Method and system for generating satellite stereo remote sensing ortho-nadine image and three-dimensional reconstruction thereof
By constructing a rectangular coordinate system for the projection space of epipolar images and correcting DEM data, orthophotos of epipolar lines are generated and densely matched, solving the problems of epipolar line generation and viewpoint distortion, and achieving efficient and high-precision 3D reconstruction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINESE ACAD OF SURVEYING & MAPPING
- Filing Date
- 2026-03-25
- Publication Date
- 2026-06-19
AI Technical Summary
Existing technologies suffer from computational complexity, sensitivity to initial values, low efficiency, and low matching accuracy due to scale differences and viewpoint distortion in high-resolution stereo image pairs when generating epipolar images, which affects the adaptability and accuracy of 3D reconstruction.
By acquiring refined geometric orientation parameters, a Cartesian coordinate system for epipolar image projection is constructed. Orthorectification of the image is performed using DEM data to generate orthorectified epipolar images. Dense matching is then performed, and the positions of corresponding points are retrieved from the DEM to achieve high-precision 3D reconstruction.
It improves the efficiency and accuracy of 3D reconstruction, especially in complex terrain areas, significantly enhancing matching efficiency and 3D reconstruction quality, eliminating terrain projection differences, and ensuring row alignment and high-precision positioning of corresponding points.
Smart Images
Figure CN122244358A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of remote sensing image photogrammetry technology, and in particular to a method and system for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images. Background Technology
[0002] With the development of high-resolution remote sensing imagery and stereo mapping technology, image-based 3D reconstruction has become a key means of acquiring large-scale, high-precision spatial information of ground features. Geometric constraints between images—especially epipolar relationships—play a fundamental role in stereo matching and 3D reconstruction: under an ideal frame-based central projection model, corresponding image points strictly fall on corresponding epipolar lines, thus constraining a two-dimensional search to a one-dimensional search along the epipolar lines, significantly improving matching efficiency and accuracy.
[0003] To address the problem of epipolar line generation in linear array imagery, several approximate and generalized models have been proposed. Common approaches include generating epipolar lines directly from image geometry (or exterior orientation elements) based on the projection trajectory method, and generating epipolar line trajectories by using an inverse / forward solution to obtain the image-object mapping. The versatility of RFM has led to its widespread adoption in high-resolution satellite imagery geometric modeling; however, its inverse solution computation is complex, and some algorithms are sensitive to initial values or ground control points, thus limiting the practicality of automated epipolar line generation and large-scale batch processing.
[0004] Existing methods for generating / approximating epipolar lines have several practical problems: First, polynomial fitting or fitting methods based on corresponding points require a large number of uniformly distributed corresponding points as samples to solve for the fitting coefficients, and the fitting result is an approximate curve, making it difficult to guarantee true one-dimensional matching across the entire image range; second, although the projection trajectory method is theoretically rigorous, it relies on accurate sensor exterior orientation or complex inverse calculations, leading to decreased efficiency and stability under conditions without ground control points or with exterior orientation disturbances; third, in the process of image registration and resampling to the epipolar line direction for one-dimensional matching, compensation for image grayscale / geometric differences must be made, and traditional methods have limited effectiveness in dealing with cross-view scale, parallax scale changes, and brightness / contrast differences.
[0005] Besides the geometric complexity of epipolar lines, high-resolution stereo image pairs also face the problems of scale differences and viewpoint distortion in practical applications: different shooting perspectives cause local scale scaling and distortion in the epipolar direction, reducing the robustness of similarity measures based on fixed windows or fixed scales (such as SAD, NCC, etc.) during dense matching; at the same time, resampling the entire image into epipolar images for row alignment, without considering the scale changes caused by viewpoint, will introduce additional registration errors, thus affecting subsequent disparity solving and 3D point accuracy. Existing research has shown that straight-line constraints can be approximated within a local range, but when the target coverage area, terrain undulation, or viewpoint difference increases, the applicability and accuracy of the approximate straight-line method will significantly decrease.
[0006] To address the aforementioned shortcomings, a novel orthophoto epipolar image generation and 3D reconstruction technique is proposed to overcome the geometric complexity of epipolar lines and the scale differences and perspective distortions faced by high-resolution stereo image pairs in practical applications, thereby improving the adaptability and accuracy of 3D reconstruction. This is indeed necessary. Summary of the Invention
[0007] The purpose of this application is to provide a method and system for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images, which can achieve high-precision and high-efficiency three-dimensional reconstruction.
[0008] To achieve the above objectives, this application provides the following solution.
[0009] In a first aspect, this application provides a method for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images, including...
[0010] The original stereo remote sensing image is acquired and preprocessed to obtain the refined geometric orientation parameters corresponding to the original stereo remote sensing image; the original stereo remote sensing image is a stereo image pair including a left original image and a right original image; the refined geometric orientation parameters include the refined geometric orientation parameters of the left original image and the refined geometric orientation parameters of the right original image.
[0011] Based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image, the transformation parameters between the epipolar image projection space rectangular coordinate system and the map projection coordinate system are calculated, and the epipolar image projection space rectangular coordinate system is constructed. The transformation parameters include the average elevation value, baseline orientation angle, and coordinates of the center point of the overlapping area on the original stereo remote sensing image in the map projection coordinate system of the left and right original images.
