Fine three-dimensional modeling method based on multi-view satellite image

Abstract

The invention discloses a fine three-dimensional modeling method based on multi-view satellite images, and belongs to the technical field of three-dimensional modeling. The fine three-dimensional modeling method based on the multi-view satellite image comprises the following steps: acquiring a multi-view image through a satellite, and preprocessing the image; performing dense matching and point cloud generation processing on the preprocessed image data to obtain point cloud data; constructing a triangulation network based on the point cloud data; performing texture mapping processing in the triangulation network to obtain a three-dimensional model; performing visual display on the three-dimensional model; according to the fine three-dimensional modeling method based on the multi-view satellite image, the steps of multi-view satellite image acquisition, image preprocessing, dense matching, point cloud generation, triangulation network construction and the like are matched with one another, so that a map three-dimensional image can be collected and constructed from multiple views, and the three-dimensional modeling method based on the multi-view satellite image is realized by utilizing an image acquisition technology of the satellite image. And the refinement degree of the ground three-dimensional model is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional modeling, and in particular to a fine three-dimensional modeling method based on multi-view satellite images. Background Art

[0002] Refined modeling of three-dimensional models refers to the process of adding details and optimizing the existing basic model. Usually, the basic model is composed of simple geometric bodies, and refined modeling makes the model more realistic and accurate by adding more geometric details and detailed textures. It is widely used in smart cities, real-life navigation, ancient building and cultural relics protection, military applications and other fields.

[0003] The existing fine 3D modeling methods mainly use aerial drone oblique photography to perform 3D modeling, but this modeling method is inefficient. Aerial photography is restricted by airspace and cannot obtain 3D models of overseas areas. It cannot meet the application needs of large-scale 3D modeling and has a limited imaging range. Summary of the invention

[0004] In view of the above problems existing in the prior art, the present invention is proposed.

[0005] Therefore, the technical problem to be solved by the present invention is a problem.

[0006] To achieve the above object, the present invention provides the following technical solutions: A method for fine three-dimensional modeling based on multi-view satellite images, comprising: acquiring multi-view images through satellites, and preprocessing the images;

[0007] The pre-processed image data is subjected to dense matching and point cloud generation processing to obtain point cloud data;

[0008] Construct triangulated network based on point cloud data;

[0009] Carry out texture mapping in the triangulated network to obtain a three-dimensional model;

[0010] Visual display of 3D models.

[0011] As a further solution of the present invention: the step of acquiring multi-view images through satellites and preprocessing the images comprises:

[0012] The satellite photographs the same target area at different angles along the orbital direction and perpendicular to the orbital direction;

[0013] Acquire multi-view satellite stereo images on the satellite platform;

[0014] Carry out radiation and atmospheric correction processing on the acquired image;

[0015] The corrected image is set in a unified geographic coordinate system to obtain image data;

[0016] Coordinate registration of image data.

[0017] As a further solution of the present invention: the satellites include Gaofen-7 satellite and Gaojing satellite.

[0018] As a further solution of the present invention: the step of performing dense matching and point cloud generation processing on the pre-processed image data to obtain point cloud data includes:

[0019] Matching common feature points between multiple image data after coordinate registration;

[0020] Generate a disparity map from common feature points using a stereo matching algorithm;

[0021] According to the disparity map and the satellite's shooting parameters, images from different perspectives are matched pixel by pixel to generate high-density three-dimensional point cloud data.

[0022] As a further solution of the present invention: the step of constructing a triangulated network based on point cloud data includes:

[0023] Select key points that present the geometric features of terrain and objects in point cloud data;

[0024] Construct a triangulated network using key points.

[0025] As a further solution of the present invention: the three-dimensional point cloud data is used to construct a triangulated network through a three-dimensional network growing method.

[0026] As a further solution of the present invention: the step of performing texture mapping in the triangulated network to obtain a three-dimensional model includes:

[0027] Select the corresponding texture image in each triangle of the triangulated network;

[0028] Select the visible image of the diagonal face as the texture image;

[0029] Map the selected texture image onto the triangulated network to construct a three-dimensional model.

[0030] As a further solution of the present invention: the three-dimensional model needs to be optimized and detailed enhanced during visual display.

[0031] As a further solution of the present invention: the model optimization adopts smoothing and noise removal methods.

[0032] As a further solution of the present invention: the three-dimensional model visualization is displayed in chart and data format.

