A method and device for regional network adjustment based on multi-source optical remote sensing satellite images

Abstract

A kind of method based on multi-source optical remote sensing satellite image block adjustment, belong to remote sensing image processing field, it is characterized by including: first, multi-source remote sensing satellite data is organized and connection point file is obtained;Weak intersection connection point is identified, and gross error detection and gross error elimination are carried out to connection point;According to the connection relationship between images obtained from connection point, parameters are corrected;The elevation value of weak intersection point is interpolated using external DEM, until the elevation value of weak intersection point is not changed twice;The accuracy of multi-source remote sensing image block adjustment is counted, and the block adjustment result is output.The present application realizes the joint processing of multi-source image, realizes the robustness of weak intersection image processing with the aid of external DEM data, ensures the compliance of multi-source data inner accuracy, improves the edge precision between multi-source data;Matrix compression storage mode is used to construct the coefficient matrix of equation, reduces the consumption of calculation memory, improves the processing efficiency of block adjustment.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing image processing, and in particular relates to a method and apparatus for regional network adjustment based on multi-source optical remote sensing satellite imagery. Background Technology

[0002] With the increasing variety of remote sensing satellite sensor types, the continuous improvement of the satellite multi-source heterogeneous space observation system, and the continuous enhancement of remote sensing data acquisition capabilities across multiple platforms and sensor modes, the processing of remote sensing satellite data also urgently needs to develop towards multi-source methods. At the same time, with the gradual increase in the area of ​​surveying and mapping work and the increasing complexity and diversity of terrain, it is difficult to achieve full coverage of the survey area imagery by relying on a single remote sensing satellite data. Therefore, it is essential to introduce remote sensing satellite imagery from different sources to complement each other and achieve joint processing of remote sensing satellite imagery.

[0003] Joint adjustment of multi-source data not only ensures the accuracy of edge matching between heterogeneous data sources but also improves the overall stability of the regional network. Utilizing joint regional adjustment techniques based on multi-source satellite imagery, even with very few or no control points, overcomes extremely poor image texture features, achieving higher accuracy and faster positioning over a wider coverage area, which has significant practical implications.

[0004] Traditional regional network adjustment of satellite remote sensing images is based on the principle of ray intersection where pairs of corresponding rays intersect. However, for multi-source satellite remote sensing images, some of these images are not stereo satellites but exist in a weak intersection manner. This can lead to non-convergence of adjustment methods using traditional regional network adjustment methods based on rational function models, resulting in low robustness and poor accuracy of the adjustment algorithm. Summary of the Invention

[0005] This invention aims to solve the above-mentioned problems by providing a joint regional network adjustment method and apparatus based on multi-source optical remote sensing satellite imagery, in order to address the issue that some images in remote sensing regional network adjustment have weak intersections, resulting in low robustness and poor accuracy of regional network adjustment.

[0006] In a first aspect, the present invention provides a method for regional network adjustment based on multi-source optical remote sensing satellite imagery, comprising:

[0007] 1) First, organize the multi-source remote sensing satellite data and obtain the DEM reference data of the corresponding survey area, and extract the tie points from the multi-source remote sensing data to obtain the tie point file;

[0008] 2) Calculate the intersection angle of each connection point using the RPC and connection point coordinates of the remote sensing image, identify weak intersection connection points, obtain the initial object space coordinates of weak intersection connection points using external DEM data, and perform gross error detection and gross error removal on the connection points.

[0009] 3) Obtain the connection relationship between images based on the connection points, and construct the relationship correspondence table of the normal equation coefficient matrix blocks; combine parallel computing to compress, fill and solve the normal equation, obtain the corrections for affine distortion, connection points and weak intersection points, and correct the parameters; calculate the image point residual mean square error, and remove gross errors for connection points and weak intersection points;

[0010] 4) Repeat step 3) until the image point residual mean error remains constant;

[0011] 5) Re-interpolate the elevation values ​​of the weak intersection points using the external DEM, and repeat step 3) until the elevation values ​​of the weak intersection points no longer change after two interpolations.

[0012] 6) Calculate the accuracy of regional network adjustment of multi-source remote sensing images, fit the RPC parameters, and output the regional network adjustment results.

[0013] Furthermore, in the method for regional network adjustment based on multi-source optical remote sensing satellite imagery described in this invention, step 2) of detecting and removing gross errors at connection points includes:

[0014] By combining the RPC parameters of the image, the intersection angle of each connection point is calculated, weak intersection connection points are identified, and the initial object space coordinates of the weak intersection connection points are obtained by combining external DEM data. The image space residual of each weak intersection connection point is calculated, and gross errors are detected and eliminated.