[0012] Obtain DEM data and convert the DEM data to the Cartesian coordinate system of the epipolar image projection space to obtain DEM data in the Cartesian coordinate system of the epipolar image projection space.
[0013] Using the DEM data in the Cartesian coordinate system of the epipolar image projection space, the original stereo remote sensing image is orthorectified to generate an orthorectified epipolar image; when the original stereo remote sensing image is the left original image, the orthorectified epipolar image is the left orthorectified epipolar image; when the original stereo image is the right original image, the orthorectified epipolar image is the right orthorectified epipolar image.
[0014] Dense matching is performed on the orthophoto epipolar image to obtain a disparity map; the disparity map includes several image points and the disparity values of the image points, and each image point corresponds to a pair of corresponding points; the dense matching is a dense matching of the left orthophoto epipolar image and the right orthophoto epipolar image along the epipolar direction; the corresponding points are a pair of image points on the left orthophoto epipolar image and the right orthophoto epipolar image, respectively, representing the same point in the geographic latitude and longitude coordinate system.
[0015] For any image point and its disparity value in the disparity map, the position of the corresponding point on the orthophoto epipolar image on the original stereo remote sensing image is inverted based on the transformation parameters between the rectangular coordinate system of the epipolar image projection space and the map projection coordinate system, thus obtaining the image point coordinates of each corresponding point on the original stereo remote sensing image.
[0016] Based on the refined orientation parameters corresponding to the original stereo remote sensing image, spatial intersection is performed on the image point coordinates on the original stereo remote sensing image corresponding to each corresponding point to obtain the positioning coordinates of each corresponding point in the geographic latitude and longitude coordinate system; the set of positioning coordinates obtained by the spatial intersection of all corresponding points constitutes a dense point cloud.
[0017] Secondly, this application provides a satellite stereo remote sensing orthophoto epipolar image generation and three-dimensional reconstruction system, which includes the following modules.
[0018] The orientation parameter acquisition module is used to acquire the original stereo remote sensing image and preprocess the original stereo remote sensing image to obtain the refined geometric orientation parameters corresponding to the original stereo remote sensing image; the original stereo remote sensing image is a stereo image pair including a left original image and a right original image; the refined geometric orientation parameters include the refined geometric orientation parameters of the left original image and the refined geometric orientation parameters of the right original image.
[0019] The transformation parameter calculation module is used to calculate the transformation parameters between the epipolar image projection space rectangular coordinate system and the map projection coordinate system based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image, and to construct the epipolar image projection space rectangular coordinate system. The transformation parameters include the average elevation value, baseline orientation angle, and coordinates of the center point of the overlapping area on the original stereo remote sensing image in the map projection coordinate system of the left and right original images.
[0020] The data conversion module is used to acquire DEM data and convert the DEM data into the rectangular coordinate system of the epipolar image projection space to obtain DEM data in the rectangular coordinate system of the epipolar image projection space.
[0021] The image generation module is used to perform epipolar image orthorectification on the original stereo remote sensing image using DEM data in the rectangular coordinate system of the epipolar image projection space, and generate an orthorectified epipolar image; when the original stereo remote sensing image is a left original image, the orthorectified epipolar image is a left orthorectified epipolar image; when the original stereo image is a right original image, the orthorectified epipolar image is a right orthorectified epipolar image.
[0022] The disparity map acquisition module is used to perform dense matching on the orthophoto epipolar image to obtain a disparity map. The disparity map includes several image points and the disparity values of the image points, with each image point corresponding to a pair of corresponding points. The dense matching is a dense matching of the left and right orthophoto epipolar images along the epipolar direction. The corresponding points are a pair of image points on the left and right orthophoto epipolar images, respectively, representing the same point in the geographic latitude and longitude coordinate system.
[0023] The original stereo image point coordinate calculation module is used to calculate the image point coordinates of each corresponding point on the original stereo remote sensing image for any image point and its disparity value in the disparity map, based on the transformation parameters between the rectangular coordinate system of the epipolar image projection space and the map projection coordinate system.
[0024] The dense point cloud generation module is used to perform spatial intersection of the image point coordinates on the original stereo remote sensing image corresponding to each corresponding point according to the refined orientation parameters of the original stereo remote sensing image, so as to obtain the positioning coordinates of each corresponding point in the geographic latitude and longitude coordinate system; the set of positioning coordinates obtained by the spatial intersection of all corresponding points constitutes the dense point cloud.
[0025] According to the specific embodiments provided in this application, this application has the following technical effects.
[0026] This application proposes a method for generating and reconstructing 3D epipolar images using satellite stereo remote sensing. A geometric benchmark is established using high-precision image orientation parameters. A rectangular coordinate system for the epipolar image projection space is constructed. The average elevation and baseline orientation angle are estimated using a DEM (Digital Image Model) to ensure alignment of corresponding points and eliminate most vertical parallax. The external DEM is converted to epipolar space and corrected during epipolar image generation to eliminate terrain projection differences, producing image pairs with both orthorectified and horizontal epipolar characteristics, laying the foundation for dense matching. Subsequently, the matching parallax map is combined with the epipolar space DEM to retrieve the original image point coordinates. Spatial forward intersection is performed using refined orientation parameters to generate geographic coordinates point by point, forming a high-precision dense 3D point cloud for 3D reconstruction. This method effectively solves the problems of low efficiency and poor accuracy in dense matching caused by terrain undulations and epipolar geometric issues in traditional processes. Attached Figure Description
[0027] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0028] Figure 1 This is a flowchart illustrating a method for generating and reconstructing a satellite stereo remote sensing orthophoto epipolar image, as provided in an embodiment of this application.
