Method and system for dynamically updating and calibrating live-action three-dimensional model based on orthoimage

Abstract

The invention provides a live-action three-dimensional model dynamic updating and calibrating method and system based on orthoimages, and the method comprises the steps: obtaining a plurality of stages of orthoimages, a plurality of stages of thermal infrared images and land surface temperature data of a target region, the target region comprising a city expansion region; based on the time sequence difference between the multi-stage orthoimage and the multi-stage thermal infrared image, extracting a land cover change area from the urban expansion area; according to the surface temperature data, identifying a local thermal anomaly region from the urban expansion region; based on the surface coverage change area and the local thermal anomaly area, in combination with a pre-trained space-time transformation model, outputting change probability information representing a surface feature change trend; and taking the change probability information as a driving signal, adjusting the Gaussian sputtering model, and updating and calibrating the geometric structure of the live-action three-dimensional model in the latest period according to the adjusted Gaussian sputtering model. According to the method, the synchronous characterization precision of the three-dimensional model of the urban expansion area on the surface heat island effect and the building change is improved.

Description

Technical Field

[0001] The present application relates to the technical field of orthophotos, and in particular to a method and system for dynamically updating and calibrating a real-scene three-dimensional model based on orthophotos. Background Art

[0002] With the rapid development of cities, the layout of buildings and the thermal environment on the ground are changing dynamically, necessitating the development of realistic models that simultaneously reflect both three-dimensional geometry and thermal properties. Existing technologies struggle to coordinate the updating of architectural changes and the urban heat island effect, resulting in temporal and spatial deviations between model data and the real world, hindering the refined management of smart cities.

[0003] The current mainstream approach uses periodic aerial photogrammetry combined with oblique photography technology to generate a 3D point cloud through multi-view image matching. Time-series point cloud comparison is then used to detect building changes. Furthermore, the surface temperature distribution is inverted from thermal infrared imagery, and data interpolation methods are used to map the temperature field onto the 3D model surface, enabling static thermal attribute annotation.

[0004] This solution relies on regular aerial data collection, and the update cycle is subject to flight operation arrangements; the temperature field mapping uses a static interpolation algorithm, which fails to reflect the dynamic impact of building geometry changes on thermal radiation; point cloud comparison only detects geometric changes and lacks quantitative analysis of the coupling relationship between building form and thermal environment, resulting in a disconnect between model updates and actual physical processes. Summary of the Invention

[0005] The present application provides a method and system for dynamically updating and calibrating a real-scene three-dimensional model based on orthophotos, which is used to solve the problem of low accuracy in the synchronous representation of surface heat island effects and building changes in three-dimensional models of urban expansion areas in the existing technology.

[0006] In a first aspect, the present application provides a method for dynamically updating and calibrating a real-scene three-dimensional model based on orthophotos, comprising: Acquiring multi-period orthophotos, multi-period thermal infrared images, and surface temperature data of a target area, wherein the target area includes an urban expansion area; Extracting a land cover change area within the urban expansion area based on a temporal difference between the multiple orthophoto images and the multiple thermal infrared images; identifying a local thermal anomaly region within the urban expansion region based on the land surface temperature data; Based on the surface cover change area and the local thermal anomaly area, combined with a pre-trained spatiotemporal transformation model, outputting change probability information representing the change trend of the ground feature; The change probability information is used as a driving signal to adjust the Gaussian sputtering model, and the geometric structure of the latest real-scene three-dimensional model is updated and calibrated according to the adjusted Gaussian sputtering model.

[0007] Optionally, the adjusting the Gaussian sputtering model using the change probability information as a driving signal, and updating and calibrating the geometric structure of the latest real-scene three-dimensional model according to the adjusted Gaussian sputtering model, includes: Initializing control parameters of a Gaussian sputtering model, wherein the control parameters include a particle diffusion rate and a particle distribution density; Dynamically adjusting the particle diffusion rate and the particle distribution density according to the numerical value of the change probability information; The geometric structure of the latest realistic 3D model is updated and calibrated based on the adjusted particle diffusion rate and the adjusted particle distribution density.

[0008] Optionally, updating and calibrating the geometric structure of the latest real-scene three-dimensional model based on the adjusted particle diffusion rate and the adjusted particle distribution density includes: Based on the adjusted particle diffusion rate and the adjusted particle distribution density, a Gaussian sputtering process is performed to make the particles diffuse in three-dimensional space; During the diffusion process, the trajectory of particles is recorded in real time; Based on the motion trajectory, determining the spatial coordinates at which the particle finally stops; Mapping the spatial coordinates to a surface area of ​​the latest real-scene three-dimensional model, and obtaining a particle deposition density distribution by accumulating the number of particles per unit area in the surface area; The geometric structure of the latest real-scene three-dimensional model is updated and calibrated according to the particle deposition density distribution.

[0009] Optionally, updating and calibrating the geometric structure of the latest real-scene three-dimensional model according to the particle deposition density distribution includes: Dividing the surface area of ​​the real-scene three-dimensional model into a plurality of grid units, and counting the particle deposition density value in each grid unit according to the particle deposition density distribution; Determining, based on the particle deposition density value, a density threshold interval to which each of the grid cells belongs, the density threshold interval being a first interval in which the particle deposition density value is less than a first preset threshold value, a second interval in which the particle deposition density value is between the first preset threshold value and the second preset threshold value, or a third interval in which the particle deposition density value is greater than the second preset threshold value, and the first preset threshold value is less than the second preset threshold value; Based on the processing method corresponding to the density threshold interval, differentiated geometric operations are performed on different grid cells, and the geometric structure of the real-scene three-dimensional model is reconstructed based on the operation results of all grid cells.

[0010] Optionally, extracting the surface cover change area within the urban expansion area based on the temporal difference between the multiple-period orthophoto images and the multiple-period thermal infrared images includes: Calculating the time series feature difference of pixels at the same spatial position of the multi-period orthophoto images and the multi-period thermal infrared images to obtain the pixel feature difference value of each period; Selecting a spatial region in the urban expansion area where the pixel feature difference value exceeds a preset natural variation threshold for two consecutive periods; The area in the spatial region that meets the geometric features of the newly built buildings or the vegetation cover attenuation features is determined as the surface cover change area.

[0011] Optionally, identifying a local thermal anomaly area in the urban expansion area based on the surface temperature data includes: Converting the surface temperature data into thermophysical distribution information, wherein the thermophysical distribution information includes a thermal radiation intensity value for each spatial unit; Setting a dynamic temperature anomaly determination boundary based on a historical thermal environment baseline of the urban expansion area; Target spatial units having thermal radiation intensity values ​​greater than the dynamic temperature anomaly determination boundary are screened, and at least three spatially continuous target spatial units are combined into a local thermal anomaly area.

