Task-driven three-dimensional reconstruction air-ground multi-source data adaptive collaborative acquisition planning method and system

Abstract

The invention discloses a task-driven three-dimensional reconstruction air-ground multi-source data adaptive collaborative acquisition planning method and system, and belongs to the field of three-dimensional modeling. The method solves the problem that an existing three-dimensional model reconstruction method is difficult to consider the modeling coverage range and the precision of local details at the same time when facing a complex environment. According to the technical scheme, the method comprises the following steps of obtaining and preprocessing initial reference data of a target area, performing semantic segmentation and structural analysis on the preprocessed initial reference data, establishing a quantitative acquisition task model based on semantic partitioning and structural analysis results, planning an air-ground collaborative acquisition path for an unmanned aerial vehicle and ground equipment, and establishing an air-ground collaborative acquisition task model for the unmanned aerial vehicle and the ground equipment. The method comprises the following steps: acquiring a current acquired area, estimating the coverage integrity of the current acquired area in real time, completing online quality evaluation, and when an evaluation result shows that coverage vulnerabilities exist, triggering a dynamic re-planning mechanism to carry out supplementary acquisition until the coverage integrity of the current acquired area reaches a preset coverage integrity parameter value in an acquisition task model; the method is applied to three-dimensional modeling.

Description

Technical Field

[0001] This invention relates to the field of 3D modeling technology, specifically to a task-driven adaptive collaborative acquisition and planning method and system for multi-source aerial and ground data in 3D reconstruction. Background Technology

[0002] In 3D reconstruction of complex scenes, single data acquisition methods have significant limitations. For example, in areas with tall, dense buildings, narrow alleyways, complex building facades, areas obscured by vegetation, large substations, and train stations, relying solely on multi-rotor UAV oblique photography technology makes it difficult to effectively acquire high-precision geometric and textural information about the building's base, obscured areas, and interior spaces. This can lead to problems such as floating bodies or missing bases during 3D model reconstruction. Conversely, relying solely on terrestrial laser scanning technology is inefficient and cannot acquire data on rooftop structures.

[0003] At present, when reconstructing 3D models of complex areas, the core of using air-to-ground data fusion modeling is to integrate a large amount of data from aerial multi-rotor UAVs and ground equipment. Specifically, based on UAV oblique photogrammetry, data from the top and some sides is acquired by changing the heading direction, increasing the number of flights, adjusting the sensor orientation, and changing the altitude, in order to obtain more texture information. On the other hand, based on ground laser scanning technology, the ground measurement range is expanded by encrypting the acquisition path and using methods such as lifting mast support equipment scanning to ensure ground data coverage.

[0004] However, traditional air-ground collaborative data acquisition relies heavily on the experience of operators for path planning, lacking scientific theoretical guidance and automated tools, which easily leads to three major problems: First, there are blind spots in data acquisition, where some key areas are not covered by any sensors due to insufficient experience; second, there is data redundancy, with excessive acquisition in simple areas, wasting storage and computing resources; and third, it increases the difficulty of registration, as improper planning results in insufficient feature points in the overlapping areas of air and ground data, affecting the registration accuracy of subsequent data fusion.

[0005] Therefore, there is an urgent need for an intelligent pre-planning method to solve the problems of reliance on human experience, low efficiency, and loss of local model details or data redundancy caused by improper planning in the traditional 3D modeling data acquisition process. Summary of the Invention