[0015] Using the image's RPC, forward intersection is performed on the connection points to obtain their initial object-space coordinates. The image-space residuals of each connection point are then calculated, and gross errors are detected and eliminated.

[0016] Furthermore, in the method for regional network adjustment based on multi-source optical remote sensing satellite imagery described in this invention, when obtaining the initial object space coordinates of weak intersection connection points, for weak intersection connection points with two or more degrees, the corresponding three-dimensional coordinates of the connection points on each image are obtained using the RPC of each image, and the average value is taken as the initial value of the object space coordinates of that point.

[0017] Furthermore, in the method for adjusting regional networks based on multi-source optical remote sensing satellite imagery described in this invention, the gross error detection and removal process first involves performing forward intersection on the connection points to obtain their object-space coordinates. Then, using single-image resection, the image point residuals of the connection points on the image are calculated, and the residual mean square error is statistically analyzed. Points with a mean square error greater than a preset value are identified as gross error points and are deleted.

[0018] The connection points are divided into two parts: the part inside the stereo image pair and the part between the stereo image pairs. Since the part between the image pairs is captured simultaneously during satellite imaging, the internal distortion and timeliness are consistent, so the connection point matching error is small. However, the part between the image pairs has a larger timeliness and distortion due to the difference in the shooting time, so the connection point error is also larger. Therefore, different gross error removal thresholds are set for these two parts of the points to remove gross errors.

[0019] Furthermore, in the method for regional network adjustment based on multi-source optical remote sensing satellite imagery described in this invention, step 3) specifically includes the following processes:

[0020] 31) Number the images, obtain the pairwise connection relationships between the images based on the connection points and image numbers, and mark them;

[0021] 32) Parallel computing is used to construct the coefficient matrix block of the normal equation in parallel using the connection points and weak intersection points, and the image number where the connection points are located is marked.

[0022] 33) Fill the coefficient matrix blocks into the corresponding positions in the matrix according to the image number of the coefficient matrix blocks of the non-zero normal equation;

[0023] 34) The conjugate gradient method is used to solve the normal equations and obtain the affine distortion parameter correction for each image.

[0024] 35) Parallel computation is used to calculate the object-space corrections for the three-dimensional coordinates of the connection points and the planar coordinate corrections for the weak intersection points;

[0025] 36) Correct the parameters and calculate the mean error of the image point residuals. Based on the mean error, delete the connection points and weak intersection points whose errors are greater than the preset mean error.

[0026] 37) Iterative solution until the mean square error of the image point residual satisfies the condition.

[0027] In a second aspect, the present invention provides a system for regional network adjustment based on multi-source optical remote sensing satellite imagery, including an electrically connected weak intersection point identification module, an initial coordinate calculation module, and an adjustment solution module;

[0028] The weak intersection point identification module is used to organize multi-source remote sensing satellite data and obtain DEM reference data for the corresponding survey area, and to extract connection points from the multi-source remote sensing data to obtain connection point files.

[0029] Then, the intersection angle of each connection point is calculated using the RPC and connection point coordinates of the remote sensing image, weak intersection connection points are identified, the initial object space coordinates of the weak intersection connection points are obtained using external DEM data, and gross error detection and gross error removal are performed on the connection points.

[0030] The initial coordinate calculation module is used to obtain the connection relationship between images based on the connection points, construct the relationship correspondence table of the normal equation coefficient matrix blocks; combined with parallel computing, it compresses, fills and solves the normal equation, obtains the corrections for affine distortion, connection points, and weak intersection points, and corrects the parameters; it calculates the image point residual mean square error and removes gross errors from connection points and weak intersection points; it repeats this process until the image point residual mean square error remains unchanged.

[0031] The adjustment and calculation module is used to reuse the external DEM to interpolate the elevation values ​​of weak intersection points, and call the initial coordinate calculation module to repeat the calculation until the elevation values ​​of the weak intersection points no longer change after two interpolations; then, the accuracy of the regional network adjustment of the multi-source remote sensing image is statistically analyzed, the RPC parameters are fitted, and the regional network adjustment results are output.

[0032] Furthermore, the system for regional network adjustment based on multi-source optical remote sensing satellite imagery described in this invention includes the following steps for detecting and eliminating gross errors at connection points: calculating the intersection angle of each connection point using the RPC parameters of the imagery, identifying weak intersection connection points, obtaining the initial object-space coordinates of the weak intersection connection points using external DEM data, and calculating the image-space residuals of each weak intersection connection point to detect and eliminate gross errors; and using the RPC parameters of the imagery to perform forward intersection on the connection points, obtaining the initial object-space coordinates of the connection points, and calculating the image-space residuals of each connection point to detect and eliminate gross errors at the connection points.