[0012] Optionally, the outputting of change probability information representing the change trend of land features based on the land cover change area and the local thermal anomaly area in combination with a pre-trained spatiotemporal transformation model includes: Calculating the overlapping area ratio between the surface cover change region and the local thermal anomaly region, and calculating the spatial correlation intensity coefficient based on the overlapping area ratio; splicing the spatial correlation intensity coefficient with the texture features of the multi-phase orthophotos and the thermal radiation gradient features of the multi-phase thermal infrared images to form an input tensor; The input tensor is input into the spatiotemporal transformation model, and change probability information is output.

[0013] In a second aspect, the present application provides a system for dynamically updating and calibrating a real-scene three-dimensional model based on orthophotos, comprising: an acquisition module, configured to acquire multi-period orthophoto images, multi-period thermal infrared images, and surface temperature data of a target area, wherein the target area includes an urban expansion area; An extraction module, configured to extract a land cover change area within the urban expansion area based on a temporal difference between the multiple orthophoto images and the multiple thermal infrared images; an identification module, configured to identify a local thermal anomaly area in the urban expansion area based on the surface temperature data; An output module is configured to output change probability information representing a trend of land feature changes based on the land cover change area and the local thermal anomaly area in combination with a pre-trained spatiotemporal transformation model; The adjustment module is used to adjust the Gaussian sputtering model using the change probability information as a driving signal, and update the geometric structure of the latest calibrated real-scene three-dimensional model according to the adjusted Gaussian sputtering model.

[0014] In a third aspect, the present application provides a computing device comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute a method for dynamically updating and calibrating a real-life three-dimensional model based on orthophotos as described in any one of the first aspects.

[0015] In a fourth aspect, the present application provides a computer storage medium having computer program instructions stored thereon, which, when executed by a processor, implements a method for dynamically updating and calibrating a real-scene three-dimensional model based on orthophotos as described in any one of the first aspects.

[0016] The present application provides a method for dynamically updating and calibrating a real-life three-dimensional model based on orthophotos, the method comprising: acquiring multi-period orthophotos, multi-period thermal infrared images, and surface temperature data of a target area, the target area including an urban expansion area; extracting a surface coverage change area from the urban expansion area based on temporal differences between the multi-period orthophotos and the multi-period thermal infrared images; identifying a local thermal anomaly area from the urban expansion area based on the surface temperature data; outputting change probability information representing a trend of ground feature change based on the surface coverage change area and the local thermal anomaly area in combination with a pre-trained spatiotemporal transformation model; adjusting a Gaussian sputtering model using the change probability information as a driving signal, and updating and calibrating the geometric structure of the latest real-life three-dimensional model based on the adjusted Gaussian sputtering model.

[0017] The technical solution provided by this application has the following beneficial effects: This application provides multi-dimensional and multi-temporal basic data support for the dynamic update of the model by synchronously collecting multiple orthophotos, thermal infrared images and surface temperature data. By using the time series difference analysis of multiple images, the changing characteristics of surface cover in the urban expansion area can be accurately identified, including typical change types such as new buildings and vegetation degradation. A thermal physical characteristic field is constructed based on surface temperature data to effectively capture local thermal radiation anomaly areas caused by human activities. The surface cover changes and thermal anomaly characteristics are integrated through the spatiotemporal transformation model to predict the evolution trend of the land objects and output a change probability field with physical meaning. The Gaussian sputtering process is driven by the change probability to achieve automatic synchronous calibration of the three-dimensional model geometry and the actual scene changes.

[0018] Furthermore, the present application initializes the particle diffusion rate and distribution density parameters of the Gaussian sputtering model, dynamically adjusts these parameters according to the change probability information, and then controls the particle motion characteristics, ultimately achieving accurate update and calibration of the geometric structure of the real-life three-dimensional model.

[0019] Moreover, this method realizes the organic unity of geometric changes and physical processes in the process of 3D model updating. Through dynamic parameter adjustment, the model update is more in line with the actual evolution law of land objects, which improves the accuracy and timeliness of model updating and avoids the subjective bias caused by human intervention.

[0020] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0022] Figure 1 A flowchart of a method for dynamically updating and calibrating a real-scene three-dimensional model based on orthophotos provided in an embodiment of the present application; Figure 2 A schematic diagram of the structure of a system for dynamically updating and calibrating a real-scene 3D model based on orthophotos provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0024] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0025] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0026] Current real-life 3D model update technology mainly relies on periodic aerial photogrammetry and point cloud reconstruction methods. Although it can identify changes in building geometry, it has limitations: the model update cycle is restricted by the frequency of aerial data collection, making it difficult to respond to the needs of rapid urban expansion in a timely manner; the static mapping method of temperature field data and 3D models cannot reflect the dynamic coupling relationship between building morphological changes and the heat island effect; the point cloud comparison method only focuses on changes in geometric features and lacks quantitative analysis of the evolution of building physical properties, resulting in temporal and spatial deviations between model updates and actual environmental changes, making it difficult to meet the urgent needs of smart city refined management for highly timely and high-precision 3D models.

[0027] In response to the above problems, this application proposes a dynamic update and calibration method for real-life three-dimensional models based on orthophotos. By fusing multi-period orthophotos, thermal infrared images, and surface temperature data, it constructs the spatiotemporal correlation characteristics of surface cover changes and thermal anomaly areas, and uses the change probability information to drive the dynamic adjustment of Gaussian sputtering model parameters. This method breaks through the limitations of the traditional static update mode and achieves synchronous and precise calibration of the three-dimensional model's geometric structure and thermal physical properties. It not only captures the real-time changes in building form through image time series difference analysis, but also dynamically optimizes the model's physical parameters based on thermal radiation characteristics, so that the model update process is highly consistent with the actual urban expansion and thermal environment evolution. It effectively solves the problems of model update lag and insufficient physical property characterization in existing technologies, and provides reliable three-dimensional data support for urban heat island effect monitoring and planning management.

[0028] Figure 1 A flowchart of a method for dynamically updating and calibrating a real-scene 3D model based on orthophotos is provided in an embodiment of the present application. Figure 1 As shown, the method includes: Step 101: Acquire multiple orthophotos, multiple thermal infrared images, and surface temperature data of a target area, wherein the target area includes an urban expansion area.

[0029] In step 101, the target area refers to the geographic area for which the 3D model needs to be updated, encompassing urban expansion areas (i.e., areas under development) and surrounding built-up areas. Multi-period orthophotos represent vertical perspective images of the same area captured at different times. These images have been geometrically corrected to have a unified coordinate reference and are used to analyze changes in land cover. Multi-period thermal infrared images represent thermal radiation data acquired simultaneously with orthophotos, reflecting surface temperature distribution and used to identify thermal anomalies. Surface temperature data represent temperature values ​​obtained through thermal infrared image inversion or ground-based temperature measurement equipment and are used to construct thermophysical property fields.