[0029] Figure 2 This is a schematic diagram of the functional modules of a satellite stereo remote sensing orthophoto epipolar image generation and 3D reconstruction system provided in an embodiment of this application. Detailed Implementation
[0030] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0031] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0032] The satellite stereo remote sensing orthorectified epipolar image generation and 3D reconstruction method provided in this application preprocesses the original stereo remote sensing image to obtain refined geometric orientation parameters corresponding to the original stereo remote sensing image. Based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image, the transformation parameters between the epipolar image projection space rectangular coordinate system and the map projection coordinate system are calculated, and the epipolar image projection space rectangular coordinate system is constructed. DEM data is acquired and converted to the epipolar image projection space rectangular coordinate system to obtain DEM data in the epipolar image projection space rectangular coordinate system. Using the DEM data in the epipolar image projection space rectangular coordinate system, the original stereo remote sensing image is orthorectified to generate an orthorectified epipolar image. Dense matching is performed on the orthorectified epipolar image to obtain a disparity map. For any image point and its disparity value in the disparity map, based on the transformation parameters between the epipolar image projection space rectangular coordinate system and the map projection coordinate system, the position of the corresponding point on the orthophoto epipolar image is inverted onto the original stereo remote sensing image, obtaining the image point coordinates of each corresponding point on the original stereo remote sensing image. Based on the refined orientation parameters corresponding to the original stereo remote sensing image, spatial intersection is performed on the image point coordinates of each corresponding point on the original stereo remote sensing image, obtaining the location coordinates of each corresponding point in the geographic latitude and longitude coordinate system; the set of location coordinates obtained from the spatial intersection of all corresponding points constitutes a dense point cloud.
[0033] In one exemplary embodiment, such as Figure 1 As shown, a method for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images is provided. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method includes the following steps A1 to A7.
[0034] A1. Acquire the original stereo remote sensing image and preprocess the original stereo remote sensing image to obtain the refined geometric orientation parameters corresponding to the original stereo remote sensing image; the original stereo remote sensing image is a stereo image pair including a left original image and a right original image; the refined geometric orientation parameters include the refined geometric orientation parameters of the left original image and the refined geometric orientation parameters of the right original image.
[0035] A2. Based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image, calculate the transformation parameters between the epipolar image projection space rectangular coordinate system and the map projection coordinate system, and construct the epipolar image projection space rectangular coordinate system; the transformation parameters include the average elevation value, baseline orientation angle, and coordinates of the center point of the overlapping area on the original stereo remote sensing image in the map projection coordinate system of the left and right original images.
[0036] A3. Obtain DEM data and convert the DEM data to the rectangular coordinate system of the epipolar image projection space to obtain DEM data in the rectangular coordinate system of the epipolar image projection space.
[0037] A4. Using the DEM data in the rectangular coordinate system of the epipolar image projection space, perform epipolar image orthorectification on the original stereo remote sensing image to generate an orthorectified epipolar image; when the original stereo remote sensing image is the left original image, the orthorectified epipolar image is the left orthorectified epipolar image; when the original stereo image is the right original image, the orthorectified epipolar image is the right orthorectified epipolar image.
[0038] A5. Perform dense matching on the orthophoto epipolar image to obtain a disparity map; the disparity map is the positional difference between a left epipolar image point and its corresponding right epipolar image point along the epipolar direction; the disparity map includes several image points and the disparity values of the image points, with each image point corresponding to a pair of corresponding points; the dense matching is a dense matching of the left and right orthophoto epipolar images along the epipolar direction; the corresponding points are a pair of image points representing the same point in the geographic latitude and longitude coordinate system on the left and right orthophoto epipolar images respectively.
[0039] A6. For any image point in the disparity map and the disparity value of the image point, based on the transformation parameters between the rectangular coordinate system of the epipolar image projection space and the map projection coordinate system, the position of the corresponding point on the orthophoto epipolar image on the original stereo remote sensing image is inverted to obtain the image point coordinates of each corresponding point on the original stereo remote sensing image.
[0040] A7. Based on the refined orientation parameters corresponding to the original stereo remote sensing image, perform spatial intersection on the image point coordinates of each corresponding point on the original stereo remote sensing image to obtain the positioning coordinates of each corresponding point in the geographic latitude and longitude coordinate system; the set of positioning coordinates obtained from the spatial intersection of all corresponding points constitutes a dense point cloud.
[0041] By implementing steps A1 to A7 above, orthorectification and disparity map acquisition can be achieved. Vertical disparity is eliminated based on parameters such as accurate average elevation and baseline orientation angle, ensuring alignment of corresponding points. The DEM data is converted to a Cartesian coordinate system in the epipolar image projection space, simulating terrain undulations and correcting the epipolar image to eliminate most terrain projection differences, generating image pairs with both orthorectified and horizontal epipolar characteristics. A disparity map is obtained through dense matching of the orthorectified epipolar images. The coordinates of corresponding points on the left and right orthorectified epipolar images are inverted onto the original image. Spatial intersection is then performed using orientation parameters to obtain the location coordinates in the geographic latitude and longitude coordinate system, forming a dense point cloud and completing 3D reconstruction. Furthermore, this application can improve the efficiency and accuracy of dense matching, especially in areas with complex terrain, significantly improving the quality of 3D reconstruction.
