An unmanned aerial vehicle oblique photography flight route planning method and system under complex terrain

Abstract

The present application relates to the technical field of unmanned aerial vehicle remote sensing surveying and mapping, and provides a method and system for planning a tilt photography flight route of an unmanned aerial vehicle in complex terrain, comprising: obtaining polygon boundary data and slope data of a target survey area, and calculating a geometric center point coordinate of the target survey area based on the polygon boundary data; constructing a radial terrain framework based on the geometric center point coordinate and a pre-configured azimuth coordinate system; planning a flight detection route based on the radial terrain framework and the slope data, and controlling the unmanned aerial vehicle to perform a detection flight along the flight detection route to obtain elevation prior data of the target survey area; the flight detection route comprises a plurality of two-dimensional waypoints; and rearranging the plurality of two-dimensional waypoints in three-dimensional space based on the elevation prior data to obtain a three-dimensional spiral ascending photography flight route.The present application realizes adaptive fitting of a flight route to complex terrain, improves modeling accuracy and collection efficiency, and enhances flight safety.

Description

Technical Field

[0001] This invention relates to the field of UAV remote sensing and mapping technology, specifically to a method and system for UAV oblique photography flight path planning in complex terrain. Background Technology

[0002] In the field of UAV oblique photogrammetry, detailed 3D modeling of complex terrains such as karst peak clusters with steep slopes and irregular conical shapes is a significant technical challenge. Existing technologies, such as conventional parallel flight path planning, struggle to effectively cover steep slopes, resulting in missing side textures and insufficient detail in the model. While terrain-following flight paths correct waypoint elevations to maintain relative ground level through altimetry systems, they are prone to flight safety risks due to sudden altitude changes in dense peak cluster areas and exhibit poor adaptability to local micro-topography. Columnar flight paths, although adding circling flight to acquire lateral information, rely on manual experience for parameter settings and have a fixed vertical sampling density, failing to dynamically adjust according to slope changes, leading to insufficient detailed information collection in areas with large elevation differences. In summary, existing flight path planning schemes generally suffer from difficulties in balancing modeling accuracy and data acquisition efficiency, as well as limitations in flight safety when dealing with complex terrain. Therefore, there is an urgent need for a UAV photogrammetry flight path planning method that can adapt to terrain features and balance high accuracy and high efficiency. Summary of the Invention

[0003] To address the problems of low accuracy, poor acquisition efficiency, and insufficient flight safety in existing technologies for UAV oblique photogrammetry modeling under complex terrain, this application proposes a method and system for UAV oblique photogrammetry flight path planning under complex terrain, which achieves high-precision, high-efficiency, and safe automated 3D modeling of complex terrain.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: A method for planning flight paths for UAV oblique photography in complex terrain includes: Acquire polygon boundary data and slope data of the target survey area, and calculate the coordinates of the geometric center point of the target survey area based on the polygon boundary data; A radial terrain framework is constructed based on the coordinates of the geometric center point and a pre-configured azimuth coordinate system. Based on the radial terrain framework and slope data, a flight exploration route is planned, and the UAV is controlled to perform exploration flights along the flight exploration route to obtain the prior elevation data of the target survey area; the flight exploration route includes multiple two-dimensional waypoints; Based on prior elevation data, multiple two-dimensional waypoints are rearranged in three-dimensional space to obtain a three-dimensional spiral ascent photography route.

[0005] The above scheme generates an initial exploration route by constructing a radial terrain framework and combining it with slope data. Then, it rearranges the two-dimensional waypoints into a three-dimensional spiral ascending trajectory using measured elevation prior data. This achieves adaptive fitting of the route to complex terrain geometry features and solves the problem of insufficient sampling in steep areas by traditional routes.

[0006] As one implementation method, the planning of flight reconnaissance routes based on radial terrain framework and slope data includes: The slope data is matched with the preset slope-waypoint spacing mapping table to obtain the waypoint spacing; Coordinate calculations were performed based on the radial terrain framework and waypoint spacing to obtain multiple two-dimensional waypoints and their corresponding coordinates. Based on the optimal path objective function, a flight route map is planned using multiple two-dimensional waypoints to obtain the flight route map. Acquire the altitude data of the target survey area, add the altitude data to the preset safe altitude, and obtain the flight altitude data; By associating and fusing flight path maps and flight altitude data, a flight reconnaissance route is obtained.

[0007] This implementation method ensures uniform and effective sampling density in different slope areas by introducing a slope-spacing mapping mechanism and optimal path planning, while optimizing the continuity of the flight path and energy consumption.

[0008] As one implementation method, the steps for constructing the slope-waypoint spacing mapping table include: The slope range of the target measurement area is divided into multiple continuous slope intervals, and a representative slope angle is determined for each slope interval based on the statistical characteristics of the pixel slope within each slope interval. The slope interval corresponding to the largest representative slope angle is selected as the benchmark interval, and the benchmark waypoint spacing is set for the benchmark interval based on the obtained sampling distance requirements. Based on the baseline waypoint spacing, the initial waypoint spacing for the remaining slope intervals is calculated according to the equal slope sampling spacing principle; the equal slope sampling spacing principle means that the quotient of the waypoint spacing for different slope intervals and the cosine value of the corresponding slope angle remains constant. Based on the micro-topographical complexity of each slope range and the constraints of UAV flight imaging, the initial waypoint spacing is corrected to obtain the corrected waypoint spacing. Each slope interval is paired with the corresponding corrected waypoint spacing, and discretization, rounding, and smoothing of adjacent intervals are performed to obtain a slope-waypoint spacing mapping table.

[0009] This implementation method is based on the principle of equal slope sampling, which mathematically ensures that the amount of image acquisition per unit area is consistent under different slopes. Furthermore, through micro-topography complexity correction, it further improves the flight path's adaptability to sudden changes in local terrain.

[0010] As one implementation method, the correction of the initial waypoint spacing includes: Calculate the micro-topographic complexity index for each slope interval; the micro-topographic complexity index is the mean of the local standard deviations of the slope of all pixels within the corresponding slope interval. The complexity adjustment interval is obtained by multiplying the initial waypoint spacing by a preset adjustment coefficient; the adjustment coefficient is negatively correlated with the micro-topography complexity index. Based on the attitude angle changes and camera field of view parameters of the UAV when maintaining a constant altitude flight at the corresponding slope, the upper limit of the spacing allowed to meet the preset minimum forward overlap and minimum lateral overlap in each slope interval is calculated, and a lower limit of the spacing to avoid sampling redundancy is set. By constraining the complexity adjustment spacing to the interval formed by the upper and lower limits of the spacing, the corrected waypoint spacing is obtained.

[0011] This implementation method takes into account the impact of micro-terrain undulations on imaging quality as well as physical constraints such as UAV attitude and camera field of view. It avoids invalid and redundant shooting while ensuring that the overlap meets the standard, thus balancing modeling quality and operational efficiency.

[0012] As one implementation method, the coordinate calculation based on the radial terrain framework and waypoint spacing includes: Construct an azimuth coordinate system containing multiple principal and auxiliary directions with the coordinates of the geometric center point as the origin; Using the coordinates of the geometric center point as the center, multiple concentric circles with different radii are determined according to the spacing between waypoints; the radius is not greater than the minimum straight-line distance between the coordinates of the geometric center point and the boundary coordinates of the target survey area; Calculate the intersection points of each concentric circle with each azimuth line in the azimuth coordinate system to obtain multiple two-dimensional waypoints and their corresponding coordinates.

[0013] This implementation utilizes the geometric properties of the polar coordinate system to generate a comprehensive and regularly distributed gridded waypoints, providing structured basic data support for subsequent path optimization.

