A method and apparatus for generating 3D meshes for stereo satellite 3D reconstruction

Abstract

A method for generating 3D meshes for stereo satellite 3D reconstruction, belonging to the field of satellite image 3D reconstruction production, is characterized by: intersecting DSM imagery and building vector shape data to obtain roof DSM and ground DEM; processing the building vector shape data separately with the roof DSM and ground DEM to obtain 3D vector feature points; performing raster-to-point cloud processing on the ground DEM and roof DSM; constructing Delaunay triangulation meshes in different regions; and merging and outputting the final mesh. By using high-resolution DSM imagery and corresponding regional building vector shape data as input, and extracting terrain features from the reference DEM and input DSM using slope calculation methods, this satellite large-area 3D mesh generation method based on building geometric features and DSM effectively overcomes the technical problems of traditional methods where building sides are too fragmented, have obvious seams, and unnatural texture transitions in large-area satellite 3D scenes. It fully utilizes existing terrain features, effectively improving the accuracy and aesthetics of the 3D mesh.

Description

Technical Field

[0001] This invention belongs to the field of satellite image 3D reconstruction production, and particularly relates to a method and apparatus for generating 3D meshes for stereo satellite 3D reconstruction. Background Technology

[0002] Satellite imagery has advantages over drones, such as a wider coverage area and lower cost. With the continuous improvement of satellite resolution, 3D reconstruction based on stereo satellites is suitable for a wide range of rapid 3D reconstruction applications, such as smart cities, virtual reality applications, and emergency disaster relief.

[0003] Currently, the mainstream methods for obtaining 3D meshes include: the Poisson surface reconstruction method, which requires the normal vectors of the point cloud, but the reconstructed mesh often exhibits issues such as bulging; the ball-pivoting algorithm, suitable for processing scanned data, has high requirements for initial point cloud data, such as pre-constructing a kd-tree, increasing algorithm complexity, and also suffers from difficulties in seed point selection; the moving cube algorithm suffers from significant information loss during reconstruction, resulting in low accuracy, and the large number of triangular faces increases storage requirements, while its inherent ambiguity may lead to inaccurate reconstruction results; the greedy triangulation algorithm for point clouds relies on the smoothness and uniformity of the point cloud, and cannot repair holes; and the 3D Delaunay triangulation algorithm suffers from high computational complexity, large computational and memory requirements, and difficulty in algorithm optimization. Considering the characteristic of incomplete point cloud information on the sides of buildings in stereo satellite 3D reconstruction, none of the above algorithms can effectively solve the problems of straight building edges, narrow side TINs leading to obvious texture seams and streaks during 3D mesh generation. Summary of the Invention

[0004] The present invention aims to solve the above problems and provides a method and apparatus for generating three-dimensional meshes for three-dimensional reconstruction of stereo satellites.

[0005] In a first aspect, the present invention provides a method for generating a 3D mesh for 3D reconstruction of a stereo satellite, comprising the following steps:

[0006] Step 1) Intersect the DSM (Digital Surface Model) image and building vector Shp data of the same area to obtain the roof DSM and ground DEM (Digital Elevation Model).

[0007] Step 2) Process the building vector Shp data with the roof DSM and ground DEM respectively to obtain a three-dimensional vector, and mark the points on the Shp vector line to obtain the three-dimensional vector feature points;

[0008] Step 3) Perform raster-to-point cloud processing on the ground DEM and roof DSM;

[0009] Step 4) Perform point insertion processing on the ground DEM and roof DSM from the lower left corner point respectively. When the difference between the elevation of the inserted point and the current point is greater than the threshold p1 or the distance between the two is greater than the threshold p2, then insert the point set.

[0010] Step 5) When the distance between two 3D vector feature points is greater than the threshold p2, insert several points so that the distance between the two points is not greater than p2, and construct the point cloud of the building side.

