A UAV real-scene 3D data automatic update system

Abstract

The present invention discloses an automatic updating system for real-scene three-dimensional data of an unmanned aerial vehicle, and relates to the technical field of three-dimensional modeling. The system comprises: a data acquisition module, a change detection module, an incremental updating module and a data management module: the data acquisition module plans the parameters of the flight route, height, speed and shooting time of the unmanned aerial vehicle according to the characteristics and requirements of the target area. The present invention uses the change detection technology to automatically and accurately identify the newly added, removed and modified parts in the three-dimensional scene. This capability not only greatly reduces the need for manual intervention, but also ensures the accuracy and timeliness of change detection, and performs incremental updates on the detected changed areas, thus avoiding the huge data processing volume and time-consuming problems brought about by the traditional comprehensive reconstruction method. This targeted updating strategy significantly reduces the consumption of computing resources, so that the system can complete the updating task in a shorter time, thereby improving the overall work efficiency and response speed.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional modeling, and in particular to an automatic updating system for real-scene three-dimensional data of an unmanned aerial vehicle. Background Art

[0002] With the rapid development and widespread application of drone technology, drone real-life 3D data collection has become one of the important means of obtaining geospatial information. The high-precision cameras and lidar sensors carried by drones can quickly capture detailed 3D information of the surface, providing unprecedented data support for urban planning, environmental monitoring, and disaster assessment. However, with the continuous expansion of application scenarios and the growing demand, how to efficiently manage and update these massive 3D data has become an urgent problem to be solved. Therefore, it is particularly important to develop a system that can automatically and efficiently update drone real-life 3D data.

[0003] In the traditional method of updating real-life 3D data of drones, a comprehensive reconstruction strategy is often adopted, that is, regular complete data collection and processing of the target area is carried out to obtain the latest 3D model. Although this method can ensure the comprehensiveness and accuracy of the data, it has significant shortcomings. First, comprehensive reconstruction requires a lot of computing resources and time, especially when processing large-scale 3D data, its efficiency is extremely low. Second, comprehensive reconstruction cannot capture and reflect subtle changes in the 3D scene in a timely manner, resulting in limited timeliness and practicality of the data. In addition, since comprehensive reconstruction involves a large amount of repeated calculations, it also causes a waste of computing resources. In response to these problems, traditional 3D data update technology is obviously difficult to meet the needs of modern applications, and a more intelligent and efficient update method is urgently needed.

[0004] Therefore, developing an automatic update system for drone real-scene three-dimensional data can not only significantly reduce the data processing volume and update time and improve system efficiency, but also ensure the timeliness and accuracy of the data, meeting the high requirements of modern applications for three-dimensional data updates. Summary of the invention

[0005] The purpose of the present invention is to make up for the shortcomings of the prior art and provide an automatic update system for real-life three-dimensional data of unmanned aerial vehicles. It can realize automatic identification and efficient update of newly added, removed and modified parts in three-dimensional scenes by integrating intelligent change detection and incremental update technology. Compared with the traditional comprehensive reconstruction method, the present invention only performs incremental processing on the changed areas, significantly reducing the data processing volume and update time, greatly improving the system efficiency, not only improving the timeliness and accuracy of three-dimensional data, but also optimizing the use of computing resources, reducing energy consumption and costs.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a system for automatically updating real-life three-dimensional data of unmanned aerial vehicles, the system comprising: a data acquisition module, a change detection module, an incremental update module and a data management module:

[0007] The data acquisition module: plans the parameters of the drone's flight route, altitude, speed, and shooting time according to the characteristics and needs of the target area, and sets the resolution of the collected data. The drone takes off according to the planned route, and the high-definition camera and lidar sensor on board start working, transmitting the collected data to the ground station for storage, and preliminarily classifying and sorting the data;

[0008] The change detection module: removes noise from the three-dimensional data, and converts data collected at different times, viewing angles and from different sensors to the same coordinate system, seamlessly splices the registered three-dimensional data sets into a complete three-dimensional model with the help of voxel grid fusion technology, converts the new and old three-dimensional models into a depth image format, and compares the graphical representations of the new and old three-dimensional models using pixel-level difference analysis, thereby identifying potential new, removed and modified areas, and verifies and confirms the initially identified change areas using the set thresholds, and uses the edge smoothing formula to smooth the edges of the detected change areas, and uses the hole filling method to automatically fill the holes in the change areas, and combines manual review to perform quality assessment on the optimized change areas;

