A DSM generation method and system based on long-line-array large-width bidirectional swing-scanning images

Abstract

The present invention provides a DSM generation method based on a long linear array and wide swath bidirectional swing-scanning image pair, which relates to the technical field of remote sensing mapping. Aiming at the complex imaging relationship of the linear array bidirectional swing-scanning, first, the rational function RFM model is used to describe the imaging geometric relationship of the swing-scanning image, and the relative spatial position of the forward and backward scan image pairs at the moment of photography is restored according to the RPC parameters to complete relative orientation; subsequently, the local orientation of the image pair at the imaging moment is determined based on ground control points to complete the absolute orientation of the stereo model; then, the approximate epipolar image is determined by the projection trajectory method, and the corresponding points are obtained through one-dimensional correlation matching; finally, the geodetic coordinates of the target points are obtained according to the RFM model to complete the DSM generation. By using the approximate epipolar image for one-dimensional correlation matching, the matching cost of the image pair is reduced, and the problem of DSM generation for the bidirectional swing-scanning image pair is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of remote sensing mapping. It mainly uses the forward and reverse scanned images obtained by a bidirectional swing-scanning imaging system to generate a digital elevation model, obtain the undulating shape and texture features of the regional ground, which helps to update large and medium-scale topographic maps and assist satellite data processing. Background Art

[0002] A digital surface model (DSM) is an entity model that describes ground elevation in the form of three-dimensional numerical point clouds and is widely used in national defense construction such as surveying and mapping, hydrology, and geomorphology. At present, the commonly used means for obtaining DSM mainly include: field measurement, laser method, digital photogrammetry method, etc. Among them, field measurement relies on high labor costs and scattered measurement areas; the laser method has all-weather active ranging and high accuracy, but it is time-consuming and laborious; the digital photogrammetry method uses aerial / space stereo image pairs and observes and measures through computer simulation of binocular vision of the human eye, with advantages such as low cost, short production cycle, large data volume, and high efficiency.

[0003] In recent years, space remote sensing imaging technology has been continuously developing, and the image resolution, swath width, and base-height ratio have been greatly improved. At the same time, data acquisition is stable and the cycle is short, providing important support for the rapid generation and real-time update of DSM. Satellite imaging systems mainly include two categories: swing-scanning type and push-broom type. For example, Figure 1 , the push-broom linear array is placed perpendicular to the track, and continuous image data is collected by relying on satellite flight; for example, Figure 2 , the swing-scanning linear array is placed along the track, and image acquisition is completed by cooperating with the uniform scanning of the reflector. Both imaging methods can achieve stereoscopic imaging within a short time interval, but the introduction of the swing mirror increases the imaging swath width and also makes the imaging relationship more complex. Many scholars at home and abroad have conducted research on DSM generation based on high-resolution satellite data such as ASTER, SPOT, QuickBird, and IKONOS. However, most of these studies are based on push-broom image pairs. Although Landsat1-7, SDGSAT-1 and other swing-scanning satellites can theoretically also achieve DSM generation by obtaining image pairs through the reciprocating scanning of the reflector, there are few related studies.

[0004] In summary, in view of the future demand for wide-swath and high-precision mapping, it is urgent to study a DSM generation method and system based on bidirectional swing-scanning remote sensing image pairs to meet the actual application requirements. Summary of the Invention

[0005] In view of this, the present invention provides a DSM generation method based on long-line-array wide-swath bidirectional swing-scanning images. To achieve the above object, the present invention is implemented by the following steps:

[0006] Step 1: Perform relative orientation of the forward and reverse scanned image pairs of the swing-scanning images;

[0007] The rational function RFM model is used to describe the geometric relationship of the swing scanning image, and the image space coordinate system S of the normal scanning image is selected. 1 -X 1 Y 1 Z 1 As an auxiliary coordinate system of image space, the forward scan film is fixed, and the relative spatial position of the forward and reverse scan image pairs at the moment of photography is restored according to the sensor / RPC parameters to complete the relative orientation;