[0014] As one implementation method, the objective function of the optimal path is a weighted cost function, including: The path length cost is calculated as the sum of the Euclidean distances between all adjacent two-dimensional waypoints in the flight path diagram, and is used to constrain the total flight path to be the shortest. The turning angle cost is calculated as the sum of the absolute values ​​of the directional change angles between all adjacent segments in the flight path diagram. It is used to constrain the flight to have the fewest turns and the highest proportion of straight flight segments. Resource consumption cost is calculated as the product of path length cost and energy consumption per unit distance, plus the product of turning angle cost and energy consumption per unit angle of turn, used to constrain the resource consumption of flight exploration missions to the minimum.

[0015] This implementation method, through multi-objective weighted optimization, significantly reduces the maneuvering frequency and battery consumption of the UAV while meeting coverage requirements, and extends the flight time of a single operation.

[0016] As one implementation method, the three-dimensional spatial rearrangement of multiple two-dimensional waypoints based on prior elevation data includes: Based on prior elevation data and preset safe flight altitude, multiple two-dimensional waypoints are mapped to a pre-constructed three-dimensional space to obtain multiple three-dimensional waypoints; Multiple 3D waypoints are grouped and sorted based on the horizontal straight-line distance between the coordinates of the 3D waypoints and the geometric center point to obtain a grouped coordinate sequence; The grouped coordinate sequences are connected sequentially in a clockwise or counterclockwise direction to form a three-dimensional spiral trajectory that spirals upward along the terrain surface of the target survey area. Exposure points are inserted on the three-dimensional spiral trajectory according to the preset encryption interval to obtain the three-dimensional spiral ascent photography route.

[0017] This implementation transforms discrete detection points into a continuous spiral ascent trajectory, enabling the UAV to ascend steadily along the terrain contours, ensuring both the acquisition angle of side-view images and avoiding the flight risks caused by frequent altitude changes.

[0018] As one implementation method, the method for obtaining the safe flight altitude includes: Acquire vegetation type distribution data and corresponding prior knowledge of vegetation growth height in the target survey area; Based on prior knowledge of vegetation growth height, the dominant vegetation type with the maximum maturity height in the target survey area is selected, and the corresponding maximum vegetation barrier height is determined. The safe flight altitude is obtained by adding the maximum vegetation obstacle height to the preset basic safety margin; the basic safety margin is the sum of the allowable vertical positioning error of the UAV when flying at a fixed altitude and the preset additional buffer distance.

[0019] This implementation combines vegetation biological characteristics with UAV positioning error models to dynamically determine safety thresholds adapted to specific environments, effectively preventing collisions caused by vegetation obstruction or positioning drift.

[0020] As one implementation method, the acquisition of the prior elevation data includes: Control the UAV to perform reconnaissance flight along the flight reconnaissance route, and use the onboard altimeter system to obtain the relative height of each two-dimensional waypoint relative to the corresponding ground projection point; Subtracting the relative altitude from the flight altitude data in the flight exploration route yields the prior elevation data of various ground projection points in the target survey area.

[0021] This implementation method uses a two-stage strategy of "first exploration, then precision flight" to obtain real terrain elevation through low-cost exploration flights, eliminating the impact of poor timeliness or insufficient accuracy of publicly available DEM data, and providing a reliable data foundation for the accurate generation of the final spiral route.

[0022] Furthermore, this invention also provides a UAV oblique photography flight path planning system for complex terrain, used to implement the UAV oblique photography flight path planning method for complex terrain as described above, including: The data acquisition and preprocessing module is used to acquire polygon boundary data and slope data of the target survey area, and calculate the coordinates of the geometric center point of the target survey area based on the polygon boundary data; The terrain framework building module is used to build radial terrain frameworks based on the coordinates of the geometric center point and a pre-configured azimuth coordinate system. The reconnaissance route planning module is used to plan flight reconnaissance routes based on radial terrain framework and slope data, and control the UAV to perform reconnaissance flights along the flight reconnaissance routes to obtain prior elevation data of the target survey area; the flight reconnaissance route includes multiple two-dimensional waypoints; The 3D route generation module is used to rearrange multiple 2D waypoints in 3D space based on prior elevation data to obtain a 3D spiral ascent photography route.

[0023] The system described above achieves automated processing of the entire route planning process through modular design, improving the efficiency and standardization of complex terrain surveying operations.

[0024] Beneficial effects: The present invention provides a method and system for planning flight paths for UAV oblique photography in complex terrain, which has the following advantages compared with the prior art: First, it significantly improves the accuracy of 3D modeling of complex terrain. This invention overcomes the shortcomings of traditional parallel flight paths, which suffer from sparse sampling on steep slopes, by constructing a radial terrain framework and dynamically adjusting the spacing between waypoints based on the principle of equal slope sampling. Simultaneously, the 3D spiral ascent photography flight path generated using prior measured elevation data closely follows the terrain undulations, ensuring complete acquisition of information on the sides and details of peak clusters. Experiments show that the average deviation of the digital elevation model (DEM) constructed using the flight path planned by this invention is superior to that of conventional horizontal flight paths, contour flight paths, and columnar flight paths, effectively solving the problem of insufficient detail in complex terrain modeling.

[0025] Secondly, it significantly improves data acquisition and processing efficiency. The slope-adaptive spacing strategy and optimal path planning algorithm adopted in this invention eliminate redundant waypoints while ensuring imaging quality; in particular, the design of the three-dimensional spiral ascent trajectory avoids frequent sudden altitude increases and decreases in terrain-following routes and repeated coverage of multi-layered circular flights in columnar routes. Under the same modeling accuracy requirements, this invention requires significantly fewer images than existing mainstream solutions, thereby shortening field flight time and office data processing cycle.

[0026] Finally, the invention enhances the flight safety of UAVs in complex environments. It incorporates prior knowledge of vegetation obstacle height and a micro-terrain complexity correction mechanism during the flight path planning stage, and considers positioning errors and buffer distances in the safe flight altitude calculation. Combined with a smooth-transition spiral ascent flight mode, it effectively avoids potential risks such as tree obstruction, abrupt terrain changes, and attitude instability, reducing the need for human intervention and the probability of crashes, and improving the reliability and automation of operations. Attached Figure Description

[0027] Figure 1 A flowchart illustrating a method for planning flight paths for UAV oblique photography in complex terrain, provided as an embodiment of this application; Figure 2 A schematic diagram of a radial terrain framework constructed based on an azimuth coordinate system, provided in an embodiment of this application; Figure 3 This is a schematic diagram of the distribution of two-dimensional waypoints provided in the embodiments of this application; Figure 4 The flight path map provided in this application is based on the optimization of the optimal path objective function. Figure 5 A schematic diagram of a three-dimensional spiral ascent photography route provided in an embodiment of this application; Figure 6 A schematic diagram of a two-dimensional photographic flight path provided in an embodiment of this application; Figure 7 A system architecture diagram of an unmanned aerial vehicle (UAV) oblique photography flight path planning system under complex terrain is provided in the embodiments of this application; Figure 8 DEM deviation distribution map of the three-dimensional spiral ascent photography route provided in the embodiments of this application; Figure 9 DEM deviation distribution diagram of a conventional planar flight path provided in the embodiments of this application; Figure 10 The DEM deviation distribution map of the terrain-following route provided in the embodiments of this application; Figure 11 The DEM deviation distribution diagram of the columnar flight path provided in the embodiments of this application. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0029] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0030] Example 1 like Figure 1 As shown in the figure, this embodiment provides a method for UAV oblique photogrammetry flight path planning in complex terrain. This method solves the problem of poor matching degree between flight path planning and terrain in complex terrain by constructing a radial framework adapted to terrain features and combining it with measured data for three-dimensional reconstruction. The method mainly includes the following steps.