[0011] The distance between two points in a 3D vector feature is the side length of the 3D vector. When the side length exceeds a threshold p2, one or more points are added to that side so that the distance between points on the side is no greater than p2. This involves dividing the side length into segments based on distance p2, obtaining the coordinates (x, y, z) of each segment point. A grid is then created on the side of the 3D vector based on these segment points to obtain grid points, thus constructing a point cloud of the building's side profile.

[0012] Step 6) Construct the Delaunay triangulation by partitioning the network;

[0013] Step 7) Merge and output the final Mesh.

[0014] Furthermore, the three-dimensional mesh generation method for stereo satellite three-dimensional reconstruction described in this invention, step 2), which involves processing the building vector Shp data with the roof DSM and ground DEM respectively to obtain a three-dimensional vector and marking the points on the Shp vector line, specifically includes the following steps:

[0015] Step 21) Overlay the roof DSM and building vector Shp data, traverse the DSM, and if the grid point is outside the building vector polygon, assign the pixel value of the corresponding DEM coordinate point at the grid point; if the grid point is inside the building vector polygon, assign the pixel value of the corresponding DSM coordinate point at the grid point.

[0016] Step 22) Copy the ground DEM to DEM2, and use the inverse distance weighting method to fill in the blank areas in DEM2;

[0017] Step 23) Overlay the building vector Shp and DEM2, record the intersection points of the endpoints of the line segments in Shp and DEM2, record the plane coordinates and elevation values, and obtain the point set s1; if the line segment length is greater than the preset threshold, add a new point on the line segment, record the elevation value corresponding to the new point, and add it to s1.

[0018] Step 24) Overlay the building vector Shp and DSM, obtain any edge of the Shp vector polygon as the current edge, and traverse the points on the current edge; if the elevation difference between the two endpoints of the current edge is greater than a preset threshold, add the point to the current edge and store it in the point set s2; then obtain any edge of the remaining edges of the Shp vector polygon as the current edge, and perform the same process until all the points on all edges of the vector polygon have been traversed.

[0019] Step 25) Each side of the vector polygon represents a side view. Record the intersection point and elevation value of each side endpoint with DSM and DEM2 respectively, and denote it as point set b. Collect the point sets of each side to obtain the point set B of the building side view.

[0020] Step 26) The pixel values ​​in DSM and DEM represent elevation values. The horizontal coordinate of the raster image is used as the X value, the vertical coordinate as the Y value, and the pixel value as the Z value. The raster is converted into DSM point set Q1 and DEM point set Q2.

[0021] Step 27) Add s1 to Q1 and s2 to Q2 respectively, and label them as building vector points.

[0022] Furthermore, in the three-dimensional mesh generation method for stereo satellite three-dimensional reconstruction described in this invention, step 4) involves inserting points from the lower left corner of the ground DEM and the roof DSM respectively. When the difference between the elevation of the inserted point and the current point is greater than a threshold p1 or the distance between them is greater than a threshold p2, a point set is inserted. Specifically, based on the maximum elevation tolerance threshold p1 and the maximum distance threshold p2, the point sets P1 and P2 of the DSM point set and the DEM point set are selected from the triangulation point set and the point set P2 respectively.

[0023] Step 41) Insert the first point in the DSM point set into P1;

[0024] Step 42) Using the newly added point p in the P1 point set as the center and the planar distance D as the radius, search the DSM point set for points whose elevation difference from the center point p is greater than h, and add them to the P1 point set. Repeat this process until all points that meet the conditions are added to the P1 point set, and mark point p as completed with the mark f. Points located inside the circle with an elevation difference less than h are removed from the DSM point set. Points marked as building vector points must be added to the point set P1.

[0025] Step 43) Iterate through all unmarked points in the P1 point set until the DSM point set is empty.

