Unmanned Aerial Vehicle (UAV) Topographic Mapping System and Methods

Abstract

This application discloses a terrain mapping system and method using an unmanned aerial vehicle (UAV), belonging to the field of terrain mapping technology. The system includes the following modules: a data acquisition module, responsible for collecting high-precision terrain and environmental parameter data of the target area using the UAV platform; a data fusion and correction module, responsible for fusing multi-source data and meteorological information to eliminate errors between sensors; a multi-dimensional feature analysis and recognition module, which extracts and classifies features from various sensor data through multi-dimensional feature analysis, identifying and separating environmental interference signals from real building deformation signals; an interference cancellation and deformation extraction module, which performs pattern recognition and quantitative analysis on periodic non-deformation interference, accurately extracting real building deformation features from complex dynamic noise; a real-time analysis module, which realizes real-time data processing and feedback, ensuring the rapid provision of high-precision building change information; and a visualization module, providing multi-dimensional dynamic visualization displays, including building change heatmaps, interference distribution maps, and time-series curves.

Description

Technical Field

[0001] This application relates to the field of topographic mapping technology, and more specifically, to a topographic mapping system and method for unmanned aerial vehicles (UAVs). Background Technology

[0002] With the rapid development of drone technology and its widespread application in various industries, the potential of drones in fields such as topographic mapping, building monitoring, and disaster early warning is gradually being explored. Traditional topographic mapping methods mostly rely on manual data collection or the use of ground equipment for measurement. These methods are not only inefficient but also prone to errors, making them difficult to cope with the challenges of complex terrain and dynamic environments. Especially in the mapping of large-scale, complex terrain, the accuracy and speed of traditional methods often fall short of the requirements.

[0003] As an aerial platform, unmanned aerial vehicles (UAVs) offer advantages such as flexibility, low cost, and rapid operation, leading to their widespread application in topographic mapping. Equipped with various high-precision sensors, including LiDAR, optical cameras, and infrared sensors, UAVs can quickly collect topographic data of target areas. However, with the surge in mapping data volume and the increasing complexity of terrain environments, ensuring high-precision data acquisition and effective data fusion, eliminating interference signals, and improving data processing efficiency have become pressing technical challenges.

[0004] Currently, while many UAV mapping systems can achieve data acquisition and preliminary processing, they generally suffer from the following problems: First, the fusion of multi-source data is not accurate enough, and it cannot eliminate errors between sensors; second, there is a lack of effective identification and processing mechanisms for interference signals that appear in complex terrain and dynamic environments; third, existing systems rely heavily on ground-based data processing, which cannot fully utilize the real-time processing capabilities of the UAV platform, resulting in delays in real-time feedback and emergency response; and fourth, existing visualization technologies are relatively simple and cannot intuitively present terrain deformation and its changing trends, thus failing to provide effective auxiliary support for decision-makers.

[0005] In conclusion, how to improve the overall performance and application value of UAV topographic mapping systems while ensuring high-precision terrain data acquisition, and by utilizing advanced data fusion, interference cancellation, deformation extraction, and real-time data processing technologies, has become an urgent technical problem to be solved. Summary of the Invention

[0006] In order to overcome a series of defects in the existing technology, the purpose of this application is to provide a terrain mapping system for UAVs, characterized by comprising the following modules:

[0007] The data acquisition module is responsible for synchronously collecting high-precision terrain and environmental parameter data of the target area using the UAV platform;

[0008] The data fusion and correction module is responsible for fusing multi-source data and meteorological information, eliminating errors between sensors, and ensuring that all data are accurately aligned under a unified spatiotemporal reference framework.

[0009] The multidimensional feature analysis and recognition module extracts and classifies features from various sensor data through multidimensional feature analysis, effectively identifying and separating environmental interference signals from real building deformation signals.

[0010] The interference cancellation and deformation extraction module performs pattern recognition and quantitative analysis on periodic non-deformation interference, accurately extracting the true deformation characteristics of the building from complex dynamic noise.

[0011] The real-time analysis module deploys optimized, lightweight data processing algorithms on the drone platform to achieve real-time data processing and feedback, ensuring the rapid provision of high-precision building change information;

[0012] The visualization module provides multi-dimensional dynamic visualization displays, including building change heatmaps, interference distribution maps, and time-series curves.

[0013] The purpose of this application is also to provide a terrain mapping method for unmanned aerial vehicles (UAVs), comprising the following steps:

[0014] Step 1: Use a drone platform to synchronously and accurately collect terrain and environmental parameters of the target area;

[0015] Step 2: Integrate multi-source data and meteorological information to eliminate errors between sensors and ensure that all data are accurately aligned under a unified spatiotemporal reference framework;

[0016] Step 3: Through multi-dimensional feature analysis, feature extraction and classification are performed on the data from each sensor to effectively identify and separate environmental interference signals from real terrain deformation signals;

[0017] Step 4: Perform pattern recognition and quantitative analysis on periodic non-deformation disturbances to accurately extract the true deformation features of the terrain from complex dynamic noise.

[0018] Step 5: Deploy optimized lightweight data processing algorithms on the drone platform to achieve real-time data processing and feedback, ensuring the rapid provision of high-precision terrain change information in emergency monitoring and disaster early warning scenarios;

[0019] Step 6 provides a multi-dimensional dynamic visualization display, including terrain change heatmaps, disturbance distribution maps, and time-series curves.

[0020] Furthermore, step 1 includes the following steps:

[0021] Choose a suitable multi-rotor drone platform and configure an RTK-GPS positioning system to ensure that the positioning accuracy of the flight path reaches the centimeter level;

[0022] Preset the optimal flight altitude, speed, and overlap rate to meet the requirements of surveying data acquisition;

[0023] Equipped with a high-precision LiDAR system to collect terrain point cloud data, and by selecting appropriate laser pulse frequency and scanning angle, high-resolution characterization of complex terrain can be achieved.

[0024] Integrating a multispectral camera, it collects surface reflectance data, covering visible light, near-infrared and short-wave infrared bands. By recording the reflectance differences of different bands, it accurately identifies environmental characteristics such as surface material, vegetation coverage and humidity distribution.

[0025] Deploy miniature weather station modules to record temperature, humidity, air pressure, wind speed, and wind direction parameters in real time;

[0026] The design of a data synchronization acquisition framework adopts a unified time base and precise timestamp marking to ensure the synchronization of data from each sensor at millisecond-level accuracy.

[0027] Based on preset flight missions and real-time data analysis, the adaptive sampling strategy is dynamically adjusted to increase sampling density in complex terrain areas and appropriately reduce sampling frequency in flat areas, so as to balance data quality and processing efficiency and maximize the use of UAV flight time and payload capacity.

[0028] Furthermore, the drone platform has the following parameters: maximum flight altitude of 250 meters, cruising speed of 12-18 meters per second, maximum flight time of no less than 40 minutes, and wind resistance of level 5.

[0029] The RTK-GPS positioning system on the drone has a horizontal positioning accuracy of better than ±2.5 cm and a vertical positioning accuracy of better than ±5 cm.

[0030] The LiDAR system has a point cloud acquisition density of no less than 80 points / square meter, a ranging accuracy of better than ±3 cm, and a scanning angle range of 360° horizontal × 60° vertical.

[0031] The multispectral camera has a spatial resolution of no less than 5 cm / pixel and a spectral resolution covering 5 bands, including blue light, green light, red light, near infrared and short-wave infrared.

[0032] The measurement accuracy of the mini weather station module is as follows: temperature ±0.5℃, relative humidity ±3%, air pressure ±0.8hPa, wind speed ±0.5 m / s, and wind direction ±8°.

[0033] Furthermore, step 2 includes the following steps:

[0034] All sensor data were converted to the local coordinate system, and feature point matching and spatial interpolation methods were used to correct the geometric position offset caused by GPS drift and attitude deviation.