[0009] The incremental update module: receives the change detection result output by the change detection module, accurately locates the specific position of the changed area in the original three-dimensional model, extracts the vertices, edges, patches, texture coordinates, and material attribute information of the changed area from the original three-dimensional model according to the positioning result, performs difference analysis between the extracted changed area data and the corresponding part of the original model, formulates an intelligent fusion strategy according to the difference analysis result, fuses the changed area data with the original model according to the fusion strategy, performs texture mapping and optimizes lighting rendering and performance optimization on the fused model, and performs quality evaluation on the updated model;

[0010] The data management module verifies the elements and details of the updated 3D model, performs accuracy and consistency checks, compares the old and new models to ensure that the changed areas are updated accurately and the internal data structure of the model is consistent, organizes the updated 3D model data, encrypts it according to standard requirements and securely transmits it to the cloud service platform, publishes data on the platform and provides an API interface, sets an automatic backup plan, and regularly backs up data to storage media.

[0011] Furthermore, the specific steps of splicing the three-dimensional model by voxel grid fusion technology in the change detection module are:

[0012] (1) Perform voxel grid division on the registered 3D data set to divide the 3D space into multiple small cubes;

[0013] (2) Fusing the overlapping areas of adjacent data sets and merging the voxel values ​​in different data sets;

[0014] (3) To solve the gap problem, an interpolation method is used to generate new 3D points in the gap area.

[0015] (4) Simplify, smooth and denoise the fused voxel grid;

[0016] (5) Generate a complete 3D model using 3D modeling software based on the optimized voxel grid data;

[0017] (6) Perform a quality assessment on the generated 3D model. If there are any problems, return to the above steps to make corrections until the quality standards are met.

[0018] Furthermore, the change detection module uses a pixel-level difference analysis method to compare the new and old 3D models to identify the changed areas:

[0019] (1) Convert the new and old 3D models into grayscale representations and preprocess the image representations of the new and old 3D models;

[0020] (2) Use the pixel difference calculation formula to compare the preprocessed new and old images pixel by pixel and calculate the difference value;

[0021] (3) Map the difference values ​​back to the image space to generate a difference map;

[0022] (4) Setting thresholds based on the statistical characteristics of the difference map, extracting potential change regions, and performing morphological operations and connectivity analysis on the change regions;

[0023] (5) Extract the features of the changed area and determine the type of change through manual review;

[0024] (6) Output the detected changed areas and their types in the form of images, lists, or reports.

[0025] Furthermore, the change detection module uses a pixel difference calculation formula to calculate the pixel difference value. Assume that the old image is at coordinates The pixel value at , the new image is at coordinates The pixel value at , define the difference value for: ,in, and is the weight coefficient, p and q are the exponential parameters, It is context information.

[0026] Furthermore, the change detection module extracts the change area by setting the difference value threshold, and the difference value of all pixels in the difference map is , calculate the mean and standard deviation respectively, the mean calculation formula is: , where N is the total number of pixels in the difference map, and the standard deviation is calculated as: , where N is the total number of pixels in the difference map, μ is the mean, is the coordinate According to the statistical characteristics of the difference map and prior knowledge, a difference threshold T is set for each pixel in the difference map. , compare their difference values The pixel point is marked as a change point according to the relationship with the set threshold T, and the pixel points marked as change points are regionally connected to form potential new addition, removal and modification areas.

[0027] Furthermore, the change detection module uses an edge smoothing formula to smooth the edge of the change area. Suppose the coordinates of a point in the change area are , whose original three-dimensional coordinates are , the point set in the neighborhood is ,in , define the smoothed coordinate values for: , where k is the smoothing strength coefficient, is the weight coefficient.

[0028] Furthermore, the change detection module uses a hole filling formula to automatically fill the hole. Assume that the three-dimensional point cloud with holes in the change area is , where the point set in the hole area is marked as , is the index set of hole points, and the filled point cloud is , for the points in the hole , its filled coordinate value for: ,in is the index set of non-hole points, is the weight coefficient, , , is a parameter to adjust the weight.

