Construction method and system for 3D model of smart city

Abstract

The invention relates to the technical field of city model construction, and discloses a smart city 3D model construction method and system, and the method comprises the steps: S1, enabling an unmanned aerial vehicle to fly over a city according to a preset flight path, dynamically calculating a proper flight height according to the height of a building and the pitch angle of a camera in the flight process, and automatically adjusting the flight height; the flight height is dynamically adjusted according to the building height and the camera pitch angle in the flight process of the unmanned aerial vehicle, and the coverage range and the resolution of image acquisition are ensured; in combination with the real-time analysis capability of the edge calculation module, data defects can be quickly identified and laser radar supplementary acquisition can be triggered, so that the data acquisition efficiency and integrity are remarkably improved; and finally, through multi-source data fusion and automatic modeling of the cloud processing platform, a high-precision and high-reality smart city 3D model is constructed, and a reliable three-dimensional digital basis is provided for city management and planning.

Description

Technical Field

[0001] The present application relates to the technical field of city model construction, and in particular to a method and system for constructing a smart city 3D model. Background Art

[0002] With the acceleration of urbanization, smart city construction has become an important direction for promoting the modernization of urban governance and improving the level of public services. Three-dimensional (3D) city models, as the core digital foundation of smart cities, can provide high-precision spatial data support for scenarios such as urban planning, traffic management, and emergency response. However, existing methods for constructing smart city 3D models still face many technical bottlenecks, which restrict their large-scale application and real-time requirements.

[0003] Traditional 3D modeling technology mainly relies on drone aerial photography and lidar scanning, but the following key problems exist in actual applications: the fixed flight altitude of drones is difficult to adapt to the height differences of urban buildings, resulting in redundant collection of low areas and missing data on the tops of high-rise buildings. The image quality relies on manual post-inspection, and defective areas cannot be identified in real time. Repeated re-collection is required, which is time-consuming and labor-intensive; existing methods often turn on the lidar throughout the entire process, and its high energy consumption characteristics shorten the drone's flight life, and redundant point cloud data increases the transmission and storage burden; there is a lack of real-time collaborative processing of images and point cloud data, and alignment errors and blind spots are prone to occur during cloud-based modeling. Summary of the Invention

[0004] In order to monitor the quality of images collected by drones and reduce redundant point cloud data, the present application provides a method and system for constructing a smart city 3D model.

[0005] In the first aspect, the present application provides a method for constructing a 3D model of a smart city, which adopts the following technical solutions: A method and system for constructing a smart city 3D model, the method comprising: S1. Use the drone to fly over the city using a preset flight path. During the flight, the appropriate flight altitude is dynamically calculated and automatically adjusted based on the building height and camera pitch angle. S2, collecting image data of the city through the collection module; S3. Analyze the collected image data through the edge computing module to determine whether to generate instructions to control the operation of the lidar; S4. Supplementary acquisition of defective areas in the image data is performed through the laser radar module to obtain point cloud data; S5. Analyze and model the collected image data and point cloud data through the cloud processing platform to obtain a 3D model of the smart city.

[0006] By adopting the above technical solutions, the image coverage and resolution are improved, the data integrity is enhanced, high-precision 3D modeling is achieved, and urban digital management is supported.

[0007] Optionally, the process of obtaining the appropriate flight altitude includes: The flight altitude h of the UAV is obtained through the above formula analysis and calculation; where H is the building height, is the camera pitch angle.

[0008] By adopting the above technical solutions, the flight safety of the UAV and the acquisition of image details are balanced, the geometric accuracy of the modeling is ensured, and diverse urban scenarios are adapted.

[0009] Optionally, in step S3, the process of analyzing the collected image data by the edge computing module includes: S31. Extract image feature points, analyze the number of feature points in each preset area of the image. If the number of feature points in any area does not meet the standard, generate an instruction to control the operation of the lidar. Otherwise, execute step S2; S32. Evaluate the overall texture clarity of the image. If the overall texture clarity of the image does not meet the standard, generate an instruction to re-collect the image data at this location; S33. Evaluate the texture clarity of each preset area of the image. If the texture clarity of the image in any area does not meet the standard, generate an instruction to control the operation of the lidar.

[0010] By adopting the above technical solutions, local and global data defects are accurately identified, invalid transmission is reduced, and the edge computing efficiency and model reliability are improved.