[0031] Step S100: Obtain polygon boundary data and slope data of the target survey area, and calculate the coordinates of the geometric center point of the target survey area based on the polygon boundary data. Specifically, the polygon boundary data defines the geographical range of the UAV operation, which can be obtained by importing GIS vector files, KML files, or manually drawing on a ground station map. The slope data reflects the undulation of the survey area and can be derived from an existing digital elevation model (DEM) or field measurement data. When calculating the reference point, this invention deliberately chooses the geometric center point instead of the traditional centroid or the center of the smallest circumscribed rectangle. This is because in complex landforms such as karst peak clusters, the boundary of the survey area often presents an extremely irregular shape. If the centroid is used, the center may be offset due to local protruding areas, causing the subsequently constructed radial lines to be prematurely truncated or insufficiently covered in some directions. The geometric center point is determined based on the arithmetic mean of the boundary vertex coordinates, which can more robustly represent the spatial distribution centroid of irregular polygons, thereby ensuring the coverage balance of the subsequently generated radial frame in all directions and reducing blind spots.

[0032] Step S200: Construct a radial terrain framework based on the coordinates of the geometric center point and a pre-configured azimuth coordinate system. Specifically, the radial terrain framework is a topological structure radiating outwards from the geometric center point, used to replace traditional parallel grids or simple ring structures. The azimuth coordinate system defines the extension direction of the radial lines; in this embodiment, as shown... Figure 2 As shown, the azimuth coordinate system includes four main directions—north, east, south, and west—and four auxiliary directions—northeast, southeast, southwest, and northwest—for a total of eight directions. However, in other implementations, more or fewer azimuth lines can be configured according to the shape of the survey area or accuracy requirements, such as twelve or sixteen azimuth lines. The azimuth angle can even be dynamically adjusted according to the terrain orientation, as long as the function of radiating outwards from the center to cover the survey area is met. This radial framework naturally fits the geometric characteristics of peak clusters and conical mountains. Compared to parallel flight paths, it can arrange sampling paths along the direction of the most significant slope change, fundamentally avoiding the structural defect of sparse sampling on steep sides in parallel flight paths, thus laying a geometric foundation for subsequent high-precision modeling.

[0033] Step S300 involves planning a flight path based on the radial terrain framework and slope data, and controlling the UAV to perform a reconnaissance flight along the flight path to obtain prior elevation data for the target survey area. The flight path includes multiple two-dimensional waypoints. It is important to note that the flight path here differs fundamentally from the final photographic operation path in terms of function and accuracy requirements. The primary task of the flight path is "coarse exploration," i.e., quickly acquiring the true ground elevation information of the survey area. Therefore, its waypoint density and image overlap requirements are relatively low, focusing on efficiency and safety. Two-dimensional waypoints refer to the projected coordinates on the horizontal plane; at this stage, the final three-dimensional operational height has not yet been determined. By executing this reconnaissance flight, the relative height of ground points is directly obtained using airborne sensors (such as lidar, millimeter-wave radar, or visual altimeter modules), and then the prior elevation data is retrieved. This mechanism constitutes a crucial bridge connecting two-dimensional planning and three-dimensional operation, effectively overcoming the problems of outdated timeliness, insufficient resolution, or excessive errors in vegetation-covered areas of existing publicly available DEM data, ensuring that the terrain benchmark upon which subsequent fine-grained route planning relies is real-time and accurate.

[0034] Step S400 involves rearranging multiple two-dimensional waypoints in three-dimensional space based on prior elevation data to obtain a three-dimensional spiral ascent photography flight path. Specifically, this three-dimensional rearrangement utilizes the real elevation information obtained in step S300 to map two-dimensional waypoints that were originally at the same horizontal plane or approximate height onto three-dimensional space. The connection order of the waypoints is then reorganized according to the magnitude of their elevation values, forming a continuous trajectory that spirals upwards along the terrain surface. This rearrangement process is not merely a simple superposition of coordinate values, but a fundamental transformation of the flight path's topology. The three-dimensional spiral ascent photography flight path generated in this way allows the UAV to maintain a stable relative altitude to the ground throughout flight, while continuously acquiring texture information from the mountainside in a circling manner. Compared to layered circling flight or terrain-following flight, the three-dimensional spiral trajectory avoids frequent acceleration / deceleration and sudden altitude changes, ensuring uniformity in image quality while significantly improving flight safety and battery endurance, achieving a unified approach to high-quality data acquisition and efficient operation in complex terrain.

[0035] Example 2 This embodiment, based on Embodiment 1, further provides a specific implementation method for planning flight reconnaissance routes based on a radial terrain framework and slope data. This approach aims to address the problem of uneven sampling density in complex terrain caused by traditional fixed-spacing flight paths. By introducing a slope adaptive mechanism, it ensures that the UAV can acquire effective data that meets modeling requirements in different slope areas. Specifically, the planning process includes the following steps.

[0036] Step S301 involves matching the slope data with a preset slope-waypoint spacing mapping table to obtain the waypoint spacing. This slope-waypoint spacing mapping table is not a simple linear lookup table, but a non-linear mapping relationship constructed based on the principles of geometric optics and topographic statistics. Its core purpose is to compensate for projection distortion caused by terrain tilt, ensuring that the effective coverage area of ​​the final acquired image on the ground surface remains relatively constant, rather than just constant on the horizontal projection plane.

[0037] Step S302 involves calculating coordinates based on the radial terrain framework and waypoint spacing to obtain multiple two-dimensional waypoints and their corresponding coordinates. Specifically, after determining the adaptive spacing on each azimuth line, the system generates a waypoint sequence by sequentially accumulating the spacing values ​​along each azimuth line of the radial terrain framework, starting from the geometric center point. This waypoint placement method based on polar coordinates naturally adapts to the geometric characteristics of centrally symmetric or radial terrains such as peak clusters, avoiding the truncation effect or redundant coverage caused by parallel routes at irregular boundaries.

[0038] Step S303 involves planning a flight path map based on multiple two-dimensional waypoints using the optimal path objective function. Since the waypoints generated by the radial frame are spatially discrete, direct connections would lead to numerous invalid turnarounds. Therefore, this embodiment reorders the waypoints using the optimal path objective function, transforming them into a continuous, smooth, and energy-efficient traversal path. This step is crucial in converting the geometrically sampled point set into an engineering-executable flight trajectory.

[0039] Step S304: Obtain the altitude data of the target survey area, and add the altitude data to the preset safe altitude to obtain the flight altitude data. This altitude data can be rough DEM data or an estimate of the highest point in the survey area, mainly used for obstacle avoidance during the reconnaissance flight phase. It is important to emphasize that the flight altitude setting for the reconnaissance flight prioritizes safety and wide-area coverage; its accuracy requirement is lower than that of the final photographic flight path. This provides a tolerance margin for subsequent refined 3D rearrangement using measured data.

[0040] Step S305 involves associating and fusing the flight path map and flight altitude data to obtain the flight reconnaissance path. By binding the optimized two-dimensional path with the initially determined safe flight altitude, a four-dimensional reconnaissance command set containing position and altitude information is generated. This reconnaissance path ensures both the effectiveness of data acquisition and the basic flight safety of the UAV in unknown and detailed terrain.

[0041] As a preferred implementation, the steps for constructing the slope-waypoint spacing mapping table specifically include the following sub-steps.

[0042] Step S3011 involves dividing the slope range of the target measurement area into multiple continuous slope intervals, and determining a representative slope angle for each interval based on the statistical characteristics of the pixel slope within each interval. For example, the range from 0° to 90° can be divided into four intervals: [0°, 30°), [30°, 50°), [50°, 70°), and [70°, 90°]. The representative slope angle can be the median, mean, or mode of the interval. In this embodiment, the median of the interval is preferred as the representative value to balance computational efficiency and representation accuracy. This discretization process transforms continuous slope changes into a finite number of control levels, facilitating real-time table lookup calculations by the embedded system.

