Method, system and device for constructing live-action three-dimensional model with real ground object side texture based on first-level satellite image and storage medium

Abstract

The invention discloses a method, a system and equipment for constructing a live-action three-dimensional model with real ground object side surface textures based on a first-level satellite image and a storage medium, belongs to the technical field of remote sensing surveying and mapping, and solves the problems that the textures of an existing satellite image live-action three-dimensional model are abnormal and lack of sense of reality. Comprising the following steps: S1, producing a DSM (Digital Subscriber Module) by using a first-level stereopair; s2, producing a live-action three-dimensional white mold in an obj format by using a DSM (Digital Subscriber Module); and S3, carrying out texture mapping on the real scene three-dimensional white model in the obj format by using a first-level image, and producing a real scene three-dimensional model in the obj format with a real ground object side texture.

Description

Technical Field

[0001] The present invention relates to the field of remote sensing mapping technology, and in particular to a method, system, device and storage medium for constructing a real-scene three-dimensional model with real object side textures based on primary satellite images. Background Art

[0002] Satellite imagery 3D models are three-dimensional geographic information models generated from high-resolution satellite remote sensing data. Based on real geographic coordinates, they intuitively display the three-dimensional spatial distribution and texture characteristics of terrain, buildings, vegetation, and other elements. They offer key advantages such as wide coverage, freedom from airspace restrictions, and excellent cost and efficiency.

[0003] Traditionally, real-world 3D models based on satellite imagery are produced using digital surface models (DSMs) and digital orthophotos (DOMs). DOMs, with their vertical perspective, only capture texture information on the tops of features, but fail to provide texture data for areas like the sides of buildings and steep terrain. This results in model texture anomalies and a lack of realism.

[0004] Currently, texture correction for real-world 3D models relies primarily on multi-source data fusion, such as multi-angle aerial imagery, laser point clouds, and close-range photogrammetry. This approach is generally costly and time-consuming. In the future, it may be possible to use deep learning to predict side details and correct the side texture of real-world 3D models. This approach is less expensive but lacks fidelity.

[0005] In summary, the existing real-scene 3D models of satellite images have abnormal textures and lack realism. Summary of the Invention

[0006] The present invention solves the problem that the existing satellite image real scene three-dimensional model has abnormal texture and lacks realism.

[0007] The method of constructing a real-scene three-dimensional model with real object side texture based on primary satellite imagery according to the present invention comprises the following steps:

[0008] Step S1, generating a DSM using a primary stereo pair;

[0009] Step S2, using DSM to produce a real-scene 3D white model in obj format;

[0010] Step S3: Using the primary image to perform texture mapping on the real-scene 3D white model in the obj format, a real-scene 3D model in the obj format with the side texture of the real object is produced.

[0011] Furthermore, in one embodiment of the present invention, in the step S1, the DSM is produced using the primary stereo image pair, wherein the block adjustment module comprises the following steps:

[0012] Step S101, selecting a reference image and a registration image, and preprocessing the reference image and the registration image respectively;

[0013] Step S102, generating a set of same-name point pairs based on the reference image and the registered image;

[0014] Step S103, generating a control point file, wherein the control point file includes plane coordinates and elevation values;

[0015] Step S104, establishing an image compensation parameter group based on each satellite image;

[0016] Step S105, constructing an error equation including a set of point pairs with the same name, a control point file, and an image space compensation parameter group, and performing a solution;

[0017] Step S106: Save the calculated image-space compensation parameter group.

[0018] Furthermore, in one embodiment of the present invention, in step S2, the step of using DSM to produce a real-scene 3D white model in obj format includes the following steps:

[0019] Step S201, filtering the DSM;

[0020] Step S202 , dividing the filtered DSM into blocks to form multiple tile DSMs;

[0021] Step S203, reading multiple tile DSM data, respectively obtaining three-dimensional coordinate information and three-dimensional coordinate information after multiple tiles are offset;

[0022] Step S204, constructing a Delaunay triangulation;

[0023] Step S205 , writing the offset 3D coordinate information of each tile as the vertex v and the Delaunay triangulation as the face f into the obj file in sequence, thereby generating a real-scene 3D white model in the obj format.