[0011] Optionally, in step S31, the process of analyzing the number of feature points in each preset area of the image includes: S311. Grayscale the image to locate areas with drastic brightness changes; S312. Identify significant feature points in the image and remove redundant feature points in low-contrast and high-density areas; S313. Divide the image into each preset area equally, count the number of valid feature points in each preset area, compare the number of valid feature points in each preset area with the critical number of feature points in a preset single area. If the number of feature points in any area is less than the critical number of feature points in a preset single area, generate an instruction to control the operation of the lidar, and perform laser scanning on this area through the lidar.

[0012] By adopting the above technical solutions, the recognition accuracy of feature points in each preset area is improved, the spatial continuity of the point cloud is ensured, and modeling blind spots are avoided.

[0013] Optionally, in step S2, the process of evaluating the overall texture clarity of the image includes: The overall texture clarity feature value of the image is obtained by combining the above formulas ; in, is the mean gradient of all pixels in the image, is the average contrast of all pixels in the image, and is the weight coefficient corresponding to the texture clarity eigenvalue influencing factor. The image has M pixels in the horizontal direction and N pixels in the vertical direction. The total number of pixels is indivual, , is the horizontal gradient of the i-th pixel, is the vertical gradient of the i-th pixel, is the gray value of the i-th pixel, is the average grayscale value of the pixels around the i-th pixel; The overall texture of the image is clear feature value With the preset first texture clear feature value Make a comparison; like , then generate an instruction to re-collect image data for the position; Otherwise, execute step S33.

[0014] By adopting the above technical solutions, the overall image quality is quantitatively evaluated, the risk of texture blur in each preset area is reduced, and the authenticity and detail integrity of the model are guaranteed.

[0015] Optionally, in step S33, the process of evaluating the texture clarity of each preset area of the image includes: S331, respectively calculating the mean gradient value and the mean contrast value of all pixels in each preset area; S332, calculating the texture clarity feature value of each preset area; S333, comparing the texture clarity feature value of each preset area with a preset second texture clarity feature value; If there is any preset area whose texture clear feature value is less than or equal to the second texture clear feature value, an instruction to control the operation of the laser radar is generated to perform laser scanning on the area through the laser radar.

[0016] By adopting the above technical solutions, the local image quality is quantitatively evaluated, the energy consumption of redundant scanning is reduced, and the overall consistency of the model is improved.

[0017] Optionally, in step S5, the process of the cloud processing platform analyzing and modeling the collected image data and point cloud data includes: S51. Perform spatial registration on the image data and the point cloud data to eliminate the scale difference and coordinate system deviation between the data, and generate a fused data set with a unified spatial reference; S52. Construct the topological relationship of urban elements, and combine the surface texture features of buildings to realize the automatic recognition and semantic 3D reconstruction of each urban element; S53. Enhance the details of the model, automatically repair the acquisition blind area, verify the rationality of the model structure, and output the urban 3D model.

[0018] By adopting the above technical solutions, multi-modal data is fused for accurate modeling, the structural defects are automatically repaired, and the urban 3D model is output.

[0019] In a second aspect, the present application provides a construction system for a smart city 3D model, adopting the following technical solutions: A construction method and system for a smart city 3D model, the system includes: An unmanned aerial vehicle (UAV), which is used to automatically fly along a preset path, dynamically calculate the flight altitude, carry a collection module, and adjust the attitude in real time to optimize the data collection perspective; A collection module, which is used to collect urban image data and point cloud data of a specified area; An edge computing module, which is used to analyze the image quality in real time and automatically trigger the lidar rescan instruction or the re-acquisition instruction; A communication module, which is used for data transmission between the UAV, the edge device and the cloud device, and supports low-latency interaction of images, point clouds and control instructions; A cloud processing platform, which is used for multi-source data fusion, semantic 3D modeling, model detail enhancement and lightweight output to generate an urban 3D model.

[0020] By adopting the above technical solutions, efficient and high-precision modeling is realized through the cooperation of multiple modules in a closed loop, the data blind area and delay are eliminated, and a real and complete urban 3D model is output.

[0021] Optionally, the collection module includes: A high-resolution camera, which is used to collect multi-angle image data of the city; A lidar module, which is used to collect point cloud data of a specified area.

[0022] In a third aspect, the present application provides a storage medium, adopting the following technical solutions: A storage medium storing a program for the method for constructing a smart city 3D model as described in any one of the above.

