Method for reconstructing water surface area for three-dimensional model data of drone oblique photography

Abstract

The present invention relates to a method for reconstructing a water surface area from three-dimensional model data of drone oblique photography. The three-dimensional model data of drone oblique photography is converted into point cloud data and subjected to meshing processing. By using the density and elevation characteristics of the point cloud data in the water surface area, a method for extracting the water surface area boundary is designed to quickly screen the water area boundary point cloud. Based on the results of the water area boundary extraction, an irregular triangular network model of the water surface area is constructed. Based on the EXIF information of the drone down-view image data, intelligent selection of the water surface area texture image is realized, and combined with the spatial inclusion relationship between the texture image and the TIN model, an intelligent processing algorithm is designed to generate the final water surface area texture image data, realizing the reconstruction of the water surface area. The content of the present invention can improve the existing three-dimensional modeling technology system of drone oblique photogrammetry, realize the rapid and automatic reconstruction of real water surface information, and has high popularization and application value in fields such as twin river basins and twin cities.

Description

Technical Field

[0001] The present invention belongs to the field of 3D modeling, and particularly relates to a method for rapidly reconstructing a water surface area for 3D model data of unmanned aerial vehicle (UAV) oblique photography. Background Art

[0002] The UAV oblique photogrammetry technology provides a new and effective means for the rapid 3D reconstruction of large-area geospatial elements, and has been widely applied in 3D modeling work at the urban and basin scales. Compared with other geospatial elements, the water surface area is easily affected by factors such as wind waves and reflections, resulting in the inability of the existing technology to achieve the rapid reconstruction of the water surface area. Therefore, in the existing 3D visualization systems, the true water surface information of rivers and lakes is rarely displayed. In general application scenarios, the true water surface information is rarely concerned, and a preset (hypothetical) water surface model is usually used to replace the true water surface information to obtain a good visualization effect. For some special application fields (such as ecological environment monitoring, water pollution supervision, etc.), the true water surface information (water color, floating objects, etc.) is an important decision-making basis, and the preset water surface model cannot meet the requirements. Although some commercial software (such as 3ds Max, DP-Modeler, etc.) can perform post-repair on the water surface area in the UAV oblique photography 3D model, the entire process requires a large amount of manual participation and is very inefficient; more importantly, such software usually uses methods such as texture interpolation and texture sampling to realize the reconstruction of water surface information, and it is still very difficult to visually express the true water surface information in the 3D scene. Therefore, realizing the rapid reconstruction of the water surface area in the UAV oblique photography 3D model data has important practical significance and application value. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for rapidly reconstructing a water surface area for 3D model data of UAV oblique photography, so as to supplement and improve the existing UAV oblique photography 3D modeling technology system.

[0004] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0005] A method for reconstructing a water surface area for 3D model data of UAV oblique photography includes:

[0006] Converting the oblique photography 3D model into point cloud data, and performing grid processing on the point cloud data; designing an eight-neighborhood traversal method to extract the point cloud data at the water area boundary with the point cloud density and elevation information in the grid as constraint conditions;

[0007] Based on the extraction result of the point cloud data at the water area boundary, constructing an irregular triangular mesh model of the water surface area;

[0008] Traverse the EXIF information of the downward-looking images of the UAVs, extract the coordinate values of the images, convert them into Gauss plane rectangular coordinates, and select the downward-looking image of the UAV with the center point coordinates closest to the center point coordinates of the irregular triangular network model as the preliminary selection result of the texture image;

[0009] Conduct a spatial inclusion relationship analysis on the selected downward-looking images of the UAVs and the irregular triangular network model, and perform image stitching on the downward-looking images of the UAVs until the selected downward-looking images of the UAVs completely contain the irregular triangular network model, that is, obtain the final texture image of the water area;

[0010] Assign the final texture image of the water area to the irregular triangular network model of the water area to achieve water surface reconstruction.

[0011] As a preferred implementation manner, after converting the oblique photography three-dimensional model into point cloud data, first perform straight-through filtering processing on the point cloud data, and then perform meshing processing, which can reduce the number of point clouds to be processed and improve the execution efficiency of the algorithm.

[0012] Furthermore, the threshold of the straight-through filtering is dynamically determined by the elevation of the point cloud.

[0013] As a preferred implementation manner, during meshing processing, the grid is set as a square, and the side length of the grid is dynamically determined according to the point cloud density.

[0014] As a preferred implementation manner, the starting point cloud grid for eight-neighborhood traversal is located at the boundary of the water area.