[0024] Furthermore, in one embodiment of the present invention, in step S3, texture mapping is performed on the obj format real-scene 3D white model using the primary image to produce an obj format real-scene 3D model with a real object side texture, including the following steps:

[0025] Step S301: Using the RPC parameters and image space compensation parameters of the primary image, all 3D coordinate points in the obj white model are inversely calculated to the row and column numbers of the primary image. The primary image is cropped according to the row and column number range and converted into a JPG format image.

[0026] Step S302, repeating step S301 for multiple primary images of the stereo pair, and then stitching the multiple cropped jpg images;

[0027] Step S303: compress the spliced images to produce a jpg format image;

[0028] Step S304, generate mtl file;

[0029] Step S305, generating a new obj file, and writing the vertex v information in the obj white model of step S205 into the new obj file in sequence;

[0030] Step S306, calculating the UV coordinates of each vertex in the texture image;

[0031] Step S307: write the UV coordinates of each vertex in the texture image as vt information into the new obj file in sequence;

[0032] Step S308, reading the face f information in the obj white model file of step S205;

[0033] Step S309: Perform occlusion analysis based on the pitch and roll angles of the multiple first-level images, and determine the vt index number that each face f needs to reference based on the occlusion analysis results;

[0034] Step S310 , writing the updated surface f information into a new obj file, thereby generating a real-scene 3D model in obj format with real object side textures.

[0035] Furthermore, in one embodiment of the present invention, the following steps are also included:

[0036] Step S4: convert the real-scene 3D model in obj format with the side texture of the real object into osgb format.

[0037] Furthermore, in one embodiment of the present invention, in step S4, converting the obj format real-scene 3D model with the real object side texture into osgb format includes the following steps:

[0038] Step S401, setting a simplified scale of a real-scene 3D model mesh with a real object side texture;

[0039] Step S402, setting the display range ratio of the real scene 3D model with the real object side texture at different levels;

[0040] Step S403, setting the minimum resolution level of the real-scene 3D model with the real object side texture;

[0041] Step S404, calculating the highest resolution level of the real-life 3D model with the real object side texture, recursively simplifying it step by step, using the LOD technology of the OSG library to display the models at different levels, and generating the real-life 3D model with the real object side texture in the OSGb format corresponding to each level;

[0042] Step S405 , writing the coordinate system information and the offset information into a metadata file metadata.xml.

[0043] The system of the present invention for constructing a real-scene three-dimensional model with real object side texture based on primary satellite imagery includes the following modules:

[0044] Module S1, producing DSM using primary stereo pairs;

[0045] Module S2, using DSM to produce real-scene 3D white model in obj format;

[0046] Module S3 uses the first-level image to perform texture mapping on the real-scene 3D white model in obj format to produce a real-scene 3D model in obj format with the side texture of the real object.

[0047] An electronic device according to the present invention comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0048] Memory for storing computer programs;

[0049] The processor is used to implement the method steps described above when executing the program stored in the memory.

[0050] The computer-readable storage medium of the present invention stores a computer program, which implements the method steps described above when executed by a processor.

[0051] The present invention solves the problem that existing satellite imagery real-scene 3D models have abnormal textures and lack of realism. Specific beneficial effects include:

[0052] The method of constructing a real-scene 3D model with real object side textures based on primary satellite images of the present invention first uses a primary stereo image pair to produce a DSM; then uses the DSM to produce a real-scene 3D white model in the obj format; then, texture mapping is performed on the white model using the primary image to produce a real-scene 3D model in the obj format with real object side textures; finally, the obj format real-scene 3D model is converted into a more general osgb format real-scene 3D model;

[0053] The method described in this paper, which constructs a realistic 3D model of a city scene with realistic side textures based on primary satellite imagery, can significantly enhance the texture fidelity of the 3D model and effectively improve the commercial value of the 3D visualization results. The method is simple in principle, easy to implement, and applicable to practical engineering applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0055] Figure 1 Schematic diagram of base-to-height ratio and overlapping area according to embodiment 6;

[0056] Figure 2 This is a schematic diagram of a primary image captured by a satellite tilted back 15 degrees as described in Implementation Method 6;