[0006] To address the technical challenge of existing 3D model reconstruction methods in balancing the coverage of modeling with the accuracy of local details in complex environments, this invention proposes a task-driven adaptive collaborative acquisition and planning method and system for multi-source aerial and ground data in 3D reconstruction.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a task-driven adaptive collaborative acquisition and planning method for multi-source aerial and ground data in 3D reconstruction, comprising the following steps: Step S1: Obtain and preprocess the initial reference data for the target area; Step S2: Perform semantic segmentation and structural analysis on the preprocessed initial reference data, divide it into multiple different types of sub-regions and extract scene structural features; Step S3: Based on the semantic partitioning and structural analysis results, establish a quantitative acquisition task model for the entire target area scene and each sub-region according to the preset parameters; Step S4: Based on the data acquisition task model, plan an air-ground collaborative data acquisition path for the UAV and ground equipment; In step S5, while the drone and ground equipment are collecting scene data of the target area along the planned collection path, the coverage integrity of the currently collected area is estimated in real time. Step S6: Compare the coverage integrity of the currently collected area with the preset coverage integrity parameters in the collection task model to complete the online quality assessment. When the assessment result shows that there are coverage gaps, trigger the dynamic replanning mechanism to perform timely supplementary collection. Step S7: Repeat steps S4 to S6 until the coverage integrity of the currently collected area reaches the preset coverage integrity parameter value in the collection task model.

[0008] Furthermore, the initial reference data includes two-dimensional maps, low-precision digital elevation models, and existing coarse-grained three-dimensional models.

[0009] Furthermore, the types of sub-regions shown include at least open areas, tall building areas, dense low-rise areas, alleyway areas, and vegetation-covered areas.

[0010] Furthermore, the scene structural features include at least one of the following structural information: building outlines, main road networks, and key terrain feature lines.

[0011] Furthermore, the preset parameters in step S3 include geometric accuracy, texture resolution, coverage integrity parameters, and heterogeneous data overlap.

[0012] Furthermore, the air-ground collaborative data acquisition path in step S4 includes UAV path planning and ground equipment path planning. During the planning process, it is ensured that the overlapping areas of air and ground data have rich common features to avoid the same facade being observed from the ground and the air from the same perspective.

[0013] Furthermore, the UAV path planning includes generating regular flight routes that ensure overlap for open areas and tall building rooftops; planning hovering points above entrances or courtyards for dense low-rise areas and narrow alleys, and controlling the gimbal to take pictures at a non-vertical specific tilt angle; and using the A-Star search algorithm to plan flight routes to achieve intelligent obstacle avoidance.

[0014] Furthermore, the ground equipment path planning includes generating basic coverage paths along the main road network; generating bow-shaped or circular in-depth paths for alleys that cannot be covered by drones; generating circular paths for important individual buildings; and acquiring ground area scanning point cloud data through a handheld LiDAR scanner.

[0015] Furthermore, in step S6, the dynamic replanning mechanism is used to assign a drone or ground equipment to perform supplementary data collection tasks; wherein, the supplementary data collection task includes at least one of instructing the drone to fly to a designated location to perform oblique supplementary photography or instructing the ground equipment to detour to a designated location to perform close-range data collection, in order to make up for the coverage gap.

[0016] A system for implementing the task-driven adaptive collaborative acquisition and planning method for multi-source aerial and ground data in 3D reconstruction as described above includes: Data preprocessing module: used to acquire and process initial reference data; The scene analysis module is used to perform semantic segmentation and structural analysis on the preprocessed initial reference data, and to complete the scene partitioning and structure extraction of the target area. The task modeling module is used to construct a quantitative data collection task model based on scenario analysis results and preset parameters. The path planning module is used to plan air-ground collaborative data acquisition paths for both UAVs and ground equipment based on the data acquisition task model. The online quality assessment and feedback control module is used to evaluate coverage quality in real time during the data acquisition process and trigger a dynamic replanning mechanism to perform timely supplementary data acquisition when coverage gaps are detected.

[0017] The advantages of this invention over the prior art are as follows: 1. Achieved a shift from passive to proactive path planning: Compared to the traditional passive data collection mode that relies on experience, the method of this invention is based on a task modeling module to construct a hierarchical, closed-loop, and adaptive planning method for path planning, online quality assessment, and dynamic replanning. This transforms the collection of multi-source air and ground data from an isolated, experience-dependent process into a unified, task-demand-driven, and dynamically adjustable intelligent process that effectively improves the relevance and rationality of the data collection path.

