An unmanned aerial vehicle close-up photography path planning method and system for landslide monitoring

Abstract

The application provides a UAV close-up photography flight path planning method and system for landslide monitoring. The method comprises the following steps: obtaining an initial three-dimensional model of a target area, and extracting three-dimensional parameters of multiple profiles of a target surface based on the initial three-dimensional model; adaptively generating a flight path planning surface that is close to the target surface according to the three-dimensional parameters of the multiple profiles; planning a snake-shaped flight path on the flight path planning surface, and controlling a UAV to fly along the snake-shaped flight path; and calculating the best image acquisition view angle of each shooting position according to the flight path planning surface and the target geometry, and generating a flight path planning result containing position and view angle information. The technical scheme provided by the application can adaptively extract a flight path planning surface that is close to complex terrains such as landslides, improve image coverage integrity and resolution consistency, simultaneously optimize the snake-shaped flight path and the best shooting view angle, and generate a high-quality flight path planning result that can support multi-period repeated flights and subsequent deformation monitoring.

Description

Technical Field

[0001] This application relates to the field of UAV close-up photography trajectory planning technology, and in particular to a UAV close-up photography trajectory planning method and system for landslide monitoring. Background Technology

[0002] Unmanned aerial vehicle (UAV) close-up photogrammetry, as an important technical means in the field of geological disaster monitoring, is widely used for refined 3D modeling and deformation monitoring of complex terrains such as landslides and steep slopes. Among related technologies, a "coarse-to-fine" operational strategy is employed, combining coarse background information acquisition, approximate description of regular shapes, and serpentine or circular flight path planning to construct a complete workflow from data acquisition to automated flight. Specifically, this technology covers key aspects such as preliminary oblique photogrammetry, flight path parameter calculation, and flight control system import, aiming to acquire high-resolution, high-overlap surface images.

[0003] However, existing close-up photogrammetry trajectory planning methods directly approximate complex targets such as landslides using single, regular shapes like planes and cylinders, without adaptively extracting profile parameters based on the actual target shape. This can lead to low alignment between the planned trajectory surface and the terrain, incomplete image coverage, or inconsistent resolution, thus affecting the accuracy of subsequent 3D modeling and the reliability of deformation monitoring. Furthermore, fixed overlap settings struggle to accommodate the shooting needs of different terrains, further exacerbating data redundancy or insufficient coverage. Summary of the Invention

[0004] The present invention aims to at least partially solve one of the technical problems in the related art.

[0005] Therefore, the first aspect of this application proposes a method for planning UAV close-up photography flight paths for landslide monitoring, the method comprising:

[0006] Obtain an initial three-dimensional model of the target area, and extract three-dimensional parameters of multiple cross-sections of the target surface based on the initial three-dimensional model; Based on the three-dimensional parameters of the multiple profiles, an adaptive trajectory planning surface that fits the target surface is generated. Plan a serpentine flight path on the flight path planning surface and control the UAV to fly along the serpentine flight path; Based on the trajectory planning surface and target geometry, the optimal image acquisition angle for each shooting position is calculated, and a trajectory planning result containing position and angle information is generated.

[0007] Preferably, the step of obtaining an initial three-dimensional model of the target region and extracting three-dimensional parameters of multiple cross-sections of the target surface based on the initial three-dimensional model includes: Oblique photogrammetry is performed through conventional flight path missions to acquire multi-view images of the target area, and an initial 3D model is constructed based on the multi-view images. The initial three-dimensional model is subjected to profile fitting to extract the three-dimensional parameters of multiple profiles of the target surface.

[0008] Furthermore, the three-dimensional parameters include the normal vector of the profile and the boundary points.

[0009] Furthermore, the adaptive generation of a trajectory planning surface that fits the target surface based on the three-dimensional parameters of the plurality of profiles includes: Based on the three-dimensional parameters of the multiple profiles, the target surface is decomposed into multiple basic units, including elevations and slopes; Generate the trajectory planning surface for each basic unit separately, and combine the trajectory planning surfaces of all basic units to form a complete trajectory planning surface.

[0010] Furthermore, when the serpentine flight path is executed by a quadcopter UAV, smooth turns are achieved at the turning points through the coordination between the rotors, and the flight path planning results include turning control parameters.

[0011] Preferably, the optimal image acquisition angle is the camera optical axis direction that minimizes the deformation of the target surface in the image and provides the richest information. The gimbal attitude parameters include pitch angle, yaw angle, and roll angle.

[0012] Furthermore, the method also includes: Based on the same flight path planning results, the UAV is controlled to fly at the same waypoint position and gimbal attitude at different times to obtain the same image sequence from the same perspective, which is used for subsequent image matching and deformation analysis.

[0013] Furthermore, the oblique photogrammetry has a forward overlap of 80% and a lateral overlap of 30%.