[0008] Step 2: Determine the position of the image at the time of imaging based on the ground control points to complete the absolute orientation of the stereo model;

[0009] Relative orientation generates image space auxiliary coordinate system S 1 -X 1 Y 1 Z 1 The three-dimensional model under the condition needs to be transformed into the corresponding geodetic coordinate system S by three-dimensional space similarity transformation. tp -X tp Y tp Z tp The actual ground model is downloaded, and the 7 absolute orientation parameters including scaling, rotation, and translation are solved using the ground control points (GCPs) to determine the actual object space position of the 3D model, thus completing the absolute orientation of the 3D model.

[0010] Step 3: Use the projection trajectory method to determine the approximate epipolar image pair;

[0011] Under the premise that the epipolar line is a straight line in the local range of the linear array image, the epipolar line model based on the projection trajectory method is used to rearrange the epipolar line of the image pair model after absolute orientation to generate an approximate epipolar line image pair;

[0012] Step 4: Obtain the geodetic coordinates of the target point according to the RFM model and complete DSM generation;

[0013] The forward and reverse scan approximate epipolar image pair eliminates the parallax in the vertical track direction. According to the maximum elevation of the survey area, the maximum parallax in the along-track direction is calculated. The maximum parallax value is added or subtracted from the pixel coordinates of the forward scan epipolar image to obtain the reverse scan epipolar image search range. The one-dimensional search obtains the coordinates of the same-name points of the approximate epipolar image pair, and then the grid mapping is calculated and converted to the coordinates of the original forward and reverse scan image pair, and brought into the RFMf 1 The coordinates of the ground points with the same name are obtained by mapping, and the elevation of the target point is used instead of the image grayscale output to obtain the DSM.

[0014] Optionally, the rational function RFM model in step 1 is

[0015] Recorded as

[0016] Its inverse form is expressed as (x, y) is the image point coordinates, (B, L, H) are the object point coordinates in the geodetic coordinate system, latitude B, longitude L, altitude H, Num(B, L, H), and Den(B, L, H) are all cubic polynomials, and the general formula is F(B, L, H) = (a 1 + a 2 L + a 3 P + a 4 H + a 5 LP + … + a 20 H 3 ), and the polynomial coefficients are RPC parameters.

[0017] Optionally, the three-dimensional space similarity transformation formula described in step two is

[0018]

[0019] where, [B tp L tp H tp T is the geodetic coordinate of the ground point corresponding to the target point, [B L H] T is the coordinate of the target point in the auxiliary image space coordinate system S 1 -X 1 Y 1 Z 1 coordinates, λ is the model scaling factor, a 1 , a 2 , a 3 , b 1 , b 2 , b 3 , c 1 , c 2 , c 3 are the direction cosines corresponding to the three rotation angles (α, β, γ) of the two coordinate axes, [ΔB ΔL ΔH] T is the translation amount of the coordinate system origin.

[0020] Optionally, the specific process of obtaining the homologous ground point coordinates described in step four is:

[0021] First, select a point P in the forward-scanned image I m 0 and form an imaging ray with S 1 (X S1 , Y S1 , Z S1 ). Select any ten points M i (i = 1, 2…10) on this ray. According to the elevation range of the survey area, assign different elevation values H i to M i . Map and calculate M from the RFM f 2 of the backward-scanned image​i On I N The corresponding image point m i '(i = 1, 2…10) coordinates

[0022] Secondly, within a local range, according to The conjugate line on the reverse-scanned image is fitted by least squares for m 0 ; Then, several points are selected on the conjugate line of the reverse-scanned image, and the conjugate lines corresponding to several points on the forward-scanned image are fitted according to the above steps;

[0023] Finally, resampling is performed along the conjugate line direction to make the image scan column coincide with the conjugate line direction, generating a pair of forward and reverse-scanned approximate conjugate line images, and defining the coordinate transformation between the original forward and reverse-scanned image pairs (I P / I N ) and the approximate conjugate line image pair (I P ' / I N '), and substituting into RFMf 1 The mapping obtains the coordinates of homologous ground points.

