Method, device, product and medium for real-scene modeling based on air-ground fusion of oblique photogrammetry

Abstract

This invention discloses a method, device, product, and medium for real-world modeling using oblique photogrammetry for air-ground fusion. By acquiring image data using oblique drones, low-altitude drones, and ground cameras according to specific rules and performing fusion calculations, the method satisfies the requirements for aerial triangulation while ensuring efficient field work time. This method significantly improves modeling effectiveness and accuracy, resulting in a constructed 3D model with picture-level texture and pixel-level precision. The method boasts strong feasibility, realistic texture effects, and a high degree of precision, providing reliable and authentic surveying and mapping geographic information for applications such as engineering planning, design, display, and 3D simulation.

Description

Technical Field

[0001] The present invention relates to the field of three-dimensional real scene modeling, and in particular to a method, device, product and medium for real scene modeling using oblique photogrammetry and air-ground fusion. Background Art

[0002] With the continuous advancement of 3D technology software and hardware, real-world 3D modeling is gaining increasing application in engineering fields. The most prominent example is oblique photogrammetry 3D real-world modeling. This technology utilizes a drone platform equipped with a five-lens oblique camera to obtain image data from various ground features at different angles through a planned route. Using specific measurement data processing methods, a 3D model of the real-world surface can be quickly, efficiently, and accurately obtained. This technology is widely used in 3D modeling, 3D display, urban planning, engineering surveying, and other fields.

[0003] 3D real-world models created using oblique photogrammetry offer rich textures, realistic results, excellent visual quality, and high coordinate accuracy. They are an effective method for restoring the actual appearance of cities, documenting the current state of cities at key points in time and space, and constructing real-world 3D models with spatial geographic coordinates. Compared to traditional manual modeling, these methods offer numerous advantages, including efficiency, authenticity, and comprehensiveness. They are widely applicable across various fields.

[0004] Drone-assisted photogrammetry 3D modeling has been widely used. However, a common practical problem is that the oblique photogrammetry is limited by the operating environment in densely built areas or drone photography, resulting in obstructions and poor resolution in building facades, eaves and other locations. As a result, the 3D model constructed using the acquired oblique image data has unclear textures, insufficient resolution and accuracy, and the effect of the constructed model is limited by the operating environment and multiple factors, resulting in unsatisfactory results, missing textures, and insufficient accuracy.

[0005] Most of the existing methods are conventional oblique photogrammetry 3D modeling methods, which can meet the needs of 3D modeling of large areas and large scenes, but the degree of refinement of local key areas and areas of interest is not enough. There are very few existing air-ground fusion modeling methods, and some of them also require the layout of a part of image control points on the wall or the ground to achieve the purpose of fusion. In this way, when the area of ​​interest is large or when the field conditions do not allow, when the resolution of the image taken on the ground is too different from the resolution of the image taken by the oblique drone, the existing methods are not feasible, and at the same time, they cause certain adverse environmental impacts in some key work areas. CN111540048A discloses a refined real-scene 3D modeling method based on air-ground fusion, which fuses and models the data of drone oblique photogrammetry and vehicle-mounted close-range photogrammetry, but the document cannot ensure that fusion modeling can be achieved through its solution, and cannot guarantee the accuracy of modeling. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a method, device, product and medium for real-scene modeling of air-ground fusion by oblique photogrammetry in view of the shortcomings of the existing technology, so as to improve the modeling accuracy.

[0007] To solve the above technical problems, the technical solution adopted by the present invention is: a method for real-scene modeling by air-ground fusion of oblique photogrammetry, comprising the following steps:

[0008] Obtain aerial triangulation results (i.e., aerial triangulation calculations) corresponding to image data taken by tilted drones, aerial triangulation results corresponding to image data taken by low-altitude drones, and aerial triangulation results corresponding to image data taken by ground cameras;

[0009] Fusing the aerial triangulation results corresponding to the image data taken by the oblique UAV and the aerial triangulation results corresponding to the image data taken by the low-altitude UAV to obtain a first fusion result;

[0010] fusing the first fusion result with an aerotriangulation result corresponding to the image data captured by the ground camera to obtain a second fusion result;

[0011] The real-scene three-dimensional model is reconstructed using the second fusion result.

[0012] In the present invention, low altitude refers to a space region 3 to 50 meters above the ground.