[0018] 2. Achieved coordinated and unified optimization of air and ground equipment: By treating the UAV and ground acquisition equipment as a whole for systematic collaborative planning, this invention breaks the barrier of independent path planning between the two in the past, effectively avoiding the phenomenon of data silos, thus providing a reliable guarantee for high-precision registration and fusion of subsequent multi-source data.

[0019] 3. Effectively address uncertainties during the data collection process: Through online quality assessment and dynamic replanning methods, data collection coverage gaps can be detected in real time and timely supplementary data collection can be carried out, ensuring the quality and stability of the final collected data.

[0020] 4. Significantly improve operational efficiency: While ensuring the integrity and accuracy of the 3D model of the target area, the overall data acquisition efficiency is greatly improved and the operational cost is reduced through semantic partitioning differentiated planning, air-ground collaborative optimization and avoidance of invalid acquisition. Attached Figure Description

[0021] The present invention will be further described below with reference to the accompanying drawings: Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0022] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate relative orientations or positional relationships and are used only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0023] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0024] like Figures 1 to 2 As shown, this invention provides a task-driven adaptive collaborative acquisition planning method for multi-source aerial and ground data in 3D reconstruction, comprising the following steps: Step S1: Initial Reference Data Acquisition and Processing: Acquire and preprocess the initial reference data for the target area. The initial reference data includes at least one of the following: a 2D map, a low-precision digital elevation model, and an existing coarse-grained 3D model.

[0025] Step S2, Semantic Segmentation and Scene Structure Analysis: Semantic segmentation and structure analysis are performed on the preprocessed initial reference data to divide it into multiple different types of sub-regions and extract scene structure features. The types of sub-regions include at least open areas, tall building areas, dense low-rise areas, alleyway areas, and vegetation-covered areas.

[0026] Step S3: Establishment of Quantitative Acquisition Task Model: Based on the semantic partitioning and structural analysis results, establish a quantitative acquisition task model for the entire target area scene and each sub-region according to the preset parameters. The preset parameters include: Geometric precision: Different target modeling precisions are set according to different modeling objects (target areas); Texture resolution: The preset required texture ground sampling distance; Coverage integrity parameter: The preset number of viewing angles or sensors required to observe each surface; Heterogeneous data overlap: The required overlap rate between the target area scene data collected by the drone and the target area scene data collected by the ground equipment.

[0027] Step S4, Air-Ground Collaborative Acquisition Path Generation: Based on the acquisition task model, an air-ground collaborative acquisition path is planned for the UAV and ground equipment. The air-ground collaborative acquisition path includes UAV path planning and ground equipment path planning. During the planning process, it is ensured that the overlapping areas of air and ground data have rich common features to avoid the same facade being observed from the ground and the air from the same perspective.

[0028] The drone path planning includes generating regular flight paths that ensure overlap for open areas and tall building rooftops; planning hovering points above entrances or courtyards for dense low-rise areas and narrow alleys, and controlling the gimbal to take pictures at a non-vertical specific tilt angle; and using the A-Star search algorithm to plan flight routes to achieve intelligent obstacle avoidance.

[0029] Ground equipment path planning includes generating basic coverage paths along the main road network; generating bow-shaped or circular paths for alleys that drones cannot cover; generating circular paths for important individual buildings; and acquiring ground area scanning point cloud data through a handheld LiDAR scanner.

[0030] Step S5, Online Quality Assessment: During the process of the UAV and ground equipment collecting scene data of the target area along the planned acquisition path, the coverage integrity of the currently collected area is estimated in real time. Specifically, the coverage integrity of the currently collected area is estimated using the SLAM algorithm or the structure-of-motion regression algorithm.

[0031] Step S6, Dynamic Replanning: Compare the coverage integrity of the currently collected area with the preset coverage integrity parameters in the collection task model to complete the online quality assessment. When the assessment results show that there are coverage gaps, the dynamic replanning mechanism is triggered to perform timely supplementary collection.

[0032] Specifically, a dynamic replanning mechanism is used to assign drones or ground equipment to perform supplementary data collection tasks. These tasks include at least one of the following: instructing a drone to fly to a designated location for oblique imaging or instructing ground equipment to detour to a designated location for close-range data collection, in order to compensate for the coverage gaps.