[0033] Compared with the prior art, the invention has the following beneficial effects: the method for fine 3D modeling based on multi-view satellite images can achieve the goal of obtaining multi-view satellite images, image preprocessing, dense matching and point cloud generation, and triangulation network construction through the cooperation between the steps. Figure 3 3D images can be collected and constructed from multiple perspectives, and the image acquisition technology of satellite images is used to improve the refinement of the ground three-dimensional model. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0035] Figure 1 The overall flow chart provided by the present invention.

[0036] Figure 2 The present invention provides a flowchart of the steps of acquiring multi-view images through satellites and preprocessing the images as described in the embodiments of the present invention.

[0037] Figure 3 A flowchart of the steps of performing dense matching and point cloud generation processing on pre-processed image data to obtain point cloud data as described in the embodiment provided by the present invention.

[0038] Figure 4 A flowchart of the steps of constructing a triangulated network based on point cloud data according to an embodiment of the present invention.

[0039] Figure 5 A flowchart of the steps of performing texture mapping in a triangulated network to obtain a three-dimensional model as described in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0041] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0042] Secondly, the present invention is described in detail with reference to the schematic diagram. When describing the embodiments of the present invention in detail, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0043] Furthermore, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0044] Example 1

[0045] like Figures 1-2 As shown, the present invention provides a technical solution: a method for fine three-dimensional modeling based on multi-view satellite images, comprising: S1: acquiring multi-view images through satellites and preprocessing the images; S2: performing dense matching and point cloud generation processing on the preprocessed image data to obtain point cloud data; S3: constructing a triangulated network based on the point cloud data; S4: performing texture mapping processing in the triangulated network to obtain a three-dimensional model; S5: visual display of the three-dimensional model.

[0046] Through the cooperation of the above steps, Figure 3 3D images can be collected and constructed from multiple perspectives, and the image acquisition technology of satellite images is used to improve the refinement of the ground three-dimensional model.

[0047] Specifically, S1: acquiring multi-view images through satellites and preprocessing the images; wherein the satellites include Gaofen-7 satellite and Gaojing satellite.

[0048] Furthermore, the steps of acquiring multi-view images through satellites and preprocessing the images include:

[0049] S11: The satellite photographs the same target area at different angles along the orbit direction and perpendicular to the orbit direction;

[0050] S12: Acquire multi-view satellite stereo images on the satellite platform;

[0051] It should be noted that a suitable satellite platform was selected, and two satellites with stereoscopic imaging capabilities, Gaofen-7 and Gaojing Satellite, were used. At the same time, it was ensured that the images covered the same target area, and were shot from two different angles, including along the orbit direction and perpendicular to the orbit direction, to obtain satellite stereo images from multiple perspectives. Each target area had more than three perspectives to solve the occlusion problem to the greatest extent.

[0052] S13: Perform radiation and atmospheric correction on the acquired images. The purpose of performing radiation and atmospheric correction on the images is to ensure the quality and consistency of the images and eliminate the influence of atmospheric and lighting conditions on the images. For example, the FLAASH atmospheric correction tool, based on the MODTRAN radiation transfer model, can correct wavelengths in the visible, near-infrared and short-wave infrared ranges, support a variety of sensors, and effectively remove water vapor / aerosol scattering effects.

[0053] Specifically, the steps of image radiation correction are as follows: first, perform data decomposition to create the original remote sensing image data file and the remote sensing auxiliary information data file; then perform system radiation correction based on the remote sensing image radiation error correction model. In ENVI, you can import the image and open the multispectral data according to a specific path for radiation correction; image atmospheric correction uses professional remote sensing image processing software such as ENVI or ArcGIS for atmospheric correction.

[0054] S14: Setting the corrected image in a unified geographic coordinate system to obtain image data; the purpose is to ensure the accuracy of images from different perspectives in geographic space.

[0055] Specifically, geocoding is corrected through the ellipsoid surface, which is a method of SAR image correction and may involve converting image data onto a specific ellipsoid surface to achieve unified geographic coordinates; or geocoding is corrected through terrain, which is also a type of SAR image correction and may take terrain factors into consideration to correct the image to make it consistent with the geographic coordinate system; it should be noted that the geographic coordinate system is used to determine the position of a point on the earth and is the basis of map making, navigation and other fields. The unified geographic coordinate system currently used in my country is the 2000 National Geodetic Coordinate System.

