altitude on the left side of the ridgeline is 450 meters, and on the right side is 400 meters. The introduction of this height information is crucial because it transforms the two-dimensional edge into a representative of three-dimensional terrain features, providing a spatial dimension for subsequent analysis. For example, assume the survey target is a coastal wetland. Edge detection may identify the boundary between the coastline and the inland marsh, and the DEM data shows that the height of the coastline is 2 meters and that of the marsh is 0.5 meters. The significance of this process is to screen out the key contours of the terrain through edge detection and supplement the details with the elevation accuracy of the DEM, avoiding height changes that may be overlooked by relying solely on image information. The determination of the heights on the sides of the edge not only lays the foundation for the quantification of terrain features but also provides data support for the assessment of the importance of edges in subsequent steps. Through this step, the system can extract preliminary structured information from the complex geographical environment, laying a solid foundation for the positioning of key location nodes.

[0030] In step S202, determine the edge continuous length and edge tortuosity characteristics of each edge, and determine the geographical height difference characteristics on both sides of the edge.

[0031] The principle of step S202 is to conduct feature quantification analysis on each edge recognized in step S201, determine its continuous length, tortuous feature, and the geographical height difference on both sides, so as to comprehensively describe the topographic significance of the edge and provide a basis for key evaluation. The core of this process lies in extracting computable parameters from the morphological and height characteristics of the edge to ensure that the subsequent screening can reflect the actual changes in the terrain. The continuous length of the edge refers to the distance that the edge extends in the geographical space. For example, a ridge line may continuously extend for 3 kilometers, reflecting the size of its coverage. The tortuous feature describes the shape complexity of the edge and is quantified by analyzing the degree of bending or the frequency of turning of the edge. For example, the edge of a smooth river may have a low tortuosity, while the edge of a serrated fault has a high tortuosity. The geographical height difference on both sides of the edge is calculated by comparing the height values on both sides of the edge in the DEM data. For example, one side of a ridge is 500 meters and the other side is 420 meters, with a difference of 80 meters. For example, when mapping a forest mountain area, the system may identify a valley edge with a continuous length of 1.5 kilometers, a tortuous feature manifested as multiple sharp turns (high tortuosity), and a height difference of 60 meters on both sides. The determination of these features requires comprehensive image analysis and height data processing to ensure that the results reflect both the geometric characteristics of the edge and its topographic significance. The significance of this process is to abstract the edge from a simple line into a topographic element with multi-dimensional attributes, providing a quantitative basis for subsequent key judgments. For example, a short and straight edge may only represent local undulations, while a long, tortuous, and large-height-difference edge may indicate an important topographic boundary. Through this step, the system can refine the description of the edge and provide comprehensive data support for the screening of key nodes.

[0032] Step S203: Based on the continuous length of the edge, the tortuous feature of the edge, and the geographical height difference feature on both sides of the edge, determine the edge criticality of the edge, and identify the edge whose edge criticality meets the preset standard as the key edge. Analyze the geographical height change rate of the key edge, identify the edge section with a geographical height change rate greater than or equal to the preset value as the key edge section, set several key position nodes at preset intervals for the key edge section, identify the edge section outside the key edge section as the non-key edge section, and select the center point of the non-key edge section as the key position node.

