Airborne laser radar-based civil air defense facility space surveying and mapping system and method

Abstract

The invention relates to the technical field of laser point cloud data processing and space surveying and mapping, and discloses a civil air defense facility space surveying and mapping system and method based on an airborne laser radar, and the system comprises a data collection module which is used for obtaining the original point cloud data and space pose information of a civil air defense facility region through an unmanned plane; the data processing module is used for receiving and processing the point cloud data and carrying out classification identification and three-dimensional modeling on the civil air defense facilities; and the precision evaluation module is used for carrying out precision quantitative evaluation on the civil air defense facility modeling result. By fusing vegetation index features and point cloud geometric features of a multispectral image, performing terrain curved surface segmentation and vegetation point cloud elimination, and combining with a facility preset geometric feature combination, precise separation and identification of civil air defense facilities and other ground features in a mixed region are realized.

Description

Technical Field

[0001] This invention relates to the field of laser point cloud data processing and spatial mapping technology, specifically to a spatial mapping system and method for civil defense facilities based on airborne lidar. Background Technology

[0002] In recent years, laser point cloud technology has been widely used in the field of spatial mapping. In particular, airborne lidar systems have become an important tool for topographic mapping due to their high efficiency and high precision. With the acceleration of urbanization, the demand for mapping underground spaces such as civil defense facilities is increasing. Traditional mapping methods are difficult to meet the requirements in terms of accuracy and efficiency in complex terrain and mixed ground features. Airborne lidar technology can quickly obtain the three-dimensional coordinate information of the ground surface and ground features by actively emitting laser pulses and receiving reflected signals, providing a new solution for spatial mapping. However, traditional methods often rely on manual design to perform multiple rounds of filtering and clustering operations on point clouds.

[0003] However, in current technology, manual rules are used to perform multiple rounds of filtering and clustering operations on point clouds. The changes in vegetation and facilities in the scene lead to low efficiency in point cloud data processing and insufficient feature separation accuracy in complex terrain and mixed ground features. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a spatial mapping system and method for civil defense facilities based on airborne lidar, which solves the problems of low efficiency in point cloud data processing and insufficient feature separation accuracy in complex terrain and mixed ground features.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a spatial mapping system for civil defense facilities based on airborne lidar, comprising: The data acquisition module uses drones to acquire raw point cloud data and spatial pose information of the civil defense facility area; The data processing module receives and processes the point cloud data, and performs classification, identification, and 3D modeling of civil defense facilities. The accuracy assessment module is used to quantitatively assess the accuracy of the modeling results of civil defense facilities.

[0006] By adopting the above technical solution, the original point cloud data and spatial pose information of the civil defense facility area are collected simultaneously by the UAV platform. The data processing module integrates multispectral vegetation index and point cloud geometric features to achieve terrain surface segmentation, vegetation removal and accurate identification of the pre-set geometric features of the facilities. The semantic three-dimensional model is automatically output. The accuracy evaluation module combines deformation heat map and vegetation exposure rate classification model to complete the dual quantitative evaluation of facility structure stability and concealment, thereby improving the efficiency and accuracy of terrain and feature separation in complex mixed areas.

[0007] Preferably, the data acquisition module includes: The lidar unit is installed on the bottom of the drone and scans the surface of civil defense facilities at a preset scanning frequency; The inertial measurement unit is physically fixed to the lidar unit to measure the attitude data of the UAV's angular velocity and acceleration. The satellite positioning unit obtains the spatial coordinates of the civil defense facility area through carrier phase differential positioning technology of satellite base stations; The point cloud generation unit integrates attitude data and spatial coordinates to generate a three-dimensional point cloud of civil defense facilities.

[0008] Preferably, the point cloud generation unit performs timestamp synchronization and coordinate transformation, including: Compensation for UAV attitude jitter is based on data collected by inertial measurement units; Convert laser scanning data to a geographic coordinate system.

[0009] Preferably, the sampling frequency of the inertial measurement unit is not lower than the scanning frequency of the lidar unit.

[0010] Preferably, the data acquisition module further includes a multi-machine collaboration unit, which controls the second UAV equipped with a multispectral camera to synchronously acquire images of the civil defense facility area, aligns the laser point cloud of the first UAV with the multispectral image of the second UAV by adding timestamps, and generates a spatial transformation matrix of the civil defense facility area based on SIFT feature matching of the point cloud intensity map and the multispectral image.