[0013] Through a large number of practical studies, it was found that the reason why the existing air-ground fusion modeling accuracy is not high or unstable is that it does not take into account the large difference in resolution between different data sources of air-ground fusion, which makes it impossible to use modeling software to achieve automatic modeling. The existing technology requires a lot of manual field survey work to improve the modeling accuracy to a certain extent. The present invention uses low-altitude drone aerial photography to obtain image data, and smoothly fuses the image data taken by the tilted drone through the low-altitude drone aerial photography image data and the image data obtained by ground camera photography, so that the modeling software can use the fused data to automatically complete the modeling process, without the need for manual field survey work to remember to obtain a high-precision air-ground fusion real-scene model.

[0014] The method of the present invention can be implemented whether there are image control points in the field or not. When there are no image control points, the coordinate system of the model or aerial triangulation calculation in each step can be aligned with the drone POS coordinate system; when there are image control points, the corresponding image control point coordinates can be added to the aerial triangulation calculation in each step and then the aerial triangulation calculation can be performed. Therefore, the method of the present invention is more suitable for various field operation methods.

[0015] In this invention, both tilted drone aerial photography and manual low-altitude drone aerial photography are essential. While tilted drones can capture comprehensive data about the survey or mission area, manually capturing effective modeling data with low-altitude drones would require exponentially more work. Furthermore, images captured by low-altitude drones serve as a transition between the resolution of images captured by tilted drones and those captured by ground-based cameras. The resolution difference is so great that the computer in the air-ground fusion process cannot automatically identify the same point with significantly different pixels, making fusion impossible.

[0016] In the present invention, the specific implementation process of obtaining the aerial triangulation calculation results corresponding to the image data taken by the ground camera includes: the ground camera continuously shoots the image data of the target area, at each shooting point, the target area is shot in a horizontal rotation manner with the exposure point of the ground camera as the center of the circle, the overlap of adjacent images shot at each shooting point is not less than 50%, and the overlap of images obtained between adjacent shooting points is not less than 50%; performing aerial triangulation on the image data taken by the ground camera to obtain the aerial triangulation calculation results corresponding to the image data taken by the ground camera.

[0017] The general rule for oblique photogrammetry is to have an overlap of at least 75% heading and at least 50% sideways; the CCC user manual suggests a limit of at least 80% heading and 50% sideways. This paper considers both data volume and modeling accuracy, and based on practical engineering experience, considers that an overlap of at least 50% heading and sideways will neither affect modeling results nor result in excessive data volume due to excessive overlap.

[0018] The specific implementation process of obtaining the aerial triangulation results corresponding to the image data of low-altitude drone aerial photography includes: planning a layer of routes, using the drone to continuously shoot the target area on the route, performing aerial triangulation on the image data obtained by the continuous shooting, and obtaining the aerial triangulation results corresponding to the image data of low-altitude drone aerial photography; or planning multiple layers of routes within the target area, using the drone to continuously shoot the target area, with the overlap of adjacent images within each layer of routes being no less than 50%, and the overlap of images between two adjacent layers of routes being no less than 50%; performing aerial triangulation on the image data corresponding to each layer of routes, and obtaining the aerial triangulation results corresponding to the multiple layers of image data of low-altitude drone aerial photography.

[0019] The process of obtaining the first fusion result includes:

[0020] When planning a single-layer route, the aerial triangulation results corresponding to the image data taken by the oblique drone and the aerial triangulation results corresponding to the image data taken by the low-altitude drone are fused to obtain a first fusion result.

[0021] In the case of planning a multi-layer route, the aerial triangulation results corresponding to the image data taken by the oblique drone are fused with the aerial triangulation results corresponding to the multi-layer image data taken by the low-altitude drone, and the lowest resolution aerial triangulation result is obtained.

[0022] Performing a fusion operation on the first preliminary fusion result and the aerial triangulation result with the second lowest resolution among the aerial triangulation results corresponding to the multi-layer image data taken by the low-altitude UAV to obtain a second preliminary fusion result;

[0023] This process is deduced in this way until all the aerial triangulation calculation results corresponding to the multi-layer image data of the low-altitude UAV aerial photography are fused to obtain the first fusion result.

