A tessellation line joint edge precision detection method based on spatial topological relations

Abstract

The application discloses a mosaic line joint edge precision detection method based on spatial topological relation. The method firstly carries out standardization treatment on multiple remote sensing images, and automatically generates a mosaic line vector file, and obtains a total mosaic line area C by set operation. Then, a total effective area A of the image effective area is calculated, and a difference area of A and C is used to complete the hollow area in C. The same name point pair of the adjacent overlapping area of the completed mosaic line is subjected to object coordinate positioning precision difference calculation, and geometric correction is carried out according to the result, and finally, a seamless image map product is obtained. The method of the application realizes automatic identification and positioning for large-scale, heterogeneous satellite image mosaic leakage detection, can directly output leakage area information, compared with the traditional manual checking method, effectively avoids subjective limitations, significantly improves the efficiency and precision of processing massive heterogeneous data, and greatly reduces the labor cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of satellite remote sensing image mosaic line joint edge precision detection, and particularly relates to a mosaic line joint edge precision detection method based on spatial topological relationship. BACKGROUND

[0002] Satellite remote sensing image mosaic technology refers to a key process of splicing a plurality of remote sensing images with overlapping areas into a comprehensive image with wider coverage, no seams and consistent color through a series of accurate processing. The core of the technology is to overcome the inconsistency of radiation and geometry caused by imaging time, sensor angle and atmospheric condition difference. The typical processing flow includes: firstly, performing radiation correction and orthographic correction on the source image to eliminate the distortion caused by the sensor and the terrain, and to ensure accurate matching of each image in the geometric space; then, through image registration algorithm, feature points are extracted in the overlapping area and accurately matched to establish a transformation model between images; then, a color balance algorithm is used to adjust the brightness, contrast and color deviation between adjacent images to eliminate the visible seams; finally, the image mosaic line is automatically generated to avoid as much as possible the ground objects such as clouds, water bodies and buildings that are easy to produce defects, and to realize the natural and seamless fusion of the boundary. The technology is the basis of large-scale remote sensing monitoring and application, and provides core support for generating high-quality regional or global digital base map.

[0003] At present, many domestic academic and engineering applications focus on the research of automatic generation of image joint edge line, which can be mainly divided into the following categories: first, the shortest path algorithm based on graph theory (such as Dijkstra algorithm), which regards the difference value (such as spectral, texture difference) of the overlapping area pixels as the cost, and finds a path with the minimum cost as the joint edge line, which can effectively avoid houses, vehicles and other obvious ground objects. Second, the intelligent optimization algorithm (such as ant colony algorithm, genetic algorithm) is used to simulate natural evolution or group intelligence behavior to search the optimal joint edge line in the overlapping area, which is suitable for complex scenes. Third, the method combined with auxiliary data is used to use digital surface model (DSM), building vector and other prior information to actively guide the joint edge line to bypass high-rise buildings and other prominent targets, which has the highest precision but depends on additional data.

[0004] The current trend is to integrate multiple methods and introduce deep learning to intelligently identify features and differences through model training, so as to achieve more efficient and adaptive automatic generation of edge lines. However, with the development of the times, the automatically generated image mosaic lines have the following technical defects: 1) The current automatic production of image mosaic lines mainly finds a path with the minimum cost as the edge line, which can effectively avoid obvious features such as houses and vehicles, and does not check whether the topological position relationship of the image mosaic line is correct; 2) The current automatic production of image mosaic lines mainly provides data support for subsequent image mosaic, and does not reasonably check the edge accuracy of the image, which often leads to serious misplacement at the image edge; 3) Most of the current automatic production of image mosaic line algorithms stop at the level of mathematical model and algorithm research, and often rely on a large amount of manual inspection in the actual processing of remote sensing data. The above technical defects are not convenient for engineering application and promotion.

[0005] For example, the application number 202410572199.9 discloses a method for generating a remote sensing orthographic image mosaic line. The high-level mosaic line network generated by the application scheme can effectively avoid buildings and roads and other areas, and can perform large-scale image mosaic. The result of the mosaic is independent of the processing order, and the mosaic processing has flexibility and efficiency, and avoids error accumulation. However, the scheme also has the problem that the topological position relationship of the image mosaic line is not checked correctly, which often leads to serious misplacement at the image edge.

[0006] Based on the above reasons, the present application discloses a mosaic line edge accuracy detection method based on spatial topological relationship, which provides a quality inspection and analysis means for large-scale, heterogeneous satellite image mosaic data processing. SUMMARY

[0007] In view of the above problems, the purpose of the present application is to provide a mosaic line edge accuracy detection method based on spatial topological relationship, which realizes automatic quality inspection and analysis of large-scale, heterogeneous satellite image mosaic data processing.