[0030] In the embodiment of the present application, the scope of the target area to be monitored is first determined to ensure that it includes the urban expansion area; through an aerial or satellite remote sensing platform, orthophotos and thermal infrared images of the area are collected in multiple times to ensure that the temporal and spatial resolution of each image is consistent; surface temperature data is simultaneously obtained, which can be the temperature value inverted from the thermal infrared image or the ground measured data; all data are unified into the same coordinate system, and data preprocessing is completed to provide a basis for subsequent analysis.

[0031] For example, using the expansion area of ​​City A as an example, satellite remote sensing was used to obtain three consecutive summer orthophotos and thermal infrared images of the area, combined with surface temperature data from meteorological stations for the same period. The orthophotos were uniformly resolved to 0.5 meters, and the thermal infrared images to 2 meters. All data was converted to a unified coordinate system to ensure temporal and spatial consistency.

[0032] Step 102: Extracting a land cover change area from the urban expansion area based on the temporal differences between the multiple orthophoto images and the multiple thermal infrared images.

[0033] In step 102, the temporal difference represents the change in pixel characteristics at the same location in the image at different time periods, reflecting the evolution of land cover. Areas of land cover change represent areas within the urban expansion area that have undergone significant artificial transformation or natural degradation, such as new buildings, roads, or areas with reduced vegetation.

[0034] In an embodiment of the present application, a pixel-by-pixel comparison is performed on multiple periods of orthophotos and thermal infrared images to calculate the characteristic difference values ​​of the same location at different periods; a natural change threshold is set to filter out spatial areas where the difference values ​​exceed the threshold; and a morphological analysis method is combined to identify areas that meet the rectangular outline of newly built buildings or vegetation degradation characteristics from the screening results and determine them as areas of surface cover change.

[0035] For example, in the expansion area of ​​City A, by comparing the orthophotos of the second year with those of the first year, differences in pixel characteristics were found in a certain area. Further analysis confirmed that multiple regular rectangular outlines appeared in the area, which was consistent with the characteristics of newly built buildings. At the same time, thermal infrared images showed that the vegetation index in the area decreased, and it was determined to be an area of ​​surface cover change.

[0036] Step 103: Identify local thermal anomaly areas in the urban expansion area based on the surface temperature data.

[0037] In step 103 , the local thermal anomaly region represents a continuous area where the surface temperature is higher than the surrounding environment, which is usually caused by artificial buildings or surface hardening.

[0038] In an embodiment of the present application, a thermal radiation distribution map is generated based on surface temperature data, and a dynamic anomaly judgment threshold is set in combination with historical temperature data; continuous spatial blocks with temperatures exceeding the threshold are identified, and after excluding natural high temperature interference, they are marked as local thermal anomaly areas.

[0039] For example, in the expansion area of ​​City A, analysis of the surface temperature data for the third year revealed that the temperature around a newly built building had increased significantly compared with the same period in history, forming a continuous high-temperature block, which was determined to be a local thermal anomaly area.

[0040] Step 104: Based on the surface cover change area and the local thermal anomaly area, combined with a pre-trained spatiotemporal transformation model, output change probability information representing the change trend of the ground feature.

[0041] In step 104, the spatiotemporal transformation model represents a deep learning model used to analyze the changing patterns of spatiotemporal data and predict future trends. Land feature change trends refer to the predicted evolution of surface cover changes such as new construction and vegetation degradation within urban expansion areas, and the resulting thermal radiation anomalies, in future time periods, as determined by the spatiotemporal transformation model through a fusion analysis of multiple input orthophotos, thermal infrared images, and thermophysical property information. This is derived from the spatiotemporal transformation model's learning of the spatiotemporal correlation features of historical time-series image data and thermophysical field data. Change probability information represents the probability map output by the model, representing the likelihood of change in each surface unit.

[0042] In an embodiment of the present application, the spatial positions of the surface cover change area and the thermal anomaly area are superimposed and matched, and the overlap coefficient is calculated; the matched regional features are spliced ​​with multi-period image data and input into the spatiotemporal transformation model; the model outputs the change probability value of each position to form a change probability distribution map.

[0043] For example, in the expansion area of ​​city A, the newly built area is superimposed on the thermal anomaly area and input into the space-time transformation model. The output shows that there is a high probability that the area will continue to expand in the next year.

[0044] Step 105: Using the change probability information as a driving signal, adjusting the Gaussian sputtering model, and updating and calibrating the geometric structure of the latest real-scene 3D model according to the adjusted Gaussian sputtering model.

[0045] In step 105, the Gaussian sputtering model represents a geometric modeling method based on particle motion, which adjusts the model morphology by controlling particle parameters. The Gaussian sputtering model is linked to the real-world 3D model in that the particle diffusion rate and density distribution parameters of the Gaussian sputtering model are adjusted to control the particle motion trajectory and deposition position in 3D space. These particle deposition results are directly used to determine the position update of building outline boundary points in the real-world 3D model, the coordinate movement of mesh vertices, and the addition and subtraction of surface units, thereby achieving dynamic update and calibration of the real-world 3D model's geometric structure.

[0046] In an embodiment of the present application, the change probability information is mapped to the particle diffusion rate and density parameters of the Gaussian sputtering model; the particle motion trajectory is adjusted according to the parameters; the building outline and grid vertex coordinates are recalculated through the particle deposition position to complete the model update.

[0047] For example, in the expansion area of ​​city A, the parameters of the Gaussian sputtering model are adjusted according to the prediction results of the high-probability area, so that the outline of the new building expands outward, and the three-dimensional model is updated synchronously.

[0048] This method achieves dynamic and accurate updating of real-life three-dimensional models through collaborative analysis of multi-source data, allowing the model's geometric structure to keep pace with actual urban expansion and thermal environment changes, improving the timeliness and accuracy of model updates and providing reliable support for urban planning and management.

[0049] To address the issue of coordinated calibration of geometric structure and thermophysical properties during dynamic updating of a 3D model, in some embodiments, step 105: adjusting the Gaussian sputtering model using the change probability information as a driving signal, and updating and calibrating the geometric structure of the latest real-life 3D model based on the adjusted Gaussian sputtering model, includes: Step 201: Initializing control parameters of a Gaussian sputtering model, wherein the control parameters include particle diffusion rate and particle distribution density.

[0050] In step 201, the control parameters of the Gaussian sputtering model are built-in parameters of the model itself, used to regulate the particle behavior characteristics of the model. The particle diffusion rate is the parameter that controls the speed of particle movement in the model; a larger value indicates a wider particle diffusion range. The particle distribution density is the parameter that determines the number of particles per unit space and directly affects the fineness of the model surface details.