[0023] In summary, this application includes at least one of the following beneficial technical effects: (1) The present invention dynamically adjusts the flight altitude of the drone according to the building height and the camera pitch angle during flight to ensure the coverage and resolution of image acquisition; combined with the real-time analysis capability of the edge computing module, it can quickly identify data defects and trigger the laser radar to supplement the acquisition, significantly improving the efficiency and integrity of data acquisition; finally, through the multi-source data fusion and automatic modeling of the cloud processing platform, a high-precision and high-realism 3D model of the smart city is constructed, providing a reliable three-dimensional digital foundation for urban management and planning.

[0024] (2) The present invention uses an edge computing module to perform a hierarchical analysis of the number of image feature points and texture clarity. First, local data loss (such as a single texture area) is detected by the number of feature points in a preset area. Second, the image quality is evaluated by the overall and regional texture clarity to achieve accurate positioning of data defects. This hierarchical verification mechanism can quickly trigger the laser radar to supplement acquisition or image reshooting instructions, reduce invalid data transmission, improve edge computing efficiency, and avoid the generation of modeling blind spots.

[0025] (3) The present invention can significantly improve the accuracy of feature point recognition by grayscale processing and feature point screening (removing low-contrast, high-density redundant points) of the image; by equally dividing the area and counting the number of valid feature points and comparing them with the preset threshold, the feature sparse areas (such as glass curtain walls and shadow areas) can be quickly located, thereby triggering the laser radar scan in a targeted manner, compensating for the shortcomings of optical images in weak texture areas, and ensuring the spatial continuity of point cloud data. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flowchart of the steps of the method for constructing a smart city 3D model disclosed in the present invention. DETAILED DESCRIPTION

[0027] Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings.

[0028] In the description of this specification, the descriptions referring to the terms "certain embodiments", "one embodiment", "some embodiments", "schematic embodiments", "examples", "specific examples", or "some examples" mean that the specific features, structures, materials, or characteristics described in connection with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiments or examples. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0029] An embodiment of the present application discloses a method for constructing a 3D model of a smart city. Referring to Figure 1 , the method includes: S1. Fly over the city by a drone along a preset flight path, and dynamically calculate an appropriate flight height according to the building height and the camera pitch angle during the flight process, and automatically adjust the flight height; S2. Collect image data of the city through a collection module; S3. Analyze the collected image data through an edge computing module to determine whether to generate an instruction for controlling the operation of a lidar; S4. Supplementarily collect the defective areas in the image data through a lidar module to obtain point cloud data; S5. Analyze and model the collected image data and point cloud data through a cloud processing platform to obtain a 3D model of a smart city.

[0030] Through the above technical solution, this embodiment provides a method for constructing a 3D model of a smart city. The method dynamically adjusts the flight height according to the building height and the camera pitch angle during the flight of the drone, ensuring the coverage range and resolution of image collection; combined with the real-time analysis ability of the edge computing module, it can quickly identify data defects and trigger lidar supplementary collection, significantly improving the data collection efficiency and integrity; finally, through the multi-source data fusion and automatic modeling of the cloud processing platform, a high-precision and high-fidelity 3D model of a smart city is constructed, providing a reliable three-dimensional digital foundation for urban management and planning.

[0031] Further, the process of obtaining the appropriate flight height includes: Analyze and calculate the flight height h of the drone through the above formula; wherein, H is the building height, and the building height data of the target area can be obtained through a pre-stored urban GIS database, is the camera pitch angle, which is the angle between the camera optical axis and the horizontal plane, and is obtained by real-time feedback from the drone gimbal control system.

[0032] Through the above technical solution, this embodiment provides a method for analyzing and calculating the flight altitude of a drone. The method calculates the flight altitude (h) based on the quantitative formula of the building height (H) and the camera pitch angle (θ), and can dynamically adapt to the shooting requirements in different scenarios: it can avoid the risk of building occlusion or collision caused by too low flight altitude, and prevent the loss of image details caused by too high altitude, so as to maximize the effective information volume of image data while ensuring safety, and provide geometric accuracy guarantee for subsequent modeling.

[0033] Further, in step S3, the process of analyzing the collected image data by the edge computing module includes: S31. Extract image feature points, analyze the number of feature points in each preset area of the image. If the number of feature points in any area does not meet the standard, generate an instruction to control the operation of the lidar; otherwise, execute step S2. S32. Evaluate the overall texture clarity of the image. If the overall texture clarity of the image does not meet the standard, generate an instruction to re-collect the image data at this position. S33. Evaluate the texture clarity of each preset area of the image. If the texture clarity of the image in any area does not meet the standard, generate an instruction to control the operation of the lidar.