[0024] Optionally, the formula for the grid mapping is

[0025]

[0026] where (x, y) are the coordinates of the original scanned image, (u, v) are the corresponding coordinates of the approximate conjugate line image, A g is the grid affine matrix, t g is the translation matrix, which is obtained by fitting the grid vertices by least squares.

[0027] On the other hand, a DSM generation system based on a long-line array and wide-width bidirectional swing-scanned image is provided, including the following modules:

[0028] Relative orientation module, performing relative orientation on the swing-scanned image pair;

[0029] Using the rational function RFM model to describe the imaging geometric relationship of the swing-scanned image, and restoring the relative spatial position of the forward and reverse-scanned image pairs at the moment of photography according to the RPC parameters to complete the relative orientation;

[0030] Absolute orientation module, determining the local orientation of the image pair at the imaging moment based on ground control points to complete the absolute orientation of the stereo model;

[0031] The relative orientation generates a stereo model in the image space auxiliary coordinate system S 1 -X 1 Y 1 Z 1 The stereo model needs to be transformed by a three-dimensional spatial similarity transformation to obtain the corresponding geodetic coordinate system S tp -X tp Ytp Z tp The actual ground model is downloaded, and the 7 absolute orientation parameters including scaling, rotation, and translation are solved using the ground control points (GCPs) to determine the actual object space position of the 3D model, thus completing the absolute orientation of the 3D model.

[0032] The module for determining the approximate epipolar image pair uses the projection trajectory method to determine the approximate epipolar image pair;

[0033] Under the premise that the epipolar line is a straight line in the local range of the linear array image, the epipolar line model based on the projection trajectory method is used to rearrange the epipolar line of the image pair model after absolute orientation to generate an approximate epipolar line image pair;

[0034] The DSM generation module obtains the geodetic coordinates of the target point according to the RFM model and completes the DSM generation;

[0035] The forward and reverse scan approximate epipolar image pair eliminates the parallax in the vertical track direction. According to the maximum elevation of the survey area, the maximum parallax in the along-track direction is calculated. The maximum parallax value is added or subtracted from the pixel coordinates of the forward scan epipolar image to obtain the reverse scan epipolar image search range. The one-dimensional search obtains the coordinates of the same-name points of the approximate epipolar image pair, and then the grid mapping is calculated and converted to the coordinates of the original forward and reverse scan image pair, and brought into the RFMf 1 The coordinates of the ground points with the same name are obtained by mapping, and the elevation of the target point is used instead of the image grayscale output to obtain the DSM.

[0036] Optionally, an object-image coordinate association module is also included, which associates the object-image coordinates using a universal sensor RFM model of a ratio polynomial.

[0037] Optionally, a three-dimensional space similarity transformation module is also included to generate an image space auxiliary coordinate system S 1 -X 1 Y 1 Z 1 The three-dimensional model under the three-dimensional space similarity transformation is used to obtain the corresponding geodetic coordinate system S tp -X tp Y tp Z tp The actual ground model.

[0038] Optionally, a ground point coordinate acquisition module with the same name is also included, which is used to acquire the coordinates of the ground point with the same name.

[0039] Optionally, a grid mapping module is also included to define the original forward and reverse scan image pairs (I P / I N ) and the approximate epipolar image pair (I P ' / I N ') coordinate conversion.

[0040] As can be seen from the above technical solutions, compared with the prior art, the present invention provides a method and system for generating DSM of a long-line array and wide-swath bidirectional swing-scanning image pair. The DSM is generated through steps such as relative orientation of image pairs, absolute orientation, and approximate epipolar images, aiming to reduce the matching cost of image pairs by performing one-dimensional correlation matching on approximate epipolar images and solve the problem of generating DSM for bidirectional swing-scanning image pairs. The present invention can ultimately achieve the generation of DSM for long-line array and wide-swath bidirectional swing-scanning image pairs. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on the provided drawings.