[0033] The principle of step S203 is to comprehensively evaluate the criticality of an edge based on the continuous length, tortuous features, and height differences on both sides of the edge, screen out the critical edges, and further analyze their height change rates to determine the critical position nodes. The core of this process lies in identifying the most representative regions in the terrain through weighted analysis of multi-dimensional features and reasonably distributing critical nodes based on this. First, the system calculates the criticality score of the edge according to preset criteria (such as a length greater than 1 km, a tortuosity higher than a certain threshold, and a height difference greater than 30 m), and identifies the edges that meet the conditions as critical edges. For example, a ridge line that is 2 km long, has obvious tortuosity, and a height difference of 50 m may be rated as a critical edge. Then, analyze the geographical height change rate of the critical edge, that is, the height change per unit distance. For example, if the height changes by 4 m every 10 m, and the change rate is greater than the preset value (such as 2 m / 10 m), then this section is identified as a critical edge section. In these sections, critical nodes are set at preset intervals (such as 50 m) to ensure capturing the sharp change points of the terrain. For non-critical edge sections (such as gentle sections), the midpoint is taken as the critical node to cover relatively flat areas. For example, when mapping the edge of a river, a node is set every 50 m in the steep riverbank section (with a large height difference and a high change rate), and the midpoint is taken in the gentle section (with a small height difference). The significance of this process is to accurately locate the terrain feature points through criticality screening and change rate analysis, avoiding over-dense node distribution or omission of important areas. For example, in mountain mapping, the dense distribution of nodes on steep slopes can reflect the terrain details, while a single node on gentle slopes is sufficient to represent the overall trend. Through this step, the system transforms the edge features into a specific set of three-dimensional nodes, providing reliable support points for the construction of the initial mapping model and being applicable to the precise mapping of complex terrains.

[0034] In some embodiments of the present invention, the method for determining the tortuous features of an edge includes: Step S2031, evenly set a number of first edge detection points on the edge, randomly select a number of first edge detection points, denoted as random detection points, and randomly combine the random detection points in pairs to obtain a number of random edge detection point groups, and connect the random edge detection points in each random edge detection point group to obtain a number of random edge detection lines.

[0035] The principle of step S2031 is to generate a set of random edge detection lines by uniformly setting detection points on the edge and making random combinations, providing a geometric basis for subsequent quantification of the edge tortuosity characteristics. The core of this process lies in discretizing the continuous form of the edge into a set of computable points and lines for analyzing its shape complexity. First, the system uniformly distributes a number of first edge detection points on the edge. For example, one point is set every 10 meters along a 1-kilometer-long ridge line, with a total of 100 detection points. The positions of these points are based on the actual path of the edge to ensure coverage of the entire edge range. Subsequently, a number of points are randomly selected from these detection points as random detection points. For example, 10 points are randomly selected and denoted as P1, P2, …, P10. Then, these random detection points are pairwise combined to form random edge detection point groups, such as (P1, P2), (P2, P3), (P5, P7), etc. A random edge detection line is generated by connecting the two points in each group. Assuming the coordinates of P1 are (0, 0), P2 are (10, 5), and P3 are (15, 15), the connecting lines P1 - P2 and P2 - P3 are respectively a straight-line segment, and a total of dozens of such detection lines may be generated. The significance of this process is to decompose the overall shape of the edge into multiple local line segments through random sampling and connection, reflecting the possibility of its direction change. For example, when mapping the edge of a meandering river, the detection points may be distributed at the bends of the riverbank, and the detection lines generated after connection will show different inclination angles, while a straight edge may generate nearly parallel line segments.

[0036] In step S2032, the random edge detection lines are randomly combined, the included angle between the random edge detection lines in the combination is determined, and the average value of all the included angles is calculated and denoted as the average included angle. The average included angle is regarded as the edge tortuosity parameter.