[0033] Step S7: Complete the acquisition: Repeat steps S4 to S6 until the coverage integrity of the currently acquired area reaches the preset coverage integrity parameter value in the acquisition task model.

[0034] This invention provides a system for implementing the above-described task-driven adaptive collaborative acquisition and planning method for multi-source aerial and ground data in 3D reconstruction, comprising: Data preprocessing module: used to acquire and process initial reference data; The scene analysis module is used to perform semantic segmentation and structural analysis on the preprocessed initial reference data, and to complete the scene partitioning and structure extraction of the target area. The task modeling module is used to construct a quantitative data collection task model based on scenario analysis results and preset parameters. The path planning module is used to plan air-ground collaborative data acquisition paths for both UAVs and ground equipment based on the data acquisition task model. The online quality assessment and feedback control module is used to evaluate coverage quality in real time during the data acquisition process and trigger a dynamic replanning mechanism to perform timely supplementary data acquisition when coverage gaps are detected.

[0035] In one specific embodiment, the application scenario is the integrated 3D modeling of the interior and exterior of a railway station. This railway station features a large-span steel structure roof, a complex railway track network, and a vast indoor waiting hall. The modeling requirement is to achieve seamless integration of the indoor and outdoor scenes with high geometric accuracy. The collaborative data acquisition and planning method for the integrated 3D modeling of this railway station includes the following steps: Step S1: Initial reference data acquisition and processing: Acquire the BIM design model or CAD drawings of the railway station as initial reference data, and preprocess the initial reference data. Step S2, Semantic Segmentation and Scene Structure Analysis: The preprocessed initial reference data is semantically segmented and structurally analyzed into four key sub-regions: "External Plaza", "Steel Structure Roof", "Platform Track Area", and "Interior View of High-Rise Waiting Hall". Step S3: Establishing a Quantitative Acquisition Task Model: Set differentiated acquisition requirements for different zones, and establish a quantitative acquisition task model for the entire target area scene and each sub-region based on preset parameters. Geometric accuracy: Target geometric accuracy of steel structure roof truss ≤ 2 cm; Target geometric accuracy of track area ≤ 1 cm; Texture resolution: The ground sampling distance for the texture of the store facade in the indoor commercial street should be ≤1 mm / pixel.

[0036] Step S4, Air-Ground Collaborative Acquisition Path Generation: Based on the acquisition task model, an air-ground collaborative acquisition path is planned for the UAV and ground equipment. The air-ground collaborative acquisition path includes UAV path planning and ground equipment path planning. During the planning process, it is ensured that the overlapping areas of air and ground data have rich common features to avoid the same facade being observed from the ground and the air from the same perspective.

[0037] Drone path planning: Plan conventional flight routes over outdoor plazas and steel structure roofs, with a focus on ensuring multi-angle, high-overlap image coverage of complex roof truss structures.

[0038] Ground equipment path planning: Plan a main indoor data collection route for the backpack-type mobile laser scanning system, running through the waiting hall, commercial area, and ticket gates. For important platforms and column-free canopy areas, plan a zigzag path to ensure comprehensive scanning of the platforms, tracks, and canopy support structures.

[0039] Collaborative planning: Extends the data collection time of ground equipment in the main entrance area of ​​the station to obtain rich data of the indoor-outdoor transition zone, providing sufficient common features for high-precision registration of subsequent air-ground data.

[0040] Step S5, Online Quality Assessment: When the ground backpack equipment is operating in the waiting hall, due to the obstruction of a large number of passengers, the online quality assessment and feedback control module detects in real time through the SLAM algorithm that there are significant holes in the point cloud data of a shop facade, and the coverage integrity does not meet the requirements of the data acquisition task model.

[0041] Step S6, Dynamic Replanning: The online quality assessment and feedback control module triggers the dynamic replanning mechanism. Considering the indoor environment and safety, it sends a prompt to the ground equipment operator's terminal: "Insufficient coverage in area X ahead, please perform local rescanning." Following the prompt, the operator performs a small back-and-forth movement in place to complete the rescanning task.