[0034] Through the above technical solution, this embodiment provides a method for preprocessing image data through an edge technology module. The edge computing module performs hierarchical analysis on the number of image feature points and texture clarity: first, detect local data loss (such as areas with single texture) through the number of feature points in the preset area, and secondly, evaluate the image quality through the overall and regional texture clarity to achieve precise positioning of data defects. This hierarchical verification mechanism can quickly trigger the lidar to supplement the collection or image re-shooting instruction, reduce the transmission of invalid data, improve the edge computing efficiency, and avoid the generation of modeling blind spots.

[0035] Further, in step S31, the process of analyzing the number of feature points in each preset area of the image includes: S311. Perform grayscale processing on the image to locate the areas with drastic brightness changes. S312. Identify the significant feature points in the image and remove the redundant feature points in the areas with low contrast and high density. S313. Divide the image into equal parts of preset areas. The preset areas are obtained based on experience. It is generally recommended to divide them into 50*50 pixel windows. Count the number of valid feature points in each preset area. Compare the number of valid feature points in each preset area with the preset critical number of feature points in a single area. If the number of feature points in any area is less than the preset critical number of feature points in a single area, generate an instruction to control the operation of the laser radar, and perform laser scanning on the area through the laser radar.

[0036] Through the above technical solution, this embodiment provides a method for analyzing the number of feature points in each preset area of an image. The method can significantly improve the accuracy of feature point recognition by grayscale processing and feature point screening (removing low-contrast, high-density redundant points) of the image; by equally dividing the area to count the number of valid feature points and comparing them with the preset threshold, feature-sparse areas (such as glass curtain walls and shadow areas) can be quickly located, thereby triggering lidar scanning in a targeted manner, making up for the shortcomings of optical images in weak texture areas, and ensuring the spatial continuity of point cloud data.

[0037] Furthermore, in step S2, the process of evaluating the overall texture clarity of the image includes: The overall texture clarity feature value of the image is obtained by combining the above formulas ; in, is the mean gradient of all pixels in the image, is the average contrast of all pixels in the image, and is the weight coefficient corresponding to the texture clarity eigenvalue influencing factor, which can be obtained through empirical selection and is generally taken as =0.6, =0.4, the image has M pixels in the horizontal direction and N pixels in the vertical direction, and the total number of pixels is The number of M and N can be obtained from the camera specification parameters. , is the horizontal gradient of the i-th pixel, which can be obtained by convolution calculation using the Sobel operator. The vertical gradient of the i-th pixel can be obtained by convolution using the Sobel operator. is the gray value of the i-th pixel point, which can be obtained by calculating the color of the pixel point through the Luminosity method. Generally, it is obtained by the formula Y = 0.299*R + 0.587*G + 0.114*B. R, G, and B are the red, green, and blue channel values (0 - 255) of the pixel, and the weight coefficients (0.299, 0.587, 0.114) are set based on the biological characteristics of the human eye's sensitivity to different colors. is the average gray value of the pixel points around the i-th pixel point, which can be obtained by averaging the gray values of the pixel points around the i-th pixel point. Generally, the number of surrounding pixel points is taken as 8 or 16. The overall texture clarity eigenvalue of the image is compared with the preset first texture clarity eigenvalue The preset first texture clarity eigenvalue can be obtained according to experience. If , then an instruction to re - collect the image data at this position is generated. Conversely, step S33 is executed.

[0038] Through the above technical solution, this embodiment provides a method for evaluating the overall texture clarity of an image. The method calculates the overall texture clarity eigenvalue of the image based on the weighted formula of the pixel gradient mean and the contrast mean , which can quantitatively evaluate the image quality and avoid subjective judgment errors. When the overall texture clarity eigenvalue of the image is lower than the first texture clarity eigenvalue, a reshooting instruction is automatically triggered to ensure that the collected image data meets the texture detail requirements for modeling, and reduce the problem of model surface distortion caused by blurring, overexposure, or underexposure.

[0039] Further, in step S33, the process of evaluating the texture clarity of each preset area of the image includes: S331: Calculate the mean of all pixel gradients and the mean of all pixel contrasts in each preset area respectively; S332: Calculate the texture clarity eigenvalue of each preset area; S333: Compare the texture clarity eigenvalues of each preset area with the preset second texture clarity eigenvalue, and the preset second texture clarity eigenvalue can be preset according to experience; If there is any texture clarity eigenvalue of a preset area less than or equal to the second texture clarity eigenvalue, an instruction to control the lidar to work is generated, and the lidar is used to perform laser scanning on this area.