[0037] The principle of step S2032 is to quantify the curvature of the edge by calculating the line angles between the random edge detection lines generated in step S2031 and taking the average value as the edge tortuosity parameter. The core of this process is to use the change of geometric angles to reflect the tortuosity characteristics of the edge, providing an important basis for the determination of key position nodes. First, the system randomly combines the random edge detection lines generated in step S2031, for example, selects several pairs from 10 detection lines, such as (P1-P2, P2-P3), (P3-P4, P5-P6). For each pair of detection lines, the line angle between them is calculated. For example, the direction vector of the P1-P2 line segment is (10, 5), and the direction vector of the P2-P3 line segment is (5, 10). The angle is calculated to be about 45 degrees by the vector angle formula (such as the cosine theorem). Then, the system counts the line angles of all combinations and calculates the average value, which is recorded as the average line angle. For example, if the edge of a ridge is mapped and the angles of 5 pairs of detection lines are generated, which are 30°, 60°, 45°, 20°, and 50° respectively, then the average angle between the lines is (30+60+45+20+50) / 5=41°, which is defined as the edge tortuosity parameter. The significance of this parameter is that the larger the angle, the more dramatic the change in the direction of the edge and the higher the degree of tortuosity. For example, the edge of a straight road may generate an average angle of nearly 0°, with a low tortuosity parameter; while the edge of a winding mountain road may have an average angle of 60°, with a high tortuosity parameter, reflecting its terrain complexity. This process converts the morphological characteristics of the edge into a single value through geometric analysis, which is convenient for combining with other features (such as length and elevation difference) to evaluate the criticality of the edge. For example, when mapping the edge of a forest, if the average angle between the lines is 55°, it means that the boundary is tortuous and changeable, and may contain multiple turning points, which requires special attention. Through this step, the system realizes the quantification of edge tortuosity characteristics, providing a scientific basis for subsequent key edge screening.

[0038] In some embodiments of the present invention, the method for determining the geographical height difference characteristics on both sides of the edge includes: Step S2033, a plurality of second edge detection points are evenly set on the edge, and height detection points are set on both sides of each second edge detection point, and the connecting line between the height detection points is perpendicular to the edge; Step S2034, calculate the height difference between the relative height detection points, and screen out the height difference that is greater than or equal to the preset value, record it as the height difference that needs attention, and identify the second edge detection point corresponding to the height difference that needs attention as the edge detection point that needs attention, calculate the ratio of the number of edge detection points that need attention to all second edge detection points, and identify the number ratio as the geographical height difference parameter.

[0039] The principle of step S2034 is to screen out significant difference regions by calculating the height difference between height detection points, and quantify the geographical height difference characteristics on both sides of the edge in the form of quantity proportion, providing a basis for edge criticality assessment. The core of this process lies in extracting feature points with topographic significance from local height data and converting them into global parameters. First, the system calculates the height difference between height detection points on both sides of each second edge detection point. For example, if the height on the left side of a detection point is 450 meters and the height on the right side is 420 meters, the difference is 30 meters. Then, a preset value (such as 20 meters) is set, and the height differences greater than or equal to this value are screened out and recorded as the height differences to be concerned. This detection point is identified as an edge detection point to be concerned. For example, among 50 detection points, if the height differences of 10 points exceed 20 meters (such as 30 meters, 25 meters, etc.), then these 10 points are the points to be concerned. Subsequently, calculate the proportion of the edge detection points to be concerned among all second edge detection points. For example, 10 / 50 = 20%, and this proportion is the geographical height difference parameter. The significance of this parameter is that it reflects the significant degree and distribution frequency of height changes on both sides of the edge. For example, when mapping the edge of a gentle ridge, if only 2 points have height differences exceeding 20 meters, accounting for only 4%, it indicates that the height difference is not significant; while at the edge of a steep cliff, if 40 points have exceeded the standard height differences, accounting for 80%, it indicates a drastic terrain change. Through this process of screening and proportion calculation, local height differences are converted into a feature description of the overall edge. For example, in coastal cliff mapping, the detection points may show that the height difference between the cliff top (50 meters) and the bottom (5 meters) reaches 45 meters, and most points are marked as points to be concerned, with a high proportion, reflecting a strong terrain boundary. Through this step, the system quantifies the height difference characteristics on both sides of the edge, providing a scientific basis for the subsequent identification of critical edges and being applicable to the analysis of complex terrains.

[0040] In some embodiments of the present invention, the method for determining the edge criticality of an edge based on the edge continuous length, edge tortuosity characteristics, and geographical height difference characteristics on both sides of the edge includes: Step S2035, perform a parametric definition conversion on the edge tortuosity characteristics to obtain an edge tortuosity parameter, and perform a parametric definition conversion on the geographical height difference characteristics on both sides of the edge to obtain a geographical height difference parameter.

[0041] Step S2036, determine the edge critical degree of the edge based on the edge continuous length, edge tortuosity parameter, and edge height difference parameter of the edge.