[0043] Step S3012: Select the slope interval corresponding to the largest representative slope angle as the baseline interval, and set the baseline waypoint spacing for the baseline interval based on the obtained sampling distance requirements. The steepest slope interval is chosen as the baseline because this area has the most stringent requirements for sampling density. If the gentlest slope is used as the baseline, it will inevitably lead to sparse sampling on steep slopes; while the spacing determined by using the steep slope as the baseline will slightly increase the sampling density in other gentler areas, but will never be lower than the minimum threshold required for modeling, thus eliminating the risk of missing data from the source. For example, if the slope sampling spacing in areas with steep slopes above 70° is required to be no more than 5 meters, then the baseline waypoint spacing is set to 5 meters.

[0044] Step S3013: Based on the baseline waypoint spacing, calculate the initial waypoint spacing for the remaining slope intervals according to the equal slope sampling spacing principle. The equal slope sampling spacing principle means that the quotient of the waypoint spacing in different slope intervals and the cosine of the corresponding slope angle remains constant. The geometric meaning of this principle is to ensure a constant sampling density per unit slope area. Specifically, when the UAV flies a horizontal distance D, its sampling density at an inclination angle of... The actual projected length on the slope To ensure that the actual slope sampling length S remains consistent across different slope gradients (i.e., equal to the reference slope length), the horizontal route point spacing D must be adjusted accordingly. The initial waypoint spacing should be dynamically adjusted according to changes. For example, if the representative angle of the reference interval [70°, 90°] is 80° and the reference spacing is 5m; for the interval [30°, 50°], the representative angle is 40°, then the initial waypoint spacing should be calculated as follows: Through this mathematical constraint, the present invention fundamentally eliminates the fluctuation in modeling accuracy caused by differences in slope.

[0045] Step S3014 involves correcting the initial waypoint spacing based on the micro-topographical complexity of each slope range and the UAV flight imaging constraints, resulting in the corrected waypoint spacing. The simple geometric projection model assumes the slope is an ideal plane, but actual terrain often exhibits local undulations. Furthermore, the UAV's attitude stability and the camera's field of view also impose physical limitations on effective sampling. Therefore, a correction factor must be introduced based on the theoretical calculations to make the flight path planning more closely reflect the real physical environment.

[0046] Step S3015: Pair each slope interval with the corresponding corrected waypoint spacing, and perform discretization rounding and smooth transition processing between adjacent intervals to obtain a slope-waypoint spacing mapping table. Discretization rounding is to adapt to the command precision of the UAV flight control system, while smooth transition processing is to avoid abrupt changes in waypoint spacing at the boundaries of slope intervals, and to prevent the UAV from affecting imaging quality or consuming extra power due to frequent acceleration and deceleration.

[0047] Furthermore, this embodiment provides a specific implementation method for the correction process in step S3014.

[0048] First, the micro-topographic complexity index is calculated for each slope interval. The micro-topographic complexity index is the mean of the local standard deviations of the slope of all pixels within the corresponding slope interval. The local standard deviation reflects the degree of undulation of the terrain at a micro scale. The larger the standard deviation, the more developed the micro-topography such as gullies and rocks in the area, and the rougher the surface. Using this index as a correction basis allows the flight path to automatically adapt to the texture details of the surface, rather than simply responding to the macro-slope.

[0049] Secondly, the initial waypoint spacing is multiplied by a preset adjustment coefficient to obtain the complexity adjustment spacing; the adjustment coefficient is negatively correlated with the micro-topographic complexity index. This means that the more complex the terrain, the smaller the adjustment coefficient, and the denser the waypoint spacing. For example, when the micro-topographic complexity index exceeds a preset threshold, the adjustment coefficient can be set to 0.8 or even lower to force denser sampling points to capture subtle terrain features; conversely, in flat and smooth areas, the adjustment coefficient can be close to 1.0 to maintain the theoretical calculation spacing and improve efficiency.

[0050] Furthermore, based on the attitude angle changes and camera field-of-view parameters of the UAV maintaining a constant altitude at corresponding slopes, the upper limit of the spacing allowed to meet the preset minimum forward overlap and minimum lateral overlap for each slope interval is calculated, and a lower limit of the spacing to avoid sampling redundancy is set. The upper limit of the spacing is a hard constraint determined by image quality: if the spacing is too large, considering the attitude tilt of the UAV and the deformation of the camera field of view when flying on the slope, the effective overlap area of ​​adjacent images may be lower than the minimum threshold required by the 3D reconstruction algorithm (such as 60%), leading to matching failure. The lower limit of the spacing is a soft constraint determined by operational efficiency: excessively dense sampling not only wastes storage and computing power, but may also lead to increased elevation calculation errors due to an excessively small base-to-height ratio. By adjusting the complexity and constraining the spacing within this dynamic range, the optimal balance between quality and efficiency is achieved.

[0051] To more intuitively illustrate the technological advancements of this embodiment, a comparative example is provided below. If a common fixed horizontal spacing strategy (e.g., uniformly set to 20 meters) is adopted in existing technologies, it may meet the requirements in gentle terrain. However, on the sides of steep peaks exceeding 70°, a 20-meter horizontal spacing corresponds to an actual slope distance exceeding 57 meters (20 / cos70°). Such a large sampling interval leads to severely insufficient image overlap in steep slope areas, making the 3D model prone to texture stretching, holes, or even complete irreconstruction in these regions. In contrast, this embodiment, through a dual mechanism of equal-slope sampling and micro-topographic complexity correction, ensures that the number of effective sampling points per unit area remains within the optimal range required for modeling, whether in gentle valleys or steep peaks, thereby significantly improving the integrity and detail of the 3D model of complex terrain.

[0052] Example 3 This embodiment, based on Embodiment 2, further provides a specific implementation method for coordinate calculation based on a radial terrain framework and waypoint spacing. This method aims to transform the abstract radial topology into a precise geographic coordinate sequence recognizable by the UAV flight control system, while ensuring that the generated waypoints strictly fall within the survey area. Specifically, the coordinate calculation process includes the following steps.

[0053] Step S3021: Construct an azimuth coordinate system with the geometric center point coordinates as the origin, containing multiple main directions and auxiliary directions. In this embodiment, the azimuth coordinate system preferably adopts an eight-directional system, that is, it includes four main directions: due north, due east, due south, and due west, and four auxiliary directions: northeast, southeast, southwest, and northwest. This configuration can ensure coverage density while taking into account computational efficiency and flight trajectory smoothness.

[0054] Step S3022: Using the geometric center point coordinates as the center, determine multiple concentric circles with different radii based on the waypoint spacing; the radius must not exceed the minimum straight-line distance between the geometric center point coordinates and the boundary coordinates of the target survey area. This radius constraint is a key boundary condition ensuring the safety and effectiveness of the flight path. Specifically, the system first calculates the minimum Euclidean distance from the geometric center point to each vertex and perpendicular segment of the polygon boundary, using this as the maximum allowable radius threshold. When generating the concentric circle sequence, once the current cumulative radius exceeds this threshold, outward expansion stops. This mechanism fundamentally eliminates the possibility of waypoints being generated in invalid areas outside the survey area, avoiding the search overhead of subsequent path planning algorithms in invalid spaces, and also ensuring the sampling integrity in the narrowest direction. For example, if the survey area is an irregular star shape, and the minimum boundary distance in a certain direction is only 50 meters, then the radius of all concentric circles will not exceed 50 meters, thus ensuring that all generated waypoints are within the safe operating envelope.

[0055] Step S3023, as follows Figure 3 As shown, the intersections of each concentric circle with each azimuth line in the azimuth coordinate system are calculated, resulting in multiple two-dimensional waypoints and their corresponding coordinates. Specifically, this is a mathematical mapping process that converts polar coordinate parameters (radius, azimuth) into Cartesian coordinates (X, Y). By traversing all combinations of effective radii and azimuth lines, the system can automatically generate a structured grid point set. Compared to traditional parallel scanline point placement, this discretized point placement method based on the polar coordinate system naturally fits the geometric characteristics of centrally symmetric or radial terrain such as peak clusters, making the spatial distribution of waypoints more consistent with the terrain's morphological patterns, providing high-quality basic data support for subsequent path optimization.