[0057] Figure 3 Schematic diagram of filtering out spike anomalies in DSM according to the third embodiment;

[0058] Figure 4 This is a schematic diagram of constructing the Delaunay triangulation described in the third embodiment;

[0059] Figure 5 This is a schematic diagram of picture stitching according to the fourth embodiment;

[0060] Figure 6 This is an example diagram of the metadata.xml file content described in Implementation Method 6;

[0061] Figure 7 This is the real texture rendering of the side of the feature described in the sixth embodiment;

[0062] Figure 8 This is a block diagram of the algorithm structure described in Implementation Method 6. DETAILED DESCRIPTION

[0063] The following will clearly and completely describe various embodiments of the present invention in conjunction with the accompanying drawings. The embodiments described with reference to the accompanying drawings are exemplary and intended to be used to explain the present invention, but should not be understood as limiting the present invention.

[0064] Implementation method 1: This implementation method includes the following steps:

[0065] Step S1, generating a DSM using a primary stereo pair;

[0066] Step S2, using DSM to produce a real-scene 3D white model in obj format;

[0067] Step S3: Using the primary image to perform texture mapping on the real-scene 3D white model in the obj format, a real-scene 3D model in the obj format with the side texture of the real object is produced.

[0068] The existing real-scene 3D models of satellite images have abnormal textures and lack realism.

[0069] To solve the above technical problems, this embodiment proposes a method for constructing a realistic three-dimensional model with real object side textures based on primary satellite imagery, comprising the following steps:

[0070] Step S1, using the first-level stereo image pair to produce DSM, mainly including regional block adjustment, epipolar image generation, dense matching and other modules;

[0071] Step S2, using DSM to produce a real-scene 3D white model in obj format;

[0072] Step S3: Using the primary image to perform texture mapping on the real-scene 3D white model in the obj format, a real-scene 3D model in the obj format with the side texture of the real object is produced.

[0073] Therefore, this embodiment first uses a primary stereo image pair to produce a DSM; then, uses the DSM to produce a real-life 3D white model in the obj format; then, uses the primary image to perform texture mapping on the white model to produce a real-life 3D model in the obj format with the side texture of the real object; finally, converts the real-life 3D model in the obj format into a more general real-life 3D model in the osgb format.

[0074] Embodiment 2. This embodiment further limits the method for constructing a real-scene three-dimensional model with real object side texture based on primary satellite imagery described in embodiment 1. In step S1, the DSM is produced using the primary stereo image pair, and the block adjustment module includes the following steps:

[0075] Step S101, selecting a reference image and a registration image, and preprocessing the reference image and the registration image respectively;

[0076] Step S102, generating a set of same-name point pairs based on the reference image and the registered image;

[0077] Step S103, generating a control point file, wherein the control point file includes plane coordinates and elevation values;

[0078] Step S104, establishing an image compensation parameter group based on each satellite image;

[0079] Step S105, constructing an error equation including a set of point pairs with the same name, a control point file, and an image space compensation parameter group, and performing a solution;

[0080] Step S106: Save the calculated image-space compensation parameter group.

[0081] In this implementation, since the image square compensation parameters of each satellite image solved by the block adjustment need to be used when subsequently constructing the mapping relationship between the three-dimensional coordinate points and the primary image pixels, the implementation process of the block adjustment module is mainly described below:

[0082] Step S101: Select the higher-quality image in the stereo pair as the reference image, and the other image as the registration image. Adaptive meshing is applied to the reference image, using a dynamic block algorithm to generate a regular grid covering the entire image. The density of feature point extraction is dynamically adjusted in each block based on the complexity of the terrain, ensuring sufficient spatially evenly distributed tie points for subsequent adjustment calculations.

[0083] Step S102, using the Fast Fourier Mellin Transform matching algorithm to generate a set of homonymous point pairs that meet the accuracy requirements;

[0084] Step S103: A high-precision digital elevation model and a reference base map are introduced as spatial constraints, and a control point file containing plane coordinates and elevation values is generated for successfully matched feature points through a feature matching algorithm.