[0024] The present invention uses a preliminary model built by tilting drone images to obtain geographic coordinate information. By picking coordinate points on the model as control points for images taken by a ground camera, the model participates in the second aerial triangulation of the ground camera image data. This effectively pulls the relative aerial triangulation results of the ground camera image data into the same coordinate system as the drone image data, overcoming the defect in the prior art of being unable to fuse two sets of aerial triangulation results with large coordinate deviations.

[0025] The specific process of obtaining the second fusion result includes:

[0026] A three-dimensional model is established using aerial triangulation results corresponding to the image data captured by the tilted drone, and a number of characteristic coordinate points are selected from the three-dimensional model, where the selected characteristic coordinate points are also points captured by a ground camera; the coordinate system of the three-dimensional model is consistent with the coordinate system of the image position positioning data acquired by the tilted drone, that is, the POS data coordinate system;

[0027] The coordinate values ​​of the characteristic coordinate points are added to the aerial triangulation results corresponding to the image data taken by the ground camera in the form of image control points, and the aerial triangulation results corresponding to the image data taken by the ground camera with the characteristic coordinate points added are again subjected to aerial triangulation to obtain the aerial triangulation results with coordinate information;

[0028] Perform air-ground fusion air triangulation on the air triangulation result with coordinate information and the first fusion result to obtain a second fusion result.

[0029] The resolution of the image data taken by the tilted drone is 1 to 10 times the resolution of the image data taken by the low-altitude drone; the resolution of the image data taken by the low-altitude drone is 1 to 10 times the resolution of the image data taken by the ground camera.

[0030] The resolution of the image data taken by the low-altitude UAV is 1 to 5 times the resolution of the image data taken by the tilted UAV; the resolution of the image data taken by the ground camera is 1 to 5 times the resolution of the image data taken by the low-altitude UAV.

[0031] In practice, the resolutions of two adjacent data sources often exceed five times that of each other. Existing technology, no matter how many image control tie points are added, cannot automatically model the image. This is because the pixel differences are so great, requiring extensive field rework and thus preventing existing technology from achieving a good fusion modeling approach. This invention addresses this issue by utilizing image data captured by low-altitude drones to achieve data source transition, effectively fusing the three data sources and producing a highly accurate fusion model.

[0032] The present invention also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory; the processor executes the computer program to implement the steps of the method of the present invention.

[0033] The present invention also provides a computer program product, comprising a computer program / instruction; when the computer program / instruction is executed by a processor, the steps of the method of the present invention are implemented.

[0034] The present invention also provides a computer-readable storage medium having a computer program / instruction stored thereon; when the computer program / instruction is executed by a processor, the steps of the method of the present invention are implemented.

[0035] Compared with the existing technology, the beneficial effects of the present invention are as follows: the present invention uses tilted drones, low-altitude drones, and ground cameras to obtain image data according to certain rules, and performs fusion calculations through a fusion calculation method, which not only meets the conditions for aerial three-dimensional fusion calculations, but also ensures efficient field operation time. At the same time, the modeling effect and accuracy are greatly improved, and the constructed three-dimensional model has picture-level texture and pixel-level fineness. The method of the present invention has the characteristics of strong feasibility, realistic texture effects, and high degree of fineness, and can provide reliable and authentic surveying and mapping geographic information data for applications such as engineering planning, design, display, and three-dimensional simulation. The present invention adds data from low-altitude drone aerial photography, which is more in line with the application scenario of multi-source data in the fusion modeling process, and solves the problem in the existing technology that fusion modeling of images from two types of data sources cannot guarantee that the data can be fused and modeled. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of the real scene modeling method for air-ground fusion of oblique photogrammetry of the present invention;

[0037] Figure 2 To obtain image data according to the planned route, the POS information is used to calculate the aerial triangulation results;

[0038] Figure 3 Aerial triangulation results using POS information for low-altitude UAV aerial images;

[0039] Figure 4 The data captured by the ground camera does not contain position information and the preliminary aerial triangulation results are relative positioning;

[0040] Figure 5 A partial zoom of the preliminary aerial triangulation results of relative positioning of data taken by ground cameras without position information;

[0041] Figure 6 Pick up coordinate points from the tilted model generated in step A;

[0042] Figure 7 The air-ground fusion method is used to obtain the aerial triangulation results;

[0043] Figure 8 A partial image of a 3D real-life model built for oblique drone images;

[0044] Figure 9 A partial image of the 3D real scene model constructed by the air-ground fusion method. DETAILED DESCRIPTION

[0045] The present invention proposes a method for air-ground fusion modeling that can avoid the need for external image control points, avoids the shortcomings of traditional oblique photography 3D modeling, and is also suitable for the construction period requirements of actual engineering projects. It has the characteristics of simple and feasible operation, high modeling accuracy, and true texture. It can provide reliable and accurate geographic information 3D real-scene model data for engineering project planning and design, 3D modeling, GIS platform construction and application, etc.