[0042] Figure 1 It is a schematic diagram of linear array pushbroom imaging;

[0043] Figure 2 It is a schematic diagram of linear array swing-scanning imaging;

[0044] Figure 3 It is a flow chart of the method for obtaining DSM from a stereo image pair;

[0045] Figure 4 It is a schematic diagram of relative orientation of forward and reverse scan images;

[0046] Figure 5 It is a schematic diagram of absolute orientation of a stereo model;

[0047] Figure 6 It is a schematic diagram of the generation of approximate epipolar image pairs;

[0048] Figure 7 It is a schematic diagram of the epipolar geometric relationship of forward and reverse swing-scanning image pairs. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0050] Embodiment 1 of the present invention discloses a method for generating a DSM based on a long linear array and wide swath bidirectional swing scanning image pair. Aiming at the complex imaging relationship of the linear array bidirectional swing scanning, first, the rational function RFM model is used to describe the imaging geometric relationship of the swing scanning image, and the relative spatial position of the forward and reverse scan image pairs at the moment of photography is restored according to the RPC parameters to complete relative orientation; then, based on the ground control points, the local orientation of the image pair at the imaging moment is determined to complete the absolute orientation of the stereo model; then, the approximate epipolar image is determined by the projection trajectory method, and the corresponding points are obtained through one-dimensional correlation matching; finally, the geodetic coordinates of the target points are obtained according to the RFM model to complete the DSM generation. The method flow is as Figure 3 . The present invention mainly includes the following steps:

[0051] Step 1: Relative orientation of the forward and reverse scan image pairs

[0052] The swing scanning imaging follows the central projection along the track direction and the parallel projection relationship along the cross-track direction. The positioning model parameters are numerous and highly correlated. At the same time, according to the sensor information confidentiality requirements, here the general sensor RFM model based on the ratio polynomial is used to correlate the object and image coordinates. To enhance the parameter stability, the coordinates are regularized to [-1, 1].

[0053] Furthermore, the rational function RFM model is

[0054] Denoted as

[0055] Its inverse calculation form is denoted as (x, y) are the image point coordinates, (B, L, H) are the object point coordinates in the geodetic coordinate system, latitude B, longitude L, elevation H, Num(B, L, H), and Den(B, L, H) are all cubic polynomials. The general formula is F(B, L, H) = (a 1 + a 2 L + a 3 P + a 4 H + a 5 LP + … + a 20 H 3 ). The polynomial coefficients are the RPC parameters.

[0056] Such as Figure 4 , S 1 , S 2 are the photography centers of the forward and reverse scan images at the imaging moment. The image space coordinate system S P of the forward scan image I 1 -X 1 Y 1 Z 1 is selected as the image space auxiliary coordinate system, and the forward scan image I P is fixed. By adjusting the 5 orientation elements of the reverse scan image I N ​ Make the homologous rays of the image pair intersect pairwise, restore the relative spatial position of the forward and backward scanned images, and complete the relative orientation of the continuous image pair, where B X , B Z is the projection component of the photographic baseline B on the X and Z coordinate axes of the auxiliary coordinate system in the image space. ω, κ are the rotation components of the backward scanned image I N around the three coordinate axes.

[0057] Step 2: Absolute orientation of the stereo model

[0058] As Figure 5 , the relative orientation generates a stereo model in the auxiliary coordinate system S 1 -X 1 Y 1 Z 1 needs to be transformed by a three-dimensional spatial similarity transformation to obtain the corresponding actual ground model in the geodetic coordinate system S tp -X tp Y tp Z tp . Use the ground control points GCPs to solve 7 absolute orientation parameters including scaling, rotation, and translation to determine the actual object space position of the stereo model and complete the absolute orientation of the stereo model..