[0014] Preferably, the number of the plurality of cross-sections is 14.

[0015] The second aspect of this application proposes a UAV close-up photography flight path planning system for landslide monitoring, comprising: The acquisition module is used to acquire an initial three-dimensional model of the target area and extract three-dimensional parameters of multiple cross-sections of the target surface based on the initial three-dimensional model. The generation module is used to adaptively generate a trajectory planning surface that fits the target surface based on the three-dimensional parameters of the multiple profiles. The control module is used to plan a serpentine flight path on the flight path planning surface and control the UAV to fly along the serpentine flight path; The acquisition module is used to calculate the optimal image acquisition angle for each shooting position based on the trajectory planning surface and target geometry, and generate trajectory planning results containing position and angle information.

[0016] The technical solutions provided by the embodiments of this application bring at least the following beneficial effects: This application proposes a method and system for UAV close-up photography trajectory planning for landslide monitoring. The method includes: acquiring an initial 3D model of the target area and extracting 3D parameters of multiple profiles of the target surface based on the initial 3D model; adaptively generating a trajectory planning surface that conforms to the target surface based on the 3D parameters of the multiple profiles; planning a serpentine flight path on the trajectory planning surface and controlling the UAV to fly along the serpentine flight path; calculating the optimal image acquisition angle for each shooting position based on the trajectory planning surface and the target geometry, and generating a trajectory planning result containing position and angle information. The technical solution proposed in this application can adaptively extract a trajectory planning surface that conforms to complex terrain such as landslides, improve image coverage integrity and resolution consistency, while optimizing the serpentine flight path and the optimal shooting angle. The generated high-quality trajectory planning result can support multiple repeated flights and subsequent deformation monitoring.

[0017] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0018] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a method for planning drone close-up photography flight paths for landslide monitoring according to an embodiment of this application; Figure 2 This is a structural diagram of a UAV close-up photography flight path planning system for landslide monitoring, provided according to an embodiment of this application. Detailed Implementation

[0019] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0020] This application proposes a method and system for UAV close-up photography trajectory planning for landslide monitoring. The method includes: acquiring an initial 3D model of the target area and extracting 3D parameters of multiple profiles of the target surface based on the initial 3D model; adaptively generating a trajectory planning surface that conforms to the target surface based on the 3D parameters of the multiple profiles; planning a serpentine flight path on the trajectory planning surface and controlling the UAV to fly along the serpentine flight path; calculating the optimal image acquisition angle for each shooting position based on the trajectory planning surface and the target geometry, and generating a trajectory planning result containing position and angle information. The technical solution proposed in this application can adaptively extract a trajectory planning surface that conforms to complex terrain such as landslides, improve image coverage integrity and resolution consistency, while optimizing the serpentine flight path and the optimal shooting angle. The generated high-quality trajectory planning result can support multiple repeated flights and subsequent deformation monitoring.

[0021] The following describes, with reference to the accompanying drawings, a method and system for planning drone close-up photography flight paths for landslide monitoring, according to an embodiment of this application.

[0022] Example 1 Figure 1 This is a flowchart illustrating a method for planning drone close-up photography flight paths for landslide monitoring, according to an embodiment of this application. Figure 1 As shown, the method includes: Step S1: Obtain an initial three-dimensional model of the target area, and extract three-dimensional parameters of multiple cross-sections of the target surface based on the initial three-dimensional model; In this embodiment of the disclosure, step S1 specifically includes: Oblique photogrammetry is performed through conventional flight path missions to acquire multi-view images of the target area, and an initial 3D model is constructed based on the multi-view images. The initial three-dimensional model is subjected to profile fitting to extract the three-dimensional parameters of multiple profiles of the target surface.

[0023] The three-dimensional parameters include the normal vector of the profile and the boundary points.

[0024] The oblique photogrammetry has a forward overlap of 80% and a lateral overlap of 30%.

[0025] The number of the multiple cross-sections is 14.

[0026] It should be noted that after obtaining the initial 3D model of the target area, 3D parameters of multiple profiles of the target surface are extracted based on this model. This step aims to provide a basic geometric basis for the subsequent generation of adaptive trajectory planning surfaces. Specifically, firstly, rough 3D information of the target area is obtained through preliminary measurement methods. This information can come from existing terrain data or be obtained through low-precision aerial surveying, thereby constructing an initial 3D model reflecting the morphology of the target surface. Subsequently, based on this initial 3D model, the target surface is subjected to profile analysis. Multiple profile 3D parameters are extracted through fitting or calculation. These parameters describe the local geometric features of the target surface at different locations, including but not limited to the normal vector of the profile and the coordinates of the boundary points of the profile. The normal vector reflects the orientation of the plane in which the profile is located, while the boundary point coordinates define the range and position of the profile in space. By obtaining the 3D parameters of multiple profiles, the complex and irregular target surface can be decomposed into multiple local units with clear geometric features, thereby providing accurate geometric constraints and data support for the subsequent adaptive generation of trajectory planning surfaces that fit the target surface. As one implementation method, oblique photogrammetry can be performed using conventional flight path missions to obtain multi-view images of the target area. The three-dimensional parameters of multiple profiles can be extracted by continuous fitting, such as extracting the normal vectors and boundary point coordinates of 14 profiles for subsequent trajectory planning.