[0046] The embodiment of the present invention includes the following steps:

[0047] Step A: Tilt the drone according to the planned route to acquire image data. This process also acquires the image position data (POS data). Use software (e.g., ContextCapture Center, Photoscan, etc.) to perform aerial triangulation and create a 3D model. The model coordinate system is consistent with the POS coordinate system.

[0048] Step B. Continuous image data of the target area is captured by a ground camera using a walking, intermittent shooting method. Each stop during the shooting process constitutes a set of camera exposure points. At each stop, the target area is captured in a horizontal rotation around the camera exposure point. The overlap of adjacent images captured at that stop is guaranteed to be greater than 50%. Furthermore, the maximum overlap of images captured between successive stop points must be greater than 50%. Software is used to perform aerial triangulation of the ground portion, ensuring that the aerial triangulation results of the captured elements are continuous and uninterrupted.

[0049] Step C. Select several feature coordinate points from the model built in Step A. The selection rule is to ensure that the image acquired in Step B captures the corresponding feature coordinate points. The coordinate values ​​of the coordinate points are added to the aerial triangulation results of Step B in the form of image control points. The aerial triangulation is performed again to obtain the aerial triangulation results with coordinate information.

[0050] Step D. Low-altitude drone aerial image data. This process involves manually controlling a small rotorcraft drone to continuously photograph the target area. If the target area is complex, multiple routes can be planned using a simulated tilted drone route. The overlap between adjacent captured images within each route layer is guaranteed to be greater than 50%, and the overlap between images between two adjacent manually flown routes is also greater than 50%. The captured image data is then aerially triangulated to obtain continuous aerial triangulation results.

[0051] Step E. Perform a fused aerotriangulation calculation on the results of step A and step D (for fused aerotriangulation calculation, see CN111540048A) to obtain a preliminary fused aerotriangulation result.

[0052] Step F. Perform air-ground fusion air triangulation on the air triangulation results with coordinate information from step C and the preliminary fusion air triangulation results from step E to obtain the air-ground fusion air triangulation results, and use software to reconstruct the model to obtain the final high-precision real-scene 3D model.

[0053] In step A, a five-lens oblique camera is equipped with an oblique drone. The corresponding software is used to plan the route of the entire modeling area according to the specific terrain and requirements. Normally, several important parameters such as altitude, heading overlap, lateral overlap, and cruising range are guaranteed to be correct. The altitude reflects the average resolution of the photographed objects (the value Re is used here to represent the resolution, that is, the ground size represented by each pixel of the camera imaging, and the unit is usually m / pixel, cm / pixel, mm / pixel). According to the planned route, conventional oblique photogrammetry field work is carried out, and the obtained image data and drone POS data are imported into the oblique photogrammetry aerial triangulation modeling software CCC for aerial triangulation calculation ( Figure 2 ) and build a three-dimensional model ( Figure 8 ), the model coordinate system is consistent with the POS coordinate system.

[0054] In step B, the ground camera is operated manually to manually shoot the building facades, characteristic landforms, and original street appearances that are of particular concern within the survey area. The shooting method is manual walking interval shooting. Figure 4 、 Figure 5The result of aerial triangulation using a ground camera is shown. Each walking stop during the capture process forms a set of camera exposure points. At each stop, the target area is captured using a horizontal rotation method, with the camera exposure point as the center. The overlap between adjacent images captured at each stop is guaranteed to be at least 50%. Furthermore, the maximum overlap between images captured at the preceding and subsequent stops must be at least 50%. This means that the overlap between the two closest images between the two closest exposures is also at least 50%. The resolution is controlled between Re / 100 and Re / 10 (generally Re / 25 to Re / 5 is suitable). Software is used to perform aerial triangulation on the ground image data, ensuring continuous and uninterrupted triangulation results for all captured features. Any discontinuities require on-site retakes according to the specified guidelines. The key to this step is ensuring a certain degree of overlap during the ground capture process. This requires determining the distance between the capture point and the target based on the specific situation and hardware; a range of 5-20 meters is generally appropriate.