[0042] Step S7: Complete the data collection: Repeat steps S4 to S6 until the coverage integrity of the currently collected area reaches the preset coverage integrity parameter value in the data collection task model. The online quality assessment and feedback control module will then indicate that the coverage of the area has reached the preset standard.

[0043] This embodiment successfully constructed a high-precision 3D model of a railway station, integrating macroscopic structure and microscopic details, and achieving seamless indoor and outdoor integration, through precise task modeling based on BIM design models or CAD drawings as spatial benchmarks, targeted collaborative planning for indoor-outdoor transition areas, and dynamic acquisition and control based on real-time quality assessment. This embodiment specifically verifies that the method and system of this invention can effectively ensure the integrity and quality of data acquisition even in dynamic and complex environments such as those with large passenger flows.

[0044] In another specific embodiment, the application scenario involves 3D modeling of a substation to provide a high-precision data foundation for the establishment of an automated substation inspection system. This substation environment presents challenges such as strictly defined safety restricted areas, complex internal equipment and architecture, and strong electromagnetic interference, severely impacting the positioning of satellite-dependent devices like GPS. The collaborative data acquisition and planning method includes the following steps: Step S1: Initial reference data acquisition and processing: Acquire the two-dimensional design drawing, wiring diagram or old model of the substation as initial reference data, and preprocess the initial reference data; Step S2, Semantic Segmentation and Scene Structure Analysis: Based on the electrical safety distance, the preprocessed initial reference data is semantically segmented as follows: "Absolute no-fly zone": such as above a live busbar; "Restricted Flight Zone": Such as the airspace above electrical equipment areas, where flight altitude and path are limited; "Safe working area": ​​such as the area above roads and lawns; At the same time, the structural features of the scene are extracted, such as large equipment like transformers and circuit breakers, fine structures at high altitudes like insulator strings and lightning rods, and underground facilities like cable trench covers.

[0045] Step S3: Establishing a Quantitative Acquisition Task Model: Set differentiated acquisition requirements for different zones, and establish a quantitative acquisition task model for the entire target area scene and each sub-region based on preset parameters. Geometric accuracy: The geometric accuracy of key connection points such as transformer bushings and equipment clamps is ≤3mm; the geometric accuracy of the outer shell and structure of large equipment is ≤1cm; the geometric accuracy of the ground and roads is ≤5cm.

[0046] Step S4, Air-Ground Collaborative Acquisition Path Generation: Based on the acquisition task model, an air-ground collaborative acquisition path is planned for the UAV and ground equipment. The air-ground collaborative acquisition path includes UAV path planning and ground equipment path planning. During the planning process, it is ensured that the overlapping areas of air and ground data have rich common features to avoid the same facade being observed from the ground and the air from the same perspective.

[0047] Drone path planning: Within the "safe operating area," a conventional flight path is planned to obtain an overall overview. For large equipment such as transformers and circuit breakers, a circular, close-proximity data acquisition path is planned to ensure that three-dimensional information is obtained from multiple angles. For insulator strings, an "orthogonal photography" path is planned, i.e., data is acquired from both the front and sides to ensure that their delicate structure is not obstructed.

[0048] Ground equipment path planning: Plan basic data acquisition paths along the internal roads of the station to collect cloud data from the bottom and inside of electrical equipment. For blind spots under transformers, at the base of the structure, and other areas where the drone has limited visibility, plan a figure-eight movement path to compensate for the lack of detail.

[0049] Collaborative planning: In key areas such as transformers, drones and ground equipment are planned to collect a large amount of overlapping data. The high-precision point cloud of the ground equipment is used to perform joint adjustment of drone imagery to improve the overall geometric accuracy.

[0050] Step S5, Online Quality Assessment: When the pose estimation of the UAV drifts due to strong electrical interference during flight, the online quality assessment and feedback control module uses the SLAM algorithm to process the generated sparse point cloud in real time and match it with the pre-loaded coarse-grained reference model to achieve visual-assisted positioning and deviation correction, ensuring flight safety and data space consistency.