[0029] Furthermore, the specific steps of performing difference analysis between the extracted changed area data and the original model in the incremental update module are:

[0030] (1) Align the extracted change area data with the original 3D model in the same coordinate system;

[0031] (2) Locate the corresponding part in the original 3D model based on the boundary information and identification information of the changed area;

[0032] (3) Based on the geometric similarity criterion, find the corresponding element in the original model for each data element in the changed area;

[0033] (4) For successfully matched data element pairs, calculate the differences in geometric position and attributes;

[0034] (5) record and classify these differences as significant, minor, or insignificant;

[0035] (6) Based on the difference calculation results, evaluate the impact of the changed area on the overall structure and visual effect of the original model;

[0036] (7) Generate a difference analysis report based on the evaluation results.

[0037] Furthermore, the incremental update module calculates the difference in geometric position and attribute of the successfully matched data element pairs through a formula. Suppose the data elements in the original three-dimensional model are , the corresponding data element in the change area is , for vertex coordinate differences, let The vertex coordinates are The vertex coordinates are , the vertex coordinate difference is: ,in , , for the edge length difference, let The edge length is The edge length is , edge length difference , for the patch normal difference, let The patch normal vector is , The patch normal vector is , the patch normal vector difference is: ,set up The material properties are , The material properties are , the material difference value is: ,in and is the value of each material attribute, n is the total number of material attributes, and for the texture coordinate difference, let The texture coordinates are The texture coordinates are , the texture coordinate difference value is: , for color difference, let The color is The color is , the color difference value is: .

[0038] Furthermore, the specific steps of formulating the fusion strategy in the incremental update module are:

[0039] (1) Collect and integrate all the difference information obtained in the difference analysis step;

[0040] (2) Formulate data replacement rules for significantly changed data and determine the elements to be replaced and their priorities;

[0041] (3) Design an insertion method for the new data and determine the best insertion position;

[0042] (4) Plan the data deletion process for the data elements that need to be deleted, and identify and mark the elements that need to be deleted;

[0043] (5) When formulating integration strategies, consider conflict situations and design conflict resolution and optimization solutions;

[0044] (6) Evaluate the formulated fusion strategy, check its effectiveness and feasibility through simulating the update process, and optimize it.

[0045] Compared with the prior art, the automatic updating system for real-scene 3D data of unmanned aerial vehicles has the following beneficial effects:

[0046] 1. The present invention uses change detection technology to automatically and accurately identify newly added, removed and modified parts in a three-dimensional scene. This capability not only greatly reduces the need for manual intervention, ensures the accuracy and timeliness of change detection, but also performs incremental updates on detected changed areas, avoiding the huge data processing volume and time-consuming problems brought about by traditional comprehensive reconstruction methods. This targeted update strategy significantly reduces the consumption of computing resources, allowing the system to complete update tasks in a shorter time, thereby improving overall work efficiency and response speed.

[0047] 2. The present invention ensures the real-time and accuracy of the data by updating only the changed areas, providing more reliable information support for decision makers. At the same time, since unnecessary comprehensive reconstruction processes are reduced, data updates can be performed more frequently, thereby capturing more dynamic change information, providing more comprehensive and detailed data support for urban planning, environmental protection, and emergency response. This efficient and accurate data update method will undoubtedly promote the application and development of drone real-life 3D data in more fields.

[0048] Other advantages, objectives and features of the present invention will be set forth in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0050] Figure 1 It is a flow chart of an automatic updating system of real-scene 3D data of UAV;

[0051] Figure 2 is the flow chart of the change detection module;

[0052] Figure 3 Flowchart for incremental update module. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] Embodiment 1:

[0055] This embodiment provides an automatic update system for real-scene three-dimensional data of unmanned aerial vehicles, including four major modules: data collection, change detection, incremental update and data management. The unmanned aerial vehicle is equipped with a high-definition camera and a laser radar to automatically collect three-dimensional data of the target area and monitor the quality in real time. The change detection module uses advanced algorithms to identify the differences between new and old models and automatically processes the changed areas. The incremental update module accurately locates the changes and intelligently integrates the new and old data to ensure accurate model updates. The data management module ensures data security and provides cloud services and API interfaces. The entire system realizes the automation, efficient updating and management of three-dimensional data.

[0056] In the specific implementation, the data acquisition module plans the optimal flight route according to the shape, size and obstacle distribution of the target area, sets the appropriate flight altitude and speed in combination with ground features and safety requirements, selects the best shooting time in combination with lighting conditions and weather forecasts, sets the shooting resolution and image overlap rate of the high-definition camera according to application requirements, and ensures that the high-definition camera and lidar sensor carried by the drone are in good working condition and the parameters match. After the drone takes off according to the planned parameters, the high-definition camera continuously captures high-definition images of the target area, and the lidar synchronously acquires point cloud data. At the same time, the flight status and data acquisition quality are monitored in real time through sensors, and the collected data is transmitted to the ground station in real time for storage, and is preliminarily classified and sorted to facilitate subsequent processing.