[0027] This step extracts three-dimensional parameters from multiple profiles of the target surface, enabling a precise quantitative description of the complex terrain geometry. This provides a reliable geometric basis for the subsequent adaptive generation of a trajectory planning surface that fits the target surface, thereby improving the matching degree between the trajectory planning and the target surface and ensuring the integrity and consistency of image coverage.

[0028] In this embodiment, the specific implementation of obtaining the initial 3D model of the target area and extracting the 3D parameters of multiple cross-sections of the target surface in step S1 is as follows. First, oblique photogrammetry is performed through a conventional flight path mission to obtain multi-view images of the target area. Specifically, flight path planning software such as DJI Pilot can be used to set flight parameters according to specifications, where the forward overlap is set to 80% and the lateral overlap is set to 30%, thereby designing and executing the oblique photogrammetry flight path. The UAV flies along this flight path and takes pictures of the target area from multiple angles, thereby obtaining multi-view image data covering the target area. These multi-view images are used as input for subsequent 3D reconstruction processing. Next, based on the acquired multi-view images, an initial 3D model of the target area is constructed through photogrammetry processing workflows such as Structure from Motion (SfM) or Multi-View Stereo (MVS). This initial 3D model represents the rough geometry of the target surface in the form of point clouds or meshes. Then, the constructed initial 3D model is subjected to cross-section fitting processing to extract the 3D parameters of multiple cross-sections of the target surface. Specifically, along the extension direction of the target surface, multiple profile positions are selected on the initial 3D model at certain intervals or based on terrain variation features. At each profile position, a fitting algorithm (e.g., least squares method) is used to perform planar fitting on the local point cloud or mesh data of the model surface at that location, thereby obtaining the 3D parameters of the profile. In this embodiment, the extracted 3D parameters specifically include the normal vector and boundary point coordinates of the profile. The normal vector characterizes the orientation of the profile in 3D space, and the boundary points define the spatial extent of the profile. As an example, 3D parameters of 14 profiles can be extracted. These parameters collectively constitute a quantitative description of the local geometry of the target surface and serve as input data for the adaptive generation of the trajectory planning surface in the subsequent step S2.

[0029] This specific implementation method acquires multi-view images and constructs an initial three-dimensional model by using oblique photogrammetry with high overlap. Then, it performs profile fitting on the model to extract three-dimensional parameters including normal vectors and boundary points. This achieves accurate quantification of the geometric shape of complex terrain surfaces such as landslides, providing a reliable data foundation for the subsequent adaptive generation of a trajectory planning surface that fits the target surface, thereby improving the accuracy of trajectory planning and terrain matching.

[0030] In this embodiment, when extracting the three-dimensional parameters of multiple profiles of the target surface in step S1, the number of multiple profiles is specifically set to 14. This number is not arbitrary but based on adaptive analysis of the complex terrain of the target area. Specifically, after acquiring the initial three-dimensional model of the target area, the system first performs geometric feature analysis on the model to identify areas with drastic terrain changes, such as steep slopes, slope inflections, and boundary regions of landslides. Subsequently, the system automatically determines the cutting position and direction of the profiles based on the distribution of these feature regions to ensure that each profile can effectively represent the local geometry of the target surface at that location. In this embodiment, after multiple fitting and verifications of typical landslide scenarios, it was determined that extracting 14 profiles can balance the precision of data acquisition with computational efficiency, fully covering the main geometric changes of the target surface while avoiding data redundancy and processing burden caused by too many profiles. The three-dimensional parameters of each profile include the normal vector of the profile and the coordinates of the boundary points. The normal vector describes the orientation of the plane containing the profile, while the coordinates of the boundary points define the specific location and range of the profile in three-dimensional space. The 3D parameters of these 14 profiles collectively constitute a discretized description of the target surface geometry, providing accurate input data for the subsequent adaptive generation of a flight path planning surface that conforms to the target surface. In one possible implementation, the system uses an iterative fitting algorithm to progressively extract the parameters of these 14 profiles from the initial 3D model, ensuring that the normal vector and boundary points of each profile most accurately reflect the surface orientation and extent of the local area. These parameters are then output and used as input to step S2 to generate a flight path planning surface that conforms to the target surface.