[0051] In this example, we first establish the basic framework of a Gaussian sputtering model, setting the initial particle velocity and spatial distribution. Based on the spatial scale of the target area, we determine the initial diffusion rate to ensure that the particles cover the entire area to be updated, while the initial distribution density ensures that there are sufficient sampling points on the model surface. The initial values ​​of these two parameters are set based on the existing accuracy requirements of the real-world 3D model.

[0052] Step 202: Dynamically adjust the particle diffusion rate and the particle distribution density according to the numerical value of the change probability information.

[0053] In step 202, dynamic adjustment refers to the process of modifying model parameters in real time according to the probability value.

[0054] In this embodiment, the probability values ​​at each location in the change probability map are converted into parameter adjustment coefficients. For high-probability regions, the particle diffusion rate is increased to allow particles to cover the area more quickly, while the particle density is increased to capture finer geometric changes. For low-probability regions, the parameter values ​​are correspondingly reduced. This adjustment is achieved through a pre-set mapping relationship, ensuring that parameter changes are positively correlated with the probability values.

[0055] Step 203: Based on the adjusted particle diffusion rate and the adjusted particle distribution density, the geometric structure of the latest calibration real-scene 3D model is updated.

[0056] In step 203, the updating process refers to the process of modifying the surface shape of the model according to the particle deposition results. The calibration refers to the adjustment operation to make the model consistent with the actual scene.

[0057] In this embodiment, a Gaussian sputtering process is run with adjusted parameters, causing particles to diffuse and move in three-dimensional space. The final particle deposition locations are recorded, and model vertices are added to construct an extended structure in densely deposited areas. Existing vertices are removed or adjusted in sparsely deposited areas. By mapping the particle deposition distribution to the model surface, a precise update of the geometric structure is achieved.

[0058] Here's a specific example: In the example of urban expansion area A, based on three consecutive years of acquired summer orthophoto and thermal infrared imagery data, as well as simultaneously collected surface temperature data, a comparative analysis first revealed significant changes in land cover in the northeastern portion of the area, manifested as multiple newly constructed buildings with regular rectangular outlines. Thermal infrared data also indicated that the temperature in this area was significantly higher than the surrounding area, forming a thermal anomaly with a diameter of approximately 300 meters. These two areas were superimposed and input into a spatiotemporal transformation model, which calculated a high probability of continued expansion in the area within the next year. Based on this probability, the Gaussian sputtering model was initialized with a basic particle diffusion rate of 5 meters per second and a basic particle density of 100 particles per cubic meter. Based on the probability value, the diffusion rate was proportionally increased to 8 meters per second and the particle density was increased to 150 particles per cubic meter. When the adjusted model was run, particles were concentrated in the predicted expansion area, with a density of 200 particles per square meter at the edges of building outlines. Based on this distribution, the outlines of the new buildings in the 3D model were expanded outward by 15 meters, with building heights added. The direction and width of surrounding roads were also adjusted to align the entire model with the actual construction situation. The correspondence between particle deposition density and model vertex updates is determined by pre-set conversion rules. When the particle density in a certain area exceeds a threshold, new model vertices are automatically generated, while areas with density below the threshold retain the existing structure. This process ensures that the 3D model is accurately updated in sync with the actual progress of urban expansion.

[0059] In the embodiment of the present application, the method achieves accurate synchronous updating of the three-dimensional model and real-scene changes by intelligently adjusting particle parameters, so that the model can not only quickly respond to large-scale construction changes, but also maintain the detail accuracy of key areas, providing real-time and reliable three-dimensional data support for urban planning and management.

[0060] To address the accuracy issue of dynamically updating the geometric structure of the real-scene 3D model, in some embodiments, step 203: updating and calibrating the geometric structure of the latest real-scene 3D model based on the adjusted particle diffusion rate and the adjusted particle distribution density, includes: Step 301: Based on the adjusted particle diffusion rate and the adjusted particle distribution density, a Gaussian sputtering process is performed to cause the particles to diffuse in a three-dimensional space.

[0061] In step 301, the Gaussian sputtering process represents a method for reconstructing three-dimensional geometry by simulating particle motion. The relationship between three-dimensional space and the real-world three-dimensional model is as follows: the three-dimensional space is the virtual computational domain for Gaussian sputtering particle motion, while the real-world three-dimensional model is a digital surface representation of that space with actual geographical significance. The two establish a positional correspondence through spatial coordinate system transformation. Particle diffusion motion represents the propagation of particles in three-dimensional space at a set rate and density.

[0062] In this embodiment, a particle motion simulation is initiated based on the adjusted diffusion rate and distribution density parameters. Starting from a preset initial position, particles diffuse in all directions at a speed determined by the diffusion rate, while the density of particles in space is controlled based on the distribution density parameters. This process continues until the preset simulation time has expired.

[0063] Step 302: During the diffusion process, the movement trajectory of the particles is recorded in real time.

[0064] In step 302 , the particle trajectory refers to the path record independently generated by each particle during the diffusion movement. The movement trajectories of each particle are calculated in parallel but finally aggregated and counted, which not only retains the individual movement characteristics but also forms the overall distribution law.

[0065] In the present embodiment, as particles diffuse, the system continuously tracks and records the spatial coordinates of each particle at different time points. These coordinate data are stored in chronological order to form a complete motion path, providing basic data for subsequent analysis.

[0066] Step 303: Based on the motion trajectory, determine the spatial coordinates where the particle finally stops.

[0067] In step 303, the spatial coordinates of the final stop point represent the spatial position of the particle at the end of its motion. The determination method is to determine the position of the end point of the motion trajectory.

[0068] In the embodiment of the present application, when the particle diffusion movement reaches a preset termination condition, the last position coordinate of each particle trajectory is extracted as its final stagnation position. These coordinate points reflect the final distribution state of the particles in three-dimensional space.

[0069] Step 304: Map the spatial coordinates to the surface area of ​​the latest real-scene 3D model, and obtain the particle deposition density distribution by accumulating the number of particles per unit area in the surface area.

[0070] In step 304, spatial coordinate mapping establishes a correspondence between particle positions and the model surface. The association between surface regions and spatial coordinates is as follows: each spatial coordinate is mapped to a specific location on the model surface using the geographic projection of the 3D model. The size of this surface region is determined by the model meshing accuracy and is typically a preset fixed-size grid cell (e.g., 1m×1m) or an adaptive subdivision patch. The volume density distribution represents a distribution diagram reflecting the degree of particle aggregation in various regions of the model surface.

[0071] In this embodiment, the final stagnation coordinates of the particles are matched to the surface grid of the real-world 3D model. The number of particles contained in each grid cell is counted and then divided by the grid area to obtain the deposition density value of that cell. The density values ​​of all grid cells together constitute a complete deposition density distribution map.

[0072] Step 305: updating and calibrating the geometric structure of the latest real-scene 3D model according to the particle deposition density distribution.