[0056] S15: coordinate registration of image data; specifically, coordinate registration of image data collected by satellite stereo from multiple perspectives can make the same-name image points in the multi-perspective stereo imaging be located on more images; it should be noted that the same-name image points refer to the image points corresponding to the same ground object on different images in the field of photogrammetry and remote sensing, also known as corresponding image points. In stereo photogrammetry, accurate determination and measurement of the same-name image points are crucial to ensure the quality and accuracy of stereo observation and measurement; through stereo observation, the image points of the same object point on different photos can be found, and then the information can be used to perform elevation measurement and three-dimensional reconstruction.

[0057] The steps of coordinate registration include: data preparation: ensure that all data are in the same reference system and have been projected to find reference points of common features; use registration tools: select appropriate registration tools in GIS software, such as "spatial correction" or "geo-registration"; perform registration: according to the selected tool, follow the software prompts to operate, such as setting control points, performing adjustment calculations, etc.; after completing the registration, check the registration results to ensure that the data are accurately aligned.

[0058] First, multi-view satellite images are acquired and calibrated to ensure consistency. Then, the image data from multiple viewpoints are aligned so that the same-name image points in the multi-view stereoscopic imaging are located on more images. After the algorithm finally generates 3D point cloud data, a 3D network is constructed using the 3D point cloud data using the 3D network growing method. The selected texture image is mapped onto the triangulated network to give the 3D model a realistic appearance and color. Finally, appropriate optimization is performed to obtain a complete and detailed 3D model. The model can be output as a visualization file or database format for application in geographic information systems, urban planning, resource management and other fields.

[0059] Example 2

[0060] Combined with Figure 3 The difference between this embodiment and the previous embodiment is that this embodiment further records the steps of densely matching and point cloud generation processing the pre-processed image data to obtain point cloud data, specifically S21: matching common feature points between multiple image data after coordinate alignment; S22: generating a disparity map from the common feature points through a stereo matching algorithm; S23: matching images of different perspectives pixel by pixel according to the disparity map and the shooting parameters of the satellite to generate high-density three-dimensional point cloud data.

[0061] Furthermore, the SIFT feature point matching algorithm is used to match common feature points between two or more image data after coordinate alignment; it should be noted that scale-invariant feature matching (SIFT) is a feature point matching algorithm based on image invariant feature quantities, and has a good effect on scale space, image scaling, rotation and even affine invariance.

[0062] Specifically, feature point extraction is done through 1. constructing scale space: generating images of different scales through Gaussian convolution kernel to simulate the multi-scale features of image data; 2. detecting extreme points: finding extreme points in Gaussian difference pyramid as candidate feature points; 3. positioning and screening of feature points: removing points with low contrast and strong edge response to determine the position and scale of feature points; 4. feature description: generating a 128-dimensional feature vector based on the gradient direction and amplitude of pixels around the feature point to describe the local image information of the feature point; 5. feature matching: using methods such as Euclidean distance or angle cosine similarity to compare the feature vectors of feature points in different images; selecting the feature point pairs with the highest similarity as matching points to achieve feature point matching between images.

[0063] It should be noted that the SIFT algorithm uses a Gaussian blur function

[0064]

[0065] DoG calculation

[0066] D(x,y,σ)=(G(x,y,kσ)-G(x,y,σ))* / (x,y)

[0067]

[0068] Among them, l(x, y) is the original image, k is the scale factor;

[0069] Gradient calculation

[0070]

[0071] Where m(x, y) is the gradient magnitude and θ(x, y) is the gradient direction.

[0072] Furthermore, the common feature points are used to generate a disparity map through a stereo matching algorithm, which can reflect the relative displacement between images. The specific steps of the stereo matching algorithm to generate a disparity map from common feature points are: 1. Image input: load the left and right eye images; 2. Image preprocessing: grayscale the image or perform other necessary processing; 3. Stereo correction: load the calibration parameters and perform epipolar correction on the image to more accurately match the corresponding points; 4. Calculate disparity: use the stereo matching algorithm to calculate the disparity value of each pixel in the left and right images; 5. Generate disparity map: generate a disparity map based on the calculated disparity value, and the value of each pixel in the disparity map represents the position difference of the point in the left and right images.

[0073] Furthermore, according to the disparity map and the shooting parameters of the satellite, the images of different perspectives are matched pixel by pixel to generate high-density three-dimensional point cloud data, which provides rich geometric information for subsequent three-dimensional modeling. Specifically, the satellite shooting parameters (such as camera baseline, focal length, etc.) and the disparity map are used to convert the disparity values ​​into depth values ​​to generate a depth map. The depth map and the pixel information of the original image are combined and the principle of triangulation is used to convert the two-dimensional pixel coordinates into three-dimensional world coordinates to generate three-dimensional point cloud data.