[0042] Among them, the expression for calculating the edge critical degree is: .

[0043] Among them, G is the edge critical degree of the edge, L is the edge continuous length, is the weight adjustment coefficient of the edge tortuosity parameter, is the weight adjustment coefficient of the edge height difference parameter, is the edge tortuosity parameter, is the edge height difference parameter, and b is the edge feature influence adjustment coefficient.

[0044] Among them, the edge continuity length L represents the extension scale of the edge in space (such as the continuous length of a mountain ridge or a valley), which is a global multiplier that directly determines the benchmark magnitude of the criticality; the edge tortuosity parameter q describes the bending complexity of the edge morphology, which is usually calculated by the mean of the angles between adjacent line segments or the curvature integral. The larger the value, the more drastic the change in the edge direction (such as the q of a jagged gully is significantly higher than that of a gentle edge); the edge height difference parameter h characterizes the elevation mutation of the terrain on both sides of the edge, which can be quantified by the absolute value of the height difference or the gradient change rate (such as the height difference h of a cliff boundary is much greater than that of a gentle slope); the weight adjustment coefficient and Control the contribution ratio of tortuosity and height difference respectively (e.g. =0.7, =0.3, more emphasis is placed on morphological complexity); the adjustment coefficient b is used as a bias term for the exponential term to compensate for the basic differences between different terrain categories (for example, b=-0.5 is set in plain areas to reduce the criticality of the edge of flat areas). The formula converts the parameters of the linear combination into nonlinearly growing criticality values ​​through exponential operations. For example, when an edge of 800m long has q=0.6, h=1.0, and =0.5, =0.4, b=0.2, G≈1968 is calculated, and when it is significantly higher than the threshold, it is determined to be a key edge. This design can adaptively enhance the weight of the edges of complex terrains such as steep mountains and dense gullies, while suppressing redundant interference on flat terrains, providing accurate feature screening basis for UAV mapping and 3D model reconstruction.

[0045] In some embodiments of the present invention, the method for constructing an initial surveying and mapping three-dimensional model and a ground three-dimensional point cloud model includes: Step S204, constructing a three-dimensional coordinate system, and mapping the key position nodes into the three-dimensional coordinate system based on the positions and altitudes of the key position nodes, to form an initial surveying and mapping three-dimensional model.

[0046] Step S205 , analyzing the laser detection data, constructing a laser point cloud in a three-dimensional coordinate system, and forming a three-dimensional point cloud model of the ground.

[0047] In some embodiments of the present invention, the method for determining the initial matching degree between the ground 3D point cloud model and the initial surveying 3D model includes: Step S301: Overlap the initial surveyed 3D model and the ground 3D point cloud model, determine the laser mapping points in the ground 3D point cloud model that are closest to each key position node, denote the combination of the two as the model mapping point group, and calculate the distance between points.

[0048] Step S302: Determine the initial matching degree based on the positional relationship between key position nodes and the distance between points in the model mapping point group corresponding to each key position node.

[0049] Among them, the expression for calculating the initial matching degree is: .

[0050] Among them, is the initial matching degree, is the matching parameter determination function corresponding to the i-th key position node, Based on the preset distance interval to which the distance between points corresponding to the i-th key position node belongs, output the corresponding matching parameter. Among them, the smaller the distance between points, the larger the matching parameter. n is the total number of key position nodes. is the continuous matching judgment function corresponding to the i-th key position node. If the distance between the i-th key position node and the closest adjacent key position node is less than or equal to the preset value, and the distances between points corresponding to both are less than or equal to the preset value, then output 1, otherwise output 0. R is the continuous matching influence adjustment coefficient, and c is the continuous matching influence adjustment constant.