[0056] Furthermore, regarding the process of planning the flight path map based on multiple two-dimensional waypoints using the optimal path objective function in step S303, this embodiment provides a specific implementation method of the weighted cost function. Since the waypoints generated by the radial frame are spatially discrete, directly connecting them in the order of generation would lead to a large number of invalid radial backtracking steps. Therefore, a multi-objective optimization algorithm must be used to rearrange them into a continuous, efficient traversal path that conforms to the dynamic characteristics of the UAV. The flight path map obtained through weighted cost function optimization is as follows: Figure 4 As shown. The objective function of this optimal path is a weighted cost function, which specifically includes the following three components.

[0057] The first component is the path length cost, calculated as the sum of the Euclidean distances between all adjacent two-dimensional waypoints in the flight path graph, used to constrain the shortest total flight path. This component directly corresponds to the drone's endurance. Under the constraint of limited battery capacity, shortening the total flight distance means being able to cover a larger survey area or reserve more power for a safe return. At the algorithmic level, this is usually manifested as solving the Traveling Salesman Problem (TSP) or its variants, aiming to find the shortest loop visiting all waypoints.

[0058] The second component is the turning angle cost, calculated as the sum of the absolute values ​​of the directional change angles between all adjacent segments in the flight path diagram. This component is used to constrain the flight to minimize the number of turns and maximize the proportion of straight flight segments. This component not only affects flight efficiency but also directly impacts imaging quality and equipment lifespan. Frequent sharp turns cause severe fluctuations in the drone's attitude, leading to gimbal jitter and image blurring; simultaneously, acceleration and deceleration significantly increase motor wear and energy consumption. By introducing a turning penalty term, the optimization algorithm tends to generate smooth trajectories composed of long straight segments and large-radius curves, rather than jagged, broken paths. For example, when two adjacent waypoints are close but require a 180-degree turn to reach, the algorithm might choose to visit other convenient waypoints first and then circle back to the original waypoint, in exchange for smoother overall trajectories.

[0059] The third component is the resource consumption cost, calculated as the product of path length cost and energy consumption per unit distance, plus the product of turning angle cost and energy consumption per unit angle of turn. This is used to constrain the resource consumption of flight reconnaissance missions to be minimized. This is a comprehensive performance index that transforms geometric distance and angle into physical energy consumption. Energy consumption per unit distance reflects the power characteristics during cruise, while energy consumption per unit angle of turn reflects the additional power overhead during maneuvering. By weighted summing these two components, the cost function can accurately reflect the actual battery resource utilization of different flight path options.

[0060] It should be noted that the weighting coefficients of the three components mentioned above are not fixed, but rather adjustable parameters that can be dynamically configured according to actual operational needs. For example, in scenarios with limited power or extremely large survey areas, the weights of path length cost and resource consumption cost can be significantly increased, causing the generated flight path to prioritize the shortest distance, even at the cost of sacrificing some smoothness. Conversely, in scenarios requiring high-precision modeling or extremely high image quality, the weight of turning angle cost can be significantly increased, forcing the algorithm to generate a smoother flight trajectory to ensure the clarity of each frame. This parameterized design allows the invention to flexibly adapt to diverse application scenarios, enhancing the universality and practicality of the technical solution.

[0061] Example 4 This embodiment, based on embodiments 1 to 3, further provides a specific implementation method for three-dimensional spatial rearrangement of multiple two-dimensional waypoints based on prior elevation data. This method aims to transform the discrete two-dimensional detection points obtained in the preceding steps into continuous, smooth, and terrain-conforming three-dimensional operational trajectories, solving the problems of abrupt attitude changes and discontinuous sampling in traditional layered circumferential flight or terrain-following flight paths under complex terrain. Specifically, the three-dimensional spatial rearrangement process includes the following steps.

[0062] Step S401: Based on prior elevation data and a preset safe flight altitude, multiple two-dimensional waypoints are mapped to a pre-constructed three-dimensional space to obtain multiple three-dimensional waypoints. This mapping is not a simple coordinate assignment, but rather a mapping that integrates the two-dimensional waypoints generated in Example 3 with the measured ground elevation obtained in step S300. and safe flight altitude The data is then fused to calculate the absolute flight altitude of each waypoint. In this way, two-dimensional waypoints that were originally on the same horizontal plane are "stretched" into three-dimensional space, initially forming a point cloud set that follows the terrain's undulations. This step realizes the transformation from geometric planning to physical space, providing a precise spatial reference for subsequent trajectory generation.

[0063] Step S402 involves grouping and sorting multiple 3D waypoints based on the horizontal straight-line distance between their coordinates and the geometric center point, resulting in a grouped coordinate sequence. Since the waypoints generated by the radial frame are distributed concentrically in space, the system first calculates the Euclidean distance from each waypoint to the geometric center point, grouping points with similar distances into the same group (i.e., the same concentric circle). Subsequently, within each group, the points are sorted according to their azimuth angles, forming sequences that facilitate clockwise or counterclockwise connections. This grouping and sorting mechanism reconstructs the unordered scattered point set into a hierarchical sequence with clear topological relationships, ensuring that subsequent connections do not involve jumps across layers or reverse reversals, thus laying the data structure foundation for generating continuous spiral trajectories.

[0064] Step S403: Connect the grouped coordinate sequences sequentially in a clockwise or counterclockwise direction to form a three-dimensional spiral trajectory that ascends along the terrain surface of the target survey area. In the specific connection process, the system adopts a unidirectional spiral strategy, starting from the innermost (or outermost) layer and extending outwards (or inwards) in a fixed direction, while using smooth curves at the transition points between layers instead of abrupt radial jumps. The spiral trajectory maintains the continuity of the UAV's attitude and the stability of its power output, avoiding gimbal jitter and power waste caused by frequent acceleration / deceleration and sudden altitude changes. Simultaneously, the continuous spiral path makes the vertical overlap between adjacent images more uniform, effectively eliminating texture misalignment or voids that may occur at layer stitching points, significantly improving the precision and integrity of the three-dimensional model.

[0065] Step S404, as follows Figure 5 As shown, exposure points are inserted along the three-dimensional spiral trajectory according to a preset density interval to obtain a three-dimensional spiral ascent photography path. To facilitate understanding of the connection method of the exposure points, in... Figure 6 The image shows a sequentially connected photographic flight path on a two-dimensional plane. It's important to note that the exposure point density along the spiral trajectory is designed to ensure the overlap of the captured images meets preset requirements. According to sampling requirements, in this example, the density spacing is set to half the original spacing between waypoints, effectively doubling the density of the original three-dimensional spiral trajectory. The flight altitude of the inserted exposure points is set to the average flight altitude of adjacent waypoints. The resulting flight path not only contains positional information but also implicitly includes flight control parameters such as speed and heading, allowing direct import into the UAV flight control system for high-precision oblique photography operations.

[0066] Furthermore, regarding the acquisition of the preset safe flight altitude in step S401 above, this embodiment provides an environment-aware dynamic determination method. In complex terrain, especially in densely vegetated karst peak cluster areas, a fixed safe altitude often fails to balance safety and imaging resolution. Therefore, the method for acquiring the safe flight altitude specifically includes the following sub-steps.

[0067] Step S4011 involves acquiring vegetation type distribution data and corresponding prior knowledge of vegetation growth height for the target survey area. Vegetation type distribution data can be obtained through remote sensing interpretation, land use classification maps, or field quadrat surveys; prior knowledge of vegetation growth height comes from forestry survey data, ecological research literature, or local experience databases. For example, for the karst region of Southwest China, it can be found that the dominant tree species in this area are eucalyptus trees aged 5-8 years, with a mature height typically between 20m and 30m. This approach of introducing prior environmental knowledge transforms flight path planning from a purely geometric calculation into one with ecological awareness, enabling it to proactively adapt to land cover characteristics.