[0042] In another exemplary embodiment of this application, in order to obtain the refined orientation parameters corresponding to the original stereoscopic image, a corresponding preprocessing method can be selected according to the category of the original stereoscopic image. The above step A1 specifically includes: steps A11 to A12.
[0043] A11. When the original stereo remote sensing image contains only stereo image pairs, perform relative orientation adjustment or absolute orientation adjustment on the stereo image pairs to obtain the refined geometric orientation parameters corresponding to the stereo image pairs.
[0044] A12. When the original stereo remote sensing image is a component of the multiview image, perform regional network adjustment on the multiview image, and simultaneously obtain the geometric orientation parameters of the original stereo remote sensing image and other refined images.
[0045] In another exemplary embodiment of this application, the transformation parameter between the epipolar image projection space rectangular coordinate system and the map projection coordinate system further includes: the coordinates of the center point of the overlapping area on the original stereo remote sensing image in the map projection coordinate system. To calculate the transformation parameter between the epipolar image projection space rectangular coordinate system and the map projection coordinate system, step A2 specifically includes steps A21 to A23.
[0046] A21. Calculate the four-corner coordinates based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image; the four-corner coordinates include the four-corner coordinates of the left original image in the map projection coordinate system and the four-corner coordinates of the right original image in the map projection coordinate system.
[0047] A22. Calculate the overlapping area of the left and right original images in the map projection coordinate system based on the four corner coordinates.
[0048] A23. Estimate the average elevation value of the DEM data of the overlapping area of the left and right original images in the map projection coordinate system, and reduce the average elevation value by a certain height as the elevation setpoint. In this embodiment, the elevation setpoint is set according to the average elevation value, specifically, the elevation setpoint is the average elevation value minus 50.
[0049] In another exemplary embodiment of this application, after step A22 described above, steps B1 to B are also included.
[0050] B1. Based on the refined geometric orientation parameters of the left original image, estimate the position of the center point of the overlapping region on the left original image. .
[0051] B2. Based on the image point position corresponding to the center point of the overlapping region on the left original image. The average elevation value and the elevation setting value are used to obtain the image point position of the center point of the overlapping area on the left original image. Positioning coordinates in a geographic latitude and longitude coordinate system and .
[0052] B3. Based on the positioning coordinates in the geographic latitude and longitude coordinate system The geometric orientation parameters of the refined original image on the right are obtained. The corresponding image point coordinates on the right original image .
[0053] B4. Calculate the coordinates of the image points on the right original image based on the refined geometric orientation parameters and average elevation values. When the elevation is the average elevation value, the positioning coordinates in the geographic latitude and longitude coordinate system .
[0054] B5. When the elevation is the average elevation value, the image points on the left original image are... Location point in the geographic latitude and longitude coordinate system Image points on the right original image Location point in the geographic latitude and longitude coordinate system Projecting onto the map projection coordinate system yields planar coordinate points. and points .
[0055] B6. The plane coordinate points and points Connect to form a vector .
[0056] B7. Calculate vectors The baseline direction angle is obtained by forming a direction angle α with the x-axis of the map projection coordinate system.
[0057] In another exemplary embodiment of this application, the origin of the x-axis and y-axis of the epipolar image projection spatial rectangular coordinate system in step 202 above is a planar coordinate point. Furthermore, the origin of the z-axis of the epipolar image projection space rectangular coordinate system is the zero point of elevation, and the positive direction of the z-axis of the epipolar image projection space rectangular coordinate system is consistent with the positive direction of elevation; the x-axis of the epipolar image projection space rectangular coordinate system lies in the plane formed by the XY axes of the map projection coordinate system, and the angle between the x-axis of the epipolar image projection space rectangular coordinate system and the X-axis of the map projection coordinate system is the baseline direction angle; the y-axis of the epipolar image projection space rectangular coordinate system lies in the plane formed by the XY axes of the map projection coordinate system according to the right-hand rule, and is perpendicular to the x-axis of the epipolar image projection space rectangular coordinate system; the zero point of elevation is located on the ground.
[0058] In another exemplary embodiment of this application, in order to generate an orthophoto epipolar image, step A4 above further includes steps A41 to A46 instead.
[0059] A41. Transform the four-corner coordinates corresponding to the left original image to the rectangular coordinate system of the epipolar image projection space using the forward transformation formula, and calculate the range of the left original image in the rectangular coordinate system of the epipolar image projection space; the forward transformation formula is: ; in, It is a point x-coordinate It is a point The ordinate, It is a point elevation, [ The three-dimensional coordinates of the object point in the map projection coordinate system; It is a point x-coordinate It is a point The ordinate, It is a point elevation, [ The coordinates of the object point are the three-dimensional coordinates in the Cartesian coordinate system of the epipolar image projection space. In this embodiment, transforming the four corner coordinates corresponding to the left original image to the Cartesian coordinate system of the epipolar image projection space specifically includes: transforming the coordinates of the four corners of the left original image in the map projection coordinate system to the Cartesian coordinate system of the epipolar image projection space.