[0085] Step S104: Create an independent compensation parameter set for each satellite image, a0, a l ,a s ,b0,b l ,b s , the calculation formula is:

[0086]

[0087] Among them, (l′, s′) is the accurate image coordinate corresponding to a certain point T (X, Y, Z) on the ground on the satellite image, and (l, s) is the image coordinate of point T calculated according to the RPC parameters of the satellite image.

[0088] Step S105: construct an error equation including the same-name points, control points, and image space compensation parameters, and solve it using the least squares method, specifically:

[0089]

[0090] Among them, (X i ,Y i ,Z i ) are the ground coordinates corresponding to the points with the same name and the control points, R s ,R0,C s , C0 is the PRC parameter of satellite image, p i is the positive solution form of the RPC polynomial, the specific form is:

[0091]

[0092] Among them, a0-a 20 The corresponding parameter value in the RPC parameter file.

[0093] Step S106: Save the calculated image-side compensation parameters.

[0094] Implementation method 3: This implementation method further defines the method for constructing a real-scene 3D model with real object side texture based on primary satellite imagery described in implementation method 1. In step S2, the method of using DSM to produce a real-scene 3D white model in obj format includes the following steps:

[0095] Step S201, filtering the DSM;

[0096] Step S202 , dividing the filtered DSM into blocks to form multiple tile DSMs;

[0097] Step S203, reading multiple tile DSM data, respectively obtaining three-dimensional coordinate information and three-dimensional coordinate information after multiple tiles are offset;

[0098] Step S204, constructing a Delaunay triangulation;

[0099] Step S205 , writing the offset 3D coordinate information of each tile as the vertex v and the Delaunay triangulation as the face f into the obj file in sequence, thereby generating a real-scene 3D white model in the obj format.

[0100] In this implementation, the DSM is a dataset of plane coordinates and elevations of regular grid points within a certain range, describing the spatial distribution of regional topography and surface. After obtaining the DSM, the following steps are performed to generate a real-world 3D white model in obj format:

[0101] Step S201: filter the DSM to eliminate the spike-like anomalies in the DSM. Figure 3 As shown;

[0102] Step S202: Divide the DSM after step S201 into tiles. Usually, the size of each tile is set between 100m*100m and 300m*300m.

[0103] Step S203: Read the DSM data of each tile after step S202 to obtain the 3D coordinate information. Subtract the X and Y coordinates of the top left corner of the first DSM tile from the X and Y coordinates, and store the X and Y coordinates of the top left corner of the first DSM tile (with the Z coordinate set to 0) in a file as the coordinate offset file of the model.

[0104] Step S204: Figure 4 The coordinate points of the positional relationship are sequentially constructed to construct the Delaunay triangulation;

[0105] Step S205 , writing the offset 3D coordinate information of each tile as vertex v and the Delaunay triangulation information as face f into the obj file in sequence, thereby producing multiple real-scene 3D white models in obj format.

[0106] Implementation 4. This implementation further defines the method for constructing a real-scene 3D model with real object side texture based on a primary satellite image described in Implementation 1. In step S3, texture mapping is performed on the real-scene 3D white model in obj format using the primary image to produce the real-scene 3D model in obj format with real object side texture, including the following steps:

[0107] Step S301: Using the RPC parameters and image space compensation parameters of the primary image, all 3D coordinate points in the obj white model are inversely calculated to the row and column numbers of the primary image. The primary image is cropped according to the row and column number range and converted into a JPG format image.

[0108] Step S302, repeating step S301 for multiple primary images of the stereo pair, and then stitching the multiple cropped jpg images;

[0109] Step S303: compress the spliced images to produce images in jpg format;

[0110] Step S304, generate mtl file;

[0111] Step S305, generating a new obj file, and writing the vertex v information in the obj white model of step S205 into the new obj file in sequence;

[0112] Step S306, calculating the UV coordinates of each vertex in the texture image;

[0113] Step S307: write the UV coordinates of each vertex in the texture image as vt information into the new obj file in sequence;

[0114] Step S308, reading the face f information in the obj white model file of step S205;

[0115] Step S309: performing occlusion analysis based on the pitch and roll angles of the multiple first-level images, and determining the vt index number to be referenced for each face f based on the results of the occlusion analysis;

[0116] Step S310 , writing the updated surface f information into a new obj file, thereby generating a real-scene 3D model in obj format with real object side textures.