[0057] The change detection module first removes the noise in the three-dimensional data and unifies the data from different times, perspectives and sensors into the same coordinate system. It uses the voxel grid fusion technology to splice the three-dimensional model, and divides the registered three-dimensional data set into voxel grids, dividing the three-dimensional space into multiple small cubes. Each voxel contains a certain number of three-dimensional points. The overlapping areas of adjacent data sets are fused, and the voxel values ​​in different data sets are merged. For the gap problem, the interpolation method is used to generate new three-dimensional points in the gap area to achieve seamless connection. The fused voxel grid is simplified, smoothed, and denoised without destroying the topological structure and important features. Based on the optimized voxel grid data, a complete three-dimensional model is generated through three-dimensional modeling software. The quality of the generated three-dimensional model is evaluated. If there is a problem, return to the above steps to correct it until the quality standard is met.

[0058] The new and old 3D models are converted into grayscale images and preprocessed. The pixel difference calculation formula is used to compare the new and old images pixel by pixel after preprocessing to calculate the difference value. The old image is assumed to be at the coordinate The pixel value at , after preprocessing , the new image is at coordinates The pixel value at , after preprocessing , define the difference value for: ,in, and is the weight coefficient, which is adjusted according to the actual situation to balance the contribution of different factors to the difference value; p and q are exponential parameters, which adjust the difference value to the absolute difference of pixel values ​​and some specific context information. It is context information. The difference value is mapped back to the image space to generate a difference map, which reflects the difference between the new and old images. The threshold is set according to the statistical characteristics of the difference map, and the change area is extracted through the set difference value threshold. For the difference value of all pixels in the difference map, , calculate the mean and standard deviation respectively, the mean calculation formula is: , where N is the total number of pixels in the difference map, and the standard deviation is calculated as: According to the statistical characteristics of the difference map and prior knowledge, a difference threshold T is set for each pixel in the difference map. , compare their difference values If the pixel point is related to the set threshold T, the pixel point is marked as a change point, and regional connections are performed on the pixel points marked as change points to form potential new additions, removals, and modifications. The features of the change areas are extracted, and the change type is determined by manual review. The detected change areas and their types are output in the form of images, lists, and reports.

[0059] For the detected change area, use the edge smoothing formula to smooth its edges. Suppose the coordinates of a point in the change area are , whose original three-dimensional coordinates are , the point set in the neighborhood is ,in , define the smoothed coordinate values for: , where k is the smoothing strength coefficient, is the weight coefficient, according to the point The distance to point p is determined by the color similarity and normal vector similarity factors. The calculation formula is: ,in Yes With point The Euclidean distance of is the color similarity function, is the normal vector similarity function. The holes in the changing area are automatically filled using the hole filling method. The hole filling formula is used to automatically fill the holes in the changing area. Suppose the 3D point cloud with holes in the changing area is , where the point set in the hole area is marked as , is the index set of hole points, and the filled point cloud is , for the points in the hole , its filled coordinate value for: ,in is the index set of non-hole points, is the weight coefficient, and the calculation formula is: , Yes and Point The Euclidean distance between is the color similarity, measured by the distance in color space, is the normal vector similarity, which is determined by calculating the angle between normal vectors. , , It is to adjust the weight parameters and conduct quality assessment on the changed areas after optimization by combining manual review.

[0060] The incremental update module receives the output from the change detection module, accurately locates the specific position of the changed area in the original 3D model, extracts the vertices, edges, patches, texture coordinates, and material attribute information of the changed area from the original 3D model based on the positioning result, performs difference analysis on the extracted changed area data and the corresponding part of the original model, aligns the extracted changed area data with the original 3D model in the same coordinate system, locates the corresponding part in the original 3D model based on the boundary information and identification information of the changed area, searches for the corresponding element in the original model for each data element in the changed area based on the geometric similarity criterion, calculates the difference in geometric position and attribute for the successfully matched data element pairs, and assumes that the data element in the original 3D model is , the corresponding data element in the change area is .