[0060] A42. Transform the four corner coordinates of the right original image in the map projection coordinate system to the rectangular coordinate system of the epipolar image projection space using the positive transformation formula, and calculate the range of the right original image in the rectangular coordinate system of the epipolar image projection space.
[0061] A43. In the rectangular coordinate system of the epipolar image projection space, calculate the overlapping area of the left and right original images based on the range of the left original image and the range of the right original image, and record the coordinates of the upper left corner of the overlapping area. and the coordinates of the bottom right corner .
[0062] A44. Obtain the spatial resolution of the original stereo remote sensing image, and set the spatial resolution of the epipolar image in the rectangular coordinate system of the epipolar image projection space according to the spatial resolution of the original stereo remote sensing image.
[0063] A45. Based on the coordinates of the upper left and lower right corners of the overlapping area and the spatial resolution of the rectangular coordinate system of the epipolar image projection space, calculate the corrected width and height of the orthophoto epipolar image.
[0064] A46. Based on the width and height of the corrected orthophoto epipolar image, the original stereo remote sensing image is corrected to the rectangular coordinate system of the epipolar image projection space using the indirect correction method to generate the orthophoto epipolar image.
[0065] In another exemplary embodiment of this application, step A46 above further includes steps A461 to A462.
[0066] A461. Based on the width and height of the corrected orthophoto epipolar image, set an image matrix; the width of the image matrix is the width of the corrected orthophoto epipolar image; the height of the image matrix is the height of the corrected orthophoto epipolar image.
[0067] A462. In the image matrix, each pixel is resampled to obtain an orthophoto epipolar image.
[0068] In another exemplary embodiment of this application, step A462 further includes steps A4621 to A4625.
[0069] A4621. Based on the coordinates of the top left corner of the overlapping area Based on the spatial resolution of the original stereo remote sensing image, the image coordinates of the current image point are transformed into the planar coordinates of the current image point in the Cartesian coordinate system of the epipolar image projection space. The elevation value of the current image point is extracted from the DEM data in the epipolar image projection space rectangular coordinate system. .
[0070] A4622. Based on the plane coordinates of the current image point in the rectangular coordinate system of the epipolar image projection space. Elevation value of the current image point Using the inverse transformation formula, we obtain the coordinates of the current image point in the map projection coordinate system. .
[0071] A4623. The coordinates of the current image point in the map projection coordinate system... Convert to a geographic latitude and longitude coordinate system to obtain the current image point's location coordinates in the geographic latitude and longitude coordinate system. The inverse transform formula is: ; in, It is a point x-coordinate It is a point The ordinate, It is a point Elevation, For object point Three-dimensional coordinates in a map projection coordinate system; It is a point x-coordinate It is a point The ordinate, It is a point elevation, [ Let be the three-dimensional coordinates of the object point in the rectangular coordinate system of the epipolar image projection space.
[0072] A4624. Based on the positioning coordinates of the current image point in the geographic latitude and longitude coordinate system. Using the refined orientation parameters, the position of the current image point on the original stereo remote sensing image is calculated. .
[0073] A4625. Using bilinear interpolation, based on the position of the current image point on the original stereo remote sensing image. The grayscale value of the current image point is resampled from the original stereo remote sensing image, and the grayscale value is assigned to the current image point to complete the resampling of the current image point.
[0074] In another exemplary embodiment of this application, the calculation of the three-dimensional coordinates corresponding to each left orthophoto epipolar image point is achieved through the following steps.
[0075] Using the image point coordinates of the left orthophoto epipolar image, calculate the plane coordinates of the image points in the rectangular coordinate system of the orthophoto epipolar projection space. The elevation value corresponding to the image point is extracted from the DEM data in the orthophoto epipolar projection space rectangular coordinate system. The spatial coordinates in the rectangular coordinate system of the orthophoto epipolar image projection space are obtained. .
[0076] Using the inverse transformation formula, calculate the coordinates of image points in the left orthophoto epipolar image in the map projection coordinate system. .
[0077] Calculate the location coordinates of the point in the geographic coordinate system using the coordinates and elevation values in the map projection coordinate system. .
[0078] Using the refined geometric orientation parameters of the left original image, calculate the image point position in the geographic coordinate system on the left original image. .
[0079] Using the image point coordinates of the left orthophoto epipolar image and the disparity map obtained by dense matching, the image point coordinates of the corresponding matching points in the right orthophoto epipolar image are retrieved.
[0080] Based on the image point coordinates of the right orthophoto epipolar image, calculate the planar position of the image point of the left orthophoto epipolar image in the rectangular coordinate system of the epipolar image projection space. Based on this location, the elevation value is extracted from the DEM in the rectangular coordinate system of the epipolar image projection space. The coordinates of the corresponding points of the right epipolar image in the epipolar projection space coordinate system are obtained. .
[0081] Calculate the coordinates of the left orthophoto epipolar image point in the map projection coordinate system using the inverse transformation formula. And further convert it to a geographic coordinate system to obtain its geographic coordinates. .
[0082] Using the refined geometric orientation parameters of the right original image, calculate the positioning coordinates in the previous geographic coordinate system, and then calculate its image point position on the right original image. .
[0083] Using the image point coordinates of the left original image Its refined geometric orientation parameters, and the image point coordinates of the original right image. The refined geometric orientation parameters are used to perform forward intersection of corresponding points to obtain a three-dimensional coordinate point in the geographic latitude and longitude coordinate system. This point is the three-dimensional reconstructed coordinate value corresponding to the current left core line image point.