[0055] In step C, several characteristic coordinate points are selected from the model built in step A, such as Figure 6 The coordinate system of the coordinate point is consistent with the model coordinate system in step A. The selection rule is to ensure that the image acquired in step B captures the corresponding feature coordinate point. The coordinate value of the coordinate point is then added to the aerial triangulation result of step B as an image control point. Aerial triangulation is performed again to obtain an aerial triangulation result with coordinate information. At this point, the coordinate system of the ground image is rigidly brought into line with the coordinate system of the image captured by the tilted drone, meeting the basic requirements for fusion calculation. However, the current resolution difference between ground images and images acquired by tilted drones is significant, making direct fusion triangulation impossible.

[0056] In step D, low-altitude drone aerial image data is obtained. This process involves manually controlling a small rotor drone to continuously shoot the target area. The resolution is controlled between Re / 10 and Re (generally, between Re / 5 and Re is appropriate). If the target area is complex, a multi-layer route planning method can be used to simulate the tilted drone route planning method to ensure that the overlap of adjacent captured images in each layer of route is above 50%, and the image overlap between two adjacent layers of manual flight routes is also above 50%. The captured image data is subjected to aerial triangulation to obtain continuous aerial triangulation results, such as Figure 3 This step is to obtain image data with an excessive resolution between the images taken by the tilted UAV and the ground camera, which is convenient for fusion calculation.

[0057] In step E, the aerial triangulation calculation results in step A and the aerial triangulation calculation results in step D are fused to obtain a preliminary fused aerial triangulation calculation result.

[0058] In step F, the coordinate-containing aerial triangulation results from step C are combined with the preliminary fused aerial triangulation results from step E to perform an air-ground fusion triangulation. Software is then used to reconstruct the model, resulting in a final, high-precision, real-world 3D model. The air-ground fusion triangulation in this step is successful due to the over-resolution triangulation results from the oblique drone data in step E, which incorporates low-altitude drone imagery. Furthermore, the ground image coordinate system in step C has been unified with the oblique drone image coordinate system.

[0059] After the above steps, the aerial triangulation results of multi-source data air-ground fusion calculation can be obtained, such as Figure 7 This result not only ensures the integrity of the modeling area but also satisfies the key areas by integrating the high-precision low-altitude UAV and the high-resolution image data taken by the ground camera. The air-ground fusion aerial triangulation calculation result is the key result obtained by the method of the present invention.

[0060] The present invention can be used in the case of non-image control point operation, because all image data coordinate systems are uniformly incorporated into the POS coordinate system obtained by the tilted drone. Even if the absolute coordinate accuracy of the overall position is insufficient, the overall relative position is unified, which can also construct an air-ground fusion real-scene three-dimensional model.

[0061] If the project has specific absolute coordinate accuracy requirements, the coordinates of the image control points measured in the field can also be added to each aerial triangulation calculation in the implementation steps, which does not affect the specific implementation of the present invention.

[0062] The resolution differences between the oblique drone images and the low-altitude drone images obtained in steps A and D, and between steps D and B, as well as the resolution differences between the low-altitude drone images and the ground camera images, should be controlled within 10 times (generally controlled between 1-5 times). In addition, the manually captured low-altitude drone image data in step D is a transition between the resolution of the oblique drone and the image captured by the ground camera, and is indispensable. Otherwise, the resolution differences between the different data sources will be too large, and the computer software will not be able to directly fuse them during the aerial triangulation calculation.

[0063] The final modeling calculation is performed on the air-ground fusion aerial triangulation calculation results to obtain the air-ground fusion high-precision real scene 3D model data of the method of the present invention, such as Figure 9 shown.