[0015] As a preferred implementation manner, the method for extracting the point cloud data at the water area boundary by designing an eight-neighborhood traversal method with the point cloud density and elevation information in the grid as constraint conditions includes:

[0016] ① Calculate the point cloud density in the eight neighborhood grids of the starting grid, and compare them with the point cloud density in the starting grid respectively. If the difference between a certain neighborhood grid and the point cloud density in the starting grid is less than the preset threshold, record the grid number; if the differences between all neighborhood grids and the point cloud density in the starting grid are not less than this threshold, record the grid number of the neighborhood grid with the smallest difference from the point cloud density in the starting grid;

[0017] ② Traverse each point cloud grid according to the neighborhood grid number recorded in step ①, calculate the average elevation value of the point cloud in each grid respectively, and compare it with the average elevation value of the point cloud in the starting grid. Select the grid with the smallest difference between the average elevation of the point cloud and the average elevation of the point cloud in the starting grid, and regard it as the newly determined point cloud grid at the water area boundary;

[0018] ③ If the point cloud grid at the boundary of the newly determined water surface area has not been traversed, it is regarded as a new starting grid, and ①-② are repeated; otherwise, the water area boundary extraction program ends, and the set of point clouds at all searched water surface area boundaries is the water surface area boundary extraction result.

[0019] As a preferred implementation manner, based on the three-dimensional point cloud data of the extracted water surface area boundary, set all the point cloud elevation values to a unified value, randomly add several points inside it as vertices, and use the Delaunay algorithm to construct an irregular triangular network model of the water surface area. Considering the characteristics of the water surface, set all the point cloud elevation values at the boundary of the extracted water surface area to a unified value (the average value of all point cloud elevations), and the number of randomly added points in the water surface area is dynamically determined according to the coordinate range of the point clouds at the boundary of the extracted water surface area.

[0020] As a preferred implementation manner, the analysis of the spatial inclusion relationship between the selected UAV downward-looking image and the irregular triangular network model includes:

[0021] Judge the inclusion relationship between the selected UAV downward-looking image and the irregular triangular network model. If the spatial range of the selected UAV downward-looking image completely contains the irregular triangular network model: regard this UAV downward-looking image as the finally determined water surface area texture image; otherwise, search for the image closest to the UAV downward-looking image from other UAV downward-looking images, splice it with the current UAV downward-looking image to form a new UAV downward-looking image, and judge the spatial inclusion relationship between the new UAV downward-looking image and the irregular triangular network model, and repeat until the spatial range of the obtained UAV downward-looking image completely contains the irregular triangular network model.

[0022] As a preferred implementation manner, analyze the spatial inclusion relationship between the selected UAV downward-looking image and the irregular triangular network model based on the following formula:

[0023]

[0024]

[0025] In the formula, x min , y min are respectively the minimum horizontal and vertical coordinates of the vertices in the irregular triangular network model, x max , y max are respectively the maximum horizontal and vertical coordinates of the vertices in the irregular triangular network model, X min , Y min are respectively the minimum horizontal and vertical coordinates of the pixels in the UAV downward-looking image, X max , Y max are respectively the maximum horizontal and vertical coordinates of the pixels in the UAV downward-looking image; (X′ i ,Y′i ) represents the rectangular coordinates of the center point of the downward-looking image of the UAV. W and H are the width and height values of the downward-looking image of the UAV respectively. Res represents the spatial resolution of the downward-looking image of the UAV. is the floor function;

[0026] When all the conditions in the above formula are satisfied, it is considered that the selected spatial range of the downward-looking image of the UAV completely contains the irregular triangular network model of the water surface area.

[0027] As a preferred implementation, for obtaining the final texture image of the water surface area, use the vector boundary of the irregular triangular network model of the water surface area to crop it. After generating the image data containing only the texture information of the water surface area, assign it to the irregular triangular network model of the water surface area to achieve water surface reconstruction.

[0028] The method for quickly reconstructing the water surface area for the three-dimensional model data of UAV oblique photography of the present invention aims at the problem that the existing UAV oblique photography three-dimensional modeling technology system cannot realize the automatic reconstruction of real water surface information, and designs a solution. First, using the point cloud density and elevation characteristics of the water surface area and non-water surface area, a method for extracting the water surface area boundary based on eight-neighborhood traversal is designed; then, based on the water area boundary extraction result, use the Delaunay method to construct an irregular triangular network (TIN) model of the water surface area; finally, design an intelligent selection and processing method for the texture image of the water surface area, and realize the rapid reconstruction of real water surface information. The content of the present invention can improve the existing UAV oblique photogrammetry three-dimensional modeling technology system, realize the rapid and automatic reconstruction of real water surface information, and has high popularization and application value in fields such as twin basins and twin cities.