[0033] Furthermore, in the system for regional network adjustment based on multi-source optical remote sensing satellite imagery described in this invention, when performing gross error detection and removal on connection points, the connection points are divided into two parts: the part inside a stereo pair and the part between stereo pairs. Different gross error removal thresholds are set for each part to remove gross errors. First, forward intersection is performed on the connection points to obtain their object-space coordinates. Then, using single-image back intersection, the image point residuals of the connection points on the image are calculated, and the residual mean square error is statistically analyzed. Points with a mean square error greater than a preset value are identified as gross error points and deleted.

[0034] Thirdly, the present invention provides an electronic device for regional network adjustment based on multi-source optical remote sensing satellite imagery, comprising a processor and a memory electrically connected to each other; the memory is used to store a computer program; when the processor executes the aforementioned computer program, it can implement the method for regional network adjustment based on multi-source optical remote sensing satellite imagery as described in the first aspect.

[0035] Fourthly, the present invention provides a computer-readable storage medium storing a computer program thereon; when the computer program is executed, it can implement the method for regional network adjustment based on multi-source optical remote sensing satellite imagery as described in the first aspect.

[0036] The method and apparatus for regional network adjustment based on multi-source optical remote sensing satellite imagery described in this invention have the following technical effects:

[0037] (1) The joint processing of multi-source images was realized. With the help of external DEM data, the robustness of weak intersection image processing was achieved, ensuring the accuracy of the multi-source data and improving the edge accuracy between multi-source data.

[0038] (2) For the processing of large-area multi-source remote sensing satellite image data, a matrix compression storage method is adopted to construct the coefficient matrix of the normal equation, thereby reducing the consumption of computing memory;

[0039] (3) The parallel filling method of normal equation coefficient matrix blocks is adopted to improve the processing efficiency of regional network adjustment. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the process of adjusting the regional network of multi-source optical remote sensing satellite imagery according to an embodiment of the present invention;

[0041] Figure 2 This is a schematic diagram of the image intersection angle according to an embodiment of the present invention;

[0042] Figure 3 This is a schematic diagram illustrating the acquisition of object coordinates of weak intersection connection points according to an embodiment of the present invention;

[0043] Figure 4 This is a schematic diagram of the regional network adjustment survey area described in an embodiment of the present invention;

[0044] Figure 5 This is a diagram showing the structure of the coefficient matrix of the normal equation as described in an embodiment of the present invention;

[0045] Figure 6 This is a schematic diagram of the sparse moment construction and calculation process described in an embodiment of the present invention;

[0046] Figure 7 This is a schematic diagram of the survey area distribution and DOM results described in the embodiments of the present invention;

[0047] Figure 8 This is a schematic diagram illustrating the edge-joining accuracy of the DOM image according to an embodiment of the present invention;

[0048] Figure 9 This is a schematic diagram showing the distribution of the survey area and control points according to an embodiment of the present invention;

[0049] Figure 10 This is a schematic diagram of the system for regional network adjustment based on multi-source optical remote sensing satellite imagery as described in an embodiment of the present invention. Detailed Implementation

[0050] The method and apparatus for regional network adjustment based on multi-source optical remote sensing satellite imagery of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0051] Example 1

[0052] This embodiment discloses a method for regional network adjustment based on multi-source optical remote sensing satellite imagery, such as... Figure 1 As shown, the specific steps include the following:

[0053] 1) First, organize the multi-source remote sensing satellite data and obtain the DEM reference data of the corresponding survey area, and extract the tie points from the multi-source remote sensing data to obtain the tie point file;

[0054] 2) Calculate the intersection angle of each connection point using the RPC and connection point coordinates of the remote sensing image, identify weak intersection connection points, obtain the initial object space coordinates of weak intersection connection points using external DEM data, and perform gross error detection and gross error removal on the connection points.

[0055] 3) Obtain the connection relationship between images based on the connection points, and construct the relationship correspondence table of the normal equation coefficient matrix blocks; combine parallel computing to compress, fill and solve the normal equation, obtain the corrections for affine distortion, connection points and weak intersection points, and correct the parameters; calculate the image point residual mean square error, and remove gross errors for connection points and weak intersection points;

[0056] 4) Repeat step 3) until the image point residual mean error remains constant;

[0057] 5) Re-interpolate the elevation values ​​of the weak intersection points using the external DEM, and repeat step 3) until the elevation values ​​of the weak intersection points no longer change after two interpolations.