[0074] Example 3

[0075] Combined with Figure 4 , the present embodiment is different from the previous embodiment in that the present embodiment further records the steps of constructing a triangulated network based on the point cloud data, specifically S31: selecting key points presenting geometric features of terrain and objects in the point cloud data; S32: constructing a triangulated network using the key points;

[0076] It should be noted that the selection criteria for key points that present the geometric features of terrain and objects in point cloud data are as follows: 1. Points on contour lines: such as turning points and intersections, such as the corners of houses and the intersections of roads; 2. Center points and direction change points: such as the center points of independent objects and the direction change points of boundary lines; 3. Line changes: such as the turning points of rivers and roads, and the line type and line direction change points.

[0077] Specifically, the steps of constructing a triangulated network using key points are to select a construction method: 1. Select a triangulated network method or a triangulated network growth method based on THIESSEN triangulation; 2. Determine the initial baseline: find two points with the smallest distance in the discrete point set, and connect them as the initial baseline; 3. Construct the first triangle: use the Delaunay judgment method or THIESSEN triangulation method to find the third point of the triangle containing the initial baseline to form the first triangle; 4. Expand the triangle: use the three sides of the first triangle as the baseline, find extension points according to specific rules (such as the D-TIN judgment method), and construct new triangles until all discrete points become the endpoints of the triangle.

[0078] It should be noted that the three-dimensional point cloud data is used to construct a triangulated network through the three-dimensional network growth method; the triangulated network growth method includes the following steps: finding the initial baseline; searching for the third point of the first triangle; taking the three sides of the triangle as the baseline, continuing to search for extension points and constructing new triangles; repeating the above steps until all points are connected into triangles; the characteristics of the initial baseline: the initial baseline is usually in the form of a straight line; its length is generally not less than 0.7 to 0.5 times the length of the bridge axis; the method for determining the initial baseline: find the two points with the shortest distance among all discrete points, and connect these two points as the initial baseline; another method is to take a point near the geometric center, find the point closest to this point, and the two points connected are the initial baseline; the method for finding the initial baseline: by calculating the distance between all discrete points, find the pair of points with the smallest distance, which are the two endpoints of the initial baseline; or start from the vicinity of the geometric center and determine the two endpoints of the initial baseline by finding the nearest point.

[0079] In this embodiment, the steps of the stereo matching algorithm include matching cost calculation: calculating the similarity of corresponding pixels in the left and right images; cost aggregation: aggregating the matching costs in the local window to reduce the influence of noise; disparity calculation: selecting the optimal disparity value according to the aggregated cost; disparity optimization: further improving the accuracy and continuity of the disparity map through post-processing steps. On the basis of the SIFT matching algorithm, the NMI algorithm and the IMED algorithm are integrated into the SIFT matching algorithm. After the NMI algorithm is integrated into the SIFT algorithm, the operation process of the SIFT algorithm is simplified and the efficiency of the algorithm is improved. After the IMED algorithm is integrated into the SIFT algorithm, the association of each SIFT feature description pixel is fully considered, which effectively improves the accuracy of the algorithm.

[0080] It should be noted that in stereo vision, the left and right images represent the same scene taken from two different perspectives; by comparing and analyzing the two images, the depth information of the scene can be obtained, and then three-dimensional reconstruction can be achieved; the left and right images are usually obtained by using two cameras (such as a binocular camera) to shoot at the same time, or by a single camera at different positions.

[0081] Specifically, similarity is calculated by absolute difference: for corresponding pixels in the left and right images, the absolute difference of their grayscale values ​​or color values ​​is calculated; formula:

[0082] (AD=|I_L(×,y)-I_R(xd,V)|)

[0083] Where (I_L) and (I_R) represent the pixel values ​​of the left and right images respectively, and (d) is the disparity.

[0084] Result: According to the selected method, a value representing the similarity of corresponding pixels of the left and right images is calculated. By comparing the similarity of each pixel, the best matching pixel position can be found, thereby generating a disparity map, in which each pixel value represents the disparity of that point.