Claims

1. A method for real-scene modeling based on air-ground fusion of oblique photogrammetry, characterized in that: The following steps are involved: Obtain the aerial triangulation results corresponding to the image data taken by the oblique drone, the aerial triangulation results corresponding to the image data taken by the low-altitude drone, and the aerial triangulation results corresponding to the image data taken by the ground camera; Fusing the aerial triangulation results corresponding to the image data taken by the oblique UAV and the aerial triangulation results corresponding to the image data taken by the low-altitude UAV to obtain a first fusion result; fusing the first fusion result with an aerotriangulation result corresponding to the image data captured by the ground camera to obtain a second fusion result; Reconstructing a real-scene three-dimensional model using the second fusion result; The specific process of obtaining the second fusion result includes: A three-dimensional model is established using aerial triangulation results corresponding to the image data captured by the tilted drone, and a number of characteristic coordinate points are selected from the three-dimensional model, where the selected characteristic coordinate points are also points captured by a ground camera; the coordinate system of the three-dimensional model is consistent with the coordinate system of the image position positioning data acquired by the tilted drone, that is, the POS data coordinate system; The coordinate values ​​of the characteristic coordinate points are added to the aerial triangulation results corresponding to the image data taken by the ground camera in the form of image control points, and the aerial triangulation results corresponding to the image data taken by the ground camera with the characteristic coordinate points added are again subjected to aerial triangulation to obtain the aerial triangulation results with coordinate information; Perform air-ground fusion air triangulation on the air triangulation result with coordinate information and the first fusion result to obtain a second fusion result.

2. The method for real-scene modeling based on air-ground fusion of oblique photogrammetry according to claim 1, characterized in that: The specific implementation process of obtaining the aerial triangulation calculation results corresponding to the image data taken by the ground camera includes: the ground camera continuously shoots the image data of the target area, at each shooting point, the target area is shot in a horizontal rotation manner with the exposure point of the ground camera as the center of the circle, the overlap of adjacent images shot at each shooting point is not less than 50%, and the overlap of images obtained between adjacent shooting points is not less than 50%; performing aerial triangulation on the image data taken by the ground camera to obtain the aerial triangulation calculation results corresponding to the image data taken by the ground camera; preferably, the distance between the shooting point and the target area is 5-20m.

3. The method for real-scene modeling based on air-ground fusion of oblique photogrammetry according to claim 1, characterized in that: The specific implementation process of obtaining the aerial triangulation results corresponding to the image data of low-altitude UAV aerial photography includes: planning a layer of routes, using the UAV to continuously shoot the target area on the route, performing aerial triangulation on the image data obtained by the continuous shooting, and obtaining the aerial triangulation results corresponding to the image data of low-altitude UAV aerial photography; or planning multiple layers of routes within the target area, using the UAV to continuously shoot the target area, with the overlap of adjacent images in each layer of routes being no less than 50%, and the overlap of images between two adjacent layers of routes being no less than 50%; performing aerial triangulation on the image data corresponding to each layer of routes, and obtaining the aerial triangulation results corresponding to the multiple layers of image data of low-altitude UAV aerial photography.

4. The method for real-scene modeling based on air-ground fusion of oblique photogrammetry according to claim 3, characterized in that: The process of obtaining the first fusion result includes: When planning a single-layer route, the aerial triangulation results corresponding to the image data taken by the oblique drone and the aerial triangulation results corresponding to the image data taken by the low-altitude drone are fused to obtain a first fusion result. In the case of planning a multi-layer route, the aerial triangulation results corresponding to the image data taken by the oblique drone are fused with the aerial triangulation results corresponding to the multi-layer image data taken by the low-altitude drone, and the lowest resolution aerial triangulation result is obtained. Performing a fusion operation on the first preliminary fusion result and the aerial triangulation result with the second lowest resolution among the aerial triangulation results corresponding to the multi-layer image data taken by the low-altitude UAV to obtain a second preliminary fusion result; This process is deduced in this way until all the aerial triangulation calculation results corresponding to the multi-layer image data of the low-altitude UAV aerial photography are fused to obtain the first fusion result.

5. The method for real-scene modeling based on air-ground fusion of oblique photogrammetry according to claim 1, characterized in that: The resolution of the image data taken by the tilted drone is 1 to 10 times the resolution of the image data taken by the low-altitude drone; the resolution of the image data taken by the low-altitude drone is 1 to 10 times the resolution of the image data taken by the ground camera.

6. The method for real-scene modeling based on air-ground fusion of oblique photogrammetry according to claim 5, characterized in that: The resolution of the image data taken by the tilted drone is 1 to 5 times the resolution of the image data taken by the low-altitude drone; the resolution of the image data taken by the low-altitude drone is 1 to 5 times the resolution of the image data taken by the ground camera.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory; characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

8. A computer program product comprising a computer program / instructions; characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program / instruction stored thereon; characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images