[0035] The network time protocol is used to eliminate the time delay in data acquisition from each sensor, and combined with the UAV flight trajectory information, the spatiotemporal inconsistencies caused by platform movement are corrected to ensure that all data points have accurate four-dimensional spatiotemporal labels.

[0036] By verifying control points and comparing overlapping area data, the errors between various sensors are quantified, including position offset, attitude angle error and scale deformation, and high-precision alignment of multi-source data is achieved based on the inverse transformation algorithm.

[0037] The effects of temperature, humidity and air pressure changes on sensor performance were analyzed, and measurement deviations caused by atmospheric refraction, thermal expansion and humidity changes were dynamically compensated by combining real-time environmental parameters during flight.

[0038] Statistical analysis and spatial consistency testing methods are used to identify and mark outliers caused by equipment failure, signal interference or sudden environmental changes. At the same time, neighborhood smoothing technology is used to repair missing data and outlier areas.

[0039] Based on the spatial sampling density and accuracy characteristics of different sensors, an adaptive weight allocation strategy is designed to ensure overall data consistency while preserving high-frequency details, ultimately generating a unified high-precision terrain and environmental parameter dataset.

[0040] Furthermore, the spatial error threshold for feature point matching is set to less than 5 cm to correct geometric position offsets; the synchronization accuracy of the network time protocol is better than 0.5 milliseconds to ensure the consistency of time stamps for all data points; the distribution density of control point verification is no less than 4 per square kilometer, with a point position accuracy better than ±1.5 cm; the temperature compensation parameter is set to a measurement deviation of ±0.08 mm / m per ℃ change; the humidity compensation parameter is set to a measurement deviation of ±0.05 mm / m per 10% change in relative humidity; the air pressure compensation parameter is set to a measurement deviation of ±0.03 mm / m per 10 hPa change in air pressure; and the standard deviation threshold for outlier identification is set to 3σ, with data points exceeding this range being marked as potential anomalies.

[0041] Furthermore, step 3 includes the following steps:

[0042] Multi-scale wavelet transform is applied to decompose point cloud data and multispectral images, extract spatiotemporal features in different frequency domains, and identify the inherent frequency patterns and anomalous fluctuations of terrain structures.

[0043] By using pre-labeled terrain feature samples for training, accurate land cover classification and segmentation can be achieved, and by comparing multi-temporal data, fixed land features and temporary environmental elements can be distinguished.

[0044] The composite terrain change signal is decomposed into independent components from different sources, and seasonal changes, human interference, instrument noise and real geological movements are identified and quantified to achieve blind separation and feature reconstruction of multi-source signals.

[0045] Semantic annotation is performed on the identified terrain change signals to distinguish different types of terrain changes and assess their development stage and potential risk level.

[0046] Spatiotemporal correlation analysis was conducted to explore the causal relationship between topographic changes and environmental parameters, and conditional probability inference was used to distinguish between temporary changes induced by the environment and continuous geological movements.

[0047] By integrating multiple feature classification results, the reliability of deformation analysis is optimized, and a high-confidence separation result map of terrain deformation and environmental disturbance is finally generated.

[0048] Furthermore, step 4 includes the following steps:

[0049] The temporal terrain data is converted to the frequency domain to extract characteristic frequencies and amplitude information, accurately identify periodic patterns, and quantify various periodic disturbances.

[0050] Based on the identified periodic disturbances and real-time observation data, a set of state-space equations is constructed, and a recursive optimal estimation algorithm is used to suppress random noise and systematic errors, while ensuring the complete preservation of non-periodic real terrain change signals.

[0051] The complex terrain change signal is decomposed into a finite number of intrinsic mode functions and singular components. Through modal energy analysis and statistical significance test, the physically meaningful terrain change patterns are extracted, while effectively filtering out random noise.

[0052] By using differential interferometry and point cloud comparative analysis, the changes in topographic elevation at different spatial scales are quantified, and by combining spatial context information and prior geological knowledge, the differences between local micro-deformations and large-scale geological movements are distinguished.

[0053] By combining Bayesian inference and Markov random field theory, and utilizing temporal continuity and spatial correlation constraints, we can identify anomalous change regions that are inconsistent with the surrounding environment and historical trends.

[0054] By integrating blind source separation, sparse representation and low-rank matrix factorization techniques, this method maximizes the suppression of noise and various interferences while preserving the magnitude and morphological integrity of terrain changes, ultimately generating a terrain deformation feature map with a high signal-to-noise ratio.

[0055] Furthermore, step 5 includes the following steps:

[0056] The data processing algorithm is compressed into an edge version with low resource consumption and fast inference speed, reducing the computational complexity by more than 80% while maintaining the core functionality;

[0057] Design a streaming data processing architecture to decompose the terrain analysis task into parallel subtasks, enabling real-time flow of sensor data acquisition, preprocessing, feature extraction, and deformation analysis;

[0058] Based on the urgency of data processing and the constraints of computing resources, a multi-granularity analysis framework is constructed to dynamically adjust the algorithm accuracy and computational complexity;

[0059] The primary analysis and early warning tasks with high real-time requirements are deployed on the drone edge computing platform, while the computationally intensive deep analysis and historical data comparison tasks are offloaded to the cloud server.

[0060] By combining preset risk thresholds and expert rules, the system automatically generates tiered early warning information and marks the spatial distribution and development trend of high-risk areas.

[0061] Ensure that critical monitoring results can be transmitted to the ground command center in a timely manner under limited bandwidth conditions.

[0062] Furthermore, step 6 includes the following steps:

[0063] A 3D terrain change heat map rendering engine is constructed, which combines a high-precision digital elevation model with terrain deformation data and adopts an adaptive color mapping and transparency encoding strategy to intuitively display the spatial distribution and intensity level of terrain changes.

[0064] The identified non-deformation interference sources are mapped to independent layers to achieve differentiated display of different interference types and support multi-dimensional cross-analysis;

[0065] High temporal resolution deformation history curves are generated for key monitoring points and regions of interest, and time-series data of environmental parameters are overlaid to reveal the temporal coupling relationship between topographic changes and external factors.

[0066] By combining geological hazard sensitivity analysis and vulnerability assessment, the potential impact range and loss estimates are calculated and visualized, supporting multi-scenario simulation and emergency response decision-making;

[0067] By deeply integrating terrain monitoring results with high-definition images and 3D models, an interactive scene combining virtual and real elements can be constructed to achieve more intuitive environmental perception and analysis.

[0068] Design a collaborative analysis and knowledge sharing platform that integrates monitoring data, analysis results, and expert knowledge, providing multi-level information display and collaborative interpretation functions.

[0069] Compared with the prior art, this application has the following beneficial effects:

[0070] This application improves the accuracy, real-time performance, and intelligence of topographic mapping by using multi-source data fusion, interference elimination, deformation extraction, real-time processing, and multi-dimensional visualization to accurately monitor and analyze changes in buildings and terrain. Attached Figure Description

[0071] Figure 1 This is a schematic diagram of the structure of the unmanned aerial vehicle (UAV) terrain mapping system disclosed in the embodiments of this application.

[0072] Figure 2 This is a flowchart illustrating the terrain mapping method for unmanned aerial vehicles (UAVs) disclosed in an embodiment of this application. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be described in more detail below with reference to the accompanying drawings. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are some embodiments of this invention, but not all embodiments.

[0074] 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.

[0075] The embodiments and directional terms described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0076] like Figure 1 As shown, the terrain mapping system for UAVs includes the following modules:

[0077] The data acquisition module is responsible for synchronously collecting high-precision terrain and environmental parameter data of the target area using the UAV platform;

[0078] The data fusion and correction module is responsible for fusing multi-source data and meteorological information, eliminating errors between sensors, and ensuring that all data are accurately aligned under a unified spatiotemporal reference framework.

[0079] The multidimensional feature analysis and recognition module extracts and classifies features from various sensor data through multidimensional feature analysis, effectively identifying and separating environmental interference signals from real building deformation signals.