[0084] Based on the same inventive concept, this application also provides a system for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images as described above. The solution provided by this system is similar to the solution described in the above method; therefore, the specific limitations of one or more embodiments of the satellite stereo remote sensing orthophoto epipolar image generation and 3D reconstruction system provided below can be found in the above-described limitations of a satellite stereo remote sensing orthophoto epipolar image generation and 3D reconstruction method.
[0085] In one exemplary embodiment, such as Figure 2 As shown, a satellite stereo remote sensing orthophoto epipolar image generation and 3D reconstruction system is provided, which includes the following modules.
[0086] The orientation parameter acquisition module is used to acquire the original stereo remote sensing image and preprocess the original stereo remote sensing image to obtain the refined geometric orientation parameters corresponding to the original stereo remote sensing image; the original stereo remote sensing image is a stereo image pair including a left original image and a right original image; the refined geometric orientation parameters include the refined geometric orientation parameters of the left original image and the refined geometric orientation parameters of the right original image.
[0087] The transformation parameter calculation module is used to calculate the transformation parameters between the epipolar image projection space rectangular coordinate system and the map projection coordinate system based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image, and to construct the epipolar image projection space rectangular coordinate system. The transformation parameters include the average elevation value, baseline orientation angle, and coordinates of the center point of the overlapping area on the original stereo remote sensing image in the map projection coordinate system of the left and right original images.
[0088] The data conversion module is used to acquire DEM data and convert the DEM data into the rectangular coordinate system of the epipolar image projection space to obtain DEM data in the rectangular coordinate system of the epipolar image projection space.
[0089] The image generation module is used to perform epipolar image orthorectification on the original stereo remote sensing image using DEM data in the rectangular coordinate system of the epipolar image projection space, and generate an orthorectified epipolar image; when the original stereo remote sensing image is a left original image, the orthorectified epipolar image is a left orthorectified epipolar image; when the original stereo image is a right original image, the orthorectified epipolar image is a right orthorectified epipolar image.
[0090] The disparity map acquisition module is used to perform dense matching on the orthophoto epipolar image to obtain a disparity map. The disparity map includes several image points and the disparity values of the image points, with each image point corresponding to a pair of corresponding points. The dense matching is a dense matching of the left and right orthophoto epipolar images along the epipolar direction. The corresponding points are a pair of image points on the left and right orthophoto epipolar images, respectively, representing the same point in the geographic latitude and longitude coordinate system.
[0091] The original stereo image point coordinate calculation module is used to calculate the image point coordinates of each corresponding point on the original stereo remote sensing image for any image point and its disparity value in the disparity map, based on the transformation parameters between the rectangular coordinate system of the epipolar image projection space and the map projection coordinate system.
[0092] The dense point cloud generation module is used to perform spatial intersection of the image point coordinates on the original stereo remote sensing image corresponding to each corresponding point according to the refined orientation parameters of the original stereo remote sensing image, so as to obtain the positioning coordinates of each corresponding point in the geographic latitude and longitude coordinate system; the set of positioning coordinates obtained by the spatial intersection of all corresponding points constitutes the dense point cloud.
[0093] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0094] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A method for generating and reconstructing three-dimensional satellite stereo remote sensing orthophoto epipolar images, characterized in that, include: The original stereo remote sensing image is acquired and preprocessed to obtain the refined geometric orientation parameters corresponding to the original stereo remote sensing image; the original stereo remote sensing image is a stereo image pair including a left original image and a right original image; the refined geometric orientation parameters include the refined geometric orientation parameters of the left original image and the refined geometric orientation parameters of the right original image. Based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image, the transformation parameters between the epipolar image projection space rectangular coordinate system and the map projection coordinate system are calculated, and the epipolar image projection space rectangular coordinate system is constructed. The transformation parameters include the average elevation value, baseline orientation angle, and coordinates of the center point of the overlapping area on the original stereo remote sensing image in the map projection coordinate system of the left and right original images. Obtain DEM data and convert the DEM data to the Cartesian coordinate system of the epipolar image projection space to obtain DEM data in the Cartesian coordinate system of the epipolar image projection space; Using the DEM data in the Cartesian coordinate system of the epipolar image projection space, the original stereo remote sensing image is orthorectified to generate an orthorectified epipolar image; when the original stereo remote sensing image is the left original image, the orthorectified epipolar image is the left orthorectified epipolar image; when the original stereo image is the right original image, the orthorectified epipolar image is the right orthorectified epipolar image. Dense matching is performed on the orthophoto epipolar image to obtain a disparity map; the disparity map includes several image points and the disparity values of the image points, and each image point corresponds to a pair of corresponding points; the dense matching is a dense matching of the left orthophoto epipolar image and the right orthophoto epipolar image along the epipolar direction; the corresponding points are a pair of image points on the left orthophoto epipolar image and the right orthophoto epipolar image, respectively, representing the same point in the geographic latitude and longitude coordinate system; For any image point and its disparity value in the disparity map, the position of the corresponding point on the orthophoto epipolar image on the original stereo remote sensing image is inverted based on the transformation parameters between the rectangular coordinate system of the epipolar image projection space and the map projection coordinate system, thus obtaining the image point coordinates of each corresponding point on the original stereo remote sensing image. Based on the refined orientation parameters corresponding to the original stereo remote sensing image, spatial intersection is performed on the image point coordinates on the original stereo remote sensing image corresponding to each corresponding point to obtain the positioning coordinates of each corresponding point in the geographic latitude and longitude coordinate system; the set of positioning coordinates obtained by the spatial intersection of all corresponding points constitutes a dense point cloud.