[0058] 6) Calculate the accuracy of regional network adjustment of multi-source remote sensing images, fit the RPC parameters, and output the regional network adjustment results.

[0059] In this technical solution, the intersection angle refers to the angle formed by the rays connecting each point in the same category to its corresponding photographic center, such as... Figure 2 As shown, weak intersection images refer to images with a small intersection angle. Generally, images with an intersection angle less than 10° are considered weak intersection images, and the smaller the intersection angle, the greater the impact on elevation accuracy. In the figure, S and S' are the photographic center points of the stereo image, points a and a' are the connection points, Am is the ground object point, and the angle θ between the two light rays Sa and S'a' is the intersection angle of the two images.

[0060] In this technical solution, since three-dimensional forward intersection cannot be performed on weak intersection points, given an arbitrary elevation, the three-dimensional coordinates of the connection points at any elevation can be obtained by performing single-image forward intersection using the RPC parameters of each image.

[0061] The rational function model (RPC) converts the geodetic coordinates of ground points D(D) into... lat D lon D hei The image coordinates d(Line,Sample) and their corresponding image point coordinates are associated using a ratio polynomial. To enhance the stability of the parameter solution, the ground coordinates and image coordinates are regularized to between -1 and 1. For an image, the following ratio polynomial is defined:

[0062]

[0063] Polynomial Num L Den L Num s Den s All are polynomials of degree no more than 3, and their general form is:

[0064]

[0065] In the above formula, (P,L,H) are the regularized ground coordinates, and (L,S) are the regularized image coordinates. The regularization formula is:

[0066]

[0067]

[0068] Here, D lat_off D lat_scale D lon_off D lon_scale D hei_off and D hei_scale S is the regularization parameter for ground coordinates. off S scale L off and L scale The regularization parameter for image coordinates.

[0069] Using the RPC parameters of the image, based on the image coordinates of the connection point and the given elevations H and H1, the object coordinates A(X,Y,H), A'(X',Y',H) and A1(X1,Y1,H1), A1'(X1',Y1',H1) of the connection point can be calculated using formula (1).

[0070] According to formula (5), the intersection angle θ of the connection points can be calculated:

[0071]

[0072] When the intersection angle of the connecting points is less than 10°, the point is considered a weak intersection point.

[0073] For weak intersection connection points, this embodiment of the disclosure adopts a method based on external high-precision DEM for constraint to obtain the initial value of its object coordinates, thereby reducing the problem of poor elevation accuracy caused by excessively small intersection angles.

[0074] like Figure 3 As shown, based on the RPC of the image and the initial elevation H0, the planar coordinates of the point connecting the points are calculated. The elevation value H1 of the point on the DEM is interpolated based on the planar coordinates. The planar position of the point is recalculated based on the elevation H1. Then, the elevation H2 is interpolated again. If the difference between the two elevation values ​​is less than the limit after iterative interpolation, the last elevation value is taken as the initial elevation value of the point.

[0075] In this embodiment of the disclosure, for weak intersection connection points of two degrees or more, the corresponding three-dimensional coordinates of the connection points on each image are obtained by using the RPC of each image, and the average value is taken as the initial value of the object-side three-dimensional coordinates of the point.

[0076] In this embodiment, a high-precision connection point extraction method is used to extract connection points from multi-source images using automatic connection point extraction. The matching accuracy can reach the sub-pixel level, and it can also meet the requirement of uniform distribution of connection points in aerial triangulation. However, due to the large differences in angle and color among multi-source images, noise is present in the images, which in turn causes errors in the image matching points. Therefore, in order to improve the adjustment efficiency and enhance the robustness of regional network adjustment, it is necessary to detect and remove gross errors in the connection points.

[0077] When performing gross error detection and removal, firstly, the forward intersection of the connection points is performed using formula (1) to obtain the object coordinates (XYZ) of the connection points. Then, using single-image resection, the image point residuals of the connection points on the image are calculated respectively, and the residual mean error is statistically analyzed. Points with a mean error greater than 3 times the mean error are considered gross error points and are deleted.

[0078] Meanwhile, during gross error detection and removal, the connection points are divided into two parts: the part inside the stereo pair and the part between stereo pairs. Since the part between the stereo pairs is captured simultaneously during satellite imaging, the internal distortion and timeliness are consistent, resulting in a smaller connection point matching error. However, the part between the stereo pairs has a larger image timeliness and distortion due to the difference in imaging time, resulting in a larger connection point error. Therefore, different gross error removal thresholds are set for these two parts of the points to remove gross errors.