[0059] Furthermore, the three-dimensional spatial similarity transformation formula is

[0060]

[0061] where, [B tp L tp H tp T is the geodetic coordinate of the ground point corresponding to the target point, [B L H] T is the coordinate of the target point in the auxiliary coordinate system S 1 -X 1 Y 1 Z 1 , λ is the model scaling factor, a 1 , a 2 , a 3 , b 1 , b 2 , b 3 , c 1 , c 2 , c 3 are the direction cosines corresponding to the 3 rotation angles (α, β, γ) of the two coordinate axes, [ΔB ΔL ΔH] T is the translation amount of the coordinate system origin.

[0062] Step 3: Determine the approximate epipolar lines and generate an epipolar image pair

[0063] ​For the long linear array bidirectional swing-scanning image pair of multi-center projection, there is no unique baseline and epipolar plane, making it difficult to establish a strict epipolar line model. Assuming that the epipolar lines are straight within the local range of the linear array images, an epipolar line model based on the projection trajectory method is used to perform epipolar line rearrangement on the image pair model after absolute orientation to generate an approximate epipolar line image pair.

[0064] As Figure 6 shown, the imaging light ray from a certain ground point M(B, L, H) passes through S 1 (X S1 , Y S1 , Z S1 ) and forms an image at point m on the forward-scanning image I P . Then, each point on the light ray passes through S 2 (X S2 , Y S2 , Z S2 ) and projects onto the backward-scanning image I N . The projection point trajectory forms the epipolar line of image point m. The corresponding point m' of m on I N must be located on this epipolar line. In addition, the epipolar curves on the linear array images can be treated as straight lines within a small range, and the corresponding epipolar line pairs within the local range are also one-to-one.

[0065] Step 4: DSM generation

[0066] The forward and backward-scanning approximate epipolar line image pair almost eliminates the parallax in the across-track direction. According to the maximum elevation of the survey area, calculate the maximum parallax in the along-track direction. Add and subtract the maximum parallax value on the pixel coordinates of the forward-scanning epipolar line image to obtain the search range of the backward-scanning epipolar line image. One-dimensional search is used to obtain the coordinates of the corresponding points of the approximate epipolar line image pair, and then calculate the grid mapping and convert it to the coordinates of the original forward and backward-scanning image pair. Substitute into the RFMf 1 mapping to obtain the coordinates of the corresponding ground points, and output the DSM by replacing the image gray value with the elevation of the target point.

[0067] Furthermore, as Figure 7 shown, first select a point P m on the forward-scanning image I 0 and form an imaging light ray with S 1 (X S1 , Y S1 , Z S1 ). Select any ten points M i (i = 1, 2…10) on this light ray. According to the elevation range of the survey area, assign different elevation values H i to M i . Calculate the coordinates of the corresponding image points m 2 ' (i = 1, 2…10) of M i on I N through the mapping calculation of the RFMf i of the backward-scanning image. Secondly, within a local range, according to the least squares method, the epipolar lines on the reverse-scanned image for m 0 are fitted; then, several points are selected on the epipolar lines of the reverse-scanned image, and the corresponding epipolar lines on the forward-scanned image for several points are fitted according to the foregoing steps; finally, resampling is performed along the epipolar line direction to make the image scan columns coincide with the epipolar line direction, generating a pair of forward and reverse-scanned approximate epipolar line images. Here, the coordinate transformation between the original pair of forward and reverse-scanned images (I P / I N ) and the pair of approximate epipolar line images (I P ' / I N ) is defined through grid mapping.

[0068] Furthermore, the formula for grid mapping is

[0069]

[0070] where (x, y) are the coordinates of the original scanned image, (u, v) are the corresponding coordinates of the approximate epipolar line image, A g is the grid affine matrix, and t g is the translation matrix, which is obtained by fitting the grid vertices through the least squares method.