[0040] Through the above technical solution, this embodiment provides a method for evaluating the texture clarity of each preset area of an image. The method calculates the corresponding texture clarity feature value for each preset area of the image, and can accurately identify locally blurred or low-contrast areas (such as dense vegetation areas and dynamic object interference areas), and trigger lidar scanning in a targeted manner. This local quality monitoring mechanism avoids the overall model effect being affected by poor data quality in individual areas, while optimizing resource allocation, enabling lidar only in defective areas, and reducing energy consumption and data redundancy.

[0041] Furthermore, in step S5, the cloud processing platform performs analysis and modeling on the collected image data and point cloud data, including: S51, spatially registering the image data and the point cloud data to eliminate scale differences and coordinate system deviations between the data pieces, and generating a fused data set with a unified spatial reference; S52: Construct the topological relationship of urban elements and combine it with the surface texture characteristics of buildings to achieve automatic recognition and semantic 3D reconstruction of various urban elements; S53: Enhance model details, automatically repair acquisition blind spots, verify the rationality of the model structure, and output the city 3D model.

[0042] Through the above technical solution, this embodiment provides a method for analyzing and modeling collected image data and point cloud data. The method eliminates the scale and coordinate deviations of image and point cloud data through spatial registration to ensure the accurate fusion of multi-source data; combines the surface texture features and topological relationships of buildings to construct a semantic three-dimensional model, which can automatically identify and reconstruct urban elements (such as roads, vegetation, and buildings); further through detail enhancement and blind spot repair algorithms, fill in the missing areas of the collection (such as the bottom of the building), verify the rationality of the model structure, and finally output a smart city 3D model with high integrity, high realism and conformity to actual spatial relationships to support refined urban management applications.

[0043] The present application also discloses a system for constructing a 3D model of a smart city, the system comprising: Drones are used to automatically fly along preset paths, dynamically calculate flight altitudes, and are equipped with acquisition modules that adjust their attitude in real time to optimize the data collection perspective. Automated flight improves collection efficiency, and dynamic perspective optimization ensures data comprehensiveness, adapting to the needs of complex urban scenarios. The acquisition module is used to collect urban image data and point cloud data of designated areas. The multi-source data complementary acquisition (image + point cloud) eliminates the blind spots of a single sensor and ensures the integrity of model details. The edge computing module is used to analyze image quality in real time and automatically trigger LiDAR rescan or re-sampling instructions. Real-time quality monitoring reduces redundant data and accurately triggers rescan / re-sampling instructions, reducing cloud processing load. A communication module for data transmission between drones, edge devices and cloud components, supporting low-latency interaction of images, point clouds and control instructions. The low-latency transmission ensures real-time interaction, improving the system response speed and multi-terminal collaboration efficiency. A cloud processing platform for multi-source data fusion, semantic 3D modeling, model detail enhancement and lightweight output, generating a 3D model of the city. The multi-source fusion and intelligent modeling enhance details, and the output of a high-precision lightweight model supports efficient urban management applications.

[0044] The acquisition module includes: A high-resolution camera for collecting multi-angle image data of the city, capturing high-definition texture details, enhancing the realism of the model surface. The multi-angle coverage reduces visual blind spots and supports accurate texture mapping. A lidar module for collecting point cloud data in a specified area, accurately obtaining the three-dimensional structure through complex environments, compensating for the geometric defects of optical images, and ensuring the accuracy of the model contour and spatial relationship.

[0045] An embodiment of the present application also discloses a storage medium storing a program of the method for constructing a 3D model of a smart city described in any one of the above.

[0046] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present application.

Claims

1. A method for constructing a 3D model of a smart city, characterized in that, The method includes: S1. Fly over the city by a drone along a preset flight path. During the flight, dynamically calculate an appropriate flight height according to the building height and the camera pitch angle and automatically adjust the flight height; S2. Collect image data of the city through a collection module; S3. Analyze the collected image data through an edge computing module to determine whether to generate an instruction for controlling the lidar to work; S4. Supplementarily collect the defective areas in the image data through the lidar module to obtain point cloud data; S5. Analyze and model the collected image data and point cloud data through a cloud processing platform to obtain a 3D model of a smart city.

2. The method for constructing a 3D model of a smart city according to claim 1, wherein, The process of obtaining the appropriate flight height includes: Analyze and calculate the flight height h of the drone through the above formula; where H is the building height, is the camera pitch angle.