[0073] In this embodiment, the deposition density distribution map is analyzed, and model vertices are added to expand the geometry in high-density areas, while existing vertices are removed or adjusted in low-density areas. Through this local refinement or simplification, the geometric features of the 3D model are aligned with the actual changes in particle deposition.

[0074] Here's a specific example: In the implementation case of City A's expansion area, the system initiated the Gaussian sputtering model based on information about high-probability expansion areas obtained through preliminary analysis. Based on the predicted results, the particle diffusion rate was set to 8 meters per second and the particle distribution density was set to 150 particles per cubic meter. These parameter values ​​were calculated by multiplying the predicted probability by the base parameter value. After the model ran, a large number of particles began to diffuse from a preset initial position. The system recorded the movement path of each particle in real time, and this path data was stored as a time series. After the particle movement reached the preset simulation duration, the system extracted the coordinates of the final stopping position of all particles. These coordinates reflect the final distribution of the particles in three-dimensional space. These spatial coordinates were then matched to the surface mesh of the real-life 3D model, and the number of particles contained within each 1m×1m grid cell was counted. An average of 200 particles were detected in the grid cell at the center of the predicted expansion area, and approximately 80 particles were detected in the surrounding area. According to pre-set conversion rules, when the number of particles in a grid cell exceeds 150, the system automatically adds a new model vertex at that location to expand the building outline; when the number of particles is between 50 and 150, the position of the existing vertex is adjusted; if the number is less than 50, it remains unchanged. Through this mechanism, the outline of the newly planned commercial complex in the model is expanded outward by 15 meters, the geometric structure of three new buildings is added, and the surrounding road network is adjusted. The entire update process is completed completely automatically, and the final generated three-dimensional model accurately reflects the latest construction progress in the area, with building height data and road directions highly consistent with the actual construction situation. The density threshold used in this conversion process is an empirical value derived from statistical analysis of historical data, which can ensure the accuracy and stability of model updates.

[0075] In the embodiments of the present application, the method realizes the precise dynamic update of the geometric structure of the three-dimensional model through physical simulation and intelligent analysis of particle motion, so that the model can automatically adapt to various changes brought about by urban expansion, ensuring high-precision modeling of key areas and maintaining the consistency of the overall model, providing a reliable three-dimensional data foundation for urban planning and management.

[0076] In order to solve the problem of refining the geometric structure update of the real-scene 3D model, in some embodiments, step 305: updating and calibrating the geometric structure of the latest real-scene 3D model according to the particle deposition density distribution includes: Step 401: Divide the surface area of ​​the real-scene three-dimensional model into a plurality of grid units, and count the particle deposition density value in each grid unit according to the particle deposition density distribution.

[0077] In step 401, grid cells represent the regular small areas into which the model surface is divided, serving as the basic unit of processing. The grid cell size is typically set larger than the unit statistical area (e.g., 10m×10m). This ensures that each grid cell captures sufficient particle deposition samples (e.g., at least 50–100 particle deposition points) to improve the statistical reliability of density. For example, when simulating particle diffusion with 0.1m accuracy, a 10m×10m grid cell ensures approximately 10,000 potential particle deposition locations, meeting statistical requirements. The particle deposition density value is a quantitative indicator of the degree of particle aggregation within a grid cell.

[0078] In this embodiment, the entire model surface is first divided into a number of uniform grid cells of a fixed size. The size of each cell is determined based on the required model accuracy. All grid cells are then traversed, and the number of particles falling within each cell is counted. This number is then divided by the cell area to obtain the particle deposition density value for that cell, ultimately generating a complete density distribution map.

[0079] Step 402: Based on the particle deposition density value, determine the density threshold interval to which each grid unit belongs, the density threshold interval being a first interval in which the particle deposition density value is less than a first preset threshold, a second interval in which the particle deposition density value is between the first preset threshold and the second preset threshold, or a third interval in which the particle deposition density value is greater than the second preset threshold, and the first preset threshold is less than the second preset threshold.

[0080] In step 402, the density threshold interval represents the range interval for dividing the density value into three levels. The first preset threshold represents the critical value for distinguishing low-density areas from medium-density areas. The second preset threshold represents the critical value for distinguishing medium-density areas from high-density areas.

[0081] In this embodiment of the present application, two density thresholds are pre-set and the density value of each grid cell is compared with these two thresholds. Density values ​​below the first threshold are classified as the first interval, values ​​between the two thresholds are classified as the second interval, and values ​​above the second threshold are classified as the third interval. This classification provides a basis for subsequent differentiated processing.

[0082] Step 403: performing differentiated geometric operations on different grid cells based on the processing methods corresponding to the density threshold intervals, and reconstructing the geometric structure of the real-scene three-dimensional model based on the operation results of all grid cells.

[0083] In step 403, the differential geometric operation means taking different model update strategies according to the density level, and the reconstruction means integrating all local update results to form a complete new model.

[0084] In this embodiment, the original geometry of the mesh cells in the first interval is maintained; the local vertex positions of the cells in the second interval are adjusted; and the cells in the third interval are expanded by adding new vertices. After the update of all cells is completed, the system automatically checks and repairs the connections between adjacent cells to ensure the continuity and integrity of the entire model surface.

[0085] Here's a specific example: In the implementation case of Urban Expansion Area A, based on particle deposition data obtained from a previous Gaussian sputtering model run, the system first divided the surface of the real-life 3D model of the commercial complex into 1m x 1m grid cells. Statistics revealed that the grid cells in the main building area contained an average of 200 particles, the surrounding road area contained approximately 80 particles, and the green space area contained fewer than 30 particles. Based on historical data analysis, the first preset threshold was 50 particles, and the second preset threshold was 150 particles. The main building area, with a particle count exceeding 150, was classified as the third interval. The system added detailed vertices of the building facade to these grid cells, accurately depicting the glass curtain wall structure. The road area, with a particle count between 50 and 150, fell into the second interval. The system repositioned existing vertices within these grid cells to smooth the road edge curves. The green space area, with a particle count below 50, was classified as the first interval, maintaining its original geometry. After completing the differentiation process for all grid cells, the system automatically checked the connectivity between adjacent cells and optimized the connecting lines between the newly added vertices and existing vertices to ensure a smooth transition between the model surface. These density thresholds were optimized by analyzing data from the past 20 model updates. The first threshold was the minimum historically significant update density, while the second threshold was the average change density. This ensured that the model update was neither overly sensitive nor overlooked significant changes. The resulting 3D model not only accurately reflected the commercial complex's latest architectural details but also fully preserved the original features of unchanged areas, achieving millimeter-level update accuracy.

[0086] In the embodiment of the present application, the method realizes the intelligent update of the geometric structure of the three-dimensional model through refined grid division and differentiated processing strategies, which not only ensures high-precision modeling of key change areas, but also avoids unnecessary repeated calculations, improves the efficiency and quality of model updates, and provides reliable three-dimensional data support for urban dynamic monitoring.