[0031] By limiting the number of profiles to 14, this embodiment achieves a refined and efficient description of the target surface geometry. This ensures that the trajectory planning surface can fully conform to complex terrain while avoiding the waste of computing resources caused by an excessive number of profiles. Thus, while ensuring the integrity of image coverage, it improves the overall efficiency of trajectory planning.

[0032] Step S2: Based on the three-dimensional parameters of the multiple profiles, adaptively generate a trajectory planning surface that fits the target surface; In this embodiment of the disclosure, step S2 specifically includes: Based on the three-dimensional parameters of the multiple profiles, the target surface is decomposed into multiple basic units, including elevations and slopes; Generate the trajectory planning surface for each basic unit separately, and combine the trajectory planning surfaces of all basic units to form a complete trajectory planning surface.

[0033] It should be noted that a flight path planning surface that conforms to the target surface is adaptively generated based on the three-dimensional parameters of the multiple cross-sections. This step aims to transform the complex and irregular surface morphology of the target area into a geometric surface that can be used as a reference for UAV flight path planning. Specifically, based on the three-dimensional parameters of multiple cross-sections extracted from the initial three-dimensional model, including but not limited to the normal vectors and boundary point coordinates of each cross-section, a flight path planning surface that highly conforms to the target surface morphology can be adaptively constructed by analyzing the local geometric features of the target surface represented by these parameters. This flight path planning surface is not a single regular geometric shape, but a composite curved surface or folded surface composed of multiple basic units. These basic units can correspond to different morphological regions of the target surface, such as elevations or slopes. By fitting and generating a flight path planning surface for each basic unit separately, and combining and connecting the planning surfaces corresponding to each unit, a complete flight path planning surface covering the entire target area is finally formed. This planning surface provides an accurate reference benchmark for subsequent flight path planning, ensuring that the UAV can fly along a path conforming to the target surface. As one implementation method, complex terrain can be decomposed into multiple basic units such as facades and slopes, and then the corresponding planning surfaces can be generated and combined to achieve adaptive description of irregular landslides or steep slope surfaces.

[0034] This step significantly improves the adaptability and coverage integrity of trajectory planning to complex terrain by adaptively generating a trajectory planning surface that fits the target surface. It avoids planning deviations and image coverage loss caused by using a single regular shape to describe the target, and lays a reliable geometric foundation for subsequent high-precision, high-overlap close-up photogrammetry.

[0035] In this embodiment, step S2, which adaptively generates a trajectory planning surface conforming to the target surface based on the three-dimensional parameters of the multiple profiles, specifically includes the following sub-steps. First, the system receives the three-dimensional parameters of the multiple profiles extracted in step S1 as input. These parameters specifically include the normal vector and boundary point coordinates of each profile. Based on these parameters, the system performs geometric decomposition on the target surface, identifying and dividing complex and irregular landslide or steep slope surfaces into multiple basic units. The basic units specifically include elevations and slopes. An elevation refers to a region with an approximately horizontal surface normal vector and a steep slope, while a slope refers to a region with a significant tilt angle in the surface normal vector. The system determines the type of basic unit to which each profile belongs based on the spatial range enclosed by the normal vector direction and boundary points of each profile, thereby completing the decomposition of the target surface. Subsequently, for each identified basic unit, the system generates its corresponding trajectory planning surface. Specifically, for elevation units, the system uses its normal vector as a reference to generate a plane parallel to the elevation at a preset safe distance from the target surface as the trajectory planning surface. For slope units, the system generates an inclined plane that conforms to the slope surface shape as the trajectory planning surface, based on the slope's inclination angle and boundary points. The trajectory planning surface of each basic unit contains the spatial coordinates of the theoretical shooting position of the UAV within that unit. Finally, the system spatially combines and splices the trajectory planning surfaces generated independently by all basic units, eliminating overlaps or gaps between adjacent planning surfaces to form a complete, continuous, and closely conforming trajectory planning surface to the overall target surface. This complete trajectory planning surface serves as the direct reference surface for planning the serpentine flight path in subsequent step S3.

[0036] By adaptively decomposing complex target surfaces into basic units such as elevations and slopes and generating planning surfaces for each, this implementation method can accurately fit the actual geometry of irregular terrains such as landslides, significantly improving the matching degree between the flight path planning surface and the target surface. This provides a high-precision spatial reference for subsequent serpentine flight path planning and ensures the integrity and consistency of image coverage.

[0037] Step S3: Plan a serpentine flight path on the flight path planning surface and control the UAV to fly along the serpentine flight path; In this embodiment of the disclosure, when the serpentine flight path is executed by a quadcopter UAV, smooth turns are achieved at the turning points through the cooperation between the rotors, and the flight path planning results include turning control parameters.