[0026] Secondly, the present invention provides a three-dimensional mesh generation system for three-dimensional reconstruction of stereo satellites, including a data acquisition module, a point cloud processing module, a triangulation construction module, and a merging output module;

[0027] The data acquisition module is used to acquire DSM images and building vector Shp data of the same area and perform intersection processing to obtain roof DSM and ground DEM; the building vector Shp data is processed separately with the roof DSM and ground DEM to obtain three-dimensional vectors, and points are marked on the Shp vector lines.

[0028] The point cloud processing module is used to perform raster-to-point cloud processing on the ground DEM and roof DSM; it performs point insertion processing on the ground DEM and roof DSM from the lower left corner point respectively. When the difference between the elevation of the inserted point and the current point is greater than the threshold p1 or the distance between the two is greater than the threshold p2, a point set is inserted; when the distance between two 3D vector feature points is greater than the threshold p2, several points are inserted so that the distance between the two points is not greater than p2, thus constructing the point cloud of the building side.

[0029] The triangulation construction module is used to construct Delaunay triangulations by partition;

[0030] The merged output module is used to merge and output the final Mesh.

[0031] Furthermore, the 3D Mesh generation system for stereo satellite 3D reconstruction described in this invention, which processes building vector Shp data with roof DSM and ground DEM respectively to obtain 3D vectors and marks points on the Shp vector lines, specifically includes the following steps:

[0032] Step 21) Overlay the roof DSM and building vector Shp data, traverse the DSM, and if the grid point is outside the building vector polygon, assign the pixel value of the corresponding DEM coordinate point at the grid point; if the grid point is inside the building vector polygon, assign the pixel value of the corresponding DSM coordinate point at the grid point.

[0033] Step 22) Copy the ground DEM to DEM2, and use the inverse distance weighting method to fill in the blank areas in DEM2;

[0034] Step 23) Overlay the building vector Shp and DEM2, record the intersection points of the endpoints of the line segments in Shp and DEM2, record the plane coordinates and elevation values, and obtain the point set s1; if the line segment length is greater than the preset threshold, add a new point on the line segment, record the elevation value corresponding to the new point, and add it to s1.

[0035] Step 24) Overlay the building vector Shp and DSM, obtain any edge of the Shp vector polygon as the current edge, and traverse the points on the current edge; if the elevation difference between the two endpoints of the current edge is greater than a preset threshold, add the point to the current edge and store it in the point set s2; then obtain any edge of the remaining edges of the Shp vector polygon as the current edge, and perform the same process until all the points on all edges of the vector polygon have been traversed.

[0036] Step 25) Each side of the vector polygon represents a side view. Record the intersection point and elevation value of each side endpoint with DSM and DEM2 respectively, and denote it as point set b. Collect the point sets of each side to obtain the point set B of the building side view.

[0037] Step 26) The pixel values ​​in DSM and DEM represent elevation values. The horizontal coordinate of the raster image is used as the X value, the vertical coordinate as the Y value, and the pixel value as the Z value. The raster is converted into DSM point set Q1 and DEM point set Q2.

[0038] Step 27) Add s1 to Q1 and s2 to Q2 respectively, and label them as building vector points.

[0039] Furthermore, in the 3D Mesh generation system for stereo satellite 3D reconstruction described in this invention, the ground DEM and roof DSM are processed by inserting points from the lower left corner. When the difference between the elevation of the inserted point and the current point is greater than a threshold p1 or the distance between them is greater than a threshold p2, a point set is inserted. Specifically, based on the maximum elevation tolerance threshold p1 and the maximum distance threshold p2, the DSM point set and the DEM point set are respectively selected to obtain triangulated point sets P1 and P2.

[0040] Step 41) Insert the first point in the DSM point set into P1;

[0041] Step 42) Using the newly added point p in the P1 point set as the center and the planar distance D as the radius, search the DSM point set for points whose elevation difference from the center point p is greater than h, and add them to the P1 point set. Repeat this process until all points that meet the conditions are added to the P1 point set, and mark point p as completed with the mark f. Points located inside the circle with an elevation difference less than h are removed from the DSM point set. Points marked as building vector points must be added to the point set P1.