[0051] After the drone photographs the insulator string, the online quality assessment and feedback control module estimates the coverage integrity of the currently acquired area using a motion recovery structure algorithm. The coverage integrity of the currently acquired area is compared with the preset coverage integrity parameters in the acquisition task model to complete the online quality assessment. The assessment results show that some of the insulator skirts are blurred due to glare.

[0052] Step S6, Dynamic Replanning: The online quality assessment and feedback control module triggers the dynamic replanning mechanism, issuing a fine-tuning command to the UAV. The command instructs the UAV to move 1 meter to the left from its current position and take 3 additional photos of the insulator string from another angle to complete the supplementary data collection task. This effectively avoids reflective areas and ensures the quality of the modeling data for the target area.

[0053] Step S7: Complete the data collection: Repeat steps S4 to S6 until the coverage integrity of the currently collected area reaches the preset coverage integrity parameter value in the data collection task model. The online quality assessment and feedback control module will then indicate that the coverage of the area has reached the preset standard.

[0054] Through the aforementioned intelligent partitioning based on safety rules, differentiated task modeling, air-ground collaborative path planning, and dynamic replanning control that integrates anti-interference positioning and real-time quality feedback, the system of this invention has successfully achieved safe, automated, and high-precision 3D data acquisition in complex substation environments with strong electromagnetic interference, significantly improving the refinement of 3D modeling and overall operational efficiency.

[0055] This invention's method and system upgrade the chaotic data acquisition process into an intelligent, collaborative, and automated digital factory, providing a reliable and efficient data foundation for building high-precision realistic 3D models. Its value lies not only in generating static models but also in empowering dynamic management and intelligent decision-making throughout the entire lifecycle. In the field of smart cities, it can efficiently complete comprehensive modeling of historical blocks and modern industrial parks, from rooftops to alleyway facades, providing a precise foundation for planning, protection, and management. In the energy and industrial sectors, it can replace manual labor in high-risk and complex environments such as substations and mines, enabling safe and precise digital inspections, achieving millimeter-level deformation monitoring and safety early warning for equipment, directly serving intelligent operation and maintenance and safe production. Its core prospect lies in transforming expensive, inefficient, and highly specialized 3D reconstruction work into a standardized service that can be scalably replicated, adaptively optimized, and guarantees result quality, significantly reducing the cost and cycle time of high-quality realistic 3D modeling, thereby injecting core driving force into the digital transformation of urban governance, industrial operation and maintenance, and cultural heritage protection.

[0056] Regarding the specific structure of this invention, it should be noted that the connection relationships between the various component modules used in this invention are definite and achievable. Except as specifically described in the embodiments, their specific connection relationships can bring about corresponding technical effects and solve the technical problems proposed by this invention without relying on the execution of corresponding software programs. The models of the components, modules, and specific components appearing in this invention, the connection methods between them, and the conventional usage methods and expected technical effects brought about by the above technical features, unless specifically described, are all publicly disclosed content in patents, journal articles, technical manuals, technical dictionaries, and textbooks that can be obtained by those skilled in the art before the application date, or belong to conventional technology, common knowledge, and other existing technologies in this field. There is no need to elaborate, which makes the technical solution provided in this case clear, complete, and achievable, and can reproduce or obtain corresponding physical products based on this technical means.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A task-driven adaptive collaborative acquisition planning method for multi-source aerial and ground data in 3D reconstruction, characterized in that: Includes the following steps: Step S1: Obtain and preprocess the initial reference data for the target area; Step S2: Perform semantic segmentation and structural analysis on the preprocessed initial reference data, divide it into multiple different types of sub-regions and extract scene structural features; Step S3: Based on the semantic partitioning and structural analysis results, establish a quantitative acquisition task model for the entire target area scene and each sub-region according to the preset parameters; Step S4: Based on the data acquisition task model, plan an air-ground collaborative data acquisition path for the UAV and ground equipment; In step S5, while the drone and ground equipment are collecting scene data of the target area along the planned collection path, the coverage integrity of the currently collected area is estimated in real time. Step S6: Compare the coverage integrity of the currently collected area with the preset coverage integrity parameters in the collection task model to complete the online quality assessment. When the assessment result shows that there are coverage gaps, trigger the dynamic replanning mechanism to perform timely supplementary collection. Step S7: Repeat steps S4 to S6 until the coverage integrity of the currently collected area reaches the preset coverage integrity parameter value in the collection task model.