[0080] The interference cancellation and deformation extraction module performs pattern recognition and quantitative analysis on periodic non-deformation interference, accurately extracting the true deformation characteristics of the building from complex dynamic noise.

[0081] The real-time analysis module deploys optimized, lightweight data processing algorithms on the drone platform to achieve real-time data processing and feedback, ensuring the rapid provision of high-precision building change information;

[0082] The visualization module provides multi-dimensional dynamic visualization displays, including building change heatmaps, interference distribution maps, and time-series curves.

[0083] In summary, this building surveying system based on UAV remote sensing technology achieves high-precision, real-time monitoring of building changes by integrating multiple modules. The data acquisition module utilizes the UAV platform to simultaneously collect terrain and environmental data, providing high-precision data for subsequent analysis; the data fusion and correction module eliminates sensor errors, ensuring data alignment and improving accuracy; the multi-dimensional feature analysis and recognition module effectively extracts and classifies sensor data, accurately distinguishing between building deformation and environmental interference signals; the interference elimination and deformation extraction module accurately identifies and quantitatively analyzes periodic interference, extracting the true deformation characteristics of the building; the real-time analysis module uses lightweight algorithms to achieve rapid data processing and feedback, providing timely information on building changes; and the visualization module enhances the intuitiveness and operability of monitoring results by dynamically displaying heatmaps of building changes and interference distribution maps. The entire system improves the accuracy, real-time performance, and interference identification capabilities of building surveying, significantly enhancing the reliability and visualization analysis capabilities of building deformation monitoring.

[0084] like Figure 2 As shown, the purpose of this application is also to provide a terrain mapping method for unmanned aerial vehicles (UAVs), comprising the following steps:

[0085] Step 1: Use a drone platform to synchronously and accurately collect terrain and environmental parameters of the target area;

[0086] Step 2: Integrate multi-source data and meteorological information to eliminate errors between sensors and ensure that all data are accurately aligned under a unified spatiotemporal reference framework;

[0087] Step 3: Through multi-dimensional feature analysis, feature extraction and classification are performed on the data from each sensor to effectively identify and separate environmental interference signals from real terrain deformation signals;

[0088] Step 4: Perform pattern recognition and quantitative analysis on periodic non-deformation disturbances to accurately extract the true deformation features of the terrain from complex dynamic noise.

[0089] Step 5: Deploy optimized lightweight data processing algorithms on the drone platform to achieve real-time data processing and feedback, ensuring the rapid provision of high-precision terrain change information in emergency monitoring and disaster early warning scenarios;

[0090] Step 6 provides a multi-dimensional dynamic visualization display, including terrain change heatmaps, disturbance distribution maps, and time-series curves.

[0091] In step 1, the UAV platform is equipped with high-precision sensor devices, such as LiDAR, high-resolution cameras, multispectral cameras, and GNSS positioning systems. Leveraging the platform's stability during flight, it conducts comprehensive and accurate data collection of the target area. During flight, the UAV, through pre-planned flight paths, acquires real-time environmental parameters such as terrain, vegetation cover, and water distribution within the area, ensuring the temporal synchronization and spatial continuity of data collection. The advantages of this technology lie in its flexibility and efficiency, enabling rapid deployment in complex or inaccessible areas. The collected data not only possesses high spatial resolution but also, due to the collaborative work of multiple sensors, forms a multi-dimensional, multi-angle terrain information dataset. This high-precision data acquisition method not only reduces subjective errors associated with traditional manual surveys but also significantly improves the real-time performance and accuracy of the data. UAV data acquisition technology also utilizes advanced positioning and attitude control to ensure the sensors remain stable throughout flight, preventing data distortion due to jitter or external interference. By preprocessing the raw sensor data to remove noise and outliers, the signal-to-noise ratio and validity of the data can be further improved. This technological step is remarkably effective, particularly in emergency rescue, natural disaster monitoring, and environmental protection, where it holds immense promise. It can provide decision-makers with precise topographic and environmental information in a short time, facilitating rapid response and informed decision-making. More importantly, by simultaneously collecting topographic data and environmental parameters, a three-dimensional, multi-dimensional panoramic view of the regional environment can be constructed, providing a solid data foundation and reliable spatiotemporal reference for subsequent data fusion, feature extraction, and change monitoring.

[0092] Step 2 involves multi-source data fusion from different sensors, platforms, and even different acquisition time periods to form a unified and standardized dataset. First, advanced algorithms are used to preprocess the acquired data, employing techniques such as data filtering, correction, and registration to eliminate errors caused by differences in acquisition angles, time intervals, and hardware precision among different sensors. The terrain, environmental, and meteorological data acquired by the sensors are spatiotemporally aligned to achieve seamless integration between different data sets. Utilizing a unified spatiotemporal reference framework, not only can precise matching between data sources in time and space be ensured, but real-time monitoring and comparison of dynamic changes within the region can also be achieved. This technology fully leverages the complementary information between multi-source data, and through algorithmic optimization, it can eliminate systematic errors in positioning and calibration of individual sensors, improving the overall reliability and accuracy of the data. Especially in complex environments and severe weather conditions, by fusing meteorological information, real-time corrections are made for factors such as wind speed, temperature, and humidity, effectively compensating for the impact of environmental changes on sensor measurement accuracy. Simultaneously, the data fusion technology also possesses high fault tolerance and robustness, ensuring that even if some data is missing or abnormal, it will not significantly affect the accuracy of the overall data. Through multi-level and multi-scale data matching and alignment, the resulting unified dataset provides a solid foundation for subsequent feature extraction, pattern recognition, and deformation analysis. This enables quantitative comparison and joint analysis of data from different sources under the same reference system, thereby achieving precise monitoring of terrain changes and environmental dynamics in the target area and meeting the practical application needs of high-requirement scenarios such as disaster early warning and geological disaster assessment.

[0093] In step 3, advanced multidimensional feature analysis technology is employed to deeply mine and extract spatial, frequency, and temporal features from the UAV-collected data, enabling detailed classification and identification of various signals within the data. First, mathematical tools such as statistical analysis, Fourier transform, and wavelet transform are used to decompose and reconstruct the main signals, noise, and latent patterns in the data. By setting various feature indicators, such as texture, gradient, shape, and dynamic change characteristics, abnormal fluctuations caused by environmental interference can be identified, while information reflecting actual terrain deformation can be separated. Further, machine learning algorithms, such as Support Vector Machines (SVM), Random Forests, and Deep Neural Networks, are used to classify and label the extracted features. This process not only improves the ability to identify signals in complex backgrounds but also maintains high recognition accuracy even with high data dimensionality and significant noise interference. The advantage of using multidimensional feature analysis lies in its ability to simultaneously consider local details and overall trends in the data, thereby achieving accurate differentiation between terrain change signals and environmental noise. Especially in scenarios involving sudden regional topographic changes or local anomalies, this technology can promptly capture minute deformation features, providing crucial information for subsequent disaster early warning and geological hazard assessment. Furthermore, continuous iteration and optimization of feature extraction and classification algorithms enhance adaptability and robustness when facing data from different environments and acquisition conditions, ensuring efficient and stable monitoring performance even in complex and ever-changing application scenarios. Through the application of multi-dimensional feature analysis technology, the entire data processing workflow is systematized and standardized, thereby achieving accurate separation of real information and interference signals from sensor data.