[0011] Preferably, the data processing module includes: The point cloud filtering unit filters noise from the three-dimensional point cloud of civil defense facilities. The 3D feature extraction unit automatically extracts the geometric features of civil defense facilities directly from the filtered point cloud using a 3D convolutional neural network; the multi-source fusion classification unit fuses the spectral features of multispectral images with the geometric features of the point cloud to output semantic labels for civil defense facilities.

[0012] Preferably, the three-dimensional feature extraction unit adopts a sparse convolutional network architecture, including: The feature encoding layer extracts local structural features of the point cloud through sparse convolutional kernels; Multi-scale pyramid layers, fusing feature maps from different receptive fields to capture global topological relationships; The multi-source fusion classification unit performs: Map multispectral image pixels to point cloud spatial coordinates; Spectral features and point cloud features are fused using a graph attention mechanism; Output three types of semantic labels and confidence levels for ventilation shafts, evacuation routes, and safety doors.

[0013] Preferably, the accuracy evaluation module includes: The time-series comparison unit uses an iterative nearest-point algorithm to perform three-dimensional deformation analysis on multiple scans of point cloud data of the same civil defense facility. The concealment assessment unit simulates the visibility risk level based on the density of the surrounding vegetation point cloud.

[0014] A method for spatial mapping of civil defense facilities based on airborne lidar, the method comprising the following steps: S1. Simultaneously collect laser scanning data, inertial measurement data, and satellite positioning data of the civil defense facility area using the first UAV; S2. Simultaneously collect multispectral image data of civil defense facilities area using a second drone; S3. Fuse attitude data and spatial coordinates to generate a 3D point cloud; S4. Perform terrain and feature separation processing on the point cloud; S5. Identify civil defense facilities and construct semantic 3D models; S6. Assess the accuracy of the model and the level of hidden risks.

[0015] S1 specifically includes: Real-time monitoring of point cloud density in civil defense facility areas and generation of facility area density heat maps; When the density of the target area is lower than a set threshold, the drone's flight altitude or laser scanning frequency is dynamically adjusted. Supplementary scanning is triggered by transmitting control commands via the 5G private network.

[0016] This invention provides a spatial mapping system and method for civil defense facilities based on airborne lidar. It has the following beneficial effects: 1. In this invention, by fusing vegetation index features and point cloud geometric features from multispectral images, terrain surface segmentation and vegetation point cloud removal are performed, and then combined with the pre-set geometric feature combination of facilities, accurate separation and identification of civil defense facilities and other land features in mixed areas is achieved.

[0017] 2. In this invention, the iterative nearest point algorithm is used to register historical and real-time point clouds of civil defense facilities to generate deformation heat maps. At the same time, a risk classification model of exposure rate is established based on vegetation point cloud density, which fills the gap in the joint assessment of structural stability and concealment in large-scale surveying and mapping, and ensures accurate quantitative management of facility safety protection effectiveness.

[0018] 3. In this invention, by using sparse convolutional neural networks and end-to-end processing with real-time feedback, the spatial topological features of the facility are automatically learned to replace manual feature design. Combined with point cloud density heatmaps to dynamically adjust flight parameters and 5G command-based scanning, the processing efficiency is improved while ensuring the modeling accuracy of ventilation shafts / channels. Attached Figure Description

[0019] Figure 1 This is an architecture diagram of a civil defense facility spatial mapping system based on airborne lidar according to the present invention; Figure 2 This is a flowchart of a spatial mapping method for civil defense facilities based on airborne lidar according to the present invention. Detailed Implementation

[0020] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see the appendix Figure 1 This invention provides a spatial mapping system for civil defense facilities based on airborne lidar, comprising: The data acquisition module uses drones to acquire raw point cloud data and spatial pose information of the civil defense facility area; The data processing module receives and processes point cloud data, and performs classification, identification, and 3D modeling of civil defense facilities. The accuracy assessment module is used to quantitatively assess the accuracy of the modeling results of civil defense facilities.