[0061] For vertex coordinate differences, let The vertex coordinates are The vertex coordinates are , the vertex coordinate difference is: ,in , , for the edge length difference, let The edge length is The edge length is , edge length difference , for the patch normal difference, let The patch normal vector is , The patch normal vector is , the patch normal vector difference is: ,set up The material properties are , The material properties are , the material difference value is: ,in and is the value of each material attribute, n is the total number of material attributes, and for the texture coordinate difference, let The texture coordinates are The texture coordinates are , the texture coordinate difference value is: , for color difference, let The color is The color is , the color difference value is: , record and classify these differences into significant, minor, and insignificant, and based on the difference calculation results, evaluate the impact of the changed area on the overall structure and visual effect of the original model, and generate a difference analysis report based on the evaluation results.

[0062] According to the results of the difference analysis, formulate an intelligent fusion strategy to collect and integrate all the difference information obtained in the difference analysis step, formulate data replacement rules for significantly changed data, determine the elements to be replaced and their priorities, design an insertion method for the new data, determine the best insertion position, plan the data deletion processing process for the data elements that need to be deleted, identify and mark the elements that need to be deleted, and when formulating the fusion strategy, consider the conflict situation to design conflict resolution and optimization solutions, evaluate the formulated fusion strategy, check its effectiveness and feasibility through simulation of the update process, and optimize it, fuse the data of the changed area with the original model according to the fusion strategy, perform texture mapping and optimize lighting rendering and performance optimization on the fused model, and conduct quality evaluation on the updated model

[0063] After receiving the 3D model data transmitted by the incremental update module, the data management module starts the automated verification program to check the data integrity and accuracy, compares the old and new models to confirm the changed areas and the consistency of the internal data structure, organizes and encrypts the verified data, securely transmits it to the cloud service platform and sets up the publishing process and access permissions. At the same time, it formulates an automatic backup plan, regularly backs up the data on the cloud platform, verifies the backup data and conducts recovery drills, and sets up recovery plans to ensure data security and recoverability.

[0064] Through the above implementation steps, this UAV real-scene 3D data automatic update system can efficiently and accurately realize the automatic collection, change detection and real-time update of 3D data of the target area, provide reliable 3D data support for applications in urban planning, environmental monitoring, and disaster assessment, and help improve the scientificity and accuracy of decision-making.

[0065] Embodiment 2:

[0066] This embodiment introduces in detail a system for automatically updating real-life 3D data of unmanned aerial vehicles. The system achieves efficient recognition and accurate updating of 3D scene changes through highly integrated data acquisition, intelligent change detection, precise incremental updates, and secure data management modules, greatly reduces data processing volume and update time, and significantly improves system efficiency.

[0067] The data acquisition module accurately plans the flight route, altitude, speed and shooting time parameters of the drone according to the characteristics and needs of the target area, and sets the resolution and overlap rate of data acquisition. The drone is equipped with a high-definition camera and a lidar sensor, and flies according to the planned route, while taking high-definition images and acquiring point cloud data. The module monitors the flight status and data acquisition quality in real time to ensure the accuracy and completeness of the data. The collected data is transmitted to the ground station for storage and preliminary classification.

[0068] The intelligent change detection module first removes noise from the three-dimensional data and converts data collected at different times, perspectives and sensors into the same coordinate system. It uses voxel grid fusion technology to seamlessly splice the aligned three-dimensional data sets into a complete three-dimensional model. Then, the module converts the new and old three-dimensional models into depth image format and uses pixel-level difference analysis to compare them and identify potential change areas.

[0069] Voxel grid fusion: By performing voxel grid division on the registered 3D data set, the 3D space is divided into multiple small cubes, and data fusion, gap interpolation, grid simplification and smoothing denoising are performed to finally generate a high-quality 3D model.

[0070] Pixel-level difference analysis: Convert the old and new 3D models into grayscale images, calculate the difference values ​​and generate a difference map. Extract the changed area based on the statistical characteristics of the difference map and the set threshold. After the changed area is optimized through edge smoothing and hole filling methods, the change type is confirmed in combination with manual review.

[0071] The incremental update module receives the change detection results output by the change detection module, accurately locates the changed area in the original 3D model, extracts the vertices, edges, patches, texture coordinates and material attribute information of the changed area from the original model, performs difference analysis with the changed area data, and formulates an intelligent fusion strategy based on the difference analysis results to fuse the changed area data with the original model, while performing texture mapping, lighting rendering and performance optimization. Finally, a quality assessment is performed on the updated model to ensure that the changed area is updated accurately and does not affect the overall structure and visual effect of the model.