[0085] Specifically, in cost aggregation, the cost value of each pixel in the cost map is mainly aggregated with the cost values ​​of surrounding pixels; there are various ways of aggregation, including but not limited to mean aggregation, median aggregation, weighted average aggregation, etc.; in addition, some advanced methods such as bilateral filtering, confidence weights, cross lines, scan lines, etc. are also applied to cost aggregation; among them, the aggregation steps are: 1. Select an aggregation window: select an aggregation window for each pixel, and the pixels in the window will participate in the cost aggregation calculation of the pixel; 2. Calculate the aggregation cost: merge the cost values ​​of all pixels in the aggregation window into an aggregate cost value, which can be calculated using statistics such as mean, median or weighted average; 3. Determine the disparity: according to the aggregation cost value, select the minimum cost value as the disparity value of the pixel.

[0086] It should be noted that the criterion for selecting the optimal disparity value is that the selection of the optimal disparity value is based on the aggregation cost value, and the disparity value with the smallest aggregation cost value is selected as the optimal disparity; the steps for selecting the optimal disparity value are: 1. Aggregation window selection: First, select an aggregation window for each pixel, and the pixels in the window will participate in the cost aggregation calculation of the pixel; 2. Aggregation cost calculation: Then, for each pixel, the cost values ​​of all pixels in the aggregation window are merged into an aggregation cost value, which can be calculated using statistics such as mean, median or weighted average; 3. Disparity determination: Finally, based on the calculated aggregation cost value, select the disparity value with the smallest cost value as the optimal disparity of the pixel.

[0087] Specifically, the post-processing steps include 1. Applying WLS filtering: Using filtering technology based on weighted least squares (WLS) to optimize the pixel values ​​in the disparity map, help reduce noise, and improve the smoothness and continuity of the disparity map; 2. Adjusting contrast, sharpness, and denoising: Using image processing software such as Adobe Photoshop, adjust the contrast and sharpness of the disparity map and perform denoising; high contrast can highlight details and enhance clarity; sharpening can make edges clearer; denoising can reduce random interference signals and improve image quality.

[0088] The present invention improves the SIFT matching algorithm by using the NMI and IMED algorithms. After the NMI algorithm is integrated into the SIFT algorithm, the operation process of the SIFT algorithm is simplified and the efficiency of the algorithm is improved. The IMED algorithm is integrated into the SIFT algorithm, fully considering the association of the pixel points of each SIFT feature descriptor, and effectively improving the accuracy of the algorithm.

[0089] Example 4

[0090] Combined with Figure 5 The difference between this embodiment and the previous embodiment is that this embodiment further records the steps of performing texture mapping in the triangulated network, and obtaining the three-dimensional model includes:

[0091] S41: Selecting a corresponding texture image in each triangle patch of the triangulated mesh; this is to determine the texture image, specifically selecting a corresponding texture image for each triangle patch in the triangulated mesh, and selecting a visible image facing the triangle patch as the texture image.

[0092] S42: Select the visible image of the diagonal face patch as a texture image; this is for mapping, and the specifically selected texture image is mapped onto the triangulated network to give the three-dimensional model a real appearance and color.

[0093] S43: Map the selected texture image onto the triangulated network to construct a three-dimensional model.

[0094] Among them, the three-dimensional model needs to be optimized and enhanced in detail for visual display; the model optimization adopts smoothing and noise removal methods to smooth and remove noise from the three-dimensional model to ensure the fidelity and accuracy of the model. Specifically, the step of removing noise points is to use the inspection method to display the point cloud or generate the surface, and to check and adjust the point cloud data through a semi-interactive and semi-automatic smoothing method to remove outliers that are obviously sparsely distributed; the data smoothing step is to apply a smoothing algorithm, such as the Taubin filter, to smooth the model; set the number of iterations, lambda and mu parameters to adjust the degree of model smoothness.

[0095] Among them, detail enhancement processing is carried out through two methods: enhancing texture effects and adding shadow enhancement processing. The steps for enhancing texture effects are: 1. Select a suitable texture: select a suitable texture for the model and ensure that it matches the appearance characteristics of the model; 2. Adjust texture parameters: edit the texture and adjust parameters such as brightness and contrast to make the texture effect more prominent and delicate; the steps for adding shadow enhancement processing are: 1. Enable shadow effects: enable shadow effects in 3D software to add real shadows to the model; 2. Adjust shadow parameters: adjust the direction, intensity and granularity of the shadow to make the shadow effect more natural and realistic.

[0096] The three-dimensional model visualization is presented in chart and data formats.

[0097] The present invention cooperates with each other among texture mapping, post-processing and optimization, and output model, so that the constructed three-dimensional model can be converted into a format for output and use after optimization processing, thereby improving the quality and applicability of three-dimensional modeling.