[0087] To address the accuracy of surface change detection in urban expansion areas, in some embodiments, step 102 of extracting surface cover change areas within the urban expansion area based on the temporal differences between the multiple orthophoto images and the multiple thermal infrared images includes: Step 501: performing temporal feature difference calculation on pixels at the same spatial position of the multi-period orthophotos and the multi-period thermal infrared images to obtain pixel feature difference values ​​of each period.

[0088] In step 501, the temporal feature difference calculation is to compare the degree of feature change at the same location in images at different time periods. The pixel feature difference value is a numerical value that quantitatively represents the change of pixel features over time.

[0089] In this example, orthophotos and thermal infrared images taken at different times are first precisely aligned to ensure that pixels at the same geographic location correspond exactly. For each pixel, the magnitude of the difference in its spectral and thermal radiation signatures between the images taken at different times is calculated. These differences comprehensively reflect the degree of change in the surface cover at that location.

[0090] Step 502: Screening spatial regions in the urban expansion area where the pixel feature difference values ​​for two consecutive periods exceed a preset natural variation threshold.

[0091] In step 502, the natural variation threshold refers to the maximum pixel feature difference of surface cover (e.g., soil, natural vegetation) caused by seasonal changes or climate fluctuations, without human interference. This threshold is determined by statistically analyzing the maximum spectral change within the same time interval in undeveloped areas of historical imagery. It is used to distinguish between artificial changes caused by urban expansion and natural surface fluctuations. A spatial region represents a continuous block consisting of multiple adjacent difference pixels.

[0092] In this embodiment, a reasonable difference threshold is determined based on historical data analysis, and all pixel locations with difference values ​​exceeding this threshold are screened. Spatial clustering analysis is then performed on these pixels, aggregating adjacent abnormal pixels into continuous spatial blocks. These blocks are candidate areas where surface changes may occur.

[0093] Step 503: Determine the area in the spatial area that meets the geometric features of the newly built buildings or the vegetation cover attenuation features as the surface cover change area.

[0094] In step 503, areas with geometric features of new buildings or vegetation cover attenuation features are further identified from the screened spatial areas. This means that the geometric regularity (such as rectangular outlines and right-angle structures) in the spatial areas where the pixel feature difference values ​​exceed the threshold is analyzed to determine whether they are new buildings, or the degree of decline in the vegetation index is used to determine whether they are vegetation degraded areas, thereby eliminating false detections caused by natural surface fluctuations.

[0095] In this embodiment, candidate change regions are analyzed for shape and their vegetation index is calculated. Regions with distinct right-angled contours and regular boundaries are identified as newly constructed buildings, while regions with a decreased vegetation index are considered degraded. This dual verification ensures that the extracted change regions are authentic and reliable.

[0096] Here's a specific example: In the implementation case of Urban Expansion Area A, the system first acquired three consecutive summer orthophotos and thermal infrared imagery of the area, converting all images to the same coordinate system. By comparing the orthophotos from the second year with those from the first, a difference calculation method was used to analyze changes in spectral characteristics at each pixel. The difference value of the pixel at the center of one area exceeded three times the average of the surrounding natural fluctuation range. The natural variation threshold, determined based on historical data analysis, was two standard deviations of the normal fluctuation range, and most pixels in this area exceeded this threshold. The system clustered these anomalous pixels into a contiguous spatial region of approximately 8,000 square meters. Morphological analysis revealed 12 regular rectangular structures within this region with aspect ratios ranging from 1:1 to 1:3, consistent with a commercial building complex. Comparison with thermal infrared imagery revealed a 60% decrease in the vegetation index compared to the first year. The difference, calculated as the difference between (near-infrared reflectance - red reflectance) / (near-infrared reflectance + red reflectance), confirmed a decrease in vegetation cover. Combining the building morphological characteristics and evidence of vegetation attenuation, it was finally determined that the 8,000 square meter area was the area of ​​surface cover change caused by the new commercial center project.

[0097] In the embodiment of the present application, the method effectively identifies the changes in surface cover in urban expansion areas through multi-feature comprehensive analysis and a dual verification mechanism, which not only avoids misjudgments caused by natural fluctuations, but also accurately captures various human-induced changes, providing reliable change area data for subsequent three-dimensional model updates.

[0098] To address the accuracy issue of urban heat island effect monitoring, in some embodiments, step 103: identifying local thermal anomaly areas in the urban expansion area based on the surface temperature data, includes: Step 601: Convert the surface temperature data into thermophysical distribution information, where the thermophysical distribution information includes a thermal radiation intensity value for each spatial unit.

[0099] In step 601, the thermophysical distribution information represents a dataset reflecting the spatial distribution of surface thermal radiation intensity. A spatial unit represents the smallest processing unit for temperature data, corresponding to a pixel area in an image. Thermal radiation intensity refers to the amount of heat energy radiated per unit area of ​​the surface per unit time. This value is calculated by applying the surface temperature data to the Stefan-Boltzmann law. Specifically, the temperature value is converted to absolute temperature and then multiplied by the fourth power of the value and the Stefan constant. This value reflects the differences in thermal radiation capacity between different surface features.

[0100] In this embodiment, the collected surface temperature data is first spatially aligned according to the image pixel position, with each pixel position corresponding to a temperature measurement. These temperature values ​​are then normalized and converted into thermal radiation intensity values ​​with a unified dimension, forming a thermophysical distribution map covering the entire urban expansion area.

[0101] Step 602: Based on the historical thermal environment baseline of the urban expansion area, a dynamic temperature anomaly determination boundary is set.

[0102] In step 602, the historical thermal environment baseline represents the normal thermal radiation intensity range for each spatial location, calculated based on multiple periods of historical surface temperature data from the target area before urban expansion. This range is derived from long-term observations of multiple periods of thermal infrared imagery. The dynamic temperature anomaly determination boundary represents the abnormal temperature threshold that varies with season and time.

[0103] In this embodiment, the temperature data for the past five years is analyzed to calculate the average temperature and normal fluctuation range for each spatial unit. Based on the historical baseline, a reasonable temperature anomaly determination standard is set based on current meteorological conditions and seasonal characteristics. This standard automatically adjusts with seasonal changes.

[0104] Step 603: Filter target spatial units whose thermal radiation intensity values ​​are greater than the dynamic temperature anomaly determination boundary, and combine at least three spatially continuous target spatial units into a local thermal anomaly area.

[0105] In step 603 , the target spatial unit represents a single detection unit whose temperature exceeds the abnormal boundary.

[0106] In this embodiment, all spatial cells whose thermal radiation intensity exceeds the dynamic boundary are first marked, and then spatial connectivity analysis is performed on these cells. At least three adjacent abnormal cells are merged into a thermal anomaly region, eliminating interference from isolated high-temperature points and ensuring that the identified thermal anomaly region is meaningful.