[0071] Embodiment 2 of the present invention discloses a DSM generation system based on a pair of long-line-array and wide-width bidirectional swing-scanned images, including the following modules:

[0072] A relative orientation module for performing relative orientation on the pair of forward and reverse-scanned swing-scanned images;

[0073] The imaging geometric relationship of the swing-scanned image is described by using the rational function RFM model, and the relative spatial position of the pair of forward and reverse-scanned images at the moment of photography is restored according to the RPC parameters to complete the relative orientation;

[0074] An absolute orientation module for determining the local orientation of the image pair at the imaging moment based on ground control points to complete the absolute orientation of the stereoscopic model;

[0075] The relative orientation generates a stereoscopic model in the image space auxiliary coordinate system S 1 -X 1 Y 1 Z 1 which needs to be transformed by a three-dimensional space similarity transformation to obtain the corresponding actual ground model in the geodetic coordinate system S tp -X tp Y tp Z tp The actual object space position of the stereoscopic model is determined by using the ground control points GCPs to solve 7 absolute orientation parameters including scaling, rotation, and translation to complete the absolute orientation of the stereoscopic model;

[0076] The module for determining the approximate epipolar image pair uses the projection trajectory method to determine the approximate epipolar image pair;

[0077] On the premise that the epipolar line is a straight line in the local range of the linear array image, the epipolar line model based on the projection trajectory method is used to rearrange the epipolar line of the image pair model after absolute orientation to generate an approximate epipolar line image pair;

[0078] The DSM generation module obtains the geodetic coordinates of the target point according to the RFM model and completes the DSM generation;

[0079] The forward and reverse scan approximate epipolar image pair eliminates the parallax in the vertical track direction. According to the maximum elevation of the survey area, the maximum parallax in the along-track direction is calculated. The maximum parallax value is added or subtracted from the pixel coordinates of the forward scan epipolar image to obtain the reverse scan epipolar image search range. The one-dimensional search obtains the coordinates of the same-name points of the approximate epipolar image pair, and then the grid mapping is calculated and converted to the coordinates of the original forward and reverse scan image pair, and brought into the RFMf 1 The coordinates of the ground points with the same name are obtained by mapping, and the elevation of the target point is used instead of the image grayscale output to obtain the DSM.

[0080] Furthermore, it also includes an object-image coordinate association module, which associates the object-image coordinates by using a universal sensor RFM model of a ratio polynomial.

[0081] Furthermore, it also includes a three-dimensional space similarity transformation module for generating an image space auxiliary coordinate system S 1 -X 1 Y 1 Z 1 The three-dimensional model under the three-dimensional space similarity transformation is used to obtain the corresponding geodetic coordinate system S tp -X tp Y tp Z tp The actual ground model.

[0082] Furthermore, it also includes a ground point coordinate acquisition module for acquiring the coordinates of the ground point with the same name.

[0083] Furthermore, a grid mapping module is included to define the original forward and reverse scan image pairs (I P / I N ) and the approximate epipolar image pair (I P ' / I N ') coordinate conversion.