[0029] It should be understood that all combinations of the foregoing concepts and additional concepts described in more detail below can be regarded as part of the inventive subject matter of the present disclosure as long as such concepts do not conflict with each other. In addition, all combinations of the claimed subject matter are regarded as part of the inventive subject matter of the present disclosure.

[0030] The foregoing and other aspects, embodiments, and features of the teachings of the present invention can be more fully understood from the following description in conjunction with the accompanying drawings. Other additional aspects of the present invention, such as the features and / or beneficial effects of exemplary embodiments, will be apparent from the following description, or will be learned through the practice of specific embodiments according to the teachings of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings are not necessarily drawn to scale. In the accompanying drawings, each identical or approximately identical component shown in each figure may be denoted by the same reference numeral. For the sake of clarity, not every component is labeled in each figure. Now, embodiments of various aspects of the present invention will be described by way of example and with reference to the accompanying drawings, wherein:

[0032] Figure 1 is the flowchart of the method of the present invention.

[0033] Figure 2 is the flowchart for extracting the boundary of the water surface area.

[0034] Figure 3 is the flowchart for constructing a triangular mesh and texture mapping of the water surface area.

[0035] Figure 4 is the conversion result of the point cloud of the 3D model of the drone oblique photography.

[0036] Figure 5 is the schematic diagram of the grid processing of the point cloud data.

[0037] Figure 6 is the extraction result of the water surface area boundary of the point cloud data.

[0038] Figure 7 is the construction result of the irregular triangular mesh (TIN) model of the water surface area.

[0039] Figure 8 is the cropping result of the texture image of the water surface area.

[0040] Figure 9 is the reconstruction result of the water surface area.

[0041] Figure 10 is the ten pieces of 3D model data of the drone oblique photography used in Embodiment 2.

[0042] Figure 11 is the qualitative evaluation result of the method of the present invention. Detailed Embodiments

[0043] To better understand the technical content of the present invention, specific embodiments are hereby given and described in conjunction with the accompanying drawings as follows.

[0044] Aspects of the present invention are described with reference to the accompanying drawings in this disclosure, in which a number of illustrative embodiments are shown. Embodiments of the present disclosure do not necessarily define all aspects of the present invention. It should be understood that the various concepts and embodiments introduced above, as well as those concepts and embodiments described in more detail below, can be implemented in any of a number of ways, because the concepts and embodiments disclosed in the present invention are not limited to any implementation. In addition, some aspects disclosed in the present invention can be used alone, or in any suitable combination with other aspects disclosed in the present invention.

[0045] Embodiment 1

[0046] This embodiment illustrates a specific implementation manner of the present invention.

[0047] This example quickly reconstructs the water surface area in the drone oblique three-dimensional model data. The specific implementation process is as follows: First, convert the original three-dimensional model data into point cloud data; then use the pass-through filtering method to process the obtained point cloud data to reduce the number of point clouds to be processed to improve the algorithm execution efficiency, and perform meshing processing on the obtained point cloud data; based on the point cloud data characteristics of the water surface area, extract the water area boundary, and further construct an irregular triangular network (TIN) model of the water surface area; based on the EXIF information of the drone down-looking image, intelligently select and process to generate the texture image data of the water surface area; finally, map the texture image to the water surface area TIN model to achieve the reconstruction of real water surface information.

[0048] As an exemplary description, the implementation of the foregoing method is specifically described below with reference to the accompanying drawings.

[0049] The method process is as Figures 1 to 3 shown, including the following steps:

[0050] Step 1: Convert the original drone oblique photography three-dimensional model data into point cloud data

[0051] Call the pcl_mesh_sampling.exe tool in the open source library PCL (Point Cloud Library) to convert the oblique photography three-dimensional model data (.obj format) into point cloud data (.pcd format). The conversion result is shown in Figure 4 , where (a) is the original oblique photography three-dimensional model, and (b) is the point cloud conversion result.

[0052] Step 2: Filter the obtained point cloud data

[0053] The PassThrough() method in the PCL (Point Cloud Library) library is called to process the acquired point cloud data, eliminate the influence of irrelevant point clouds on the reconstruction of the water surface area, and improve the execution efficiency of the algorithm.