[0107] Here's a specific example: In the implementation case of the expansion area of ​​City A, the system obtained the surface temperature data of the third summer. The temperature value was first converted into the thermal radiation intensity value, and the calculation formula was E=σT 4 , where E represents the intensity of thermal radiation and σ is the Stefan-Boltzmann constant , where T is the absolute surface temperature. Thermal radiation intensity values ​​for each 10m x 10m spatial unit were calculated to form a thermophysical distribution map. Based on the region's summer temperature data from the past five years, the historical average thermal radiation intensity was determined to be 450W / m² with a standard deviation of 35W / m². Considering that the temperature that year was 2 degrees Celsius higher than usual, the dynamic temperature anomaly determination boundary was set at the historical average plus 2.5 times the standard deviation, or 535W / m². Analysis revealed that 28 contiguous spatial units surrounding the newly built commercial complex had thermal radiation intensities exceeding this boundary, with a peak value reaching 620W / m². These units clustered into a 2,800-square-meter localized thermal anomaly area, which overlapped closely with the newly built construction area identified in the orthophoto imagery. The dynamic boundary value was set by first calculating the average and fluctuation range of the historical data for the same period, then adjusting it appropriately based on the temperature fluctuations that year to ensure that the boundary value both reflected the changing climate background and effectively identified the anomaly area.

[0108] In the embodiment of the present application, the method accurately identifies thermal anomaly areas during urban expansion through dynamic threshold setting and spatial continuity verification, which not only avoids the misjudgment of temporary high-temperature points, but also reliably captures the heat island effect caused by building development, providing accurate data support for urban thermal environment management.

[0109] To address the accuracy of predicting the trend of land feature changes in urban expansion areas, in some embodiments, step 104: outputting change probability information representing the trend of land feature changes based on the land cover change area and the local thermal anomaly area in combination with a pre-trained spatiotemporal transformation model, includes: Step 701: Calculate the overlapping area ratio between the surface cover change region and the local thermal anomaly region, and calculate the spatial correlation strength coefficient based on the overlapping area ratio.

[0110] In step 701, the overlap percentage represents the percentage of the overlapping area of ​​the surface cover change region and the local thermal anomaly region within the spatial projection range, as a percentage of the total area of ​​both regions. This percentage is calculated through geospatial overlay analysis. The spatial correlation strength coefficient is an indicator that quantifies the degree of correlation between surface change and thermal anomaly.

[0111] In this embodiment, the spatial intersection area of ​​the surface cover change region and the thermal anomaly region is first calculated, and then divided by the union area of ​​the two to obtain the overlap ratio. This ratio is then mapped to a spatial correlation strength coefficient using a preset conversion function, reflecting the degree of spatial correlation between the two.

[0112] Step 702: The spatial correlation intensity coefficient is combined with the texture features of the multi-phase orthophotos and the thermal radiation gradient features of the multi-phase thermal infrared images to form an input tensor.

[0113] In step 702, texture features represent the spatial distribution pattern of pixel grayscale values ​​that reflects the structural regularity of surface objects. They are extracted by performing statistical analysis of pixels in a local window on each orthophoto image. Thermal radiation gradient features represent the rate of change of thermal radiation intensity between adjacent pixel units. They are obtained by calculating the temperature difference between each pixel and its neighboring pixels in each thermal infrared image. The input tensor represents the multidimensional data array input to the model.

[0114] In this example, surface texture features for each period are extracted from orthophotos, and temperature gradient features are calculated from thermal infrared images. These features are combined with spatial correlation strength coefficients in a pre-set dimensional order to construct a multidimensional data array that meets the model input requirements.

[0115] Step 703: Input the input tensor into the spatiotemporal transformation model and output change probability information.

[0116] In this embodiment, the constructed input tensor is fed into a pre-trained spatiotemporal transformation model. The model analyzes the correlation between spatiotemporal features and outputs the probability of each spatial unit changing in the future time period, forming a change probability distribution map covering the entire area.

[0117] Here's a specific example: In the implementation case of Urban Expansion Area A, the system first calculated the spatial relationship between the land cover change area and the thermal anomaly area in the newly built commercial district. The land cover change area was 8,000 square meters, the thermal anomaly area was 2,800 square meters, and the overlap between the two was 2,100 square meters. The overlap percentage was calculated as 24% by dividing the overlapping area by the total area of ​​the two areas (2,100 / 8,800). Based on the conversion relationship established through historical data analysis, the 24% percentage was substituted into the formula k=1-e^(-λx) to calculate a spatial correlation strength coefficient of 0.45, where λ is a fitting parameter with a value of 3.5 and x is the overlap percentage. Texture features were also extracted from three consecutive years of orthophoto imagery of the area, including indicators such as building edge clarity and surface roughness, as well as temperature gradient characteristics from thermal infrared imagery to reflect the severity of temperature changes. These feature data and the spatial correlation strength coefficient were then concatenated according to time series and spatial location to form a three-dimensional data array that served as input to the spatiotemporal transformation model. After analysis, the model outputs a probability distribution map for the expansion of the commercial district in the next year, showing that the probability of main building expansion is 78%, the probability of parking lot expansion is 65%, and the probability of green space transformation is 32%.

[0118] In the embodiment of the present application, the method achieves accurate prediction of the changing trend of land features in urban expansion areas through multi-source feature fusion and spatiotemporal correlation analysis. It takes into account both the changes in surface morphology and the evolution characteristics of the thermal environment, providing a reliable decision-making basis for the intelligent update of the three-dimensional model.

[0119] Figure 2 A structural diagram of a dynamic update calibration system for a real-scene 3D model based on orthophotos provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, the system includes: The acquisition module 21 is used to acquire multi-period orthophoto images, multi-period thermal infrared images and surface temperature data of a target area, wherein the target area includes an urban expansion area.

[0120] The extraction module 22 is configured to extract a land cover change area within the urban expansion area based on a temporal difference between the multiple orthophoto images and the multiple thermal infrared images.

[0121] The identification module 23 is configured to identify a local thermal anomaly area in the urban expansion area based on the surface temperature data.

[0122] The output module 24 is configured to output change probability information representing a trend of land feature changes based on the land cover change area and the local thermal anomaly area in combination with a pre-trained spatiotemporal transformation model.

[0123] The adjustment module 25 is configured to adjust the Gaussian sputtering model using the change probability information as a driving signal, and update the geometric structure of the latest calibrated real-scene 3D model according to the adjusted Gaussian sputtering model.