[0038] It should be noted that a serpentine flight path is planned on the flight path planning surface, and the UAV is controlled to fly along the serpentine flight path. Specifically, based on the generated flight path planning surface that conforms to the target surface, one or more serpentine flight paths in a reciprocating pattern are arranged on the planning surface. The goal of the flight path is to continuously cover the entire planning surface. By setting the starting point, turning point, and heading spacing of the flight path, the UAV can sequentially traverse each shooting area on the planning surface during flight. The planning of the serpentine flight path needs to consider the flight performance of the UAV and the geometric characteristics of the target area. For example, according to the changes in the width, length, and curvature of the planning surface, the turning position and heading spacing of the flight path are adaptively adjusted to ensure that the flight path completely covers the planning surface and avoids redundant flight. After the flight path is planned, the generated flight path information is imported into the UAV's flight control system, and the UAV is controlled to fly automatically along the serpentine flight path. During flight, the UAV performs straight-line flight and turning maneuvers according to the preset flight path parameters. The turning maneuver can be smoothly transitioned by utilizing the UAV's own maneuverability. For example, small-radius turns can be completed through coordinated control between the rotors, thereby reducing the space and time required for turning and improving aerial survey efficiency. As one implementation method, the serpentine route is applicable to large-scale aerial survey areas. By using a back-and-forth path pattern, the UAV can efficiently cover the entire planned area without frequently adjusting its course. At the same time, the turning strategy is optimized based on the flight characteristics of the UAV to avoid repeated measurements or coverage gaps caused by large turning radii.

[0039] This step, by planning a serpentine flight path and controlling the UAV to fly along it, achieves efficient and continuous coverage of the planned flight path surface, significantly improving the integrity and efficiency of aerial survey operations. At the same time, by utilizing the maneuverability of the UAV to optimize the turning process, the flight time and energy consumption are reduced. It is especially suitable for landslide monitoring scenarios with complex surfaces and large areas, providing a stable and reliable flight path foundation for subsequent high-resolution image acquisition.

[0040] In this embodiment, the specific implementation of controlling the UAV to fly along the serpentine flight path in step S3 is as follows. First, based on the trajectory planning surface adaptively generated in step S2, a serpentine flight path conforming to the target surface is planned. This flight path consists of a series of parallel or nearly parallel flight path segments and turning segments connecting adjacent flight path segments. The UAV used is a quadcopter, whose four rotors are independently controllable and can achieve flexible maneuverability through differential speed adjustment. When the UAV flies along the serpentine flight path to a turn, the flight control system calculates the speed adjustment of each rotor in real time according to the preset turning path and the current flight state. Specifically, during the turn, by increasing the speed of the inner rotor and decreasing the speed of the outer rotor, a yaw torque is generated, enabling the UAV to achieve a smooth turn while maintaining stable altitude and speed. At the same time, utilizing the symmetrical layout and rapid response characteristics of the quadcopter, the turning radius can be controlled within a very small range, without the need to reserve a large turning space as with fixed-wing or long-wheelbase UAVs. This rotor-assisted mechanism enables the UAV to continuously and efficiently complete multiple turns within a serpentine flight path, avoiding deviations or redundant measurements caused by excessively large turning radii, thus ensuring the efficiency and accuracy of the entire aerial survey process. After completing a turn, the UAV automatically returns to straight flight and continues its close-up photography mission along the next flight segment. Throughout the flight, the flight control system continuously monitors the UAV's attitude, position, and speed, and ensures the smoothness and stability of turning maneuvers through closed-loop control.

[0041] This specific implementation method significantly reduces the turning radius by using a quadcopter UAV and utilizing the rotors to achieve smooth turns at the bends of the serpentine flight path, thus avoiding repeated measurements and greatly improving the efficiency of aerial surveying. At the same time, it ensures the continuity of the flight path and the integrity of image acquisition, providing a reliable data foundation for subsequent multi-phase image matching and deformation analysis.

[0042] Step S4: Based on the trajectory planning surface and target geometry, calculate the optimal image acquisition angle for each shooting position and generate a trajectory planning result containing position and angle information.

[0043] In this embodiment of the disclosure, the optimal image acquisition angle is the camera optical axis direction that minimizes the deformation of the target surface in the image and provides the richest information. The gimbal attitude parameters include pitch angle, yaw angle, and roll angle.

[0044] In this embodiment of the disclosure, the method further includes: Based on the same flight path planning results, the UAV is controlled to fly at the same waypoint position and gimbal attitude at different times to obtain the same image sequence from the same perspective, which is used for subsequent image matching and deformation analysis.