[0051] The formula quantifies the spatial alignment quality between the ground 3D point cloud model and the reference model through the synergistic effect of discrete key point matching accuracy and continuous region matching consistency. Among them, the discrete matching term represents the sum of the matching parameters of all key position nodes (such as mountaintop points, valley points). Its core logic is that the smaller the distance between points, the larger the matching parameter β(i) (for example, when the distance d ≤ 0.5m, β(i) = 1; when 0.5 < d ≤ 1m, β(i) = 0.5), which directly reflects the positioning accuracy of a single node. The continuous matching enhancement term amplifies the contribution of the continuous matching region through an exponential function. Among them, the continuous matching judgment function a(i) needs to satisfy two conditions simultaneously: the spatial distance between the i-th node and the closest adjacent node ≤ the preset value (such as 50m), and the distances between points of both are ≤ the preset threshold (such as 1m). When satisfied, a(i) = 1, otherwise it is 0. Its average value represents the proportion of the spatial continuity of successful adjacent node matching; the adjustment coefficient R controls the amplification intensity of continuous matching on the total score (for example, when R = 0.3, every 10% of the continuous matching rate can increase the total score by about 3.5%), and the adjustment constant c is used to set the basic amplification factor (such as c = 0.5 can avoid the exponential term approaching zero when the continuous matching rate is extremely low).

[0052] In some embodiments of the present invention, the method for determining the secondary matching degree between the ground three-dimensional point cloud model and the initial surveying and mapping three-dimensional model includes: Step S401: Grayscale the ground image and the surveying and mapping block respectively to obtain a grayscale ground image and a grayscale surveying and mapping block image, and overlap the grayscale ground image and the grayscale surveying and mapping block image. Step S402: Randomly select a number of first grayscale image blocks for the grayscale ground image, and determine the first average grayscale value and the first average grayscale change rate of the first grayscale image block.

[0053] Step S403: Perform dynamic block scanning on the grayscale surveying and mapping block image, and determine the second average grayscale value and the second average grayscale change rate within the dynamic block during each dynamic block stay analysis. If the second average grayscale value of the dynamic block is the same as the first average grayscale value, and the second average grayscale change rate is the same as the first average grayscale change rate, then the dynamic block at this time is identified as the second grayscale block, calculate the block spacing between the first grayscale block and the second grayscale block, and retain the second grayscale block whose block spacing is less than or equal to the preset value.

[0054] Step 404: Statistically calculate the block spacing between the corresponding first grayscale block and the second grayscale block, and determine the proximity parameter between the first grayscale block and the second grayscale block based on the preset block spacing interval to which the block spacing belongs. Statistically calculate the sum of the proximity parameters corresponding to all block spacings, which is recorded as the secondary matching degree.

[0055] In some embodiments of the present invention, a drone surveying and mapping system based on three-dimensional point cloud data analysis is also disclosed, including: The first module is used to obtain geographical images at the same altitude of the drone by using the GIS information system, and demarcate the surveying and mapping blocks that need to be surveyed and mapped on the geographical images. The second module is used to determine the altitude of the key position nodes on the surveying and mapping blocks, and construct an initial surveying and mapping three-dimensional model based on the altitude of the key position nodes. The third module is used to analyze the laser detection data of the drone, construct a ground three-dimensional point cloud model, and perform several initial model adjustments on the ground three-dimensional point cloud model. Each initial model adjustment includes scale adjustment, rotation, and translation. After each initial model adjustment, calculate the initial matching degree between the ground three-dimensional point cloud model and the initial surveying and mapping three-dimensional model, and screen and save the initial model adjustment strategies whose initial matching degree is greater than or equal to the preset value. The fourth module is used to perform image feature comparison and analysis on the ground images captured by the drone and the surveying and mapping block, and based on the analysis results, determine the secondary matching degree between the ground three-dimensional point cloud model and the initial surveying and mapping three-dimensional model. Based on the secondary matching degree, select the ground three-dimensional point cloud model corresponding to the optimal initial model adjustment strategy, and determine the output of the three-dimensional point cloud model surveying and mapping data based on the surveying and mapping marks on the surveying and mapping block.