[0068] Step S4012: Based on prior knowledge of vegetation growth height, select the dominant vegetation type with the highest maturity height within the target survey area and determine the corresponding maximum vegetation obstacle height. The system traverses all vegetation patches within the survey area to identify the obstacle height with the greatest potential threat. For example, if both shrubs (3m) and eucalyptus trees (30m) exist in the survey area, then 30m is taken as the maximum vegetation obstacle height. This conservative strategy ensures that even in areas with the highest vegetation, the UAV can maintain a sufficient safe distance, avoiding collisions due to local underestimation of obstacles.

[0069] Step S4013: Add the maximum vegetation obstacle height to the preset basic safety margin to obtain the safe flight altitude. The basic safety margin is the sum of the allowable vertical positioning error of the UAV during altitude-holding flight and the preset additional buffer distance. The vertical positioning error depends on the positioning technology used (such as RTK, PPK, or ordinary GPS). For example, it is typically 0.1m-0.2m in RTK mode and may need to be 5m-10m in ordinary GPS mode. The additional buffer distance is used to cope with airflow disturbances, tree swaying, or sudden maneuvering needs, and is generally recommended to be set to 5m-10m. For example, when the maximum vegetation obstacle height is 30m, the RTK vertical error is 0.2m, and the additional buffer is 10m, the final determined safe flight altitude relative to the ground is 40.2m. This triple safety model, consisting of "biological features + equipment performance + environmental redundancy," is more robust than a single fixed value. It prevents image resolution loss due to excessive margin and eliminates safety risks caused by insufficient margin, making it particularly suitable for operation scenarios with complex terrain and vegetation.

[0070] Furthermore, regarding the acquisition of prior elevation data in step S300, this embodiment also provides an inversion mechanism based on real-time detection. Given that publicly available DEM data often suffers from time lag, insufficient resolution, or severe accuracy degradation in vegetated areas, this invention employs a two-stage "detect first, then fly" strategy to ensure the reliability of the topographic benchmark. Specifically, the acquisition of prior elevation data includes the following steps.

[0071] Step S3001: Control the UAV to perform a reconnaissance flight along the flight path and use the onboard altimeter system to obtain the relative height of each two-dimensional waypoint relative to the corresponding ground projection point. The altimeter system can be a lidar, millimeter-wave radar, or ultrasonic sensor, which directly measures the instantaneous distance from the sensor to the surface of the ground below. Since the reconnaissance path itself has already taken into account the basic obstacle avoidance height, this process is an active perception performed under safe conditions.

[0072] Step S3002 involves subtracting the relative altitude from the flight altitude data along the flight path to obtain the prior elevation data for each surface projection point in the target survey area. That is... Through this differential calculation, the system can reproduce the true ground elevation in real time, effectively eliminating interference from non-ground elements such as tree canopies and buildings (the effect is even better when combined with point cloud filtering algorithms). Compared to relying on external data sources, this self-collecting and self-processing data link not only ensures that the elevation information is completely synchronized with the surface state at the current operation time, but also captures dynamic factors such as seasonal vegetation changes and recent construction and renovation, providing irreplaceable data support for the accurate alignment of subsequent three-dimensional spiral flight paths, fundamentally solving the problem of flight path planning failure caused by "difference between map and reality" in complex terrain.

[0073] Example 5 like Figure 7 As shown, this embodiment provides a UAV oblique photography flight path planning system 500 for complex terrain, used to implement the UAV photography flight path planning method for complex terrain as described in the previous embodiment. The system 500, through a modular data processing architecture, encapsulates the abstract flight path planning algorithm into independent functional units, achieving fully automated conversion from raw geographic data to executable flight commands. Specifically, the system 500 includes a data acquisition and preprocessing module 501, a terrain framework construction module 502, a detection flight path planning module 503, and a 3D flight path generation module 504.

[0074] The data acquisition and preprocessing module 501 is used to acquire polygon boundary data and slope data of the target survey area, and calculate the coordinates of the geometric center point of the target survey area based on the polygon boundary data. Specifically, this module is the system's data entry point, responsible for connecting to external geographic information sources. Its input is compatible with various standard formats, such as Shapefile vector boundaries, KML / KMZ landmark files, or API interface call results from online map services. In the internal processing flow, this module not only performs simple coordinate mean calculations, but also performs integrity checks and coordinate system standardization transformations on the raw data to ensure the uniformity of the subsequent calculation benchmark. Its output, the geometric center point coordinates, is a precise two-dimensional or three-dimensional vector, which serves as the spatial reference origin for the entire system and is directly passed to downstream modules. This preprocessing mechanism effectively shields the differences between multi-source heterogeneous data and improves the robustness of the system.

[0075] The terrain framework construction module 502 is used to construct a radial terrain framework based on the coordinates of the geometric center point and a pre-configured azimuth coordinate system. Specifically, this module receives the center point coordinates from the data acquisition and preprocessing module 501 and, combined with user-configured or default-set azimuth parameters (such as octagonal, dodecagonal, etc.), generates a set of topological mesh models in memory. This model is not a solid graphic, but a set of parameters describing the direction of radial lines, spacing rules, and boundary constraints. By discretizing continuous geographic space into structured framework nodes, this module provides a standardized geometric skeleton for subsequent waypoint generation. Compared to directly processing unstructured terrain data, this intermediate-layer abstract representation significantly reduces the computational complexity of the path search algorithm, enabling the system to complete framework initialization in milliseconds.

[0076] The reconnaissance flight path planning module 503 is used to plan flight reconnaissance paths based on a radial terrain framework and slope data, and to control the UAV to perform reconnaissance flights along these paths to obtain prior elevation data for the target survey area. The flight reconnaissance path includes multiple two-dimensional waypoints. Specifically, this module is the core scheduling unit of the system, undertaking the dual functions of "strategy generation" and "execution feedback." In the strategy generation stage, it reads the terrain framework parameters and slope distribution map, and outputs a series of ordered two-dimensional waypoint sequences and corresponding safe flight altitude commands through a built-in adaptive spacing mapping algorithm and optimal path optimizer. In the execution feedback stage, this module sends commands to the UAV flight control system through a communication link and receives altimeter data transmitted back from airborne sensors in real time. After differential inversion, it generates a high-precision prior elevation dataset. This dataset contains the real ground elevation information of each reconnaissance point and is a key data link connecting two-dimensional planning and three-dimensional reconstruction. The design of this module realizes a closed loop between planning and perception, ensuring the system's dynamic adaptability to unknown environments.

[0077] The 3D flight path generation module 504 is used to rearrange multiple 2D waypoints in 3D space based on prior elevation data to obtain a 3D spiral ascent photography flight path. Specifically, this module is the final output unit of the system. It receives prior elevation data and original 2D waypoints output by the exploration flight path planning module 503, and uses 3D mapping and spiral sorting algorithms to reconstruct the planar point set into a continuous spatial trajectory that conforms to the terrain surface. In terms of data format, this module converts the generated 3D spiral ascent photography flight path into a common UAV mission file format (such as Waypoint, Mission, etc.), which not only includes latitude, longitude, and altitude information, but also embeds operational parameters such as heading angle, gimbal pitch angle, and photo trigger distance. Through this end-to-end automated generation mechanism, the system completely eliminates the tedious process of manually adjusting waypoints, significantly improving the efficiency of field preparation while ensuring modeling accuracy.