[0042] Step 43) Iterate through all unmarked points in the P1 point set until the DSM point set is empty.

[0043] Thirdly, the present invention provides a three-dimensional mesh generation apparatus for three-dimensional reconstruction of stereo satellites, comprising a memory and a processor; the memory is used to store a computer program; the processor is used to implement the three-dimensional mesh generation method for three-dimensional reconstruction of stereo satellites as described in any of the first aspects when the computer program is executed.

[0044] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the three-dimensional mesh generation method for three-dimensional reconstruction of stereo satellites as described in any of the first aspects.

[0045] The present invention describes a method and apparatus for generating 3D meshes for stereo satellite 3D reconstruction. It takes high-resolution DSM imagery and corresponding regional building vector shape data as input, extracts terrain features from the reference DEM and input DSM using slope calculation methods, and generates large-area 3D meshes based on building geometric features and DSM. This method effectively overcomes the technical problems of traditional methods where building sides are too fragmented, have obvious seams, and unnatural texture transitions in large-area 3D satellite scenes. By fully utilizing existing terrain features, it can effectively improve the accuracy and aesthetics of the 3D mesh. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the three-dimensional mesh generation method for three-dimensional reconstruction of stereo satellites according to an embodiment of the present invention;

[0047] Figure 2 This is a schematic diagram of a DSM image as described in an embodiment of the present invention;

[0048] Figure 3 This is a vector diagram of the building described in an embodiment of the present invention;

[0049] Figure 4 This is a schematic diagram of a Delaunay triangulation generated using existing technology, as described in an embodiment of the present invention.

[0050] Figure 5 This is a schematic diagram of the final output Mesh as described in an embodiment of the present invention. Detailed Implementation

[0051] The method and apparatus for generating three-dimensional meshes for three-dimensional reconstruction of stereo satellites according to the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0052] Example 1

[0053] This embodiment discloses a method for generating 3D meshes for stereo satellite 3D reconstruction, such as... Figure 1As shown, it includes the following steps:

[0054] Step 1) Intersect the DSM imagery and building vector Shp data of the same area to obtain the roof DSM and ground DEM;

[0055] In this embodiment of the disclosure, the input data is the DSM image to be processed, such as... Figure 2 As shown; and the corresponding building vector Shp data, such as Figure 3 As shown. Create two empty raster maps, DSM and DEM, with an initial value of -9999.

[0056] Step 2) Process the building vector Shp data with the roof DSM and ground DEM respectively to obtain a three-dimensional vector, and mark the points on the Shp vector line to obtain the three-dimensional vector feature points;

[0057] This step specifically includes the following processes:

[0058] Specifically, the following steps are included:

[0059] Step 21) Overlay the roof DSM and building vector Shp data, traverse the DSM, and if the grid point is outside the building vector polygon, assign the pixel value of the corresponding DEM coordinate point at the grid point; if the grid point is inside the building vector polygon, assign the pixel value of the corresponding DSM coordinate point at the grid point.

[0060] Step 22) Copy the ground DEM to DEM2, and use the inverse distance weighting method to fill in the blank areas in DEM2;

[0061] Step 23) Overlay the building vector Shp and DEM2, record the intersection points of the endpoints of the line segments in Shp and DEM2, record the plane coordinates and elevation values, and obtain the point set s1; if the line segment length is greater than the preset threshold, add a new point on the line segment, record the elevation value corresponding to the new point, and add it to s1.

[0062] Step 24) Overlay the building vector Shp and DSM, obtain any edge of the Shp vector polygon as the current edge, and traverse the points on the current edge; if the elevation difference between the two endpoints of the current edge is greater than a preset threshold, add the point to the current edge and store it in the point set s2; then obtain any edge of the remaining edges of the Shp vector polygon as the current edge, and perform the same process until all the points on all edges of the vector polygon have been traversed.