[0117] In this embodiment, step S301 uses the RPC parameters and image space compensation parameters of the primary image to reversely calculate the 3D coordinates of all points in each obj white model to the row and column numbers of the primary image. The primary image is cropped according to the row and column number range and converted into a JPG format image.

[0118] Step S302: The two primary images of the stereo pair are processed as in step S301, and then the two cropped jpg images are spliced together. The splicing effect is as follows: Figure 5 The number of rows in the spliced image is the sum of the number of rows in the two images, and the number of columns in the spliced image is the larger value of the number of columns in the two images;

[0119] Step S303: compress the merged image, and the resulting image is still in jpg format. The upper and lower sub-images in the image are derived from two primary images, and can be used to map the real-scene 3D model from two angles.

[0120] Step S304: generate an mtl file, referencing the image result of step S303;

[0121] Step S305: Generate a new obj file, reference the mtl file from step S304, and write the vertex v information in the obj white model into the new obj file in sequence;

[0122] Step S306, calculating the UV coordinates of each vertex in the texture image;

[0123] For the above sub-image, the calculation formula is:

[0124] xRatio1=

[0125] (res[0].x-x_min_No1) / (max(x_max_No1-x_min_No1,x_max_No2-x_min_No2));

[0126] yRatio1=(y_max_No1-res[0].y+y_max_No2-y_min_No2) /

[0127] (y_max_No1-y_min_No1+y_max_No2-y_min_No2);

[0128] For the following sub-image, the calculation formula is:

[0129] xRatio2=

[0130] (res[0].x-x_min_No2) / (max(x_max_No1-x_min_No1,x_max_No2-x_min_No2));

[0131] yRatio2=(y_max_No2-res[0].y) / (y_max_No1-y_min_No1+

[0132] y_max_No2-y_min_No2);

[0133] Among them, (xRatio1, yRatio1) and (xRatio2, yRatio2) correspond to the UV coordinates of the vertex in the upper and lower sub-images respectively; res[0].x and res[0].y represent the x-coordinate and y-coordinate of the vertex respectively; in step S301, all vertices of the obj white model correspond to a set of row and column numbers of the first-level image, x_min_No1 represents the minimum column number when inverting back to image 1, x_min_No2 represents the minimum column number when inverting back to image 2, x_max_No1 represents the maximum column number when inverting back to image 1, and x_max_No2 represents the maximum column number when inverting back to image 2; y_min_No1 represents the minimum row number when inverting back to image 1, y_min_No2 represents the minimum row number when inverting back to image 2, y_max_No1 represents the maximum row number when inverting back to image 1, and y_max_No2 represents the maximum row number when inverting back to image 2; max() represents the maximum value;

[0134] Step S307, write (xRatio1, yRatio1) and (xRatio2, yRatio2) as vt information into the new obj file in sequence;

[0135] Step S308, read the face f information in the obj white model;

[0136] Step S309: perform occlusion analysis based on the pitch angles of the two primary images, and determine the vt index number that each face f needs to reference based on the results of the occlusion analysis;

[0137] Step S310: Write the updated surface f information into a new obj file. At this point, an obj model with the true texture of the side of the feature can be produced.

[0138] Implementation 5: This implementation further defines the method for constructing a realistic three-dimensional model with real object side texture based on primary satellite imagery described in Implementation 1, and further includes the following steps:

[0139] Step S4: convert the real-scene 3D model in obj format with the side texture of the real object into osgb format.

[0140] In this embodiment, step S4 converts the real-scene 3D model in obj format into osgb format.