[0094] In step 4, a detailed analysis of periodic non-deformation interference signals is conducted. By establishing a mathematical model and pattern recognition algorithm for the interference signals, periodic noise in the environment is distinguished from actual terrain deformation signals. Periodic non-deformation interference is often caused by equipment vibration, external mechanical motion, or periodic meteorological fluctuations; these signals exhibit obvious periodic characteristics in the frequency domain. Through methods such as Fourier transform, wavelet analysis, and autoregressive models, the periodic components in the data can be quantitatively decomposed and their spectrum analyzed. First, all signal components with obvious periodic characteristics in the data are identified, and template matching and spectral analysis are used to compare and correct these components, effectively eliminating periodic interference unrelated to actual terrain deformation. Quantitative analysis techniques further model and compensate for the interference signals by extracting noise intensity, frequency changes, and phase information, making subsequent deformation feature extraction more accurate. Simultaneously, by combining time-series information from multi-source data, the variation patterns of interference signals in different time periods can be dynamically monitored, constructing a complete interference signal database. This allows for the prediction and compensation of potential periodic errors in future data acquisition. This technology not only theoretically solves the problem of periodic noise interfering with terrain deformation analysis, but also, in practical applications, achieves effective extraction of real signals from complex dynamic noise through data filtering and reconstruction. This not only ensures the accuracy of data processing results but also significantly improves robustness and stability in high-noise environments. Through pattern recognition and quantitative analysis of periodic non-deformation interference signals, it ultimately provides an accurate and interference-free terrain change model, offering a scientific basis for disaster early warning, environmental monitoring, and engineering construction, ensuring that every critical decision is based on real and reliable data.

[0095] Step 5 focuses on integrating advanced data processing algorithms into the UAV platform to ensure real-time processing and analysis while data is being collected, thereby providing timely response information for emergency monitoring and disaster early warning. The optimization goal of lightweight data processing algorithms is to balance computational complexity and real-time requirements. Through algorithm compression, model pruning, and edge computing technologies, high-speed data processing can be achieved even with the limited computing resources of the UAV platform. Real-time data processing requires algorithms to not only possess efficient data preprocessing, feature extraction, and model inference capabilities, but also to control the algorithm response time to the millisecond level for timely feedback of monitoring results. To this end, multi-threaded parallel computing and adaptive sampling techniques are introduced into the algorithm design to minimize latency in all stages of data transmission, processing, and feedback. Simultaneously, considering the complex environment and massive data volume in emergency scenarios, incremental learning and online update mechanisms are adopted to continuously optimize and adjust model parameters during data collection, improving the overall adaptability to dynamically changing scenarios. By deploying data processing algorithms on the UAV platform, not only is the time and bandwidth pressure of data transmission to the backend center reduced, but the real-time performance of on-site decision-making is also significantly improved, providing reliable support for emergency rescue and disaster early warning. This solution, through the tight coupling of hardware and software, achieves seamless integration of data acquisition, processing, and feedback. This enables each flight mission to generate a high-precision terrain change report, which is transmitted to the command center in real time via wireless communication technology. As a result, decision-makers can obtain accurate and detailed information at the first opportunity, formulate timely countermeasures, and safeguard the safety of people's lives and property.

[0096] Step 6 transforms the multidimensional data, processed and analyzed in real time, into an intuitive graphical interface using advanced visualization technology, providing users with a comprehensive and multi-faceted platform for displaying monitoring results. First, a terrain change heatmap is generated, presenting deformation information within the region in the form of color gradients, allowing users to intuitively observe which areas have undergone significant changes and which remain stable. Simultaneously, an interference distribution map, by marking the spatial location and intensity of interference signals, helps users distinguish between genuine terrain deformation signals and environmental noise in the data, thus providing a reference for subsequent analysis and decision-making. The time-series curves display the dynamic changes of data within the monitoring area over time; through real-time updated data curves, the trends and periodic characteristics of terrain changes can be reflected. This multidimensional dynamic visualization display, based on advanced data rendering technology and interactive graphical interface design, allows complex data to be presented in a concise and intuitive way. Users can quickly obtain key information through various formats such as graphs, charts, and maps. It also supports interactive operation, allowing users to zoom in and out, switch perspectives and time periods, and deeply analyze data changes in specific areas or at specific points in time, improving the flexibility and depth of data analysis. The entire visualization module not only achieves efficient data integration and display in terms of technology, but also features meticulous design in terms of user experience, ensuring that users of different levels and backgrounds can quickly understand the information conveyed by the data. Whether researchers, emergency decision-makers, or the general public, everyone can gain an intuitive understanding of terrain changes and environmental dynamics through this platform, providing invaluable reference for scientific research, urban planning, and disaster emergency response. Furthermore, the dynamic display of multi-dimensional data has transformed the data analysis process from traditional static reporting to an interactive, real-time monitoring mode, greatly enhancing the application effectiveness and decision support capabilities of monitoring.

[0097] In summary, this UAV-based mapping data acquisition and analysis method achieves high-precision monitoring throughout the entire process, from data collection to result display, through multiple steps and technologies. Each step fully demonstrates the advantages of advanced sensing technology, data fusion, pattern recognition, and real-time processing, ensuring data accuracy while improving processing efficiency and application flexibility. From synchronous acquisition to multi-source data fusion, to multi-dimensional feature extraction and interference removal, and finally to real-time feedback and dynamic visualization, each link is interconnected, jointly constructing a complete and efficient monitoring system. This method not only has extremely high application value in disaster early warning and emergency rescue but also provides new technical ideas and implementation paths for geographic information systems and environmental monitoring. In the future, it will undoubtedly play a greater role in fields such as urban planning, ecological protection, and resource management.

[0098] Furthermore, step 1 includes the following steps:

[0099] Choose a suitable multi-rotor drone platform and configure an RTK-GPS positioning system to ensure that the positioning accuracy of the flight path reaches the centimeter level;

[0100] Preset the optimal flight altitude, speed, and overlap rate to meet the requirements of surveying data acquisition;

[0101] Equipped with a high-precision LiDAR system to collect terrain point cloud data, and by selecting appropriate laser pulse frequency and scanning angle, high-resolution characterization of complex terrain can be achieved.

[0102] Integrating a multispectral camera, it collects surface reflectance data, covering visible light, near-infrared and short-wave infrared bands. By recording the reflectance differences of different bands, it accurately identifies environmental characteristics such as surface material, vegetation coverage and humidity distribution.

[0103] Deploy miniature weather station modules to record temperature, humidity, air pressure, wind speed, and wind direction parameters in real time;

[0104] The design of a data synchronization acquisition framework adopts a unified time base and precise timestamp marking to ensure the synchronization of data from each sensor at millisecond-level accuracy.

[0105] Based on preset flight missions and real-time data analysis, the adaptive sampling strategy is dynamically adjusted to increase sampling density in complex terrain areas and appropriately reduce sampling frequency in flat areas, so as to balance data quality and processing efficiency and maximize the use of UAV flight time and payload capacity.

[0106] In summary, by precisely configuring the UAV platform and high-precision sensors, centimeter-level positioning accuracy and an efficient data acquisition strategy are achieved, ensuring high resolution and precision in topographic mapping. RTK-GPS improves the accuracy of flight paths, while optimized flight parameters (altitude, speed, overlap rate) ensure the integrity and consistency of data coverage. High-precision LiDAR, combined with appropriate laser pulse frequencies and scanning angles, enables detailed depiction of complex terrain, while multispectral cameras accurately analyze surface material, vegetation cover, and humidity distribution through reflectivity data from different bands. An integrated micro-weather station records environmental parameters in real time, providing meteorological compensation information and improving data reliability. Furthermore, the data synchronization acquisition framework utilizes a unified time base and high-precision timestamps to ensure the spatiotemporal consistency of data from various sensors, and optimizes data acquisition under different terrain conditions through adaptive sampling strategies, improving data quality while enhancing flight efficiency. Overall, this solution maximizes the UAV's payload capacity and endurance, achieving intelligent and dynamic data optimization and management while ensuring high-precision mapping data acquisition.

[0107] Furthermore, the drone platform has the following parameters: maximum flight altitude of 250 meters, cruising speed of 12-18 meters per second, maximum flight time of no less than 40 minutes, and wind resistance of level 5.

[0108] The RTK-GPS positioning system on the drone has a horizontal positioning accuracy of better than ±2.5 cm and a vertical positioning accuracy of better than ±5 cm.