[0063] Step 25) Each side of the vector polygon represents a side view. Record the intersection point and elevation value of each side endpoint with DSM and DEM2 respectively, and denote it as point set b. Collect the point sets of each side to obtain the point set B of the building side view.

[0064] Step 26) The pixel values ​​in DSM and DEM represent elevation values. The horizontal coordinate of the raster image is used as the X value, the vertical coordinate as the Y value, and the pixel value as the Z value. The raster is converted into DSM point set Q1 and DEM point set Q2.

[0065] Step 27) Add s1 to Q1 and s2 to Q2 respectively, and label them as building vector points.

[0066] Step 3) Perform raster-to-point cloud processing on the ground DEM and roof DSM;

[0067] Step 4) Perform point insertion processing on the ground DEM and roof DSM from the lower left corner point respectively. When the difference between the elevation of the inserted point and the current point is greater than the threshold p1 or the distance between the two is greater than the threshold p2, then insert the point set.

[0068] This step specifically includes: based on the maximum elevation tolerance threshold p1 and the maximum distance threshold p2, selecting the triangulation point sets P1 and P2 from the DSM point set and the DEM point set respectively;

[0069] Step 41) Insert the first point in the DSM point set into P1;

[0070] Step 42) Using the newly added point p in the P1 point set as the center and the planar distance D as the radius, search the DSM point set for points whose elevation difference from the center point p is greater than h, and add them to the P1 point set. Repeat this process until all points that meet the conditions are added to the P1 point set, and mark point p as completed with the mark f. Points located inside the circle with an elevation difference less than h are removed from the DSM point set. Points marked as building vector points must be added to the point set P1.

[0071] Step 43) Iterate through all unmarked points in the P1 point set until the DSM point set is empty.

[0072] Step 5) When the distance between two 3D vector feature points is greater than the threshold p2, insert several points so that the distance between the two points is not greater than p2, and construct the point cloud of the building side.

[0073] Step 6) Construct the Delaunay triangulation by partitioning the network;

[0074] The specific process of this step includes:

[0075] The specific steps for calculating the triangulation of the DEM point set are as follows:

[0076] First, calculate the maximum and minimum vertices in the XY directions of the point set P. Add the four vertices to P to obtain P'. Generate two super triangles based on the four vertices, thus forming the initial triangulation T0. Since it only contains two right triangles, T0 is a Delaunay triangulation (with four vertices).

[0077] Insert each vertex in the point set P into the existing triangulation Ti, and make the following adjustments:

[0078] In the existing triangulation, the set of all triangles whose circumcircles contain vertex v constitutes a "star-shaped polygon". The meaning of a star-shaped polygon is that any line connecting any vertex of the polygon to v is inside the polygon.

[0079] For the star-shaped polygon described above, all the triangles inside it are deleted, forming a "cavity". The vertices of the cavity boundary are connected to the newly added vertex v to obtain a new triangle, which replaces the triangles deleted in the triangulation. At this point, a new Delaunay triangulation containing vertex v is obtained.

[0080] Once the last vertex is inserted into the triangulation and all edge flipping operations are completed, we obtain a Delaunay triangulation of the point set P'. Now, remove the four vertices of the bounding box added in the first step and remove all triangles connected to them. The remaining triangles then constitute the Delaunay triangulation T of the point set P.

[0081] The triangulation of the DSM point set is the same as that of the DEM point set.

[0082] The Delaunay mesh generation algorithm for building side views includes:

[0083] Iterate through the smaller set b of the point set B on the side of the building, fit the plane, and then densify the edge points at a certain distance for each side length;

[0084] Edge points are inserted one by one into the triangular mesh. After each insertion, a new Delaunay triangular mesh is obtained using the aforementioned algorithm for calculating the triangulation of the DEM point set. For triangles within the region, if their area exceeds a certain threshold, their centroids are inserted into the triangular mesh, and adjusted according to the aforementioned algorithm for calculating the DEM point set to obtain a new Delaunay triangulation. This process continues until no more centroids need to be inserted, at which point the complete Delaunay triangular mesh of the region is obtained.