[0022] Specifically, the data acquisition module uses a drone equipped with a lidar to scan the surface of civil defense facilities, and simultaneously integrates GNSS carrier phase differential positioning and IMU attitude compensation data to generate a georeferenced 3D point cloud with timestamps. Based on laser penetration of vegetation technology and a dynamic point cloud density feedback mechanism, it ensures complete coverage of concealed structures.

[0023] The data processing module automatically learns the geometric features of point clouds based on a 3D convolutional neural network, integrates the spectral reflectance of multispectral images, and outputs semantic BIM models of ventilation shafts, passages, and protective doors, thereby improving the accuracy of vegetation-covered area identification.

[0024] The accuracy assessment module uses a point cloud registration algorithm to analyze facility deformation trends and combines the vegetation point cloud density distribution to assess concealment levels, forming a concealment quantification model and facility status monitoring mechanism under safety protection requirements.

[0025] The data acquisition module includes: The lidar unit is installed on the bottom of the drone and scans the surface of civil defense facilities at a preset scanning frequency; The inertial measurement unit is physically fixed to the lidar unit to measure the attitude data of the UAV's angular velocity and acceleration. The satellite positioning unit obtains the spatial coordinates of the civil defense facility area through carrier phase differential positioning technology of satellite base stations; The point cloud generation unit integrates attitude data and spatial coordinates to generate a three-dimensional point cloud of civil defense facilities.

[0026] The point cloud generation unit performs timestamp synchronization and coordinate transformation, including: Compensation for UAV attitude jitter is based on data collected by inertial measurement units; Convert laser scanning data to a geographic coordinate system.

[0027] Specifically, the lidar unit emits laser pulses at a preset scanning frequency through a laser emitter installed on the bottom of the drone and receives the reflected signals from the surface of the civil defense facility to obtain raw ranging data, thereby achieving non-contact measurement of the geometric structure of the surface of the civil defense facility and generating a raw point cloud containing distance and intensity information. The inertial measurement unit is physically connected to the lidar unit and uses built-in sensors to measure the angular velocity and acceleration data of the UAV, outputting real-time attitude change parameters of the UAV and providing a motion reference for point cloud position compensation. The satellite positioning unit receives carrier phase signals transmitted by satellite base stations, uses differential positioning technology to calculate the spatial coordinates of the civil defense facility area, provides geographic positioning information, and establishes a spatial reference frame for point clouds. The point cloud generation unit achieves time synchronization by triggering laser scanning and attitude data acquisition through satellite positioning second pulses. It uses the pitch angle, roll angle and yaw angle output by the inertial measurement unit to construct rotational relationships to compensate for the attitude deviation of the UAV. Combined with the carrier phase differential coordinates provided by the satellite positioning unit, the laser ranging value is converted to the geographic coordinate system through the coordinate system transformation matrix. Finally, the point cloud model of civil defense facilities with three-dimensional position, reflection intensity and time stamp is generated.

[0028] The sampling frequency of the inertial measurement unit is not lower than the scanning frequency of the lidar unit.

[0029] Specifically, the sampling frequency of the inertial measurement unit is no less than that of the lidar scanning frequency, ensuring that the corresponding instantaneous attitude data of the UAV can be acquired every time a laser pulse is emitted. The point cloud position shift caused by the UAV flight vibration is compensated by real-time attitude angle integral calculation, eliminating the point cloud distortion caused by high-frequency attitude changes during the scanning process, and improving the geometric accuracy and structural integrity of the three-dimensional point cloud reconstruction of the surface of civil defense facilities.

[0030] The data acquisition module also includes a multi-machine collaboration unit, which controls a second UAV equipped with a multispectral camera to simultaneously acquire images of the civil defense facility area. By adding timestamps, the laser point cloud of the first UAV is aligned with the multispectral image of the second UAV, and a spatial transformation matrix of the civil defense facility area is generated based on SIFT feature matching of the point cloud intensity map and the multispectral image.

[0031] Specifically, the multi-drone collaborative unit controls a second UAV equipped with a multispectral camera to simultaneously collect image data of the civil defense facility area. It uses a unified timestamp to align the laser point cloud of the first UAV with the multispectral image sequence of the second UAV, and generates a spatial transformation matrix by matching the key feature points of the point cloud intensity map and the multispectral image based on the scale-invariant feature transformation algorithm. This achieves precise spatiotemporal registration of cross-platform data and significantly improves the accuracy and completeness of subsequent multi-source data fusion for civil defense facilities.