[0098] Importantly, it should be noted that the construction and arrangement of the present application shown in a number of different exemplary embodiments are only exemplary. Although only a few embodiments are described in detail in this disclosure, it should be readily understood by those who refer to this disclosure that many modifications are possible, for example, the size, scale, structure, shape and proportion of various elements, and parameter values ​​such as temperature, pressure, etc., mounting arrangements, use of materials, color, directional changes, etc., without substantially departing from the novel angles and advantages of the subject matter described in the application. For example, the element shown as integrally formed can be composed of multiple parts or elements, the position of the element can be inverted or otherwise changed, and the nature or number or position of the discrete element can be changed or changed. Therefore, all such modifications are intended to be included in the scope of the present invention. The order or sequence of any process or method steps can be changed or reordered according to alternative embodiments. In the claims, any "device plus function" clause is intended to cover the structure of the execution function described herein, and is not only structurally equivalent but also equivalent structure. Without departing from the scope of the present invention, other replacements, modifications, changes and omissions can be made in the design, operating conditions and arrangement of the exemplary embodiments. Therefore, the invention is not limited to a specific embodiment, but extends to numerous modifications still falling within the scope of the appended claims.

[0099] Furthermore, in order to provide a concise description of exemplary embodiments, all features of an actual embodiment may not be described, i.e., those features that are not relevant to the best mode presently contemplated for carrying out the invention or those features that are not relevant to implementing the invention.

[0100] It should be understood that in the development of any actual implementation, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but for those of ordinary skill having the benefit of this disclosure, the development effort will be a routine task of design, fabrication, and production without undue experimentation.

[0101] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for fine three-dimensional modeling based on multi-view satellite images, characterized by: Including steps, Acquire multi-view images through satellites and pre-process the images; The pre-processed image data is subjected to dense matching and point cloud generation processing to obtain point cloud data; Construct triangulated network based on point cloud data; Carry out texture mapping in the triangulated network to obtain a three-dimensional model; Visual display of 3D models.

2. The method for fine three-dimensional modeling based on multi-view satellite images according to claim 1, characterized in that: The steps of acquiring multi-view images through satellites and preprocessing the images include: The satellite photographs the same target area at different angles along the orbital direction and perpendicular to the orbital direction; Acquire multi-view satellite stereo images on the satellite platform; Carry out radiometric and atmospheric correction processing on the acquired images; The corrected image is set in a unified geographic coordinate system to obtain image data; Coordinate registration of image data.

3. The method for fine three-dimensional modeling based on multi-view satellite images according to claim 1 or 2, characterized in that: The satellites include Gaofen-7 satellite and Gaojing satellite.

4. The method for fine three-dimensional modeling based on multi-view satellite images according to claim 3, characterized in that: The step of performing dense matching and point cloud generation processing on the pre-processed image data to obtain point cloud data includes: Matching common feature points between multiple image data after coordinate registration; Generate a disparity map from common feature points using a stereo matching algorithm; According to the disparity map and the satellite's shooting parameters, images from different perspectives are matched pixel by pixel to generate high-density three-dimensional point cloud data.

5. The method for fine three-dimensional modeling based on multi-view satellite images according to claim 4, characterized in that: The step of constructing a triangulated network based on point cloud data includes: Select key points that present the geometric features of terrain and objects in point cloud data; Construct a triangulated network using key points.

6. The method for fine three-dimensional modeling based on multi-view satellite images according to claim 3, characterized in that: The three-dimensional point cloud data is used to construct a triangulated network through a three-dimensional network growing method.

7. The method for fine 3D modeling based on multi-view satellite images according to any one of claims 4 to 6, characterized in that: The step of performing texture mapping in the triangulated network to obtain a three-dimensional model comprises: Select the corresponding texture image in each triangle of the triangulated network; Select the visible image of the diagonal face as the texture image; Map the selected texture image onto the triangulated network to construct a three-dimensional model.

8. The method for fine three-dimensional modeling based on multi-view satellite images according to claim 7, characterized in that: The three-dimensional model needs to be optimized and detailed enhanced during visual display.

9. The method for fine three-dimensional modeling based on multi-view satellite images according to claim 8, characterized in that: The model optimization adopts smoothing and noise removal.

10. The method for fine three-dimensional modeling based on multi-view satellite images according to claim 9, characterized in that: The three-dimensional model visualization is presented in chart and data formats.

Images