[0045] It should be noted that, based on the generated trajectory planning surface conforming to the target surface, and combined with the local geometry of the target surface, for each preset shooting position, the spatial relationship between the normal direction of the trajectory planning surface and the normal direction of the target surface at that position is analyzed to determine the viewing angle that minimizes the angle between the camera optical axis and the normal direction of the target surface, while also minimizing the geometric distortion and occlusion effects of the target surface in the image. This optimal image acquisition viewing angle is taken as the optimal image acquisition viewing angle for that shooting position. The process of determining this optimal image acquisition viewing angle can be calculated based on the three-dimensional geometric parameters of the target surface and the spatial position of the trajectory planning surface. Its core is to make the camera optical axis as perpendicular as possible to the local area of ​​the target surface, thereby acquiring high-resolution, low-distortion image data. After obtaining the optimal image acquisition viewing angle for each shooting position, this viewing angle information is converted into attitude control parameters for the camera gimbal, including pitch angle, yaw angle, and roll angle, and transmitted in real time to the UAV flight control system to drive the gimbal to automatically adjust to the corresponding attitude when the UAV arrives at each shooting position, completing image acquisition. This method ensures that the acquired images closely fit the target surface in terms of spatial coverage, reducing insufficient image overlap or data redundancy caused by viewpoint deviations, while improving the geometric fidelity of the images and providing a high-quality data foundation for subsequent 3D reconstruction and deformation analysis. As one implementation method, images from the same viewpoint can be acquired based on two separate 3D models to support subsequent image matching and temporal deformation analysis.

[0046] This step, through perspective optimization calculations based on the flight path planning surface and target geometry, achieves precise matching between the camera optical axis direction and the target surface normal, significantly improving the geometric quality and coverage integrity of image acquisition, reducing occlusion and tilt distortion, and providing reliable image data support for high-precision 3D modeling and multi-stage deformation monitoring.

[0047] In this embodiment, the core of the optimal image acquisition angle described in step S4 lies in determining a specific camera optical axis direction. This direction ensures that the target surface exhibits minimal geometric deformation in the final acquired image while simultaneously capturing the richest surface texture and structural information. Specifically, the calculation process for this optimal image acquisition angle is based on the flight path planning surface generated in step S2 and the target geometric information extracted in step S1. First, the system acquires the three-dimensional coordinates of each preset shooting position on the flight path planning surface, as well as the geometric parameters of the corresponding local area of ​​the target surface, including the approximate normal vector of the local area. Then, using this normal vector as a reference, combined with the geometric principles of camera imaging, an optimal camera optical axis direction is calculated. This direction is not simply perpendicular to the target surface, but rather, through an optimization algorithm, minimizes image distortion caused by perspective projection while ensuring image coverage integrity. For example, it maximizes the projection area of ​​feature points on the target surface onto the image plane, while avoiding texture stretching or compression caused by excessively steep or gentle shooting angles. In one possible implementation, the optimization process can be based on an objective function that considers the image coverage area, overlap, and the angle between the target surface and the imaging plane. The optimal camera optical axis direction is obtained through iterative solving. The calculated optimal image acquisition angle is output as three-dimensional attitude angle data, including pitch, yaw, and roll angles. This data is directly transmitted to the camera gimbal control module in the UAV flight control system. The gimbal control module adjusts the camera attitude accordingly, ensuring that the camera optical axis precisely points in the calculated optimal direction as the UAV flies along a serpentine flight path to each preset shooting point, thus achieving high-quality image acquisition of the target surface.

[0048] This specific implementation significantly improves the quality of the acquired images by precisely calculating and controlling the camera's optimal shooting angle, ensuring the integrity and geometric fidelity of the target surface information in the images, and providing a reliable data foundation for subsequent high-precision 3D modeling and accurate matching of multiple images.

[0049] This embodiment provides a detailed description of the complete implementation of a UAV close-up photography trajectory planning method for landslide monitoring. This method follows a "coarse-to-fine" and "from scratch" operational strategy. By adaptively extracting a trajectory planning surface that fits the target surface, and combining a serpentine flight path with optimal viewing angle optimization, it achieves refined and efficient close-up photography of complex terrains such as landslides and steep slopes, supporting subsequent image matching and deformation analysis.

[0050] First, preliminary survey data acquisition and profile parameter extraction are performed. Preliminary oblique photogrammetry is conducted using conventional flight path planning software (such as DJIPilot) to obtain an initial 3D model of the target area. Specifically, flight parameters are set according to specifications, with a forward overlap of 80% and a lateral overlap of 30%, and multi-view images of the target area are acquired along this flight path. Based on the acquired multi-view images, an initial 3D model of the target area is constructed using algorithms such as Structure from Motion (SfM) or Multi-View Stereo (MVS). Subsequently, profile fitting processing is performed on this initial 3D model, extracting 3D parameters from multiple profiles of the target surface through continuous fitting. In this embodiment, 3D parameters of 14 profiles are extracted, each including the normal vector and the coordinates of its boundary points. These profile parameters describe the local geometry of the target surface, providing a data foundation for the subsequent generation of adaptive flight path planning surfaces.