[0032] The data processing module includes: The point cloud filtering unit filters noise from the three-dimensional point cloud of civil defense facilities. The 3D feature extraction unit automatically extracts the geometric features of civil defense facilities directly from the filtered point cloud using a 3D convolutional neural network; the multi-source fusion classification unit fuses the spectral features of multispectral images with the geometric features of the point cloud to output semantic labels for civil defense facilities.

[0033] The 3D feature extraction unit employs a sparse convolutional network architecture, including: The feature encoding layer extracts local structural features of the point cloud through sparse convolutional kernels; Multi-scale pyramid layers, fusing feature maps from different receptive fields to capture global topological relationships; Multi-source fusion classification unit execution: Map multispectral image pixels to point cloud spatial coordinates; Spectral features and point cloud features are fused using a graph attention mechanism; Output three types of semantic labels and confidence levels for ventilation shafts, evacuation routes, and safety doors.

[0034] Specifically, the point cloud filtering unit performs noise filtering on the three-dimensional point cloud of civil defense facilities based on the outlier removal algorithm of neighborhood point density statistics. By calculating the point density distribution within the neighborhood radius of each point, discrete noise points with a density lower than a set threshold are removed, while retaining continuous point cloud data on the surface of civil defense facilities.

[0035] Among them, the outlier detection condition is: k i <k min k i Let k be the number of points in the neighborhood of the i-th point in the point cloud. min The preset minimum number of neighborhood points threshold; Neighborhood range: starting from point p iA spherical space centered at a radius r (in meters); Point density calculation I(·) is an indicator function (1 if within the neighborhood, 0 otherwise), p j For the coordinates of other points.

[0036] The 3D feature extraction unit automatically extracts geometric features directly from the filtered point cloud by adopting a sparse convolutional network architecture. It extracts local structural features (such as the circular outline of a ventilation shaft) through the sparse convolutional kernel of the feature encoding layer, and then captures global topological relationships (such as the longitudinal continuity of the channel) by fusing different receptive field feature maps through a multi-scale pyramid layer, thereby realizing the hierarchical expression of the geometric features of civil defense facilities.

[0037] The multi-source fusion classification unit maps multispectral image pixels to point cloud spatial coordinates, and performs weighted fusion of spectral features and point cloud geometric features through graph attention mechanism. Based on the fused features, it outputs semantic labels and confidence scores for three types of civil defense facilities: ventilation shafts, evacuation routes, and protective doors.

[0038] The 3D feature extraction unit adopts a sparse convolutional network architecture, including: The feature encoding layer extracts local structural features of the point cloud through sparse convolutional kernels; Multi-scale pyramid layers, fusing feature maps from different receptive fields to capture global topological relationships; Multi-source fusion classification unit execution: Map multispectral image pixels to point cloud spatial coordinates; Spectral features and point cloud features are fused using a graph attention mechanism; Output three types of semantic labels and confidence levels for ventilation shafts, evacuation routes, and safety doors.

[0039] Specifically, in the 3D feature extraction unit, the feature encoding layer slides and calculates on the non-empty voxels of the point cloud through a 3D sparse convolution kernel to extract local structural features of civil defense facilities (such as the curvature of ventilation shaft edges and abrupt changes in the normal direction of passage walls), and outputs local feature maps. The multi-scale pyramid layer adopts a hierarchical dilated convolution architecture, constructs receptive fields of different scales through differences in dilation rate, and extracts point-level local features, component-level mesoscopic structure and facility-level global topology respectively, thereby fusing multi-scale feature maps to generate spatial geometric codes for civil defense facilities. The multi-source fusion classification unit rigidly registers the geographic coordinates of each pixel in the multispectral image with the spatial location of the point cloud, achieving a one-to-one mapping from pixel to point cloud. Furthermore, it constructs a point cloud adjacency graph through a graph attention mechanism, using spectral features as node attributes and geometric features as edge weights, and iteratively updates the feature weights to fuse spectral and geometric information.