[0109] The LiDAR system has a point cloud acquisition density of no less than 80 points / square meter, a ranging accuracy of better than ±3 cm, and a scanning angle range of 360° horizontal × 60° vertical.

[0110] The multispectral camera has a spatial resolution of no less than 5 cm / pixel and a spectral resolution covering 5 bands, including blue light, green light, red light, near infrared and short-wave infrared.

[0111] The measurement accuracy of the mini weather station module is as follows: temperature ±0.5℃, relative humidity ±3%, air pressure ±0.8hPa, wind speed ±0.5 m / s, and wind direction ±8°.

[0112] Furthermore, step 2 includes the following steps:

[0113] All sensor data were converted to the local coordinate system, and feature point matching and spatial interpolation methods were used to correct the geometric position offset caused by GPS drift and attitude deviation.

[0114] The network time protocol is used to eliminate the time delay in data acquisition from each sensor, and combined with the UAV flight trajectory information, the spatiotemporal inconsistencies caused by platform movement are corrected to ensure that all data points have accurate four-dimensional spatiotemporal labels.

[0115] By verifying control points and comparing overlapping area data, the errors between various sensors are quantified, including position offset, attitude angle error and scale deformation, and high-precision alignment of multi-source data is achieved based on the inverse transformation algorithm.

[0116] The effects of temperature, humidity and air pressure changes on sensor performance were analyzed, and measurement deviations caused by atmospheric refraction, thermal expansion and humidity changes were dynamically compensated by combining real-time environmental parameters during flight.

[0117] Statistical analysis and spatial consistency testing methods are used to identify and mark outliers caused by equipment failure, signal interference or sudden environmental changes. At the same time, neighborhood smoothing technology is used to repair missing data and outlier areas.

[0118] Based on the spatial sampling density and accuracy characteristics of different sensors, an adaptive weight allocation strategy is designed to ensure overall data consistency while preserving high-frequency details, ultimately generating a unified high-precision terrain and environmental parameter dataset.

[0119] Furthermore, the effects of temperature, humidity, and air pressure changes on sensor performance were analyzed, and measurement deviations caused by atmospheric refraction, thermal expansion, and humidity changes were dynamically compensated by incorporating real-time environmental parameters during flight. This included the following steps:

[0120] The temperature data collected in real time during flight is expressed using the formula ΔS. T =S0·α T • (T-T0) Compensate for thermal expansion error, adjust the measured value to S T =S0-ΔS T , where ΔS T S0 represents the measurement deviation caused by temperature; S0 is the original measurement value, i.e., the measurement data before any compensation; α T The temperature compensation coefficient represents the sensor's sensitivity to temperature changes; T is the current flight temperature; T0 is the standard temperature during calibration; S T These are sensor measurements after temperature compensation;

[0121] Using real-time humidity data, the formula ΔS H =S0·α H • (H-H0) corrects for deviations caused by humidity changes, yielding S H =S0-ΔS H , where ΔS H This is the humidity compensation amount, representing the sensor measurement deviation caused by changes in humidity; α H Here, is the humidity sensitivity coefficient, representing the sensor's response to changes in humidity; H is the current ambient humidity; H0 is the standard humidity at the time of calibration; S H These are the sensor measurements after humidity compensation;

[0122] Using real-time air pressure, according to the formula ΔS P =S0·α P • (P-P0) Eliminates the influence of air pressure changes; the corrected measurement value is S. P =S0-ΔS P , where ΔS P This is the air pressure compensation amount, representing the sensor measurement deviation caused by changes in air pressure; α P is the pressure sensitivity coefficient, representing the sensor's response to changes in air pressure; P is the current ambient air pressure; P0 is the standard air pressure during calibration; S P These are sensor measurements after pressure compensation;

[0123] By considering the combined effects of air pressure, temperature, and humidity, the formula is adopted. Dynamic compensation for refraction error, correction value is S ref =S0-ΔS ref , where ΔS refβ is the atmospheric refraction compensation, representing the measurement deviation caused by atmospheric refraction; β is the atmospheric refraction correction coefficient, representing the degree of sensor response to atmospheric refraction; S ref These are sensor measurements after atmospheric refraction compensation;

[0124] The effects of temperature, humidity, air pressure, and atmospheric refraction are comprehensively compensated, and the final compensated sensor output is calculated using the following formula: Among them, S comp This is the sensor measurement value after comprehensive compensation.

[0125] Furthermore, the spatial error threshold for feature point matching is set to less than 5 cm to correct geometric position offsets; the synchronization accuracy of the network time protocol is better than 0.5 milliseconds to ensure the consistency of time stamps for all data points; the distribution density of control point verification is no less than 4 per square kilometer, with a point position accuracy better than ±1.5 cm; the temperature compensation parameter is set to a measurement deviation of ±0.08 mm / m per ℃ change; the humidity compensation parameter is set to a measurement deviation of ±0.05 mm / m per 10% change in relative humidity; the air pressure compensation parameter is set to a measurement deviation of ±0.03 mm / m per 10 hPa change in air pressure; and the standard deviation threshold for outlier identification is set to 3σ, with data points exceeding this range being marked as potential anomalies.

[0126] Furthermore, step 3 includes the following steps:

[0127] Multi-scale wavelet transform is applied to decompose point cloud data and multispectral images, extract spatiotemporal features in different frequency domains, and identify the inherent frequency patterns and anomalous fluctuations of terrain structures.

[0128] By using pre-labeled terrain feature samples for training, accurate land cover classification and segmentation can be achieved, and by comparing multi-temporal data, fixed land features and temporary environmental elements can be distinguished.

[0129] The composite terrain change signal is decomposed into independent components from different sources, and seasonal changes, human interference, instrument noise and real geological movements are identified and quantified to achieve blind separation and feature reconstruction of multi-source signals.

[0130] Semantic annotation is performed on the identified terrain change signals to distinguish different types of terrain changes and assess their development stage and potential risk level.

[0131] Spatiotemporal correlation analysis was conducted to explore the causal relationship between topographic changes and environmental parameters, and conditional probability inference was used to distinguish between temporary changes induced by the environment and continuous geological movements.

[0132] By integrating multiple feature classification results, the reliability of deformation analysis is optimized, and a high-confidence separation result map of terrain deformation and environmental disturbance is finally generated.

[0133] In summary, by applying multi-scale wavelet transform to conduct in-depth analysis of point cloud data and multispectral imagery, spatiotemporal features in different frequency domains are extracted, accurately identifying the inherent frequency patterns and anomalous fluctuations of terrain structures, thereby effectively analyzing complex terrain changes. Pre-labeled terrain feature samples are used for training to achieve land cover classification and segmentation. Simultaneously, through multi-temporal data comparison, fixed features are distinguished from temporary environmental elements. Furthermore, blind source separation technology is used to decompose the composite terrain change signal into independent components from different sources, identifying seasonal changes, human interference, instrument noise, and genuine geological movements, enhancing the accuracy of data processing. Based on this, semantic annotation and potential risk assessment are performed on the terrain change signal. Combined with spatiotemporal correlation analysis, the causal relationship between terrain change and environmental parameters is explored. Conditional probability inference further distinguishes between temporary changes and continuous geological movements. Finally, by integrating multiple feature classification results, the reliability of deformation analysis is optimized, generating a high-confidence separation map of terrain deformation and environmental interference, ensuring high accuracy and reliability of terrain change analysis.

[0134] Furthermore, the multi-scale wavelet transform decomposition hierarchy is set to 4 levels, covering spatial scale features from 0.5 meters to 40 meters; the land cover classification accuracy is required to reach over 85%, and the smallest segmented unit area is no greater than 2 square meters; the time interval for multi-temporal data comparison is set to four levels: 12 hours, 24 hours, 7 days, and 30 days; the number of independent components for signal decomposition is set to 3-6, adaptively adjusted according to the signal-to-noise ratio; the period of seasonal variation is set to a range of 30-365 days, and the characteristic frequency range of human interference is 1-30 days; the confidence threshold for spatiotemporal correlation analysis is set to 90%, and the correlation coefficient threshold is 0.7.