[0056] Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented by hardware or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various implementation scenarios of the present invention.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A UAV surveying and mapping method based on three-dimensional point cloud data analysis, characterized in that: include: Step S100, using the GIS information system to obtain a geographic image at the same altitude as the drone, and delineate a surveying and mapping block that needs to be surveyed and mapped on the geographic image; Step S200, determining the altitude of the key position nodes on the surveying and mapping block, and constructing an initial surveying and mapping three-dimensional model based on the altitude of the key position nodes; Step S300, analyzing the laser detection data of the drone, constructing a ground 3D point cloud model, and performing several initial model adjustments on the ground 3D point cloud model, each initial model adjustment includes scale adjustment, rotation and translation, and after each initial model adjustment, calculating the initial matching degree between the ground 3D point cloud model and the initial mapping 3D model, and screening and saving the initial model adjustment strategies with an initial matching degree greater than or equal to a preset value; Step S400, performing image feature comparison analysis on the ground image taken by the drone and the surveying and mapping block, and based on the analysis result, determining the secondary matching degree between the ground 3D point cloud model and the initial surveying and mapping 3D model, selecting the ground 3D point cloud model corresponding to the optimal initial model adjustment strategy based on the secondary matching degree, and determining the output of the 3D point cloud model surveying and mapping data based on the surveying and mapping marks on the surveying and mapping block.

2. The method for UAV mapping based on three-dimensional point cloud data analysis according to claim 1, characterized in that: Methods for determining key location nodes on the surveying and mapping block include: Step S201, edge detection is performed on the surveying and mapping block, and the geographical height beside each edge is determined in combination with DEM data; Step S202, determining the edge continuity length and edge meandering characteristics of each edge, and determining the geographical height difference characteristics on both sides of the edge; Step S203, based on the edge continuous length, edge tortuosity characteristics and geographical height difference characteristics on both sides of the edge, determine the edge criticality of the edge, and identify the edge whose edge criticality meets the preset standard as a critical edge, analyze the geographical height change rate of the critical edge, identify the edge segment with a geographical height change rate greater than or equal to a preset value as a critical edge segment, set a number of key position nodes for the key edge segment at preset intervals, identify the edge segment outside the key edge segment as a non-critical edge segment, and select the center point of the non-critical edge segment as the key position node.

3. The method for UAV mapping based on three-dimensional point cloud data analysis according to claim 2, characterized in that: Methods for determining edge tortuosity characteristics include: Step S2031, uniformly set a number of first edge detection points on the edge, randomly select a number of first edge detection points, record them as random detection points, randomly combine the random detection points in pairs to obtain a number of random edge detection point groups, and connect the random edge detection points in each random edge detection point group to obtain a number of random edge detection lines; Step S2032, randomly combine the random edge detection lines, determine the line angles between the random edge detection lines in the combination, and calculate the average value of all line angles, recorded as the average line angle, and identify the average line angle as the edge tortuosity parameter.

4. The method for surveying and mapping by unmanned aerial vehicle based on three-dimensional point cloud data analysis according to claim 2, characterized in that: Methods for determining the geographical height difference characteristics on both sides of the edge include: Step S2033, a plurality of second edge detection points are evenly set on the edge, and height detection points are set on both sides of each second edge detection point, and the connecting line between the height detection points is perpendicular to the edge; Step S2034, calculate the height difference between the relative height detection points, and screen out the height difference that is greater than or equal to the preset value, record it as the height difference that needs attention, and identify the second edge detection point corresponding to the height difference that needs attention as the edge detection point that needs attention, calculate the ratio of the number of edge detection points that need attention to all second edge detection points, and identify the number ratio as the geographical height difference parameter.

5. The method for surveying and mapping by unmanned aerial vehicle based on three-dimensional point cloud data analysis according to claim 2, characterized in that: Methods for determining edge criticality of an edge based on the edge continuous length, edge tortuosity characteristics, and geographic height difference characteristics on both sides of the edge include: Step S2035, performing parameter definition conversion on the edge tortuosity feature to obtain the edge tortuosity parameter, and performing parameter definition conversion on the geographic height difference feature on both sides of the edge to obtain the geographic height difference parameter; Step S2036, determining the edge criticality of the edge based on the edge continuous length, the edge tortuosity parameter and the edge height difference parameter of the edge.