[0051] Next, based on the extracted 3D parameters from multiple profiles, a trajectory planning surface adaptively conforms to the target surface. Specifically, based on the normal vectors and boundary points of each profile, complex surface targets (such as landslides and steep slopes) are decomposed into multiple basic units, which mainly include elevations and slopes. For each basic unit, a trajectory planning surface conforming to its surface geometry is generated. For example, for elevation units, a planning surface parallel to the elevation is generated; for slope units, a planning surface at a set distance from the normal to the slope surface is generated. Finally, the trajectory planning surfaces of all basic units are combined to form a complete trajectory planning surface covering the entire target area. This process enables the trajectory planning surface to adaptively conform to irregular target surfaces, avoiding the poor adaptability problem caused by approximating a single regular shape (such as a plane or cylinder).

[0052] After generating the flight path planning surface, a serpentine flight path is planned on this surface. The serpentine flight path is a zigzag reciprocating flight path pattern, suitable for covering large-scale aerial survey areas. In this embodiment, a quadcopter UAV is used. Quadcopter UAVs are characterized by their flexible flight; at the turns of the serpentine flight path, they can achieve smooth turns through the differential speed coordination between the four rotors, without requiring a large turning radius. This characteristic can significantly improve aerial survey efficiency and avoid redundant measurements or invalid flight paths caused by large-radius turns.

[0053] After planning the serpentine flight path, the optimal image acquisition angle for each shooting position is calculated by combining the trajectory planning plane and the target geometry. This optimal image acquisition angle is the direction of the camera optical axis that minimizes the deformation of the target surface in the image and provides the richest information. Specifically, at each shooting point, based on the normal vector of the trajectory planning plane corresponding to that point and the local geometry of the target surface, the direction in which the camera gimbal should point is calculated to minimize the angle between the camera optical axis and the normal of the target surface, thereby reducing image occlusion and tilt distortion. This optimal angle guides the control of the camera gimbal, ensuring that the acquired images effectively cover the target surface, providing high-quality image data for subsequent high-precision 3D modeling.

[0054] Based on the aforementioned flight path planning, the UAV automatically flies along the planned serpentine route, controlling the camera gimbal to acquire images at each shooting location according to the calculated optimal image acquisition perspective. This method supports multiple repeated flights, meaning that based on the same flight path planning, images from the same perspective can be acquired at different times (e.g., two phases of 3D models). Because the flight path and perspective remain consistent, the acquired images from multiple phases have a high degree of consistency, facilitating subsequent image matching and temporal deformation analysis, and serving the long-term monitoring of geological hazards such as landslides.

[0055] This embodiment improves the integrity of image coverage by adaptively extracting flight path planning surfaces that fit complex terrains such as landslides; it enhances aerial survey efficiency by optimizing serpentine flight paths in conjunction with quadcopter flight characteristics; it ensures image quality through optimal perspective calculation, facilitating subsequent 3D modeling and deformation monitoring; and it supports multiple repeated flights to obtain consistent perspective images, serving temporal deformation analysis, thus achieving refined and efficient close-up photography of complex terrains such as landslides and slopes.

[0056] In summary, the UAV close-up photography trajectory planning method for landslide monitoring proposed in this embodiment can adaptively extract trajectory planning surfaces that conform to complex terrains such as landslides, improve image coverage integrity and resolution consistency, and optimize serpentine flight paths and optimal shooting angles. The generated high-quality trajectory planning results can support multiple repeated flights and subsequent deformation monitoring.

[0057] Example 2 Figure 2 This is a structural diagram of a UAV close-up photography flight path planning system for landslide monitoring according to an embodiment of this application, as shown below. Figure 2 As shown, the system includes: The acquisition module 100 is used to acquire an initial three-dimensional model of the target area and extract three-dimensional parameters of multiple cross-sections of the target surface based on the initial three-dimensional model; The generation module 200 is used to adaptively generate a trajectory planning surface that fits the target surface based on the three-dimensional parameters of the multiple profiles. The control module 300 is used to plan a serpentine flight path on the flight path planning surface and control the UAV to fly along the serpentine flight path; When the serpentine flight path is executed by a quadcopter UAV, smooth turns are achieved at the turning points through the cooperation between the rotors, and the flight path planning results include turning control parameters.

[0058] The acquisition module 400 is used to calculate the best image acquisition angle for each shooting position based on the trajectory planning surface and target geometry, and generate trajectory planning results containing position and angle information.

[0059] The optimal image acquisition angle is the camera optical axis direction that minimizes the deformation of the target surface in the image and provides the most information. The gimbal attitude parameters include pitch angle, yaw angle, and roll angle.