[0079] In this embodiment, when constructing and solving the joint regional network adjustment model of multi-source remote sensing satellite imagery, the influence of satellite system parameters on the geometric correction accuracy of the imagery is first analyzed. Two types of errors need to be corrected: one type of parameter corrects row-direction errors, and the other type corrects column-direction errors. Row parameters absorb the influence of orbit and attitude in the row direction, while column parameters absorb the influence of orbit and attitude in the column direction. Therefore, affine transformations defined on the image plane can be used to correct these errors.

[0080] An affine transformation on an image can be defined as follows:

[0081]

[0082] Where e0,f0 are image-side translation orientation parameters; e1,e2 and f1,f2 are image-side two-dimensional affine transformation orientation parameters. The three parameters e0,e1,e2 will correct the errors in the image column direction caused by the sensor's position and attitude errors in the scanning direction, while the three parameters f0,f1,f2 will correct the errors in the image row direction caused by the sensor's position and attitude errors in the flight direction.

[0083] Let (C',R') represent the image point coordinates after systematic error correction, and (C,R) represent the image point coordinates calculated using the RPC model. Both are unregularized parameter values. △C , △R represents the image point coordinates C , The corrections for R, e0, e1, e2, f0, f1, and f2 are the affine transformation parameters for each image. A mathematical model for regional network adjustment can be constructed (Equation 6).

[0084] Using the error model (6) and formula (1), the error equation for the regional network adjustment can be established, let:

[0085]

[0086] Using the affine transformation parameters e0, e1, e2, f0, f1, f2 of the image and the coordinates (P, L, H) of the ground points as unknowns, F S F L Expanding the Tailor series down to the first term, we obtain the error equation as follows:

[0087]

[0088] For weak intersection points, since their elevation values ​​are interpolated from the DEM, they are not corrected during the calculation. The error equation for weak intersection points is:

[0089]

[0090] The error equation can be written in matrix form as follows:

[0091]

[0092] In the formula:

[0093] V = [F S F L ] T ,

[0094]

[0095]

[0096] t = [da0 da1 da2 db0 db1 db2] T

[0097] X = [dP dL dH] T

[0098] In this matrix, B1 represents the object-side correction for weak intersection junctions.

[0099] The corresponding normal equation is:

[0100]

[0101] Where: N 11 =A T PA, N 22 =B T PB; n1 = A T PL,n2=B T PL.

[0102] Since the number of connection points in the normal equation is much greater than the affine distortion correction parameter, eliminating one type of unknown X from the above equation yields:

[0103]

[0104] Parallel construction and solution of the coefficient matrix of the normal equations are performed; first, the expansion analysis of formula (12) is performed, and A in the matrix is... T PA is the variance matrix of the affine distortion correction, which is only related to the image itself, and matrix B... T PB is the variance matrix of the object coordinate corrections of the connection points, which is related to the connection points themselves. Matrix B T PA is the covariance matrix of the affine distortion parameters and the tie points. If the tie points and the image are uncorrelated, then matrix B... T PA is a zero matrix. Therefore, the coefficient matrix of the normal equation contains a large number of zero matrices, and the larger the survey area, the more zero matrices there are.

[0105] like Figure 4As shown, the coefficient matrix of the normal equations for this survey area is constructed as follows: Figure 5 As shown in the left figure, 0, 1, 2, and 3 in the matrix represent image numbers. Each block in the matrix is ​​a 6x6 matrix unit, where black blocks are non-zero matrices. During storage, the 6x6 non-zero matrix units are stored in a one-dimensional container to reduce computational memory usage. Since the coefficient matrix of the normal equation is a non-zero positive definite matrix, only the non-zero upper triangular parts of the matrix are stored during matrix storage, such as... Figure 5 As shown in the right figure, this further reduces memory consumption.

[0106] In this embodiment of the disclosure, step 3) uses sparse moment construction and solution, and the specific process includes: 31) numbering the images, obtaining the pairwise connection relationships between the images based on the connection points and the image numbers, and marking them;

[0107] 32) Parallel computing is used to construct the coefficient matrix block of the normal equation in parallel using the connection points and weak intersection points, and the image number where the connection points are located is marked.

[0108] 33) Based on the image number of the coefficient matrix block of the non-zero normal equation, fill the coefficient matrix block into the corresponding position of the matrix (only fill the upper triangular matrix);

[0109] 34) The conjugate gradient method is used to solve the normal equations and obtain the affine distortion parameter correction for each image.