6. The method for surveying and mapping by unmanned aerial vehicle based on three-dimensional point cloud data analysis according to claim 1, characterized in that: Methods for constructing an initial surveying and mapping 3D model and a ground 3D point cloud model include: Step S204, constructing a three-dimensional coordinate system, and mapping the key position nodes into the three-dimensional coordinate system based on the positions and altitudes of the key position nodes, to form an initial surveying and mapping three-dimensional model; Step S205 , analyzing the laser detection data, constructing a laser point cloud in a three-dimensional coordinate system, and forming a three-dimensional point cloud model of the ground.

7. The method for UAV surveying and mapping based on three-dimensional point cloud data analysis according to claim 1, characterized in that: Methods for determining the initial matching degree between the ground 3D point cloud model and the initial survey 3D model include: Step S301, overlap the initial surveying 3D model and the ground 3D point cloud model, determine the closest laser mapping point of each key position node in the ground 3D point cloud model, record the combination of the two as a model mapping point group, and calculate the distance between the points; Step S302, determining the initial matching degree based on the positional relationship between the key position nodes and the distance between the points of the model mapping point group corresponding to each key position node.

8. The method for UAV mapping based on three-dimensional point cloud data analysis according to claim 1, characterized in that: Methods for determining the secondary matching degree between the ground 3D point cloud model and the initial survey 3D model include: Step S401, graying the ground image and the surveying and mapping block respectively to obtain a ground image grayscale map and a surveying and mapping block grayscale map, and overlapping the ground image grayscale map and the surveying and mapping block grayscale map; Step S402, randomly selecting a number of first grayscale image blocks for the ground image grayscale image, and determining a first average grayscale value and a first average grayscale change rate of the first grayscale image blocks; Step S403, performing dynamic block scanning on the grayscale image of the surveying block, and determining the second average grayscale value and the second average grayscale change rate in the dynamic block each time the dynamic block stops for analysis, if the second average grayscale value of the dynamic block is the same as the first average grayscale value, and the second average grayscale change rate is the same as the first average grayscale change rate, then the dynamic block at this time is identified as the second grayscale block, the block spacing between the first grayscale block and the second grayscale block is calculated, and the second grayscale block whose block spacing is less than or equal to the preset value is retained; Step 404, counting the block spacing between the corresponding first grayscale block and the second grayscale block, and determining the proximity parameter between the first grayscale block and the second grayscale block based on the preset block spacing interval to which the block spacing belongs, and counting the sum of the proximity parameters corresponding to all block spacings, which is recorded as the secondary matching degree.

9. The UAV mapping system based on 3D point cloud data analysis is characterized by: include: The first module is used to use the GIS information system to obtain geographic images at the same altitude as the drone, and to delineate the surveying and mapping blocks that need to be surveyed on the geographic images; The second module is used to determine the altitude of the key position nodes on the surveying and mapping block, and to construct an initial surveying and mapping three-dimensional model based on the altitude of the key position nodes; The third module is used to analyze the laser detection data of the drone, construct a ground 3D point cloud model, and perform several initial model adjustments on the ground 3D point cloud model. Each initial model adjustment includes scale adjustment, rotation and translation. After each initial model adjustment, the initial matching degree between the ground 3D point cloud model and the initial mapping 3D model is calculated, and the initial model adjustment strategies with an initial matching degree greater than or equal to a preset value are screened and saved. The fourth module is used to compare and analyze the image features of the ground images taken by the drone and the surveying and mapping blocks, and based on the analysis results, determine the secondary matching degree between the ground 3D point cloud model and the initial surveying and mapping 3D model, and based on the secondary matching degree, select the ground 3D point cloud model corresponding to the optimal initial model adjustment strategy, and based on the surveying and mapping marks on the surveying and mapping blocks, determine the output of the 3D point cloud model surveying and mapping data.

Images