[0078] Through the modular architecture described above, this invention decomposes the complex flight path planning task into several data processing stages with clearly defined functions and standardized interfaces. The modules interact through clearly defined data objects, ensuring both the continuity of the overall process and allowing for independent upgrades or replacements of individual modules. For example, when adapting to a new type of lidar, only the sensor driver and inversion algorithm in the flight path planning module 503 need to be updated, without modifying the code of other modules. This highly cohesive and loosely coupled design philosophy enables the system to flexibly address the challenges of different UAV models, diverse terrain scenarios, and varying accuracy requirements, demonstrating strong engineering practicality and scalability.

[0079] Example 6 To further verify the practical application effect of the present invention in complex terrain, this embodiment uses a typical karst peak cluster area in Southwest China as the application scenario to reproduce and quantitatively evaluate the entire process of the UAV photography flight path planning method described in the previous embodiment. In this application scenario, the target survey area exhibits a significant cone-shaped or pointed cone-shaped morphology, with steep mountain slopes and dense vegetation covering the surface, representing an extreme test environment where traditional flight path planning struggles to balance accuracy and safety.

[0080] In the specific implementation process, key parameters are first configured according to the method of the aforementioned embodiment. Based on the terrain characteristics of the survey area, the slope range is divided into four continuous intervals: [0°, 30°), [30°, 50°), [50°, 70°), and [70°, 90°]. Based on the principle of equal slope sampling intervals, the steepest interval [70°, 90°] is selected as the benchmark, and the benchmark waypoint spacing is set to 5 meters. Based on this, the initial spacing of the remaining intervals is calculated to be approximately 30 meters, 20 meters, and 10 meters, respectively. Simultaneously, considering that the dominant vegetation in the survey area is eucalyptus trees aged 5 to 8 years, with a maximum mature height of up to 30 meters, and combining the UAV's vertical positioning error with an additional buffer distance, the safe flight altitude relative to the ground is set to above 30 meters to ensure absolute safety when operating in vegetated areas. Based on the above parameters, the system automatically generates a three-dimensional spiral ascent photography flight path that conforms to the terrain surface.

[0081] To objectively evaluate the technical advantages of this invention, three sets of control experiments were set up, using conventional planar flight paths, terrain-following flight paths, and columnar flight paths built into mainstream UAV ground station systems for data collection. To ensure single variables, all control groups were uniformly set with a forward overlap of 80% and a lateral overlap of 60%. Simultaneously, a high-precision 3D laser scanner was used to collect point cloud data of the survey area and construct a digital elevation model as a truth benchmark to calculate the deviation distribution of the modeling results for each scheme. Figures 8 to 11 As shown, the DEM deviation distribution maps of the 3D spiral ascent photography route planned in this invention, the conventional horizontal route, the terrain-following route, and the columnar route in the survey area are displayed sequentially. Taking the present invention and the conventional horizontal route as examples, the model deviation using the present invention is extremely uniformly distributed in the survey area, with an overall whitish tone (representing low deviation), and only slight deviations exist in a very few edge areas; while the model generated by the conventional horizontal route shows large areas of dark patches on the sides of peak clusters and in valley areas, indicating significant elevation errors or even missing data in these areas. This visual difference is quantitatively reflected in the statistical values: as shown in Table 1, the average DEM deviation value of the present invention is only 0.54 meters, significantly better than the 1.24 meters of the conventional horizontal route, the 0.87 meters of the terrain-following route, and the 0.62 meters of the columnar route.

[0082] Table 1

[0083] The microscopic mechanism behind the improved accuracy is primarily attributed to the adaptive matching capability of the 3D spiral ascent photography flight path to complex terrain geometry. Conventional planar flight paths maintain a fixed spacing on the horizontal projection plane, leading to a sharp decrease in actual sampling density on steep slopes, making it impossible to acquire sufficient lateral texture information. While terrain-following flight paths can maintain relatively uniform elevation, they are susceptible to terrain abrupt changes in dense peak cluster areas and lack continuous optimization of the lateral viewing angle. Although columnar flight paths increase the number of circling flights, their vertical layering structure results in sampling blind spots in the transition areas between layers. In contrast, this invention, through a radial terrain framework and a slope-adaptive spacing mechanism, ensures a constant number of effective sampling points per unit slope area. Combined with the continuously spiraling ascent trajectory design, this allows the camera to consistently cover the mountain's lateral details at the optimal tilt angle, fundamentally eliminating modeling voids and texture stretching.

[0084] In terms of data acquisition efficiency, this invention also demonstrates significant advantages. As shown in Table 1, to complete the aforementioned modeling tasks, the method of this invention requires only 98 images, while conventional planar flight routes require 134, terrain-following flight routes require 175, and columnar flight routes require as many as 254. This order-of-magnitude reduction is not due to insufficient sampling, but rather to a precise on-demand sampling strategy. Traditional methods, in order to ensure minimum overlap in steep slope areas, are often forced to take a large number of redundant shots in gentle areas; or, in order to cover the entire survey area height, they have to set an excessively large safety envelope. This invention, through three-dimensional spatial rearrangement driven by prior elevation data, eliminates all invalid airspace and overlapping coverage areas, minimizing the amount of image acquisition. This not only significantly shortens field flight time but also multiplies the cost of indoor data processing and storage, which has important engineering value for large-scale complex terrain mapping.

[0085] Furthermore, improved flight safety is another crucial dimension verified in this embodiment. In field tests, drones using terrain-following flight paths frequently exhibited sudden altitude increases and decreases when flying over peaks, resulting in drastic fluctuations in motor load and a high risk of attitude instability or battery over-discharge. In contrast, the 3D spiral ascent photography flight path generated by this invention maintained a smooth rate of altitude change throughout, ensuring continuous and stable drone attitude without triggering any obstacle avoidance stops or manual intervention. This demonstrates that the flight path planning strategy, based on prior measured elevation data and environmentally aware safety margins, effectively avoids potential threats posed by the combined effects of complex terrain and vegetation, significantly improving the reliability of unmanned operations. In summary, this embodiment, through multi-dimensional quantitative comparisons and mechanistic explanations, fully demonstrates the substantial features and significant advancements of this invention in solving the challenges of drone photography modeling in complex terrain.

[0086] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. For example, the number of azimuths of the radial terrain frame, the granularity of the slope interval division, the specific values ​​of the distance between reference waypoints, the calculation parameters of the safe flight altitude, and the rotation direction and connection strategy of the three-dimensional spiral trajectory can all be adaptively adjusted according to the actual survey area characteristics, UAV platform performance, and modeling accuracy requirements; the physical deployment form, data interaction protocol, and algorithm implementation method of each functional module can also be replaced by equivalent means in the prior art. These non-substantial modifications made based on the concept of the present invention should not be considered as departing from the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for planning flight paths for UAV oblique photography in complex terrain, characterized in that, include: Acquire polygon boundary data and slope data of the target survey area, and calculate the coordinates of the geometric center point of the target survey area based on the polygon boundary data; A radial terrain framework is constructed based on the coordinates of the geometric center point and a pre-configured azimuth coordinate system. Based on the radial terrain framework and slope data, a flight exploration route is planned, and the UAV is controlled to perform exploration flights along the flight exploration route to obtain the prior elevation data of the target survey area; the flight exploration route includes multiple two-dimensional waypoints; Based on prior elevation data, multiple two-dimensional waypoints are rearranged in three-dimensional space to obtain a three-dimensional spiral ascent photography route. The flight exploration route planning based on the radial terrain framework and slope data includes: The slope data is matched with the preset slope-waypoint spacing mapping table to obtain the waypoint spacing; Coordinate calculations were performed based on the radial terrain framework and waypoint spacing to obtain multiple two-dimensional waypoints and their corresponding coordinates. Based on the optimal path objective function, a flight route map is planned using multiple two-dimensional waypoints to obtain the flight route map. Acquire the altitude data of the target survey area, add the altitude data to the preset safe altitude, and obtain the flight altitude data; By associating and fusing flight path maps and flight altitude data, a flight reconnaissance route is obtained; The steps for constructing the slope-waypoint spacing mapping table include: The slope range of the target measurement area is divided into multiple continuous slope intervals, and a representative slope angle is determined for each slope interval based on the statistical characteristics of the pixel slope within each slope interval. The slope interval corresponding to the largest representative slope angle is selected as the benchmark interval, and the benchmark waypoint spacing is set for the benchmark interval based on the obtained sampling distance requirements. Based on the baseline waypoint spacing, the initial waypoint spacing for the remaining slope intervals is calculated according to the equal slope sampling spacing principle; the equal slope sampling spacing principle means that the quotient of the waypoint spacing for different slope intervals and the cosine value of the corresponding slope angle remains constant. Based on the micro-topographical complexity of each slope range and the constraints of UAV flight imaging, the initial waypoint spacing is corrected to obtain the corrected waypoint spacing. Each slope interval is paired with the corresponding corrected waypoint spacing, and discretization, rounding, and smoothing of adjacent intervals are performed to obtain a slope-waypoint spacing mapping table.