[0040] The accuracy assessment module includes: The time-series comparison unit uses an iterative nearest-point algorithm to perform three-dimensional deformation analysis on multiple scans of point cloud data of the same civil defense facility. The concealment assessment unit simulates the visibility risk level based on the density of the surrounding vegetation point cloud.

[0041] Specifically, the time-series comparison unit performs three-dimensional registration of multiple scanned point cloud data of the same civil defense facility through the iterative nearest point algorithm, obtains historical point clouds and current point clouds, minimizes the distance error of corresponding points by iteratively optimizing the spatial transformation matrix, and calculates the displacement of key structural points based on the final convergence residual to generate a three-dimensional deformation heat map, thereby realizing the quantitative analysis of the structural stability of civil defense facilities. Iterative optimization includes: R is a rotation matrix, a 3×3 orthogonal matrix, representing the rotational relationship between point clouds; t is the translation vector, a three-dimensional vector (tx, ty, tz), representing the amount of translation between point clouds; P i The coordinates are historical point cloud coordinates, a three-dimensional vector (x, y, z), derived from a facility point cloud from a past scan; q i Given the coordinates of the current point in the point cloud, a 3D vector (x, y, z), P is matched using nearest neighbor search. i get; n is the number of valid point pairs, which is an integer and n≥3; That is, input historical point cloud P i ={p1,p2,...,pm} and the current point cloud Q ={q1,q2,...,qk}; Nearest neighbor search: For each pi, find the nearest point qi in Q. Solve for the rigid transformation, calculate the optimal R,t minimization coordinate transformation for P. i Calculate using transformation deformation, Δdi=||pi-qi|| The iteration terminates when the average rate of change of Δdi is less than 0.1% or when the maximum number of iterations is reached; The concealment assessment unit sets up multi-level spherical neighborhoods centered on the spatial coordinates of civil defense facilities, counts the proportion of vegetation point clouds (number of vegetation points / total number of points) within each neighborhood, and divides the visibility risk into three levels: low (exposure rate ≤ 30%), medium (30% < exposure rate ≤ 60%), and high (exposure rate > 60%) based on the exposure rate threshold, and outputs a risk spatial distribution map.

[0042] Please see the appendix Figure 2 A method for spatial mapping of civil defense facilities based on airborne lidar includes the following steps: S1. Simultaneously collect laser scanning data, inertial measurement data, and satellite positioning data of the civil defense facility area using the first UAV; S2. Simultaneously collect multispectral image data of civil defense facilities area using a second drone; S3. Fuse attitude data and spatial coordinates to generate a 3D point cloud; S4. Perform terrain and feature separation processing on the point cloud; S5. Identify civil defense facilities and construct semantic 3D models; S6. Assess the accuracy of the model and the level of hidden risks.

[0043] Specifically, the first UAV, equipped with a lidar, simultaneously collects spatial pose and laser scanning data of the civil defense facility area, while the second UAV acquires multispectral image data. The attitude and coordinate information are fused to generate a 3D point cloud. Based on the point cloud, the terrain and features are separated to extract the civil defense facility targets. A deep learning model is used to identify three types of facilities: ventilation shafts, evacuation routes, and protective doors, and a semantic 3D model is constructed. Finally, deformation analysis and vegetation shading rate assessment are performed by combining historical point clouds, and a report on the health status and concealment risk level of the civil defense facilities is output.

[0044] S1 specifically includes: Real-time monitoring of point cloud density in civil defense facility areas and generation of facility area density heat maps; When the density of the target area is lower than a set threshold, the drone's flight altitude or laser scanning frequency is dynamically adjusted. Supplementary scanning is triggered by transmitting control commands via the 5G private network.

[0045] Specifically, during data collection, the system monitors the spatial coverage integrity by generating a point cloud density heat map of the civil defense facility area in real time. When the density of a key area is found to be lower than a set threshold, the system automatically adjusts the drone's flight altitude or laser scanning frequency and triggers supplementary scanning by transmitting control commands in real time via a 5G private network, thereby achieving the integrity of densely vegetated areas or complex structural parts.

[0046] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A spatial mapping system for civil defense facilities based on airborne lidar, characterized in that, include: The data acquisition module uses drones to acquire raw point cloud data and spatial pose information of the civil defense facility area; The data processing module receives and processes the point cloud data, and performs classification, identification, and 3D modeling of civil defense facilities. The accuracy assessment module is used to quantitatively assess the accuracy of the modeling results of civil defense facilities.