2. The task-driven adaptive collaborative acquisition and planning method for multi-source aerial and ground data for 3D reconstruction according to claim 1, characterized in that: The initial reference data includes two-dimensional maps, low-precision digital elevation models, and existing coarse-grained three-dimensional models.

3. The task-driven adaptive collaborative acquisition and planning method for multi-source aerial and ground data in 3D reconstruction according to claim 1, characterized in that: The types of sub-regions shown include at least open areas, tall building areas, dense low-rise areas, alleyway areas, and vegetation-covered areas.

4. The task-driven adaptive collaborative acquisition and planning method for multi-source aerial and ground data in 3D reconstruction according to claim 1, characterized in that: The scene structural features include at least one of the following structural information: building outlines, main road networks, and key terrain feature lines.

5. The task-driven adaptive collaborative acquisition planning method for multi-source aerial and ground data in 3D reconstruction according to claim 1, characterized in that: The preset parameters in step S3 include geometric accuracy, texture resolution, coverage integrity parameters, and heterogeneous data overlap.

6. The task-driven adaptive collaborative acquisition and planning method for multi-source aerial and ground data for 3D reconstruction according to claim 1, characterized in that: In step S4, the air-ground collaborative data acquisition path includes UAV path planning and ground equipment path planning. During the planning process, it is ensured that the overlapping areas of air and ground data have rich common features to avoid the same facade being observed from the ground and the air from the same perspective.

7. The task-driven adaptive collaborative acquisition and planning method for multi-source aerial and ground data for 3D reconstruction according to claim 6, characterized in that: The UAV path planning includes generating regular flight paths that ensure overlap for open areas and tall building rooftops; planning hovering points above entrances or courtyards for dense low-rise areas and narrow alleys, and controlling the gimbal to take pictures at a non-vertical specific tilt angle; and using the A-Star search algorithm to plan flight routes to achieve intelligent obstacle avoidance.

8. The task-driven adaptive collaborative acquisition planning method for multi-source aerial and ground data in 3D reconstruction according to claim 6, characterized in that: The ground equipment path planning includes generating basic coverage paths along the main road network; generating bow-shaped or circular paths for alleys that cannot be covered by drones; generating circular paths for important individual buildings; and acquiring ground area scanning point cloud data through a handheld LiDAR scanner.

9. The task-driven adaptive collaborative acquisition and planning method for multi-source aerial and ground data in 3D reconstruction according to claim 1, characterized in that: In step S6, the dynamic replanning mechanism assigns a drone or ground equipment to perform supplementary data collection tasks; wherein, the supplementary data collection task includes at least one of instructing the drone to fly to a designated location for oblique supplementary photography or instructing the ground equipment to detour to a designated location for close-range data collection, in order to make up for the coverage gap.

10. A system for implementing the task-driven adaptive collaborative acquisition and planning method for multi-source aerial and ground data in 3D reconstruction as described in any one of claims 1-9, characterized in that: include: Data preprocessing module: used to acquire and process initial reference data; The scene analysis module is used to perform semantic segmentation and structural analysis on the preprocessed initial reference data, and to complete the scene partitioning and structure extraction of the target area. The task modeling module is used to construct a quantitative data collection task model based on scenario analysis results and preset parameters. The path planning module is used to plan air-ground collaborative data acquisition paths for both UAVs and ground equipment based on the data acquisition task model. The online quality assessment and feedback control module is used to evaluate coverage quality in real time during the data acquisition process and trigger a dynamic replanning mechanism to perform timely supplementary data acquisition when coverage gaps are detected.

Images