[0110] 35) Parallel calculation is adopted. According to formula (12), the object space correction of the three-dimensional coordinates of the connection point and the plane coordinate correction of the weak intersection point are calculated.

[0111] 36) Correct the parameters and calculate the mean error of the image point residuals. Based on the mean error, delete the connection points and weak intersection points whose errors are greater than 3 times the mean error.

[0112] 37) Iterative solution until the mean square error of the image point residual satisfies the condition.

[0113] In this embodiment of the disclosure, after the regional network adjustment is completed, the refined RPC is used to re-interpolate the new elevation values ​​of the weak intersection points, and then the adjustment is performed again until the difference between the two interpolated elevation values ​​of the weak intersection points is less than the limit.

[0114] After the adjustment is completed, the RPC is refitted, and the mean square error of the image point residuals and the fitting accuracy of the RPC parameters are calculated. For the stereo image pairs, the vertical disparity of the image pairs and the edge accuracy of the image pairs are calculated to ensure the accuracy of the joint adjustment.

[0115] In this embodiment of the disclosure, two types of data are used for experimental verification. The first set of data consists entirely of weak intersection remote sensing images, and the second set of data includes both stereo images and weak intersection images.

[0116] (1) Using 768 scenes of GF2 and 336 scenes of GF6 remote sensing satellite data, and with the aid of 30-meter SRTM data, a joint regional network adjustment was performed. The survey area distribution and adjustment results are as follows: Figure 7 As shown, the edge-matching accuracy of the GF6 and GF2 images after joint adjustment is better than 1 pixel, as... Figure 8 As shown.

[0117] (2) Joint regional network adjustment was performed using image data from ZY3 and GF2 satellites, and accuracy was checked using control points. Statistical analysis showed that after joint adjustment, the mean square error of the image point residuals of each image was better than 0.5 pixels, the vertical parallax of each image pair was better than 0.5 pixels, and the edge accuracy between image pairs was better than 1 pixel, which met the requirements for aerial triangulation accuracy.

[0118] The accuracy of the positioning was checked using control points, and the statistics of the joint regional network adjustment are shown in Table 1.

[0119] Table 1. Statistics on Control Point Accuracy

[0120] 10354 1.01 1.59 1.89 1.49 10361 -3.48 2.45 4.25 -2.18 10343 -2.94 6.24 6.90 -3.55 10344 -2.66 -0.10 2.66 -3.52 10360 3.54 4.45 5.68 -2.53 10355 -2.09 3.11 3.75 -4.77 10352 3.17 -4.00 5.11 2.05 average value -0.49 1.96 4.32 -1.86 Maximum value 3.54 6.24 6.90 2.05 Minimum value -3.48 -4.00 1.89 -4.77 Mean error 2.82 3.64 4.61 3.05

[0121] Example 2

[0122] This embodiment discloses a system for regional network adjustment based on multi-source optical remote sensing satellite imagery, such as... Figure 10 As shown, it includes a weak intersection point identification module with electrical connection, an initial coordinate calculation module, and an adjustment solution module;

[0123] The weak intersection point identification module is used to organize multi-source remote sensing satellite data and obtain DEM reference data for the corresponding survey area, and to extract connection points from the multi-source remote sensing data to obtain connection point files.

[0124] Then, the intersection angle of each connection point is calculated using the RPC and connection point coordinates of the remote sensing image, weak intersection connection points are identified, the initial object space coordinates of the weak intersection connection points are obtained using external DEM data, and gross error detection and gross error removal are performed on the connection points.

[0125] The initial coordinate calculation module is used to obtain the connection relationship between images based on the connection points, construct the relationship correspondence table of the normal equation coefficient matrix blocks; combined with parallel computing, it compresses, fills and solves the normal equation, obtains the corrections for affine distortion, connection points, and weak intersection points, and corrects the parameters; it calculates the image point residual mean square error and removes gross errors from connection points and weak intersection points; it repeats this process until the image point residual mean square error remains unchanged.

[0126] The adjustment and calculation module is used to reuse the external DEM to interpolate the elevation values ​​of weak intersection points, and call the initial coordinate calculation module to repeat the calculation until the elevation values ​​of the weak intersection points no longer change after two interpolations; then, the accuracy of the regional network adjustment of the multi-source remote sensing image is statistically analyzed, the RPC parameters are fitted, and the regional network adjustment results are output.

[0127] In this embodiment of the disclosure, the gross error detection and removal of the connection points includes: calculating the intersection angle of each connection point by combining the RPC parameters of the image, identifying weak intersection connection points, obtaining the initial object space coordinates of the weak intersection connection points by combining external DEM data, and calculating the image space residual of each weak intersection connection point to perform gross error detection and removal.