[0072] Difference Analysis: Compares the changed area data with the original model data, calculates the differences in geometry and attributes, and classifies the differences as significant, minor, and insignificant.

[0073] Fusion strategy: formulate replacement rules for significantly changed data, design insertion methods for new data, plan the process of deleting data, and design conflict resolution and optimization solutions.

[0074] The data management module is responsible for verifying the accuracy and consistency of the updated 3D model, ensuring that the changed areas are updated accurately and the internal data structure of the model is consistent, organizing and encrypting the updated 3D model data and securely transmitting it to the cloud service platform, publishing the data on the platform and providing an API interface, setting an automatic backup plan, regularly backing up data to storage media, and verifying the integrity and availability of the backup data.

[0075] Through the above embodiments, the intelligent change detection and incremental update system of drone real-scene three-dimensional data significantly improves the efficiency and accuracy of three-dimensional scene updates, and provides strong technical support for the fields of smart cities, disaster monitoring, and environmental protection.

[0076] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A UAV real-scene 3D data automatic update system, characterized in that: The system includes: data acquisition module, change detection module, incremental update module and data management module: The data acquisition module: plans the parameters of the drone's flight route, altitude, speed, and shooting time according to the characteristics and needs of the target area, and sets the resolution of the collected data. The drone takes off according to the planned route, and the high-definition camera and lidar sensor on board start working, transmitting the collected data to the ground station for storage, and preliminarily classifying and sorting the data; The change detection module: removes noise from the three-dimensional data, and converts data collected at different times, viewing angles and from different sensors to the same coordinate system, seamlessly splices the registered three-dimensional data sets into a complete three-dimensional model with the help of voxel grid fusion technology, converts the new and old three-dimensional models into a depth image format, and compares the graphical representations of the new and old three-dimensional models using pixel-level difference analysis, thereby identifying potential new, removed and modified areas, and verifies and confirms the initially identified change areas using the set thresholds, and uses the edge smoothing formula to smooth the edges of the detected change areas, and uses the hole filling method to automatically fill the holes in the change areas, and combines manual review to perform quality assessment on the optimized change areas; The incremental update module: receives the change detection result output by the change detection module, accurately locates the specific position of the changed area in the original three-dimensional model, extracts the vertices, edges, patches, texture coordinates, and material attribute information of the changed area from the original three-dimensional model according to the positioning result, performs difference analysis between the extracted changed area data and the corresponding part of the original model, formulates an intelligent fusion strategy according to the difference analysis result, fuses the changed area data with the original model according to the fusion strategy, performs texture mapping and optimizes lighting rendering and performance optimization on the fused model, and performs quality evaluation on the updated model; The data management module verifies the elements and details of the updated 3D model, performs accuracy and consistency checks, compares the old and new models to ensure that the changed areas are updated accurately and the internal data structure of the model is consistent, organizes the updated 3D model data, encrypts it according to standard requirements and securely transmits it to the cloud service platform, publishes data on the platform and provides an API interface, sets an automatic backup plan, and regularly backs up data to storage media.

2. The automatic update system of real-scene 3D data of unmanned aerial vehicle according to claim 1, characterized in that: The specific steps of splicing the 3D model by voxel grid fusion technology in the change detection module are as follows: (1) Perform voxel grid division on the registered 3D data set to divide the 3D space into multiple small cubes; (2) Fusing the overlapping areas of adjacent data sets and merging the voxel values ​​in different data sets; (3) To solve the gap problem, an interpolation method is used to generate new 3D points in the gap area. (4) Simplify, smooth and denoise the fused voxel grid; (5) Generate a complete 3D model using 3D modeling software based on the optimized voxel grid data; (6) Perform a quality assessment on the generated 3D model. If there are any problems, return to the above steps to make corrections until the quality standards are met.

3. The automatic updating system of real-scene 3D data of unmanned aerial vehicle according to claim 1, characterized in that: The specific steps of using pixel-level difference analysis method to compare the new and old 3D models to identify the changed areas in the change detection module are as follows: (1) Convert the new and old 3D models into grayscale representations and preprocess the image representations of the new and old 3D models; (2) Use the pixel difference calculation formula to compare the preprocessed new and old images pixel by pixel and calculate the difference value; (3) Map the difference values ​​back to the image space to generate a difference map; (4) Setting thresholds based on the statistical characteristics of the difference map, extracting potential change regions, and performing morphological operations and connectivity analysis on the change regions; (5) Extract the features of the changed area and determine the type of change through manual review; (6) Output the detected changed areas and their types in the form of images, lists, or reports.