[0060] In this embodiment of the disclosure, the acquisition module 100 is further configured to: Oblique photogrammetry is performed through conventional flight path missions to acquire multi-view images of the target area, and an initial 3D model is constructed based on the multi-view images. The initial three-dimensional model is subjected to profile fitting to extract the three-dimensional parameters of multiple profiles of the target surface.

[0061] The three-dimensional parameters include the normal vector of the profile and the boundary points.

[0062] The oblique photogrammetry has a forward overlap of 80% and a lateral overlap of 30%.

[0063] The number of the multiple cross-sections is 14. In this embodiment of the disclosure, the generation module 200 is further configured to: Based on the three-dimensional parameters of the multiple profiles, the target surface is decomposed into multiple basic units, including elevations and slopes; Generate the trajectory planning surface for each basic unit separately, and combine the trajectory planning surfaces of all basic units to form a complete trajectory planning surface.

[0064] In this embodiment of the disclosure, the acquisition module 400 is further configured to: Based on the same flight path planning results, the UAV is controlled to fly at the same waypoint position and gimbal attitude at different times to obtain the same image sequence from the same perspective, which is used for subsequent image matching and deformation analysis.

[0065] In summary, the UAV close-up photography trajectory planning system for landslide monitoring proposed in this embodiment can adaptively extract trajectory planning surfaces that conform to complex terrains such as landslides, improve image coverage integrity and resolution consistency, and optimize serpentine flight paths and optimal shooting angles. The generated high-quality trajectory planning results can support multiple repeated flights and subsequent deformation monitoring.

[0066] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in the specification, as well as the features of different embodiments or examples.

[0067] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0068] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for planning drone close-up photography flight paths for landslide monitoring, characterized in that, The method includes: Obtain an initial three-dimensional model of the target area, and extract three-dimensional parameters of multiple cross-sections of the target surface based on the initial three-dimensional model; Based on the three-dimensional parameters of the multiple profiles, an adaptive trajectory planning surface that fits the target surface is generated. Plan a serpentine flight path on the flight path planning surface and control the UAV to fly along the serpentine flight path; Based on the trajectory planning surface and target geometry, the optimal image acquisition angle for each shooting position is calculated, and a trajectory planning result containing position and angle information is generated.

2. The method as described in claim 1, characterized in that, The process of obtaining an initial 3D model of the target region and extracting 3D parameters of multiple cross-sections of the target surface based on the initial 3D model includes: Oblique photogrammetry is performed through conventional flight path missions to acquire multi-view images of the target area, and an initial 3D model is constructed based on the multi-view images. The initial three-dimensional model is subjected to profile fitting to extract the three-dimensional parameters of multiple profiles of the target surface.

3. The method as described in claim 2, characterized in that, The three-dimensional parameters include the normal vector of the profile and the boundary points.

4. The method as described in claim 3, characterized in that, The step of adaptively generating a trajectory planning surface that fits the target surface based on the three-dimensional parameters of the multiple profiles includes: Based on the three-dimensional parameters of the multiple profiles, the target surface is decomposed into multiple basic units, including elevations and slopes; Generate the trajectory planning surface for each basic unit separately, and combine the trajectory planning surfaces of all basic units to form a complete trajectory planning surface.

5. The method as described in claim 4, characterized in that, When the serpentine flight path is executed by a quadcopter UAV, smooth turns are achieved at the turning points through the coordination between the rotors, and the flight path planning results include turning control parameters.

6. The method as described in claim 1, characterized in that, The optimal image acquisition angle is the camera optical axis direction that minimizes the deformation of the target surface in the image and provides the richest information. The gimbal attitude parameters include pitch angle, yaw angle, and roll angle.

7. The method as described in claim 1, characterized in that, The method further includes: Based on the same flight path planning results, the UAV is controlled to fly at the same waypoint position and gimbal attitude at different times to obtain the same image sequence from the same perspective, which is used for subsequent image matching and deformation analysis.

8. The method as described in claim 2, characterized in that, The oblique photogrammetry has a forward overlap of 80% and a lateral overlap of 30%.

9. The method as described in claim 1, characterized in that, The number of the multiple cross-sections is 14.

10. A UAV close-up photography flight path planning system for landslide monitoring, characterized in that, The system includes: The acquisition module is used to acquire an initial three-dimensional model of the target area and extract three-dimensional parameters of multiple cross-sections of the target surface based on the initial three-dimensional model. The generation module is used to adaptively generate a trajectory planning surface that fits the target surface based on the three-dimensional parameters of the multiple profiles. The control module is used to plan a serpentine flight path on the flight path planning surface and control the UAV to fly along the serpentine flight path; The acquisition module is used to calculate the optimal image acquisition angle for each shooting position based on the trajectory planning surface and target geometry, and generate trajectory planning results containing position and angle information.

Images