2. The spatial mapping system for civil defense facilities based on airborne lidar according to claim 1, characterized in that, The data acquisition module includes: The lidar unit is installed on the bottom of the drone and scans the surface of civil defense facilities at a preset scanning frequency; The inertial measurement unit is physically fixed to the lidar unit to measure the attitude data of the UAV's angular velocity and acceleration. The satellite positioning unit obtains the spatial coordinates of the civil defense facility area through carrier phase differential positioning technology of satellite base stations; The point cloud generation unit integrates attitude data and spatial coordinates to generate a three-dimensional point cloud of civil defense facilities.

3. A spatial mapping system for civil defense facilities based on airborne lidar according to claim 2, characterized in that, The point cloud generation unit performs timestamp synchronization and coordinate transformation, including: Compensation for UAV attitude jitter is based on data collected by inertial measurement units; Convert laser scanning data to a geographic coordinate system.

4. A spatial mapping system for civil defense facilities based on airborne lidar according to claim 2, characterized in that, The sampling frequency of the inertial measurement unit is not lower than the scanning frequency of the lidar unit.

5. A spatial mapping system for civil defense facilities based on airborne lidar according to claim 2, characterized in that, The data acquisition module also includes a multi-machine collaboration unit, which controls a second UAV equipped with a multispectral camera to simultaneously acquire images of the civil defense facility area. By adding timestamps, the laser point cloud of the first UAV and the multispectral image of the second UAV are aligned, and a spatial transformation matrix of the civil defense facility area is generated based on SIFT feature matching of the point cloud intensity map and the multispectral image.

6. A spatial mapping system for civil defense facilities based on airborne lidar according to claim 5, characterized in that, The data processing module includes: The point cloud filtering unit filters noise from the three-dimensional point cloud of civil defense facilities. The three-dimensional feature extraction unit automatically extracts the geometric features of civil defense facilities directly from the filtered point cloud through a three-dimensional convolutional neural network. The multi-source fusion classification unit integrates the spectral features of multispectral images with the geometric features of point clouds to output semantic tags for civil defense facilities.

7. A spatial mapping system for civil defense facilities based on airborne lidar according to claim 6, characterized in that, The three-dimensional feature extraction unit adopts a sparse convolutional network architecture, including: The feature encoding layer extracts local structural features of the point cloud through sparse convolutional kernels; Multi-scale pyramid layers, fusing feature maps from different receptive fields to capture global topological relationships; The multi-source fusion classification unit performs: Map multispectral image pixels to point cloud spatial coordinates; Spectral features and point cloud features are fused using a graph attention mechanism; Output three types of semantic labels and confidence levels for ventilation shafts, evacuation routes, and safety doors.

8. A spatial mapping system for civil defense facilities based on airborne lidar according to claim 1, characterized in that, The accuracy evaluation module includes: The time-series comparison unit uses an iterative nearest-point algorithm to perform three-dimensional deformation analysis on multiple scans of point cloud data of the same civil defense facility. The concealment assessment unit simulates the visibility risk level based on the density of the surrounding vegetation point cloud.

9. A spatial mapping system and method for civil defense facilities based on airborne lidar, characterized in that, The method for a civil defense facility spatial mapping system based on airborne lidar as described in any one of claims 1-8 includes the following steps: S1. Simultaneously collect laser scanning data, inertial measurement data, and satellite positioning data of the civil defense facility area using the first UAV; S2. Simultaneously collect multispectral image data of civil defense facilities area using a second drone; S3. Fuse attitude data and spatial coordinates to generate a 3D point cloud; S4. Perform terrain and feature separation processing on the point cloud; S5. Identify civil defense facilities and construct semantic 3D models; S6. Assess the accuracy of the model and the level of hidden risks.

10. A spatial mapping method for civil defense facilities based on airborne lidar according to claim 9, characterized in that: S1 specifically includes: Real-time monitoring of point cloud density in civil defense facility areas and generation of facility area density heat maps; When the density of the target area is lower than a set threshold, the drone's flight altitude or laser scanning frequency is dynamically adjusted. Supplementary scanning is triggered by transmitting control commands via the 5G private network.

Images