4. The automatic updating system of real-scene 3D data of unmanned aerial vehicle according to claim 3, characterized in that: The change detection module uses the pixel difference calculation formula to calculate the pixel difference value, assuming that the old image is at the coordinate The pixel value at , the new image is at coordinates The pixel value at , define the difference value for: ,in, and is the weight coefficient, p and q are the exponential parameters, It is context information.

5. The automatic updating system of real-scene 3D data of unmanned aerial vehicle according to claim 3, characterized in that: The change detection module extracts the change area by setting the difference value threshold, and the difference value of all pixels in the difference map is , calculate the mean and standard deviation respectively, the mean calculation formula is: , where N is the total number of pixels in the difference map, and the standard deviation is calculated as: , where N is the total number of pixels in the difference map, μ is the mean, is the coordinate According to the statistical characteristics of the difference map and prior knowledge, a difference threshold T is set for each pixel in the difference map. , compare their difference values The pixel point is marked as a change point according to the relationship with the set threshold T, and the pixel points marked as change points are regionally connected to form potential new addition, removal and modification areas.

6. The automatic update system of real-scene 3D data of unmanned aerial vehicle according to claim 1, characterized in that: In the change detection module, the edge of the change area is smoothed by the edge smoothing formula. Suppose the coordinates of a point in the change area are , whose original three-dimensional coordinates are , the point set in the neighborhood is ,in , define the smoothed coordinate values for: , where k is the smoothing strength coefficient, is the weight coefficient.

7. The automatic update system of real-scene 3D data of unmanned aerial vehicle according to claim 1, characterized in that: The change detection module uses the hole filling formula to automatically fill the hole. Assume that the three-dimensional point cloud with holes in the change area is , where the point set in the hole area is marked as , is the index set of hole points, and the filled point cloud is , for the points in the hole , its filled coordinate value for: ,in is the index set of non-hole points, is the weight coefficient, , , is a parameter to adjust the weight.

8. The automatic update system of real-scene 3D data of unmanned aerial vehicle according to claim 1, characterized in that: The specific steps of performing difference analysis between the extracted changed area data and the original model in the incremental update module are: (1) Align the extracted change area data with the original 3D model in the same coordinate system; (2) Locate the corresponding part in the original 3D model based on the boundary information and identification information of the changed area; (3) Based on the geometric similarity criterion, find the corresponding element in the original model for each data element in the changed area; (4) For successfully matched data element pairs, calculate the differences in geometric position and attributes; (5) record and classify these differences as significant, minor, or insignificant; (6) Based on the difference calculation results, evaluate the impact of the changed area on the overall structure and visual effect of the original model; (7) Generate a difference analysis report based on the evaluation results.

9. The automatic update system of real-scene 3D data of unmanned aerial vehicle according to claim 8, characterized in that: The incremental update module calculates the differences in geometric position and attributes of the successfully matched data element pairs through a formula. Suppose the data elements in the original three-dimensional model are , the corresponding data element in the change area is , for vertex coordinate differences, let The vertex coordinates are The vertex coordinates are , the vertex coordinate difference is: ,in , , for the edge length difference, let The edge length is The edge length is , edge length difference , for the patch normal difference, let The patch normal vector is , The patch normal vector is , the patch normal vector difference is: ,set up The material properties are , The material properties are , the material difference value is: ,in and is the value of each material attribute, n is the total number of material attributes, and for the texture coordinate difference, let The texture coordinates are The texture coordinates are , the texture coordinate difference value is: , for color difference, let The color is The color is , the color difference value is: .

10. The automatic update system of real-scene 3D data of unmanned aerial vehicle according to claim 1, characterized in that: The specific steps of formulating the fusion strategy in the incremental update module are as follows: (1) Collect and integrate all the difference information obtained in the difference analysis step; (2) Formulate data replacement rules for significantly changed data and determine the elements to be replaced and their priorities; (3) Design an insertion method for the new data and determine the best insertion position; (4) Plan the data deletion process for the data elements that need to be deleted, and identify and mark the elements that need to be deleted; (5) When formulating integration strategies, consider conflict situations and design conflict resolution and optimization solutions; (6) Evaluate the formulated fusion strategy, check its effectiveness and feasibility through simulating the update process, and optimize it.

Images