[0135] Furthermore, spatiotemporal correlation analysis is conducted to explore the causal relationship between topographic changes and environmental parameters. Conditional probability inference is used to distinguish between temporary changes induced by the environment and continuous geological movements, including the following steps:

[0136] Based on the formula P(D(t),E(t))=P(D(t)∣E(t))P(E(t)), a joint probability distribution model of topographic change and environmental parameters is established, where P(D(t),E(t)) represents the joint probability distribution of topographic change D(t) and environmental parameter E(t) at time t; P(D(t)∣E(t)) represents the conditional probability of topographic change D(t) occurring given environmental parameter E(t); and P(E(t)) represents the marginal probability distribution of environmental parameter E(t), that is, the probability of environmental parameter occurring at a certain time t.

[0137] use The formula calculates the conditional probability of terrain change under different environmental conditions, quantifying the direct impact of environmental factors on terrain change;

[0138] The topographic change D(t) is decomposed into long-term trend T(t) and short-term fluctuation N(t), i.e., D(t) = T(t) + N(t), and their respective characteristics are extracted through time series analysis;

[0139] use Analyzing the relationship between short-term fluctuations N(t) and environmental parameters E(t) helps determine whether topographic changes are induced by the environment. Here, P(N(t)|E(t)) represents the conditional probability of short-term fluctuations N(t) occurring given environmental parameters E(t); σ is the standard deviation of environmental parameters E(t), reflecting the degree of fluctuation; μ... E (t) is the mean of environmental parameter E(t) at time t, representing the average level of environmental parameter at a certain moment;

[0140] The interaction between topography and environmental parameters in various regions was comprehensively analyzed, causal relationships were verified using statistical tests, and abnormal areas were marked and risk assessments were conducted.

[0141] Furthermore, step 4 includes the following steps:

[0142] The temporal terrain data is converted to the frequency domain to extract characteristic frequencies and amplitude information, accurately identify periodic patterns, and quantify various periodic disturbances.

[0143] Based on the identified periodic disturbances and real-time observation data, a set of state-space equations is constructed, and a recursive optimal estimation algorithm is used to suppress random noise and systematic errors, while ensuring the complete preservation of non-periodic real terrain change signals.

[0144] The complex terrain change signal is decomposed into a finite number of intrinsic mode functions and singular components. Through modal energy analysis and statistical significance test, the physically meaningful terrain change patterns are extracted, while effectively filtering out random noise.

[0145] By using differential interferometry and point cloud comparative analysis, the changes in topographic elevation at different spatial scales are quantified, and by combining spatial context information and prior geological knowledge, the differences between local micro-deformations and large-scale geological movements are distinguished.

[0146] By combining Bayesian inference and Markov random field theory, and utilizing temporal continuity and spatial correlation constraints, we can identify anomalous change regions that are inconsistent with the surrounding environment and historical trends.

[0147] By integrating blind source separation, sparse representation and low-rank matrix factorization techniques, this method maximizes the suppression of noise and various interferences while preserving the magnitude and morphological integrity of terrain changes, ultimately generating a terrain deformation feature map with a high signal-to-noise ratio.

[0148] In summary, by transforming time-series topographic data to the frequency domain, characteristic frequencies and amplitude information are accurately extracted, thereby effectively identifying and quantifying periodic interference signals and improving the accuracy of data analysis. Based on the identified periodic interference, a state-space equation set is constructed, and a recursive optimal estimation algorithm is used to suppress random noise and systematic errors while ensuring the integrity of the real topographic change signal. Furthermore, this method employs mode decomposition technology to break down complex topographic change signals into a finite number of intrinsic mode functions and singular components. Through modal energy analysis and statistical significance testing, physically meaningful topographic change patterns are extracted, and random noise is effectively filtered out. In terms of spatial analysis, differential interferometry and point cloud comparative analysis are combined to accurately quantify topographic elevation changes at different scales, and geological prior knowledge is used to distinguish between local micro-deformations and large-scale geological movements. Further, Bayesian inference and Markov random field theory are used, through temporal continuity and spatial correlation constraints, to accurately identify anomalous change areas and enhance the stability of topographic change detection. Finally, by integrating blind source separation, sparse representation and low-rank matrix factorization techniques, while preserving the amplitude and morphological integrity of terrain deformation, various interferences are suppressed to the maximum extent, generating a terrain deformation feature map with a high signal-to-noise ratio, thereby ensuring high accuracy, high stability and high reliability of terrain change analysis.

[0149] Furthermore, the recursive optimal estimation algorithm achieves noise suppression capability of over 15dB while ensuring a retention rate of no less than 90% for non-periodic signals; the threshold for intrinsic mode function decomposition is set to a feature energy ratio of no less than 80% of the total energy; the minimum detectable deformation of differential interferometry is 8 mm, and the elevation change detection accuracy of point cloud comparative analysis is better than 1.5 cm; the prior probability model of Bayesian inference is constructed based on historical data from the past 120 days, and the spatial correlation radius of the Markov random field is set to 20-80 meters; the rank parameter r of low-rank matrix decomposition is set to 8%-12% of the matrix dimension to ensure the efficiency of sparse signal representation.

[0150] Furthermore, by using differential interferometry and point cloud comparative analysis, the topographic elevation changes at different spatial scales are quantified. Combined with spatial context information and prior geological knowledge, the differences between localized micro-deformations and large-scale geological movements are distinguished, including the following steps:

[0151] By calculating the phase difference Δφ of the radar echo, according to the formula... Quantifying the change in surface elevation, where Δh DInSAR λ is the elevation change calculated through differential interferometry; λ is the wavelength of the radar signal; θ is the angle between the radar beam and the ground.

[0152] By calculating the elevation difference in continuous time point cloud data The local elevation changes are assessed, and the overall trend of deformation is analyzed using the mean and standard deviation to obtain detailed information on local deformation, where Δh PC This refers to the elevation change calculated based on point cloud data. Let be the elevation value of the i-th point measured at time t+1; Let be the elevation value of the i-th point measured at time t;

[0153] By weighted fusion of DInSAR and point cloud data, the formula is expressed as: Δh fusion (x,y)=w1·Δh DInSAR (x,y)+w2·Δh PC (x,y), where Δh fusion (x,y) represents the elevation change at coordinates (x,y) after fusing DInSAR and point cloud data; w1 and w2 are the weighting coefficients of the DInSAR and point cloud data, respectively; Δh DInSAR (x,y), Δh PC (x,y) represents the elevation change at coordinates (x,y) calculated from DInSAR and point cloud data, respectively.

[0154] Large-scale deformation through Δh large (x,y)=G σ *Δh fusion (x,y) is obtained, while the local deformation is Δh. local (x,y)=Δh fusion (x,y)-Δh large (x,y) effectively distinguishes deformation characteristics at different scales, where Δh large (x,y) represents the large-scale topographic changes at the spatial location (x,y); Δh local (x,y) represents a small topographical change at the spatial location (x,y);

[0155] Based on prior geological knowledge, a threshold T for large-scale deformation is set. large and the threshold T of local small deformation local If Δh large (x,y) exceeds T large If Δh local (x,y) exceeds T local This indicates a minor, localized deformation.

[0156] Furthermore, step 5 includes the following steps:

[0157] The data processing algorithm is compressed into an edge version with low resource consumption and fast inference speed, reducing the computational complexity by more than 80% while maintaining the core functionality;

[0158] Design a streaming data processing architecture to decompose the terrain analysis task into parallel subtasks, enabling real-time flow of sensor data acquisition, preprocessing, feature extraction, and deformation analysis;

[0159] Based on the urgency of data processing and the constraints of computing resources, a multi-granularity analysis framework is constructed to dynamically adjust the algorithm accuracy and computational complexity;

[0160] The primary analysis and early warning tasks with high real-time requirements are deployed on the drone edge computing platform, while the computationally intensive deep analysis and historical data comparison tasks are offloaded to the cloud server.