3. The method for constructing a 3D model of a smart city according to claim 1, wherein, In step S3, the process of analyzing the collected image data through the edge computing module includes: S31. Extract image feature points, analyze the number of feature points in each preset area of the image. If the number of feature points in any area does not meet the standard, generate an instruction for controlling the lidar to work. Otherwise, execute step S2; S32. Evaluate the overall texture clarity of the image. If the overall texture clarity of the image does not meet the standard, generate an instruction to re-collect the image data at this location; S33. Evaluate the texture clarity of each preset area of the image. If the texture clarity of any area of the image does not meet the standard, generate an instruction for controlling the lidar to work.

4. The method for constructing a 3D model of a smart city according to claim 3, characterized in that, In step S31, the process of analyzing the number of feature points in each preset area of the image includes: S311. Perform grayscale processing on the image to locate areas with drastic light and dark changes; S312. Identify the significant feature points in the image and remove the redundant feature points in areas with low contrast and high density; S313. Divide the image equally into each preset area, count the number of valid feature points in each preset area, compare the number of valid feature points in each preset area with the critical number of feature points in a preset single area. If the number of feature points in any area is less than the critical number of feature points in a preset single area, generate an instruction for controlling the lidar to work, and perform laser scanning on this area through the lidar.

5. The method for constructing a 3D model of a smart city according to claim 4, wherein, In step S2, the process of evaluating the overall texture clarity of the image includes: The overall texture clarity feature value of the image is obtained by combining the above formulas ; Among them, is the average value of the gradients of all pixels in the image, is the average value of the contrast of all pixels in the image, and are the weight coefficients corresponding to the influence factors of the texture clarity feature values. There are M pixel points in the horizontal direction and N pixel points in the vertical direction of the image, and the total number of all pixels is pieces. , is the horizontal gradient of the i-th pixel point, is the vertical gradient of the i-th pixel point, is the gray value of the i-th pixel point, is the average value of the gray values of the pixel points around the i-th pixel point; The overall texture clarity eigenvalue of the image is compared with a preset first texture clarity eigenvalue for comparison; If , an instruction for re-acquiring image data at this position is generated; Otherwise, execute step S33.

6. The method for constructing a 3D model of a smart city according to claim 5, characterized in that, In step S33, the process of evaluating the texture clarity of each preset area of the image includes: S331. Calculate the average value of all pixel gradients and the average value of all pixel contrasts in each preset area respectively; S332. Calculate the texture clarity feature value of each preset area; S333. Compare the texture clarity feature value of each preset area with the preset second texture clarity feature value; If the texture clarity feature value of any preset area is less than or equal to the second texture clarity feature value, generate an instruction for controlling the lidar to work, and perform laser scanning on this area through the lidar.

7. The construction method of the 3D model of the smart city according to claim 6, characterized in that, In step S5, the process of the cloud processing platform analyzing and modeling the collected image data and point cloud data includes: S51. Perform spatial registration on the image data and the point cloud data to eliminate the scale difference and coordinate system deviation between the data, and generate a fused data set with a unified spatial reference; S52. Construct the topological relationship of urban elements, and combine with the surface texture features of buildings to achieve the automatic recognition and semantic 3D reconstruction of urban elements; S53. Enhance the details of the model, automatically repair the acquisition blind area, verify the rationality of the model structure, and output the urban 3D model.

8. The construction system of the 3D model of the smart city, characterized in that, The system is used to execute the construction method of the smart city 3D model as described in any one of claims 1-7. The system includes: An unmanned aerial vehicle (UAV) for automatically flying along a preset path, dynamically calculating the flight altitude, carrying a collection module, and adjusting the attitude in real time to optimize the data collection perspective; A collection module for collecting urban image data and point cloud data of a specified area; An edge computing module for analyzing the image quality in real time and automatically triggering the lidar rescan instruction or the re-acquisition instruction; A communication module for data transmission between the UAV, edge devices and the cloud, supporting low-latency interaction of images, point clouds and control instructions; A cloud processing platform for multi-source data fusion, semantic 3D modeling, model detail enhancement and lightweight output to generate an urban 3D model.

9. The construction system of the 3D model of the smart city according to claim 8, characterized in that, characterized in that, The collection module includes: A high-resolution camera for collecting urban multi-angle image data; A lidar module for collecting point cloud data of a specified area.

10. A storage medium, characterized in that, A program storing the construction method of the smart city 3D model as described in any one of claims 1-7.

Images