2. The method for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images according to claim 1, characterized in that, The preprocessing of the original stereo remote sensing image to obtain the refined geometric orientation parameters corresponding to the original stereo remote sensing image specifically includes: When the original stereo remote sensing image contains only stereo image pairs, relative orientation adjustment or absolute orientation adjustment is performed on the stereo image pairs to obtain the refined geometric orientation parameters corresponding to the stereo image pairs. When the original stereo remote sensing image is a component of the multiview image, the multiview image is subjected to regional network adjustment, and the geometric orientation parameters of the original stereo remote sensing image and other refined images are obtained simultaneously.
3. The method for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images according to claim 1, characterized in that, Based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image, the transformation parameters between the epipolar image projection space rectangular coordinate system and the map projection coordinate system are calculated, including: based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image, the average elevation value of the overlapping area of the left and right original images in the map projection coordinate system is calculated, specifically including: The four-corner coordinates are calculated based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image; the four-corner coordinates include the four-corner coordinates of the left original image in the map projection coordinate system and the four-corner coordinates of the right original image in the map projection coordinate system. Calculate the overlapping area of the left and right original images in the map projection coordinate system based on the four corner coordinates; Estimate the average elevation value of the DEM data of the overlapping area of the left and right original images in the map projection coordinate system, and reduce the height by a certain amount based on the average elevation value as the elevation set value.
4. The method for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images according to claim 3, characterized in that, After estimating the average elevation value of the DEM data of the overlapping area of the left and right original images in the map projection coordinate system, and reducing the average elevation value by a certain height as the elevation setpoint, the method further includes: Based on the refined geometric orientation parameters of the left original image, estimate the image point position of the center point of the overlapping region on the left original image. ; Based on the image point position corresponding to the center point of the overlapping region on the left original image The image point is obtained by combining the average elevation value and the elevation set value. Positioning coordinates in a geographic latitude and longitude coordinate system and ; Based on the positioning coordinates in the geographic latitude and longitude coordinate system The geometric orientation parameters of the refined original image on the right are obtained. The corresponding image point coordinates on the right original image ; The coordinates of the image points are calculated based on the geometric orientation parameters and average elevation values of the refined original right image. When the elevation is the average elevation value, the positioning coordinates in the geographic latitude and longitude coordinate system ; When the elevation is the average elevation value, the image points on the left original image are... Location point in the geographic latitude and longitude coordinate system Image points on the right original image Location point in the geographic latitude and longitude coordinate system Projecting onto the map projection coordinate system yields planar coordinate points. and points ; The plane coordinate points and points Connect to form a vector ; Calculate vectors The baseline direction angle is obtained by forming a direction angle α with the x-axis of the map projection coordinate system.
5. The method for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images according to claim 4, characterized in that, The origin of the x-axis and y-axis of the nuclear radiation image projection spatial rectangular coordinate system is a plane coordinate point. Furthermore, the origin of the z-axis of the epipolar image projection space rectangular coordinate system is the zero point of elevation, and the positive direction of the z-axis of the epipolar image projection space rectangular coordinate system is consistent with the positive direction of elevation; the x-axis of the epipolar image projection space rectangular coordinate system lies in the plane formed by the XY axes of the map projection coordinate system, and the angle between the x-axis of the epipolar image projection space rectangular coordinate system and the X-axis of the map projection coordinate system is the baseline direction angle; the y-axis of the epipolar image projection space rectangular coordinate system lies in the plane formed by the XY axes of the map projection coordinate system according to the right-hand rule, and is perpendicular to the x-axis of the epipolar image projection space rectangular coordinate system; the zero point of elevation is located on the ground.
6. The method for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images according to claim 1, characterized in that, Based on the DEM data in the Cartesian coordinate system of the epipolar image projection space, the original stereo remote sensing image is orthorectified to generate an orthorectified epipolar image, specifically including: The four-corner coordinates corresponding to the left original image are transformed to the rectangular coordinate system of the epipolar image projection space using a forward transformation formula, and the range of the left original image in the rectangular coordinate system of the epipolar image projection space is calculated; the forward transformation formula is: ； in, It is a point x-coordinate It is a point The ordinate, It is a point elevation, [ The three-dimensional coordinates of the object point in the map projection coordinate system; It is a point x-coordinate It is a point The ordinate, It is a point elevation, [ The three-dimensional coordinates of the object point in the rectangular coordinate system of the epipolar image projection space; The four corner coordinates of the right original image in the map projection coordinate system are transformed to the rectangular coordinate system of the epipolar image projection space by using the positive transformation formula, and the range of the right original image in the rectangular coordinate system of the epipolar image projection space is calculated. In the rectangular coordinate system of the epipolar image projection space, the overlapping area of the left and right original images is calculated based on the range of the left original image and the range of the right original image, and the coordinates of the upper left corner of the overlapping area are recorded. and the coordinates of the bottom right corner ; Obtain the spatial resolution of the original stereo remote sensing image, and set the spatial resolution of the epipolar image in the rectangular coordinate system of the epipolar image projection space according to the spatial resolution of the original stereo remote sensing image; Based on the coordinates of the upper left and lower right corners of the overlapping region and the spatial resolution of the epipolar image, calculate the width and height of the corrected orthophoto epipolar image. Based on the width and height of the corrected orthophoto epipolar image, an indirect correction method is used to correct the original stereo remote sensing image to the rectangular coordinate system of the epipolar image projection space, thereby generating an orthophoto epipolar image.