[0128] Using the image's RPC, forward intersection is performed on the connection points to obtain their initial object-space coordinates. The image-space residuals of each connection point are then calculated, and gross errors are detected and eliminated.

[0129] In this embodiment of the disclosure, when performing gross error detection and removal on the connection points, the connection points are divided into two parts: the part inside the stereo image pair and the part between the stereo image pairs. Different gross error removal thresholds are set for each part to remove gross errors. First, forward intersection is performed on the connection points to obtain the object coordinates of the connection points. Then, using single-image back intersection, the image point residuals of the connection points on the image are calculated, and the residual error is statistically analyzed. Points with a residual error greater than a preset error value are identified as gross error points and are deleted.

[0130] Example 3

[0131] This invention provides an electronic device for regional network adjustment based on multi-source optical remote sensing satellite imagery, comprising a processor and a memory electrically connected to each other; the memory is used to store a computer program; when the processor executes the aforementioned computer program, it can implement the method for regional network adjustment based on multi-source optical remote sensing satellite imagery as described in Embodiment 1 above. The specific adjustment method steps are the same as those in Embodiment 1 above, and will not be repeated here.

[0132] Example 4

[0133] The present invention provides a computer-readable storage medium storing a computer program; when the computer program is executed, it can implement the method for regional network adjustment based on multi-source optical remote sensing satellite imagery as described in Embodiment 1 above. The specific adjustment method steps are the same as those in Embodiment 1 above, and will not be repeated here.

[0134] The computer described in this application embodiment can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. The computer-readable storage medium can be any usable medium that a computer can read, or a data storage device such as a server or data center that integrates one or more usable media. The usable medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital versatile optical disc (DVD)), or a semiconductor medium (e.g., solid-state drive (SSD)). The software formed by the computer's stored code can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other storage media that are mature in the art.

[0135] In the various embodiments of this application, the functional modules can be integrated into one processing unit or module, or each module can exist physically separately, or two or more modules can be integrated into one unit or module. In the above embodiments, they can be implemented entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, they can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated.

[0136] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for regional network adjustment based on multi-source optical remote sensing satellite imagery, characterized in that... include: 1) First, organize the multi-source remote sensing satellite data and obtain the DEM reference data of the corresponding survey area, and extract the tie points from the multi-source remote sensing data to obtain the tie point file; 2) Calculate the intersection angle of each connection point using the RPC and connection point coordinates of the remote sensing image, identify weak intersection connection points, obtain the initial object space coordinates of weak intersection connection points using external DEM data, and perform gross error detection and gross error removal on the connection points. 3) Obtain the connection relationship between images based on the connection points, and construct the relationship correspondence table of the normal equation coefficient matrix blocks; combine parallel computing to compress, fill and solve the normal equation, obtain the corrections for affine distortion, connection points and weak intersection points, and correct the parameters; calculate the image point residual mean square error, and remove gross errors for connection points and weak intersection points; 4) Repeat step 3) until the image point residual mean error remains constant; 5) Re-interpolate the elevation values ​​of the weak intersection points using the external DEM, and repeat step 3) until the elevation values ​​of the weak intersection points no longer change after two interpolations. 6) Calculate the accuracy of regional network adjustment of multi-source remote sensing images, fit the RPC parameters, and output the regional network adjustment results.

2. The method for regional network adjustment based on multi-source optical remote sensing satellite imagery according to claim 1, characterized in that: Step 2) involves gross error detection and removal at connection points, including: By combining the RPC parameters of the image, the intersection angle of each connection point is calculated, weak intersection connection points are identified, and the initial object space coordinates of the weak intersection connection points are obtained by combining external DEM data. The image space residual of each weak intersection connection point is calculated, and gross errors are detected and eliminated. Using the image's RPC, forward intersection is performed on the connection points to obtain their initial object-space coordinates. The image-space residuals of each connection point are then calculated, and gross errors are detected and eliminated.

3. The method for regional network adjustment based on multi-source optical remote sensing satellite imagery according to claim 2, characterized in that: When obtaining the initial object coordinates of weak intersection connection points, for weak intersection connection points with two or more degrees, the corresponding three-dimensional coordinates of the connection points on each image are obtained using the RPC of each image, and the average value is taken as the initial value of the object coordinates of that point.