[0085] Step 7) Merge and output the final Mesh mesh;

[0086] In this embodiment, specifically, Mesh1 is generated by merging the DEM point set, Mesh2 is generated by the DSM point set, and Mesh3 is generated by the side of the building. The origin index of Mesh2 is increased based on the point index of Mesh1, and the index of the triangular facets is also increased based on Mesh1; similarly, the new index of the building Mesh3 is obtained, and the final complete result Mesh is output.

[0087] like Figure 4 The image shown is a Delaunay triangulation generated using the existing incremental interpolation method, Bowyer-Watson. Figure 5 The image shows the final output mesh diagram using the method described in this embodiment. It can be clearly seen from the image that the accuracy and aesthetics of the 3D mesh generated using the method described in this embodiment are significantly improved.

[0088] Example 2

[0089] Based on Embodiment 1, this embodiment discloses a three-dimensional mesh generation system for stereo satellite three-dimensional reconstruction, including a data acquisition module, a point cloud processing module, a triangulation construction module, and a merging output module;

[0090] The data acquisition module is used to acquire DSM images and building vector Shp data of the same area and perform intersection processing to obtain roof DSM and ground DEM; the building vector Shp data is processed separately with the roof DSM and ground DEM to obtain three-dimensional vectors, and points are marked on the Shp vector lines.

[0091] The point cloud processing module is used to perform raster-to-point cloud processing on the ground DEM and roof DSM; it performs point insertion processing on the ground DEM and roof DSM from the lower left corner point respectively; when the difference between the elevation of the inserted point and the current point is greater than the threshold p1 or the distance between the two is greater than the threshold p2, the point set is inserted; when the distance between the three-dimensional vector feature points is greater than the threshold p2, appropriate points are inserted to construct the point cloud of the building side.

[0092] The triangulation construction module is used to construct Delaunay triangulations by partition;

[0093] The merged output module is used to merge and output the final Mesh.

[0094] The specific operation steps of the 3D Mesh generation system for stereo satellite 3D reconstruction described in this embodiment are the same as those of the 3D Mesh generation method for stereo satellite 3D reconstruction described in Embodiment 1 above, and will not be repeated here.

[0095] Example 3

[0096] This embodiment discloses a 3D mesh generation device for stereo satellite 3D reconstruction, including a memory and a processor; the memory is used to store a computer program; the processor is used to implement the 3D mesh generation method for stereo satellite 3D reconstruction as described in Embodiment 1 when the computer program is executed. The specific generation method steps are the same as those in Embodiment 1, and will not be repeated here.

[0097] Example 4

[0098] This embodiment discloses a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the three-dimensional mesh generation method for three-dimensional reconstruction of stereo satellites as described in Embodiment 1. The specific generation method steps are the same as those in Embodiment 1, and will not be repeated here.

[0099] The computer described in this application embodiment can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. The computer-readable storage medium can be any usable medium that a computer can read, or a data storage device such as a server or data center that integrates one or more usable media. The usable medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital versatile optical disc (DVD)), or a semiconductor medium (e.g., solid-state drive (SSD)). The software formed by the computer's stored code can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other storage media that are mature in the art.

[0100] In the various embodiments of this application, the functional modules can be integrated into one processing unit or module, or each module can exist physically separately, or two or more modules can be integrated into one unit or module. In the above embodiments, they can be implemented entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, they can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated.