7. The method for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images according to claim 6, characterized in that, Based on the corrected width and height of the orthorectified epipolar image, an indirect correction method is used to correct the original stereo remote sensing image to a Cartesian coordinate system in the epipolar image projection space, generating an orthorectified epipolar image. Specifically, this includes: An image matrix is set up based on the width and height of the corrected orthophoto epipolar image; the width of the image matrix is the width of the corrected orthophoto epipolar image; the height of the image matrix is the height of the corrected orthophoto epipolar image. In the image matrix, each pixel is resampled to obtain an orthophoto epipolar image.
8. The method for generating and reconstructing satellite stereo remote sensing orthophoto epipolar images according to claim 7, characterized in that, In the image matrix, resampling of each pixel specifically includes: Based on the coordinates of the top left corner of the overlapping area Based on the spatial resolution of the original stereo remote sensing image, the coordinates of the current image point are transformed into planar coordinates of the current image point in the Cartesian coordinate system of the epipolar image projection space. The elevation value of the current image point is extracted from the DEM data in the epipolar image projection space rectangular coordinate system. ; Based on the plane coordinates of the current image point in the Cartesian coordinate system of the epipolar image projection space Elevation value of the current image point Using the inverse transformation formula, we obtain the coordinates of the current image point in the map projection coordinate system. ; The coordinates of the current image point in the map projection coordinate system Convert to a geographic latitude and longitude coordinate system to obtain the current image point's location coordinates in the geographic latitude and longitude coordinate system. The inverse transform formula is: ； in, It is a point x-coordinate It is a point The ordinate, It is a point Elevation, For object point Three-dimensional coordinates in a map projection coordinate system; It is a point x-coordinate It is a point The ordinate, It is a point elevation, [ The three-dimensional coordinates of the object point in the rectangular coordinate system of the epipolar image projection space; Based on the current image point's location coordinates in the geographic latitude and longitude coordinate system Using the refined geometric orientation parameters, the position of the current image point on the original stereo remote sensing image is calculated. ; Using bilinear interpolation, the position of the current image point on the original stereo remote sensing image is determined. The grayscale value of the current image point is resampled from the original stereo remote sensing image, and the grayscale value is assigned to the current image point to complete the resampling of the current image point.
9. A satellite stereo remote sensing orthophoto epipolar image generation and 3D reconstruction system, characterized in that, include: The orientation parameter acquisition module is used to acquire the original stereo remote sensing image and preprocess the original stereo remote sensing image to obtain the refined geometric orientation parameters corresponding to the original stereo remote sensing image; the original stereo remote sensing image is a stereo image pair including a left original image and a right original image; the refined geometric orientation parameters include the refined geometric orientation parameters of the left original image and the refined geometric orientation parameters of the right original image. The transformation parameter calculation module is used to calculate the transformation parameters between the epipolar image projection space rectangular coordinate system and the map projection coordinate system based on the refined geometric orientation parameters corresponding to the original stereo remote sensing image, and to construct the epipolar image projection space rectangular coordinate system; the transformation parameters include the average elevation value, baseline orientation angle, and coordinates of the center point of the overlapping area on the original stereo remote sensing image in the map projection coordinate system of the left and right original images; The data conversion module is used to acquire DEM data and convert the DEM data to the rectangular coordinate system of the epipolar image projection space to obtain DEM data in the rectangular coordinate system of the epipolar image projection space. The image generation module is used to perform epipolar image orthorectification on the original stereo remote sensing image using DEM data in the rectangular coordinate system of the epipolar image projection space, and generate an orthorectified epipolar image; when the original stereo remote sensing image is a left original image, the orthorectified epipolar image is a left orthorectified epipolar image; when the original stereo image is a right original image, the orthorectified epipolar image is a right orthorectified epipolar image. The disparity map acquisition module is used to perform dense matching on the orthophoto epipolar image to obtain a disparity map. The disparity map includes a number of image points and the disparity values of the image points, with each image point corresponding to a pair of corresponding points. The dense matching is a dense matching of the left and right orthophoto epipolar images along the epipolar direction. The corresponding points are a pair of image points on the left and right orthophoto epipolar images, respectively, representing the same point in the geographic latitude and longitude coordinate system. The original stereo image image point coordinate calculation module is used to calculate the image point coordinates of each corresponding point on the original stereo remote sensing image based on the conversion parameters between the rectangular coordinate system of the epipolar image projection space and the map projection coordinate system, for any image point and the disparity value of the image point in the disparity map. The dense point cloud generation module is used to perform spatial intersection of the image point coordinates on the original stereo remote sensing image corresponding to each corresponding point according to the refined orientation parameters of the original stereo remote sensing image, so as to obtain the positioning coordinates of each corresponding point in the geographic latitude and longitude coordinate system; the set of positioning coordinates obtained by the spatial intersection of all corresponding points constitutes the dense point cloud.