[0054] The threshold of the pass filter is determined according to the elevation of the point cloud. In this example, the point cloud elevation range of the constructed filter is -20 to 30 (unit: meter).

[0055] Step 3: Grid the filtered point cloud data

[0056] Set a square grid, whose side length is determined by the density of the acquired point cloud; in this example, the grid side length is set to 1 meter. Divide the point cloud into corresponding grids according to the horizontal and vertical coordinates. See the schematic diagram of point cloud gridding processing Figure 5 .

[0057] Step 4: Extract the water surface area boundary based on the gridded point cloud data

[0058] Based on the point cloud features of the water surface area, an eight-neighborhood traversal method is designed to extract point cloud data at the boundary of the water area. First, the starting grid is manually determined and ensured to be located on the boundary of the water surface area; then, the next point cloud grid located on the boundary of the water surface area is searched from the eight neighboring grids around the starting grid. The search process is as follows:

[0059] ① Calculate the point cloud density (number of points / grid area) in the eight neighborhood grids and compare them with the point cloud density in the starting grid. If the difference between the point cloud density of a neighborhood grid and the starting grid is less than a given threshold (determined based on experience; the threshold in this example is 10 / square meter), record the number of the neighborhood grid; if the difference between the point cloud density of all neighborhood grids and the starting grid is not less than the threshold, record the number of the neighborhood grid with the smallest difference in point cloud density with the starting grid;

[0060] ② According to the neighborhood grid number recorded in step ①, traverse each point cloud grid, calculate the average elevation value of the point cloud in each grid, and compare it with the average elevation value of the point cloud in the starting grid. Select the grid with the smallest difference between the average elevation of the point cloud and the average elevation of the starting grid, and regard it as the point cloud grid at the boundary of the newly determined water surface area;

[0061] ③If the point cloud grid at the newly determined water surface area boundary has not been traversed, it will be regarded as a new starting grid and repeat ①-②; otherwise, the water area boundary extraction program ends, and the set of point clouds at all searched water surface area boundaries is the water surface area boundary extraction result.

[0062] The results of water surface area boundary extraction are shown in Figure 6, where (a) is the original oblique photography 3D model, (b) is the point cloud conversion result, and (c) is the water area boundary extraction result.

[0063] Step 5: Construct an irregular triangular network (TIN) model for the water area based on the water area boundary extraction result

[0064] Randomly add a certain number (200 in this example) of vertices inside the extracted water area boundary (3D point cloud), and use the Delaunay method to jointly construct a network for the water area boundary extraction result and the randomly added points. The generated result is the TIN model of the water area. The construction result of the TIN model is shown in Figure 7 , where (a) is the water area boundary extraction result and (b) is the construction result of the water area TIN model.

[0065] Step 6: Intelligently select and process to generate a texture image for the water area based on the constructed water area TIN model

[0066] ① Calculate the coordinates of the center point of the TIN model based on Equation (1) according to the point cloud coordinates in the water area TIN model. Where: (x c , y c ) is the calculation result of the center point coordinates, x max , y max , x min , y min are respectively the maximum x coordinate, maximum y coordinate, minimum x coordinate, and minimum y coordinate of the point cloud in the TIN model.

[0067]

[0068] ② Traverse the EXIF information of the drone's downward-looking image data, and use Equation (2) to convert the longitude and latitude of each image data into Gaussian plane rectangular coordinates, and then select the image with the center point coordinates closest to the center point coordinates of the TIN as the initially determined texture image;

[0069]

[0070] Among them, (x, y) is the calculation result of the Gaussian plane rectangular coordinates of the center point of the image, B and L are respectively the longitude and latitude of the center point of the image, N is the radius of curvature of the prime vertical, L0 is the longitude of the central meridian, a and b are respectively the major axis radius and minor axis radius of the earth ellipsoid, ρ = 206264.806247096355″, and X represents the ellipsoidal arc length from the equator to the projection point of the center point of the figure on the reference ellipsoid.

[0071] ③ Determine whether the initially determined spatial range of the texture image completely contains the TIN model of the water surface area. The determination method is shown in Equations (3) and (4). If this containment relationship holds, then take the initially determined texture image as the selection result of the water surface area texture image. If this containment relationship does not hold, continue to select the image with the coordinates closest to those of the initially determined texture image, splice the selected image with the initially determined texture image to form a new texture image, and take this new texture image as the initially determined texture image.