[0124] Figure 2 The real-scene 3D model dynamic update calibration system based on orthophoto can be executed Figure 1 The implementation principles and technical effects of the method for dynamically updating and calibrating a real-world 3D model based on orthophotos described in the illustrated embodiment will not be elaborated upon. The specific manner in which the various modules and units in the above-described embodiment of the system for dynamically updating and calibrating a real-world 3D model based on orthophotos perform their operations has been described in detail in the related embodiments and will not be further elaborated upon here.

[0125] In one possible design, Figure 2 The embodiment shown is a real scene 3D model dynamic update calibration system based on orthophoto image, which can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32; The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0126] The processing component 32 is used to perform the above Figure 1 The embodiment provides a method for dynamically updating and calibrating a real-scene three-dimensional model based on orthophotos.

[0127] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0128] The storage component 31 is configured to store various types of data to support operations on the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random-access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0129] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0130] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0131] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0132] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0133] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The embodiment shown is a method for dynamically updating and calibrating a real-scene three-dimensional model based on orthophotos.

[0134] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0135] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0136] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for dynamically updating and calibrating a real-scene three-dimensional model based on orthophotos, characterized in that: include: Acquiring multi-period orthophotos, multi-period thermal infrared images, and surface temperature data of a target area, wherein the target area includes an urban expansion area; Extracting a land cover change area within the urban expansion area based on a temporal difference between the multiple orthophoto images and the multiple thermal infrared images; identifying a local thermal anomaly region within the urban expansion region based on the land surface temperature data; Based on the surface cover change area and the local thermal anomaly area, combined with a pre-trained spatiotemporal transformation model, outputting change probability information representing the change trend of the ground feature; The change probability information is used as a driving signal to adjust the Gaussian sputtering model, and the geometric structure of the latest real-scene three-dimensional model is updated and calibrated according to the adjusted Gaussian sputtering model.

2. The method according to claim 1, characterized in that The method of adjusting the Gaussian sputtering model using the change probability information as a driving signal and updating and calibrating the geometric structure of the latest real-scene three-dimensional model according to the adjusted Gaussian sputtering model includes: Initializing control parameters of a Gaussian sputtering model, wherein the control parameters include a particle diffusion rate and a particle distribution density; Dynamically adjusting the particle diffusion rate and the particle distribution density according to the numerical value of the change probability information; The geometric structure of the latest realistic 3D model is updated and calibrated based on the adjusted particle diffusion rate and the adjusted particle distribution density.

3. The method according to claim 2, characterized in that The updating and calibrating of the geometric structure of the latest real-scene three-dimensional model based on the adjusted particle diffusion rate and the adjusted particle distribution density includes: Based on the adjusted particle diffusion rate and the adjusted particle distribution density, a Gaussian sputtering process is performed to make the particles diffuse in three-dimensional space; During the diffusion process, the trajectory of particles is recorded in real time; Based on the motion trajectory, determining the spatial coordinates at which the particle finally stops; Mapping the spatial coordinates to a surface area of ​​the latest real-scene three-dimensional model, and obtaining a particle deposition density distribution by accumulating the number of particles per unit area in the surface area; The geometric structure of the latest real-scene three-dimensional model is updated and calibrated according to the particle deposition density distribution.

4. The method according to claim 3, characterized in that The updating and calibrating of the geometric structure of the latest real-scene three-dimensional model according to the particle deposition density distribution includes: Dividing the surface area of ​​the real-scene three-dimensional model into a plurality of grid units, and counting the particle deposition density value in each grid unit according to the particle deposition density distribution; Determining, based on the particle deposition density value, a density threshold interval to which each of the grid cells belongs, the density threshold interval being a first interval in which the particle deposition density value is less than a first preset threshold value, a second interval in which the particle deposition density value is between the first preset threshold value and a second preset threshold value, or a third interval in which the particle deposition density value is greater than the second preset threshold value, and the first preset threshold value is less than the second preset threshold value; Based on the processing method corresponding to the density threshold interval, differentiated geometric operations are performed on different grid cells, and the geometric structure of the real-scene three-dimensional model is reconstructed based on the operation results of all grid cells.

5. The method according to claim 1, characterized in that The extracting the surface cover change area within the urban expansion area based on the time sequence difference between the multiple orthophoto images and the multiple thermal infrared images includes: Calculating the time series feature difference of pixels at the same spatial position of the multi-period orthophoto images and the multi-period thermal infrared images to obtain the pixel feature difference value of each period; Selecting a spatial region in the urban expansion area where the pixel feature difference value exceeds a preset natural variation threshold for two consecutive periods; The area in the spatial region that meets the geometric features of the newly built buildings or the vegetation cover attenuation features is determined as the surface cover change area.

6. The method according to claim 1, characterized in that The step of identifying a local thermal anomaly area in the urban expansion area based on the surface temperature data includes: Converting the surface temperature data into thermophysical distribution information, wherein the thermophysical distribution information includes a thermal radiation intensity value for each spatial unit; Setting a dynamic temperature anomaly determination boundary based on a historical thermal environment baseline of the urban expansion area; Target spatial units having thermal radiation intensity values ​​greater than the dynamic temperature anomaly determination boundary are screened, and at least three spatially continuous target spatial units are combined into a local thermal anomaly area.

7. The method according to claim 1, characterized in that The output of the change probability information representing the change trend of the ground feature based on the surface cover change area and the local thermal anomaly area in combination with the pre-trained spatiotemporal transformation model includes: Calculating the overlapping area ratio between the surface cover change region and the local thermal anomaly region, and calculating the spatial correlation intensity coefficient based on the overlapping area ratio; splicing the spatial correlation intensity coefficient with the texture features of the multi-phase orthophotos and the thermal radiation gradient features of the multi-phase thermal infrared images to form an input tensor; The input tensor is input into the spatiotemporal transformation model, and change probability information is output.

8. A dynamic update and calibration system for real-scene three-dimensional models based on orthophotos, characterized in that: include: an acquisition module, configured to acquire multi-period orthophoto images, multi-period thermal infrared images, and surface temperature data of a target area, wherein the target area includes an urban expansion area; An extraction module, configured to extract a land cover change area within the urban expansion area based on a temporal difference between the multiple orthophoto images and the multiple thermal infrared images; an identification module, configured to identify a local thermal anomaly area in the urban expansion area based on the surface temperature data; An output module is configured to output change probability information representing a trend of land feature changes based on the land cover change area and the local thermal anomaly area in combination with a pre-trained spatiotemporal transformation model; The adjustment module is used to adjust the Gaussian sputtering model using the change probability information as a driving signal, and update the geometric structure of the latest calibrated real-scene three-dimensional model according to the adjusted Gaussian sputtering model.

9. A computing device, characterized in that The method comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a method for dynamically updating and calibrating a real-life three-dimensional model based on orthophotos as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, the method for dynamically updating and calibrating a real-scene three-dimensional model based on orthophoto images as described in any one of claims 1 to 7 is implemented.

Images