[0161] By combining preset risk thresholds and expert rules, the system automatically generates tiered early warning information and marks the spatial distribution and development trend of high-risk areas.

[0162] Ensure that critical monitoring results can be transmitted to the ground command center in a timely manner under limited bandwidth conditions.

[0163] In summary, by optimizing the data processing algorithm, the computational complexity is reduced by more than 80%, making it suitable for resource-constrained UAV edge computing platforms, thereby achieving efficient and low-power data analysis. A streaming data processing architecture is adopted, decomposing the terrain analysis task into parallel sub-tasks, enabling real-time flow of sensor data acquisition, preprocessing, feature extraction, and deformation analysis, improving system response speed. The multi-granularity analysis framework dynamically adjusts algorithm accuracy and computational complexity based on task urgency and computational resource constraints, ensuring efficient operation in various application scenarios. Furthermore, this method employs a hierarchical computing strategy, deploying high-real-time primary analysis and early warning tasks on the UAV, while offloading computationally intensive deep analysis and historical data comparison tasks to cloud servers, thus balancing computational efficiency and analytical accuracy. In terms of data application, combined with preset risk thresholds and expert rules, tiered early warning information is automatically generated, marking the spatial distribution and development trends of high-risk areas to ensure the reliability of disaster monitoring.

[0164] Furthermore, step 6 includes the following steps:

[0165] A 3D terrain change heat map rendering engine is constructed, which combines a high-precision digital elevation model with terrain deformation data and adopts an adaptive color mapping and transparency encoding strategy to intuitively display the spatial distribution and intensity level of terrain changes.

[0166] The identified non-deformation interference sources are mapped to independent layers to achieve differentiated display of different interference types and support multi-dimensional cross-analysis;

[0167] High temporal resolution deformation history curves are generated for key monitoring points and regions of interest, and time-series data of environmental parameters are overlaid to reveal the temporal coupling relationship between topographic changes and external factors.

[0168] By combining geological hazard sensitivity analysis and vulnerability assessment, the potential impact range and loss estimates are calculated and visualized, supporting multi-scenario simulation and emergency response decision-making;

[0169] By deeply integrating terrain monitoring results with high-definition images and 3D models, an interactive scene combining virtual and real elements can be constructed to achieve more intuitive environmental perception and analysis.

[0170] Design a collaborative analysis and knowledge sharing platform that integrates monitoring data, analysis results, and expert knowledge, providing multi-level information display and collaborative interpretation functions.

[0171] In summary, by constructing a 3D terrain change heatmap rendering engine, combining a high-precision digital elevation model with terrain deformation data, and employing adaptive color mapping and transparency encoding strategies, the spatial distribution and intensity levels of terrain changes are intuitively displayed, improving data visualization. Simultaneously, non-deformation interference sources are mapped to independent layers, supporting differentiated display and multi-dimensional cross-analysis of different interference types, enhancing the ability to identify interference in complex environments. For key monitoring points and regions of interest, high-temporal-resolution deformation history curves are generated, overlaid with time-series data of environmental parameters to reveal the temporal coupling relationship between terrain changes and external factors. Furthermore, this method combines geological hazard sensitivity analysis and vulnerability assessment to calculate and visualize the potential impact range and loss estimation, providing data support for multi-scenario simulation and emergency response. In terms of data fusion, terrain monitoring results are deeply integrated with high-definition imagery and 3D models to construct an interactive virtual-real fusion scene, enhancing the intuitiveness of environmental perception and analysis. Finally, a collaborative analysis and knowledge-sharing platform is designed, integrating monitoring data, analysis results, and expert knowledge, supporting multi-level information display and collaborative interpretation, enhancing the application value of terrain monitoring data, and providing intelligent support for disaster early warning and emergency decision-making.

[0172] Furthermore, the spatial resolution of the three-dimensional terrain change heatmap is 1 meter, the change display accuracy is at the millimeter level, and 10 levels of color are used to represent the deformation intensity.

[0173] The interference source layers are divided into 5 categories: vegetation change, hydrological change, meteorological influence, human activities and sensor error, with each category using different legend labels;

[0174] The sampling interval for the high temporal resolution deformation history curve is 2 hours, and the time span can be adjusted to 24 hours, 7 days, 30 days, and 180 days.

[0175] The spatial resolution of the potential impact range assessment is 10 meters, and the uncertainty range of the loss estimation is controlled within ±25%.

[0176] The interactive scene that combines virtual and real elements supports stepless scaling from 1:1 to 1:5000, and the geometric accuracy of the 3D model is better than 15 centimeters.

[0177] The collaborative analysis platform supports simultaneous access from no fewer than 15 user terminals, with a data response latency of less than 3 seconds.

[0178] 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 of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A terrain mapping method for unmanned aerial vehicles (UAVs), used in a terrain mapping system for UAVs, characterized in that, The terrain mapping system of the UAV includes the following modules: The data acquisition module is responsible for synchronously collecting high-precision terrain and environmental parameter data of the target area using the UAV platform; The data fusion and correction module is responsible for fusing multi-source data and meteorological information, eliminating errors between sensors, and ensuring that all data are accurately aligned under a unified spatiotemporal reference framework. The multidimensional feature analysis and recognition module extracts and classifies features from various sensor data through multidimensional feature analysis, effectively identifying and separating environmental interference signals from real building deformation signals. The interference cancellation and deformation extraction module performs pattern recognition and quantitative analysis on periodic non-deformation interference, accurately extracting the true deformation characteristics of the building from complex dynamic noise. The real-time analysis module deploys optimized, lightweight data processing algorithms on the drone platform to achieve real-time data processing and feedback, ensuring the rapid provision of high-precision building change information; The visualization module provides multi-dimensional dynamic visualization displays, including building change heatmaps, disturbance distribution maps, and time-series curves; The terrain mapping method for the UAV includes the following steps: Step 1: Use a drone platform to synchronously and accurately collect terrain and environmental parameters of the target area; Step 2: Integrate multi-source data and meteorological information to eliminate errors between sensors and ensure that all data are accurately aligned under a unified spatiotemporal reference framework; Step 3: Through multi-dimensional feature analysis, feature extraction and classification are performed on the data from each sensor to effectively identify and separate environmental interference signals from real terrain deformation signals; Step 4: Perform pattern recognition and quantitative analysis on periodic non-deformation disturbances to accurately extract the true deformation features of the terrain from complex dynamic noise. Step 5: Deploy optimized lightweight data processing algorithms on the drone platform to achieve real-time data processing and feedback, ensuring the rapid provision of high-precision terrain change information in emergency monitoring and disaster early warning scenarios; Step 6 provides a multi-dimensional dynamic visualization display, including terrain change heatmaps, disturbance distribution maps, and time-series curves; Step 3 includes the following steps: Multi-scale wavelet transform is applied to decompose point cloud data and multispectral images, extract spatiotemporal features in different frequency domains, and identify the inherent frequency patterns and anomalous fluctuations of terrain structures. By using pre-labeled terrain feature samples for training, accurate land cover classification and segmentation can be achieved, and by comparing multi-temporal data, fixed land features and temporary environmental elements can be distinguished. The composite terrain change signal is decomposed into independent components from different sources, and seasonal changes, human interference, instrument noise and real geological movements are identified and quantified to achieve blind separation and feature reconstruction of multi-source signals. Semantic annotation is performed on the identified terrain change signals to distinguish different types of terrain changes and assess their development stage and potential risk level. Spatiotemporal correlation analysis was conducted to explore the causal relationship between topographic changes and environmental parameters, and conditional probability inference was used to distinguish between temporary changes induced by the environment and continuous geological movements. By integrating multiple feature classification results, the reliability of deformation analysis is optimized, and a high-confidence separation result map of terrain deformation and environmental disturbance is finally generated.