2. The method for planning flight paths for UAV oblique photography in complex terrain according to claim 1, characterized in that, The correction of the initial waypoint spacing includes: Calculate the micro-topographic complexity index for each slope interval; the micro-topographic complexity index is the mean of the local standard deviations of the slope of all pixels within the corresponding slope interval. The complexity adjustment interval is obtained by multiplying the initial waypoint spacing by a preset adjustment coefficient; the adjustment coefficient is negatively correlated with the micro-topography complexity index. Based on the attitude angle changes and camera field of view parameters of the UAV when maintaining a constant altitude flight at the corresponding slope, the upper limit of the spacing allowed to meet the preset minimum forward overlap and minimum lateral overlap in each slope interval is calculated, and a lower limit of the spacing to avoid sampling redundancy is set. By constraining the complexity adjustment spacing to the interval formed by the upper and lower limits of the spacing, the corrected waypoint spacing is obtained.

3. The method for planning UAV oblique photography flight paths in complex terrain according to claim 1, characterized in that, The coordinate calculation based on the radial terrain framework and waypoint spacing includes: Construct an azimuth coordinate system containing multiple principal and auxiliary directions with the coordinates of the geometric center point as the origin; Using the coordinates of the geometric center point as the center, multiple concentric circles with different radii are determined according to the spacing between waypoints; the radius is not greater than the minimum straight-line distance between the coordinates of the geometric center point and the boundary coordinates of the target survey area; Calculate the intersection points of each concentric circle with each azimuth line in the azimuth coordinate system to obtain multiple two-dimensional waypoints and their corresponding coordinates.

4. The method for planning flight paths for UAV oblique photography in complex terrain according to claim 1, characterized in that, The objective function for the optimal path is a weighted cost function, including: The path length cost is calculated as the sum of the Euclidean distances between all adjacent two-dimensional waypoints in the flight path diagram, and is used to constrain the total flight path to be the shortest. The turning angle cost is calculated as the sum of the absolute values ​​of the directional change angles between all adjacent segments in the flight path diagram. It is used to constrain the flight to have the fewest turns and the highest proportion of straight flight segments. Resource consumption cost is calculated as the product of path length cost and energy consumption per unit distance, plus the product of turning angle cost and energy consumption per unit angle of turn, used to constrain the resource consumption of flight exploration missions to the minimum.

5. The method for planning flight paths for UAV oblique photography in complex terrain according to claim 1, characterized in that, The method of rearranging multiple two-dimensional waypoints in three-dimensional space based on prior elevation data includes: Based on prior elevation data and preset safe flight altitude, multiple two-dimensional waypoints are mapped to a pre-constructed three-dimensional space to obtain multiple three-dimensional waypoints; Multiple 3D waypoints are grouped and sorted based on the horizontal straight-line distance between the coordinates of the 3D waypoints and the geometric center point to obtain a grouped coordinate sequence; The grouped coordinate sequences are connected sequentially in a clockwise or counterclockwise direction to form a three-dimensional spiral trajectory that spirals upward along the terrain surface of the target survey area. Exposure points are inserted on the three-dimensional spiral trajectory according to the preset encryption interval to obtain the three-dimensional spiral ascent photography route.

6. The method for planning UAV oblique photography flight paths in complex terrain according to claim 5, characterized in that, The methods for obtaining the safe flight altitude include: Acquire vegetation type distribution data and corresponding prior knowledge of vegetation growth height in the target survey area; Based on prior knowledge of vegetation growth height, the dominant vegetation type with the maximum maturity height in the target survey area is selected, and the corresponding maximum vegetation barrier height is determined. The safe flight altitude is obtained by adding the maximum vegetation obstacle height to the preset basic safety margin; the basic safety margin is the sum of the allowable vertical positioning error of the UAV when flying at a fixed altitude and the preset additional buffer distance.

7. The method for planning UAV oblique photography flight paths in complex terrain according to claim 1, characterized in that, The methods for obtaining the prior elevation data include: Control the UAV to perform reconnaissance flight along the flight reconnaissance route, and use the onboard altimeter system to obtain the relative height of each two-dimensional waypoint relative to the corresponding ground projection point; Subtracting the relative altitude from the flight altitude data in the flight exploration route yields the prior elevation data of various ground projection points in the target survey area.

8. A UAV oblique photography flight path planning system for complex terrain, used to implement the UAV oblique photography flight path planning method for complex terrain as described in any one of claims 1 to 7, characterized in that, include: The data acquisition and preprocessing module is used to acquire polygon boundary data and slope data of the target survey area, and calculate the coordinates of the geometric center point of the target survey area based on the polygon boundary data; The terrain framework building module is used to build radial terrain frameworks based on the coordinates of the geometric center point and a pre-configured azimuth coordinate system. The reconnaissance route planning module is used to plan flight reconnaissance routes based on radial terrain framework and slope data, and control the UAV to perform reconnaissance flights along the flight reconnaissance routes to obtain prior elevation data of the target survey area; the flight reconnaissance route includes multiple two-dimensional waypoints; The 3D route generation module is used to rearrange multiple 2D waypoints in 3D space based on prior elevation data to obtain a 3D spiral ascent photography route. The flight exploration route planning based on the radial terrain framework and slope data includes: The slope data is matched with the preset slope-waypoint spacing mapping table to obtain the waypoint spacing; Coordinate calculations were performed based on the radial terrain framework and waypoint spacing to obtain multiple two-dimensional waypoints and their corresponding coordinates. Based on the optimal path objective function, a flight route map is planned using multiple two-dimensional waypoints to obtain the flight route map. Acquire the altitude data of the target survey area, add the altitude data to the preset safe altitude, and obtain the flight altitude data; By associating and fusing flight path maps and flight altitude data, a flight reconnaissance route is obtained; The steps for constructing the slope-waypoint spacing mapping table include: The slope range of the target measurement area is divided into multiple continuous slope intervals, and a representative slope angle is determined for each slope interval based on the statistical characteristics of the pixel slope within each slope interval. The slope interval corresponding to the largest representative slope angle is selected as the benchmark interval, and the benchmark waypoint spacing is set for the benchmark interval based on the obtained sampling distance requirements. Based on the baseline waypoint spacing, the initial waypoint spacing for the remaining slope intervals is calculated according to the equal slope sampling spacing principle; the equal slope sampling spacing principle means that the quotient of the waypoint spacing for different slope intervals and the cosine value of the corresponding slope angle remains constant. Based on the micro-topographical complexity of each slope range and the constraints of UAV flight imaging, the initial waypoint spacing is corrected to obtain the corrected waypoint spacing. Each slope interval is paired with the corresponding corrected waypoint spacing, and discretization, rounding, and smoothing of adjacent intervals are performed to obtain a slope-waypoint spacing mapping table.

Images