4. The method for regional network adjustment based on multi-source optical remote sensing satellite imagery according to claim 3, characterized in that: During the gross error detection and removal process, the connection points are first forwardly intersected to obtain their object coordinates. Then, using single-image back intersection, the image point residuals of the connection points on the image are calculated, and the residual error is statistically analyzed. Points with a residual error greater than a preset error value are identified as gross error points and are deleted. The connection points are divided into two parts: the inside of the stereo image pair and the between stereo image pairs. Different gross error removal thresholds are set for each part to remove gross errors.

5. The method for regional network adjustment based on multi-source optical remote sensing satellite imagery according to claim 4, characterized in that: Step 3) The process specifically includes: 31) Number the images, obtain the pairwise connection relationships between the images based on the connection points and image numbers, and mark them; 32) Parallel computing is used to construct the coefficient matrix block of the normal equation in parallel using the connection points and weak intersection points, and the image number where the connection points are located is marked. 33) Fill the coefficient matrix blocks into the corresponding positions of the matrix according to the image number of the coefficient matrix blocks of the non-zero normal equation; 34) The conjugate gradient method is used to solve the normal equations and obtain the affine distortion parameter correction for each image. 35) Parallel computation is used to calculate the object-space corrections for the three-dimensional coordinates of the connection points and the planar coordinate corrections for the weak intersection points; 36) Correct the parameters and calculate the mean error of the image point residuals. Based on the mean error, delete the connection points and weak intersection points whose errors are greater than the preset mean error. 37) Iterative solution until the mean square error of the image point residual satisfies the condition.

6. A system for regional network adjustment based on multi-source optical remote sensing satellite imagery, characterized in that: This includes a weak intersection point identification module for electrical connections, an initial coordinate calculation module, and an adjustment solution module; The weak intersection point identification module is used to organize multi-source remote sensing satellite data and obtain DEM reference data for the corresponding survey area, and to extract connection points from the multi-source remote sensing data to obtain connection point files. Then, the intersection angle of each connection point is calculated using the RPC and connection point coordinates of the remote sensing image, weak intersection connection points are identified, the initial object space coordinates of the weak intersection connection points are obtained using external DEM data, and gross error detection and gross error removal are performed on the connection points. The initial coordinate calculation module is used to obtain the connection relationship between images based on the connection points, construct the relationship correspondence table of the normal equation coefficient matrix blocks; combined with parallel computing, it compresses, fills and solves the normal equation, obtains the corrections for affine distortion, connection points and weak intersection points, and corrects the parameters; calculates the image point residual mean square error, and removes gross errors from connection points and weak intersection points; Repeat this process until the mean square error of the image point residual remains constant; The adjustment and calculation module is used to reuse the external DEM to interpolate the elevation values ​​of weak intersection points, and call the initial coordinate calculation module to repeat the calculation until the elevation values ​​of the weak intersection points no longer change after two interpolations; then, the accuracy of the regional network adjustment of the multi-source remote sensing image is statistically analyzed, the RPC parameters are fitted, and the regional network adjustment results are output.

7. The system for regional network adjustment based on multi-source optical remote sensing satellite imagery according to claim 6, characterized in that: The gross error detection and removal of the connection points includes: calculating the intersection angle of each connection point by combining the RPC parameters of the image, identifying weak intersection connection points, obtaining the initial object space coordinates of the weak intersection connection points by combining external DEM data, and calculating the image space residual of each weak intersection connection point to perform gross error detection and removal. Using the image's RPC, forward intersection is performed on the connection points to obtain their initial object-space coordinates. The image-space residuals of each connection point are then calculated, and gross errors are detected and eliminated.

8. The system for regional network adjustment based on multi-source optical remote sensing satellite imagery according to claim 7, characterized in that: When performing gross error detection and removal on the connection points, the connection points are divided into two parts: the part inside the stereo image pair and the part between the stereo image pairs. Different gross error removal thresholds are set for each part. First, forward intersection is performed on the connection points to obtain their object coordinates. Then, using single-image back intersection, the image point residuals of the connection points on the image are calculated, and the residual error is statistically analyzed. Points with a residual error greater than a preset error value are identified as gross error points and deleted.

9. An electronic device for regional network adjustment based on multi-source optical remote sensing satellite imagery, comprising a processor and a memory electrically connected to each other; the memory is used to store computer programs; characterized in that: When the processor executes the aforementioned computer program, it can implement the method for regional network adjustment based on multi-source optical remote sensing satellite imagery as described in any one of claims 1-5.

10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program; when the computer program is executed, it can implement the method for regional network adjustment based on multi-source optical remote sensing satellite imagery as described in any one of claims 1-5.

Images