[0084] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0085] The foregoing description of the disclosed embodiments enables those skilled in the art to practice or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for generating DSM based on long-line-array wide-swath bidirectional swing-scanning images, Characterized in that, This method is implemented by the following steps: Step 1: Perform relative orientation on the forward and reverse scan image pairs of the swing-scanning images; The rational function RFM model is used to describe the imaging geometric relationship of push-broom images, and the image space coordinate system S of the forward-scanning image is selected. 1 -X 1 Y 1 Z 1 As the auxiliary image space coordinate system, the forward-scanning image is fixed, and the relative spatial position of the forward and backward-scanning image pairs at the moment of photography is restored according to the sensor / RPC parameters to complete relative orientation. Step 2: Determine the local orientation of the image pair at the imaging moment based on ground control points to complete the absolute orientation of the stereo model; Relative orientation generates the image space auxiliary coordinate system S 1 -X 1 Y 1 Z 1 The stereo model under needs to be obtained through a three-dimensional space similarity transformation to the corresponding geodetic coordinate system S tp -X tp Y tp Z tp The actual ground model under, uses the ground control points GCPs to solve 7 absolute orientation parameters including scaling, rotation, and translation to determine the actual object space position of the stereo model, and completes the absolute orientation of the stereo model; Step 3: Use the projection trajectory method to determine the approximate epipolar image pair; On the premise that the epipolar line is a straight line within the local range of the line array image, use the epipolar line model based on the projection trajectory method to perform epipolar reordering on the image pair model after absolute orientation to generate an approximate epipolar image pair; Step 4: Obtain the geodetic coordinates of the target points according to the RFM model to complete the generation of DSM; The forward and reverse scan approximate epipolar image pair eliminates the parallax in the vertical track direction. According to the maximum elevation of the survey area, the maximum parallax in the along-track direction is calculated. The maximum parallax value is added or subtracted from the pixel coordinates of the forward scan epipolar image to obtain the reverse scan epipolar image search range. The one-dimensional search obtains the coordinates of the same-name points of the approximate epipolar image pair, and then the grid mapping is calculated and converted to the coordinates of the original forward and reverse scan image pair, and brought into the RFMf 1 The coordinates of the ground points with the same name are obtained by mapping, and the elevation of the target point is used instead of the image grayscale output to obtain the DSM.

2. A method for generating DSM based on long-line-array wide-swath bidirectional swing-scanning images according to claim 1, Characterized in that, The rational function RFM model described in step 1 is Denoted as Its inverse calculation form is denoted as (x, y) are the coordinates of the image point, (B, L, H) are the coordinates of the object point in the geodetic coordinate system, latitude B, longitude L, altitude H, Num(B, L, H), and Den(B, L, H) are all cubic polynomials, and the general formula is F(B, L, H) = (a 1 + a 2 L + a 3 P + a 4 H + a 5 LP + … + a 20 H 3 ), and the polynomial coefficients are RPC parameters.

3. A method for generating DSM based on long-line-array wide-swath bidirectional swing-scanning images according to claim 1, Characterized in that, In step 2, the three-dimensional space similarity transformation formula is Among them, [B tp L tp H tp T is the geodetic coordinate of the ground point corresponding to the target point, [B L H] T is the coordinate of the target point in the auxiliary image space coordinate system S 1 -X 1 Y 1 Z 1 coordinate, λ is the model scaling factor, a 1 , a 2 , a 3 , b 1 , b 2 , b 3 , c 1 , c 2 , c 3 are the direction cosines corresponding to the three rotations (α, β, γ) of the two coordinate axes, [ΔB ΔL ΔH] T is the translation amount of the coordinate system origin.​ 4. A method for generating DSM based on long-line-array wide-swath bidirectional swing-scanning images according to claim 1, Characterized in that, The specific process of obtaining the coordinates of corresponding ground points in step 4 is: First, in the forward-scanned image I P Select a point m 0 and S 1 (X S1 , Y S1 , Z S1 ) to form an imaging ray. Arbitrarily select ten points M i (i = 1, 2... 10) on this ray. According to the elevation range of the surveyed area, assign different elevation values H i to M i . Calculate the corresponding image point m 2 ' (i = 1, 2... 10) coordinates of M i on I N by mapping with the RFM f i of the backward-scanned image Secondly, within a local range, according to the least squares method, fit the epipolar lines of m on the reverse-scanned image 0 ; Then, select several points on the epipolar line of the reverse scan image, and fit the epipolar lines corresponding to several points on the forward scan image according to the previous steps; Finally, resampling is performed along the epipolar line direction to make the image scan columns coincide with the epipolar line direction, generating a pair of approximately epipolar images with forward and reverse scans. And through grid mapping, the coordinate transformation between the original forward and reverse scan image pair (I P / I N ) and the approximately epipolar image pair (I P ' / I N ) is defined, and the RFMf 1 is substituted to map and obtain the coordinates of corresponding ground points.