[0072]

[0073]

[0074] In Equation (3), x min and y min are respectively the minimum horizontal and vertical coordinates of the vertices in the TIN model, x max and y max are respectively the maximum horizontal and vertical coordinates of the vertices in the TIN model, X min and Y min are respectively the minimum horizontal and vertical coordinates of the pixels in the texture image, X max and Y max are respectively the maximum horizontal and vertical coordinates of the pixels in the texture image. In Equation (4), (X′ i , Y′ i ) represents the rectangular coordinates of the center point of the texture image, W and H are respectively the width and height values of the image, Res represents the spatial resolution of the image, is the floor function.

[0075] ④ Repeat ③ until the initially determined spatial range of the texture image completely contains the TIN model of the water surface area; take the initially determined texture image as the selection result of the water surface area texture image.

[0076] Step 7: Map the texture image to the TIN model of the water surface area to achieve the reconstruction of the water surface area

[0077] Use the vector boundary of the TIN model of the water surface area to crop the texture image of the water surface area, further remove the interference elements in the texture image, and obtain the texture information that only retains the water surface area (see the result of the texture image cropping in Figure 8 , where (a) is the selection result of the texture image of the water surface area and (b) is the cropping result); assign the cropped texture image to the TIN model of the water surface area to achieve the reconstruction of the water surface area (see the reconstruction result in Figure 9 , where (a) is the TIN model of the water surface area, (b) is the water surface reconstruction result, and (c) is the integration result of the water surface area reconstruction result and the original oblique photography three-dimensional model).

[0078] Example 2

[0079] This embodiment illustrates the reconstruction efficiency and accuracy evaluation results of the water surface area for the 3D model data of UAV oblique photography of the present invention.

[0080] Ten pieces of 3D model data of UAV oblique photography containing water surface areas and the original UAV image data were selected (see Figure 10 ), and the 3D reconstruction of the water surface area was carried out using the method designed by the present invention. The computer configuration used is shown in Table 1.

[0081] Table 1 Computer configuration information for the experiment

[0082]

[0083] First, the effect of water surface reconstruction was tested from a qualitative perspective. As Figure 11 shown, (a) is a on-site real-shot photo of the UAV, and (b) is a computer screen capture of the water surface reconstruction result. As can be seen from the results in Figure 11 , the method designed by the present invention has a good visual effect.

[0084] In addition, the accuracy of the water surface area boundary extraction result was evaluated from a quantitative perspective. The water surface area boundary was extracted from the original oblique photography 3D model data by manual delineation, and its accuracy was compared with the water surface area boundary extracted by the method designed by the present invention. The evaluation indices include the average error of vertex coordinates (AE), the root mean square error of vertex coordinates (RMSE), the standard deviation of vertex coordinates (SD), and the difference in water surface area (EOA). The calculation methods for the four evaluation indices are shown in Equation (5), where (x i , y i ) is the coordinate of the i-th point in the water surface area boundary extraction result, (x′ i , y′ i ) is the coordinate of the point closest to (x i , y i ) in the manually delineated water surface area boundary, n is the number of points included in the water surface area boundary extraction result, S a is the area enclosed by the water surface area boundary extracted by the algorithm, and S m is the area enclosed by the manually delineated water surface area boundary.

[0085]

[0086] The accuracy evaluation results are shown in Table 2. As can be seen from the results in Table 2, the water surface area boundary extraction algorithm designed by the present invention has a high accuracy.

[0087] Table 2 Accuracy evaluation results

[0088]

[0089]

[0090] Finally, the water surface reconstruction efficiency of the design method of the present invention was tested, and the results are shown in Table 3. As can be seen from the results in Table 3, using a computer device with a conventional configuration, the method designed by the present invention has ideal efficiency.

[0091] Table 3 Efficiency test results

[0092]