[0141] Implementation 6. This implementation further limits the method for constructing a real-scene 3D model with real object side texture based on a primary satellite image described in Implementation 5. In step S4, converting the real-scene 3D model with real object side texture in obj format into osgb format includes the following steps:

[0142] Step S401, setting a simplified scale of a real-scene 3D model mesh with a real object side texture;

[0143] Step S402, setting the display range ratio of the real scene 3D model with the real object side texture at different levels;

[0144] Step S403, setting the minimum resolution level of the real-scene 3D model with the real object side texture;

[0145] Step S404, calculating the highest resolution level of the real-life 3D model with the real object side texture, recursively simplifying it step by step, using the LOD technology of the OSG library to display the models at different levels, and generating the real-life 3D model with the real object side texture in the OSGb format corresponding to each level;

[0146] Step S405 , writing the coordinate system information and the offset information into a metadata file metadata.xml.

[0147] In this embodiment, step S401 is to set the model mesh simplification ratio;

[0148] Step S402, setting the display range ratio of models at different levels;

[0149] Step S403, setting the minimum resolution level of the model;

[0150] Step S404, calculating the highest resolution level of the model, recursively simplifying it step by step, using the LOD technology of the OSG library to display models of different levels in a hierarchical manner, and generating OSGB models corresponding to each level;

[0151] Step S405: Write the coordinate system information and offset information into a metadata file metadata.xml. The metadata.xml file content is as follows: Figure 6 shown.

[0152] In summary, the method for constructing a realistic 3D model with real object side texture based on primary satellite images described in embodiments 1 to 6 takes into account that the production of DSM and DOM requires the use of primary satellite stereo image pairs with a certain base-to-height ratio, such as Figure 1 As shown, the first-level image contains certain ground feature side information, such as Figure 2 As shown in Figure 2. Using the RPC parameters of the primary image and the image square compensation parameters after adjustment, a mapping relationship between 3D coordinate points and primary image pixels can be constructed. This mapping relationship can be used to produce a realistic 3D model with the true texture of the feature's side.

[0153] In order to better illustrate the method of constructing a real-scene three-dimensional model with real object side texture based on primary satellite imagery described in embodiments 1 to 6, the following examples are described in detail:

[0154] A brief flowchart of the algorithm is as follows Figure 8 The algorithm was verified using the Jilin-1 stereo image data to create a DSM, construct the real texture of the ground feature side, and create a real-scene 3D model. Microsoft Visual Studio 2022 was used as the simulation platform and the OSG library was used.

[0155] The effect of the real scene 3D model with the real texture of the side of the object after the algorithm is used is as follows Figure 7 As shown, the first-level image used for mapping is as follows Figure 2 shown.

[0156] The simulation results show that after the algorithm is used, a realistic texture effect is created on the side of the object, and the algorithm is valuable.

[0157] Embodiment 7: This embodiment describes a system for constructing a realistic three-dimensional model with real object side textures based on primary satellite images, including the following modules:

[0158] Module S1, producing DSM using primary stereo pairs;

[0159] Module S2, using DSM to produce real-scene 3D white model in obj format;

[0160] Module S3 uses the first-level image to perform texture mapping on the real-scene 3D white model in obj format to produce a real-scene 3D model in obj format with the side texture of the real object.

[0161] Embodiment 8: An electronic device according to this embodiment includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus.

[0162] Memory for storing computer programs;

[0163] The processor is configured to implement the method steps described in any one of the first to sixth embodiments when executing the program stored in the memory.

[0164] Implementation method 9: A computer-readable storage medium described in this implementation method is characterized in that a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the method steps described in any one of implementation methods 1 to 6 are implemented.

[0165] The above is a detailed introduction to the method, system, device and storage medium proposed in the present invention for constructing a real-scene three-dimensional model with real side texture of objects based on primary satellite images. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for constructing a real-scene three-dimensional model with real object side texture based on first-level satellite imagery, characterized in that: The following steps are involved: Step S1, generating a DSM using a primary stereo pair; Step S2, using DSM to produce a real-scene 3D white model in obj format; Step S3: Using the primary image to perform texture mapping on the real-scene 3D white model in the obj format, a real-scene 3D model in the obj format with the side texture of the real object is produced.