[0101] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A three-dimensional mesh generation method for stereoscopic satellite three-dimensional reconstruction, characterized by Includes the following steps: Step 1) Intersect the DSM imagery and building vector Shp data of the same area to obtain the roof DSM and ground DEM; Step 2) Process the building vector Shp data with the roof DSM and ground DEM separately to obtain a 3D vector, and mark the points on the Shp vector lines to obtain 3D vector feature points; specifically including the following steps: Step 21) Overlay the roof DSM and building vector Shp data, traverse the DSM, and if the grid point is outside the building vector polygon, assign the pixel value of the corresponding DEM coordinate point at the grid point; if the grid point is inside the building vector polygon, assign the pixel value of the corresponding DSM coordinate point at the grid point. Step 22) Copy the ground DEM to DEM2, and use the inverse distance weighting method to fill in the blank areas in DEM2; Step 23) Overlay the building vector Shp and DEM2, record the intersection points of the endpoints of the line segments in Shp with DEM2, record the plane coordinates and elevation values, and obtain the point set s1; if the line segment length is greater than the preset threshold, add a new point on the line segment, record the elevation value corresponding to the new point, and add it to s1. Step 24) Overlay the building vector Shp and DSM, obtain any edge of the Shp vector polygon as the current edge, and traverse the points on the current edge; if the elevation difference between the two endpoints of the current edge is greater than a preset threshold, add the point to the current edge and store it in the point set s2; then obtain any edge of the remaining edges of the Shp vector polygon as the current edge, and perform the same process until all the points on all edges of the vector polygon have been traversed. Step 25) Each side of the vector polygon represents a side view. Record the intersection point and elevation value of each side endpoint with DSM and DEM2 respectively, and denote it as point set b. Combine the point sets of each side to obtain the point set B of the building side view. Step 26) The pixel values ​​in DSM and DEM represent elevation values. The horizontal coordinate of the raster image is used as the X value, the vertical coordinate as the Y value, and the pixel value as the Z value. The raster is converted into DSM point set Q1 and DEM point set Q2. Step 27) Add s1 to Q1 and s2 to Q2 respectively, and label them as building vector points; Step 3) Perform raster-to-point cloud processing on the ground DEM and roof DSM; Step 4) Perform point insertion processing on the ground DEM and roof DSM from the lower left corner point respectively. When the difference between the elevation of the inserted point and the current point is greater than the threshold p1 or the distance between the two is greater than the threshold p2, then insert the point set. Step 5) When the distance between two 3D vector feature points is greater than the threshold p2, insert several points so that the distance between the two points is not greater than p2, and construct the point cloud of the building side. Step 6) Construct the Delaunay triangulation by partitioning the network; Step 7) Merge and output the final Mesh.

2. The method of claim 1, wherein: Step 4) The point insertion process is performed on the ground DEM and the roof DSM from the lower left corner. When the difference between the elevation of the inserted point and the current point is greater than the threshold p1 or the distance between the two is greater than the threshold p2, the point set is inserted. Specifically, based on the maximum elevation tolerance threshold p1 and the maximum distance threshold p2, the point sets P1 and P2 of the triangulation are selected from the DSM point set and the DEM point set, respectively. Step 41) Insert the first point in the DSM point set into P1; Step 42) Using the newly added point p in the P1 point set as the center and the planar distance D as the radius, search the DSM point set for points whose elevation difference from the center point p is greater than h, and add them to the P1 point set. Repeat this process until all points that meet the conditions are added to the P1 point set, and mark point p as completed with the mark f. Points located inside the circle with an elevation difference less than h are removed from the DSM point set. Points marked as building vector points must be added to the point set P1. Step 43) Iterate through all unmarked points in the P1 point set until the DSM point set is empty.