Claims

1. A method for reconstructing a water surface area from three-dimensional model data of oblique photography of an unmanned aerial vehicle, characterized in that, Including: Converting the oblique photography 3D model into point cloud data and performing meshing processing on the point cloud data; Taking the point cloud density and elevation information within the grid as constraint conditions, designing an eight-neighborhood traversal method to extract the point cloud data at the water area boundary; Based on the extraction result of the point cloud data at the water area boundary, constructing an irregular triangular network model of the water surface area; Traversing the EXIF information of the UAV downward-looking image and extracting the coordinate values of the image. After converting them into Gaussian plane rectangular coordinates, selecting the UAV downward-looking image with the center point coordinates closest to the center point coordinates of the irregular triangular network model as the preliminary extraction result of the texture image; Performing spatial inclusion relationship analysis on the selected UAV downward-looking image and the irregular triangular network model based on the following formula: ； ； In the formula, x min , y min are respectively the minimum horizontal and vertical coordinates of the vertices in the triangulated irregular network model. x max , y max are respectively the maximum horizontal and vertical coordinates of the vertices in the triangulated irregular network model. X min , Y min are respectively the minimum horizontal and vertical coordinates of the pixels in the downward-looking image of the unmanned aerial vehicle. X max , Y max are respectively the maximum horizontal and vertical coordinates of the pixels in the downward-looking image of the unmanned aerial vehicle; represents the rectangular coordinates of the center point of the downward-looking image of the unmanned aerial vehicle. W and H are respectively the width and height values of the downward-looking image of the unmanned aerial vehicle, and Res represents the spatial resolution of the downward-looking image of the unmanned aerial vehicle. is the floor function; When all the conditions in the above formula are satisfied, it is considered that the spatial range of the selected UAV downward-looking image completely contains the irregular triangular network model of the water surface area; Performing image stitching on the UAV downward-looking image until the selected UAV downward-looking image completely contains the irregular triangular network model, that is, obtaining the final texture image of the water surface area; Assigning the final texture image of the water surface area to the irregular triangular network model of the water surface area to achieve water surface reconstruction.

2. The method according to claim 1, characterized in that, After converting the oblique photography 3D model into point cloud data, first perform direct filtering processing on the point cloud data, and then perform meshing processing.

3. The method according to claim 1, characterized in that The threshold of the direct filtering is dynamically determined by the point cloud elevation.

4. The method according to claim 1, characterized in that, During meshing processing, the grid is set as a square, and the side length of the grid is dynamically determined according to the point cloud density.

5. The method according to claim 1, wherein The starting point cloud grid of the eight-neighborhood traversal is located at the boundary of the water surface area.

6. The method according to claim 5, wherein The method of designing an eight-neighborhood traversal method to extract the point cloud data at the water area boundary with the point cloud density and elevation information within the grid as constraint conditions includes: ① Calculating the point cloud density in the eight neighborhood grids of the starting grid and comparing them with the point cloud density in the starting grid respectively. If the difference in point cloud density between a certain neighborhood grid and the starting grid is less than the preset threshold, record the grid number; if the difference in point cloud density between all neighborhood grids and the starting grid is not less than this threshold, record the grid number of the neighborhood grid with the smallest difference in point cloud density from the starting grid; ② Traversing each point cloud grid according to the neighborhood grid number recorded in step ①, calculating the average point cloud elevation value in each grid respectively, and comparing it with the average point cloud elevation value in the starting grid. Select the grid with the smallest difference in average point cloud elevation from the starting grid and regard it as the newly determined point cloud grid at the water surface area boundary; If the point cloud grid at the boundary of the newly determined water surface area has not been traversed, it is regarded as a new starting grid, and steps ①-② are repeated; otherwise, the water area boundary extraction program ends, and the set of point clouds at the boundaries of all the searched water surface areas is the result of the water surface area boundary extraction.

7. The method according to claim 1, characterized in that, Based on the extracted 3D point cloud data at the water surface area boundary, setting all the point cloud elevation values to a unified value, randomly adding several points inside it as vertices, and using the Delaunay algorithm to construct an irregular triangular network model of the water surface area.

8. The method according to claim 1, wherein The performing spatial inclusion relationship analysis on the selected UAV downward-looking image and the irregular triangular network model includes: Judge the inclusion relationship between the selected downward-looking UAV image and the irregular triangular network model. If the spatial range of the selected downward-looking UAV image completely contains the irregular triangular network model: use this downward-looking UAV image as the finally determined water area texture image; otherwise, search for the image closest to the downward-looking UAV image from other downward-looking UAV images, splice it with the current downward-looking UAV image to form a new downward-looking UAV image, and judge the spatial inclusion relationship between the new downward-looking UAV image and the irregular triangular network model, and repeat until the spatial range of the obtained downward-looking UAV image completely contains the irregular triangular network model.

9. The method according to claim 1, wherein For the finally obtained water area texture image, use the vector boundary of the irregular triangular network model of the water area to crop it, generate image data containing only water area texture information, and then assign it to the irregular triangular network model of the water area to realize water surface reconstruction.

Images