2. The method of constructing a real-scene three-dimensional model with real object side texture based on primary satellite imagery according to claim 1, characterized in that: In the step S1, the DSM is produced using the primary stereo image pair, wherein the block adjustment module comprises the following steps: Step S101, selecting a reference image and a registration image, and preprocessing the reference image and the registration image respectively; Step S102, generating a set of same-name point pairs based on the reference image and the registered image; Step S103, generating a control point file, wherein the control point file includes plane coordinates and elevation values; Step S104, establishing an image compensation parameter group based on each satellite image; Step S105, constructing an error equation including a set of point pairs with the same name, a control point file, and an image space compensation parameter group, and performing a solution; Step S106: Save the calculated image-space compensation parameter group.

3. The method for constructing a real-scene three-dimensional model with real object side texture based on primary satellite imagery according to claim 1, characterized in that: In the step S2, the production of a real-scene 3D white model in obj format using DSM includes the following steps: Step S201, filtering the DSM; Step S202 , dividing the filtered DSM into blocks to form multiple tile DSMs; Step S203, reading multiple tile DSM data, respectively obtaining three-dimensional coordinate information and three-dimensional coordinate information after multiple tiles are offset; Step S204, constructing a Delaunay triangulation; Step S205 , writing the offset 3D coordinate information of each tile as the vertex v and the Delaunay triangulation as the face f into the obj file in sequence, thereby generating a real-scene 3D white model in the obj format.

4. The method of constructing a real-scene three-dimensional model with real object side texture based on primary satellite imagery according to claim 1, characterized in that: In step S3, the obj format real-scene 3D white model is texture mapped using the primary image to produce an obj format real-scene 3D model with a real object side texture, including the following steps: Step S301: Using the RPC parameters and image space compensation parameters of the primary image, all 3D coordinate points in the obj white model are inversely calculated to the row and column numbers of the primary image. The primary image is cropped according to the row and column number range and converted into a JPG format image. Step S302, repeating step S301 for multiple primary images of the stereo pair, and then stitching the multiple cropped jpg images; Step S303: compress the spliced images to produce a jpg format image; Step S304, generate mtl file; Step S305, generating a new obj file, and writing the vertex v information in the obj white model of step S205 into the new obj file in sequence; Step S306, calculating the UV coordinates of each vertex in the texture image; Step S307: write the UV coordinates of each vertex in the texture image as vt information into the new obj file in sequence; Step S308, reading the face f information in the obj white model file of step S205; Step S309: Perform occlusion analysis based on the pitch and roll angles of the multiple first-level images, and determine the vt index number that each face f needs to reference based on the occlusion analysis results; Step S310 , writing the updated surface f information into a new obj file, thereby generating a real-scene 3D model in obj format with real object side textures.

5. The method of constructing a real-scene three-dimensional model with real object side texture based on primary satellite imagery according to claim 1, characterized in that: The following steps are also included: Step S4: convert the obj format real-scene 3D model with the real object side texture into osgb format.

6. The method for constructing a real-scene three-dimensional model with real object side texture based on primary satellite imagery according to claim 5, characterized in that: In the step S4, the step of converting the obj format real scene 3D model with the real object side texture into osgb format includes the following steps: Step S401, setting a simplified scale of a real-scene 3D model mesh with a real object side texture; Step S402, setting the display range ratio of the real scene 3D model with the real object side texture at different levels; Step S403, setting the minimum resolution level of the real scene 3D model with the real object side texture; Step S404, calculating the highest resolution level of the real-life 3D model with the real object side texture, recursively simplifying it step by step, using the LOD technology of the OSG library to display the models at different levels, and generating the real-life 3D model with the real object side texture in the OSGb format corresponding to each level; Step S405 , writing the coordinate system information and the offset information into a metadata file metadata.xml.

7. A system for constructing a real-scene three-dimensional model with real surface textures based on primary satellite images, characterized in that: Includes the following modules: Module S1, producing DSM using primary stereo pairs; Module S2, using DSM to produce real-scene 3D white model in obj format; Module S3 uses the first-level image to perform texture mapping on the real-scene 3D white model in obj format to produce a real-scene 3D model in obj format with the side texture of the real object.

8. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; Memory for storing computer programs; A processor, configured to implement the method steps described in any one of claims 1 to 6 when executing a program stored in a memory.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method steps of any one of claims 1 to 6 are implemented.

Images