3. A three-dimensional mesh generation system for three-dimensional reconstruction of stereo satellites, characterized in that: It includes a data acquisition module, a point cloud processing module, a triangulation construction module, and a merging output module; The data acquisition module is used to acquire DSM imagery and building vector Shp data of the same area and perform intersection processing to obtain roof DSM and ground DEM; the building vector Shp data is processed separately with the roof DSM and ground DEM to obtain three-dimensional vectors, and points on the Shp vector lines are marked; specifically, the following steps are included: Step 21) Overlay the roof DSM and building vector Shp data, traverse the DSM, and if the grid point is outside the building vector polygon, assign the pixel value of the corresponding DEM coordinate point at the grid point; if the grid point is inside the building vector polygon, assign the pixel value of the corresponding DSM coordinate point at the grid point. Step 22) Copy the ground DEM to DEM2, and use the inverse distance weighting method to fill in the blank areas in DEM2; Step 23) Overlay the building vector Shp and DEM2, record the intersection points of the endpoints of the line segments in Shp with DEM2, record the plane coordinates and elevation values, and obtain the point set s1; if the line segment length is greater than the preset threshold, add a new point on the line segment, record the elevation value corresponding to the new point, and add it to s1. Step 24) Overlay the building vector Shp and DSM, obtain any edge of the Shp vector polygon as the current edge, and traverse the points on the current edge; if the elevation difference between the two endpoints of the current edge is greater than a preset threshold, add the point to the current edge and store it in the point set s2; then obtain any edge of the remaining edges of the Shp vector polygon as the current edge, and perform the same process until all the points on all edges of the vector polygon have been traversed. Step 25) Each side of the vector polygon represents a side view. Record the intersection point and elevation value of each side endpoint with DSM and DEM2 respectively, and denote it as point set b. Combine the point sets of each side to obtain the point set B of the building side view. Step 26) The pixel values ​​in DSM and DEM represent elevation values. The horizontal coordinate of the raster image is used as the X value, the vertical coordinate as the Y value, and the pixel value as the Z value. The raster is converted into DSM point set Q1 and DEM point set Q2. Step 27) Add s1 to Q1 and s2 to Q2 respectively, and label them as building vector points; The point cloud processing module is used to perform raster-to-point cloud processing on the ground DEM and roof DSM; it performs point insertion processing on the ground DEM and roof DSM from the lower left corner point respectively. When the difference between the elevation of the inserted point and the current point is greater than the threshold p1 or the distance between the two is greater than the threshold p2, a point set is inserted; when the distance between two 3D vector feature points is greater than the threshold p2, several points are inserted so that the distance between the two points is not greater than p2, thus constructing the point cloud of the building side. The triangulation construction module is used to construct Delaunay triangulations by partition; The merged output module is used to merge and output the final Mesh.

4. The three-dimensional mesh generation system for three-dimensional reconstruction of stereo satellites according to claim 3, characterized in that: The process involves inserting points from the lower left corner of the ground DEM and the roof DSM respectively. When the difference between the elevation of the inserted point and the current point is greater than the threshold p1 or the distance between them is greater than the threshold p2, a point set is inserted. Specifically, based on the maximum elevation tolerance threshold p1 and the maximum distance threshold p2, the DSM point set and the DEM point set are respectively selected to form triangulated point sets P1 and P2. Step 41) Insert the first point in the DSM point set into P1; Step 42) Using the newly added point p in the P1 point set as the center and the planar distance D as the radius, search the DSM point set for points whose elevation difference from the center point p is greater than h, and add them to the P1 point set. Repeat this process until all points that meet the conditions are added to the P1 point set, and mark point p as completed with the mark f. Points located inside the circle with an elevation difference less than h are removed from the DSM point set. Points marked as building vector points must be added to the point set P1. Step 43) Iterate through all unmarked points in the P1 point set until the DSM point set is empty.

5. A three-dimensional mesh generation device for three-dimensional reconstruction of stereo satellites, characterized in that: It includes a memory and a processor; the memory is used to store a computer program; the processor is used to implement, when executing the computer program, the three-dimensional mesh generation method for three-dimensional reconstruction of stereo satellites as described in any one of claims 1-2.

6. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, implements the three-dimensional mesh generation method for three-dimensional reconstruction of stereo satellites as described in any one of claims 1-2.

Images