2. The terrain mapping method for unmanned aerial vehicles according to claim 1, characterized in that, Step 1 includes the following steps: Choose a suitable multi-rotor drone platform and configure an RTK-GPS positioning system to ensure that the positioning accuracy of the flight path reaches the centimeter level; Preset the optimal flight altitude, speed, and overlap rate to meet the requirements of surveying data acquisition; Equipped with a high-precision LiDAR system to collect terrain point cloud data, and by selecting appropriate laser pulse frequency and scanning angle, high-resolution characterization of complex terrain can be achieved. It integrates a multispectral camera to collect surface reflectance data, covering the visible light, near-infrared and short-wave infrared bands. By recording the reflectance differences of different bands, it can accurately identify surface material, vegetation coverage and humidity distribution. Deploy miniature weather station modules to record temperature, humidity, air pressure, wind speed, and wind direction parameters in real time; The design of a data synchronization acquisition framework adopts a unified time base and precise timestamp marking to ensure the synchronization of data from each sensor at millisecond-level accuracy. Based on preset flight missions and real-time data analysis, the adaptive sampling strategy is dynamically adjusted to increase sampling density in complex terrain areas and appropriately reduce sampling frequency in flat areas, so as to balance data quality and processing efficiency and maximize the use of UAV flight time and payload capacity.

3. The terrain mapping method for UAVs according to claim 2, characterized in that, The drone platform has the following parameters: maximum flight altitude of 250 meters, cruising speed of 12-18 meters per second, maximum flight time of no less than 40 minutes, and wind resistance of level 5. The RTK-GPS positioning system on the drone has a horizontal positioning accuracy of better than ±2.5 cm and a vertical positioning accuracy of better than ±5 cm. The LiDAR system has a point cloud acquisition density of no less than 80 points / square meter, a ranging accuracy of better than ±3 cm, and a scanning angle range of 360° horizontal × 60° vertical. The multispectral camera has a spatial resolution of no less than 5 cm / pixel and a spectral resolution covering 5 bands, including blue light, green light, red light, near infrared and short-wave infrared. The measurement accuracy of the mini weather station module is as follows: temperature ±0.5℃, relative humidity ±3%, air pressure ±0.8hPa, wind speed ±0.5 m / s, and wind direction ±8°.

4. The terrain mapping method for UAVs according to claim 1, characterized in that, Step 2 includes the following steps: All sensor data were converted to the local coordinate system, and feature point matching and spatial interpolation methods were used to correct the geometric position offset caused by GPS drift and attitude deviation. The network time protocol is used to eliminate the time delay in data acquisition from each sensor, and combined with the UAV flight trajectory information, the spatiotemporal inconsistencies caused by platform movement are corrected to ensure that all data points have accurate four-dimensional spatiotemporal labels. By verifying control points and comparing overlapping area data, the errors between various sensors are quantified, including position offset, attitude angle error and scale deformation, and high-precision alignment of multi-source data is achieved based on the inverse transformation algorithm. The effects of temperature, humidity and air pressure changes on sensor performance were analyzed, and measurement deviations caused by atmospheric refraction, thermal expansion and humidity changes were dynamically compensated by combining real-time environmental parameters during flight. Statistical analysis and spatial consistency testing methods are used to identify and mark outliers caused by equipment failure, signal interference or sudden environmental changes. At the same time, neighborhood smoothing technology is used to repair missing data and outlier areas. Based on the spatial sampling density and accuracy characteristics of different sensors, an adaptive weight allocation strategy is designed to ensure overall data consistency while preserving high-frequency details, ultimately generating a unified high-precision terrain and environmental parameter dataset.

5. The terrain mapping method for unmanned aerial vehicles according to claim 4, characterized in that, The spatial error threshold for feature point matching is set to less than 5 cm to correct geometric position offsets; the synchronization accuracy of the network time protocol is better than 0.5 milliseconds to ensure the consistency of time stamps for all data points; the distribution density of control point verification is no less than 4 per square kilometer, with a point position accuracy better than ±1.5 cm; the temperature compensation parameter is set to ±0.08 mm / m for each change in temperature; the humidity compensation parameter is set to ±0.05 mm / m for each 10% change in relative humidity; the air pressure compensation parameter is set to ±0.03 mm / m for each 10 hPa change in air pressure; the standard deviation threshold for outlier identification is set to 3σ, and data points exceeding this range are marked as potential anomalies.

6. The terrain mapping method for unmanned aerial vehicles according to claim 1, characterized in that, Step 4 includes the following steps: The temporal terrain data is converted to the frequency domain to extract characteristic frequencies and amplitude information, accurately identify periodic patterns, and quantify various periodic disturbances. Based on the identified periodic disturbances and real-time observation data, a set of state-space equations is constructed, and a recursive optimal estimation algorithm is used to suppress random noise and systematic errors, while ensuring the complete preservation of non-periodic real terrain change signals. The complex terrain change signal is decomposed into a finite number of intrinsic mode functions and singular components. Through modal energy analysis and statistical significance test, the physically meaningful terrain change patterns are extracted, while effectively filtering out random noise. By using differential interferometry and point cloud comparative analysis, the changes in topographic elevation at different spatial scales are quantified, and by combining spatial context information and prior geological knowledge, the differences between local micro-deformations and large-scale geological movements are distinguished. By combining Bayesian inference and Markov random field theory, and utilizing temporal continuity and spatial correlation constraints, we can identify anomalous change regions that are inconsistent with the surrounding environment and historical trends. By integrating blind source separation, sparse representation and low-rank matrix factorization techniques, this method maximizes the suppression of noise and various interferences while preserving the magnitude and morphological integrity of terrain changes, ultimately generating a terrain deformation feature map with a high signal-to-noise ratio.

7. The terrain mapping method for unmanned aerial vehicles according to claim 1, characterized in that, Step 5 includes the following steps: The data processing algorithm is compressed into an edge version with low resource consumption and fast inference speed, reducing the computational complexity by more than 80% while maintaining the core functionality; Design a streaming data processing architecture to decompose the terrain analysis task into parallel subtasks, enabling real-time flow of sensor data acquisition, preprocessing, feature extraction, and deformation analysis; Based on the urgency of data processing and the constraints of computing resources, a multi-granularity analysis framework is constructed to dynamically adjust the algorithm accuracy and computational complexity; The primary analysis and early warning tasks with high real-time requirements are deployed on the drone edge computing platform, while the computationally intensive deep analysis and historical data comparison tasks are offloaded to the cloud server. By combining preset risk thresholds and expert rules, the system automatically generates tiered early warning information and marks the spatial distribution and development trend of high-risk areas. Ensure that critical monitoring results can be transmitted to the ground command center in a timely manner under limited bandwidth conditions.

8. The terrain mapping method for unmanned aerial vehicles according to claim 1, characterized in that, Step 6 includes the following steps: A 3D terrain change heat map rendering engine is constructed, which combines a high-precision digital elevation model with terrain deformation data and adopts an adaptive color mapping and transparency encoding strategy to intuitively display the spatial distribution and intensity level of terrain changes. The identified non-deformation interference sources are mapped to independent layers to achieve differentiated display of different interference types and support multi-dimensional cross-analysis; High temporal resolution deformation history curves are generated for key monitoring points and regions of interest, and time-series data of environmental parameters are overlaid to reveal the temporal coupling relationship between topographic changes and external factors. By combining geological hazard sensitivity analysis and vulnerability assessment, the potential impact range and loss estimates are calculated and visualized, supporting multi-scenario simulation and emergency response decision-making; By deeply integrating terrain monitoring results with high-definition images and 3D models, an interactive scene combining virtual and real elements can be constructed to achieve more intuitive environmental perception and analysis. Design a collaborative analysis and knowledge sharing platform that integrates monitoring data, analysis results, and expert knowledge, providing multi-level information display and collaborative interpretation functions.

Images