5. A method for generating DSM based on long-line-array wide-swath bidirectional swing-scanning images according to claim 4, Characterized in that, The formula for grid mapping is Among them, (x, y) are the coordinates of the original scanned image, (u, v) are the corresponding coordinates of the approximate epipolar image, and A g is the grid affine matrix, and t g is the translation matrix, which is obtained by fitting the grid vertices through least squares.

6. A system for generating DSM based on long-line-array wide-swath bidirectional swing-scanning images, Characterized in that, It includes: A relative orientation module that performs relative orientation on the forward and reverse scan image pairs of the swing-scanning images; Use the rational function RFM model to describe the imaging geometric relationship of the swing-scanning images, and restore the relative spatial position of the forward and reverse scan image pairs at the moment of photography according to the RPC parameters to complete the relative orientation; An absolute orientation module that determines the local orientation of the image pair at the imaging moment based on ground control points to complete the absolute orientation of the stereo model; Relative orientation generates the image space auxiliary coordinate system S 1 -X 1 Y 1 Z 1 The stereoscopic model under needs to be obtained through a three-dimensional space similarity transformation to the corresponding geodetic coordinate system S tp -X tp Y tp Z tp The actual ground model below. Use the ground control points GCPs to solve 7 absolute orientation parameters including scaling, rotation, and translation to determine the actual object space position of the stereoscopic model and complete the absolute orientation of the stereoscopic model; An approximate epipolar image pair determination module that uses the projection trajectory method to determine the approximate epipolar image pair; On the premise that the epipolar line is a straight line within the local range of the line array image, use the epipolar line model based on the projection trajectory method to perform epipolar reordering on the image pair model after absolute orientation to generate an approximate epipolar image pair; A DSM generation module that obtains the geodetic coordinates of the target points according to the RFM model to complete the generation of DSM; The forward and reverse scan approximate epipolar image pair eliminates the parallax in the vertical track direction. According to the maximum elevation of the survey area, the maximum parallax in the along-track direction is calculated. The maximum parallax value is added or subtracted from the pixel coordinates of the forward scan epipolar image to obtain the reverse scan epipolar image search range. The one-dimensional search obtains the coordinates of the same-name points of the approximate epipolar image pair, and then the grid mapping is calculated and converted to the coordinates of the original forward and reverse scan image pair, and brought into the RFMf 1 The coordinates of the ground points with the same name are obtained by mapping, and the elevation of the target point is used instead of the image grayscale output to obtain the DSM.

7. A system for generating DSM based on long-line-array wide-swath bidirectional swing-scanning images according to claim 6, Characterized in that, It further includes an object-image coordinate association module that associates the object-image coordinates using the general sensor RFM model of the ratio polynomial.

8. A system for generating DSM based on long-line-array wide-swath bidirectional swing-scanning images according to claim 6, Characterized in that, It further includes a three-dimensional space similarity transformation module, which is used to perform a three-dimensional space similarity transformation on the stereo model under the image space auxiliary coordinate system S 1 -X 1 Y 1 Z 1 to obtain the corresponding actual ground model under the geodetic coordinate system S tp -X tp Y tp Z tp below.

9. A system for generating DSM based on long-line-array wide-swath bidirectional swing-scanning images according to claim 6, Characterized in that, It further includes a module for obtaining the coordinates of corresponding ground points, which is used to obtain the coordinates of corresponding ground points.

10. A system for generating DSM based on long-line-array wide-swath bidirectional swing-scanning images according to claim 9, It is characterized in that It further includes a grid mapping module for defining the coordinate transformation between the original forward and backward scan image pair (I P / I N ) and the approximate epipolar image pair (I P ' / I N ').

Images