[0008] First aspect: a mosaic line edge accuracy detection method based on spatial topological relationship, comprising the steps of:

[0009] S1, receiving multiple remote sensing images and performing standardization processing;

[0010] S2, automatically generating a remote sensing image mosaic line vector file, and obtaining a total mosaic line area C by taking the union set of all mosaic vector faces in the mosaic line;

[0011] S3, calculating the effective vector faces of the effective areas of each remote sensing image, taking the union set of all effective vector faces, and obtaining a total effective area A;

[0012] S4, calculate the difference area of the total effective area A and the total tessellation line area C, and complement the hollow area of the total tessellation line area C by using the difference area;

[0013] S5, perform difference calculation of object coordinate positioning accuracy on homonymic point pairs of the adjacent overlapping area of the complemented total tessellation line area C';

[0014] S6, perform geometric correction on the complemented total tessellation line area C' according to the difference result of the positioning accuracy, and obtain a seamless linked image map product.

[0015] In an embodiment of the present application, the standardization processing in S1 comprises:

[0016] The images in different formats are converted into a unified GeoTIFF format, the metadata information of each image is parsed, and orthographic correction processing is performed on each image, and orthographic images with the same resolution are output.

[0017] In an embodiment of the present application, the metadata information comprises RPC geometric model parameters carried by the image.

[0018] In an embodiment of the present application, the automatic generation of the remote sensing image tessellation line in S2 comprises:

[0019] Based on the shortest path algorithm of graph theory, the difference value of the overlapping area pixels is regarded as a cost, a path with the minimum cost is found as a joint edge line, the joint edge position relationship of the image is described based on the joint edge line, and a tessellation line vector file is obtained.

[0020] In an embodiment of the present application, the total effective area A obtained in S3 comprises:

[0021] Invalid pixels of each image are removed, and an effective area vector graph of the image is generated. The effective area vector graphs of all images are performed set operation, and the total effective area A is obtained.

[0022] In an embodiment of the present application, the complement of the hollow area of the total tessellation line area C by using the difference area in S4 comprises:

[0023] The difference area of the total effective area A and the total tessellation line area C is calculated, and the formula is expressed as:

[0024] W1=A-C

[0025] Wherein, W1 is the difference area;

[0026] If the difference area W1 is greater than a set tolerance threshold, the set operation of the difference area W1 and the adjacent tessellation vector surface is performed.

[0027] In one embodiment of the present application, the S5 comprises the steps of:

[0028] S51, complementing the total mosaic line area C', selecting adjacent image pairs in sequence, using an image feature-based matching algorithm in the image pair overlap area to establish a connection between the images, and extracting a large number of uniformly distributed homonymic point pairs;

[0029] S52, converting the image coordinates of the homonymic points into object coordinates based on the geometric model parameters of each image;

[0030] S53, for each homonymic point pair, calculating the homonymic point pair coordinate difference ΔX, ΔY, ΔZ in the object coordinates;

[0031] S54, counting the coordinate difference of all homonymic point pairs of the image pair, denoted as the positioning accuracy difference W2.

[0032] In one embodiment of the present application, the condition for performing geometric correction in the S6 comprises:

[0033] Comparing the accuracy difference W2 with the accuracy threshold value, and if the accuracy difference W2 exceeds the accuracy threshold value, performing geometric correction.

[0034] In one embodiment of the present application, the step of geometric correction comprises:

[0035] S61, selecting a reference image;

[0036] S62, using the homonymic points between the image to be corrected and the reference image as control points to construct a correction model;

[0037] S63, applying the correction model to resample the image to be corrected, eliminating the systematic geometric deviation between the image to be corrected and the reference image, and outputting the corrected image;

[0038] S64, using the corrected image as the image to be corrected, repeating the steps S62 and S63 until the accuracy difference W2 of the image to be corrected meets the requirements.

[0039] Second aspect: an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method provided in the first aspect.

[0040] Third aspect: a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the steps of the method provided in the first aspect.

[0041] The present application has the following beneficial effects:

[0042] 1.The method is aimed at mosaic gap detection of large-scale and heterogeneous satellite images, realizes full-automatic identification and positioning, and directly outputs the region information of the gap. Compared with the traditional manual gap detection method, the method not only completely eliminates the limitations of manual subjective judgment, but also efficiently processes massive heterogeneous image data, significantly improves the efficiency and accuracy of gap detection, and greatly reduces the labor cost.

[0043] 2.The method can quantitatively calculate the edge joint precision of large-scale and heterogeneous satellite images through automatic quality inspection function, and can generate a standardized edge joint precision report, so that the edge joint precision state is clear and traceable. At the same time, the automatic early warning is triggered for the case of non-standard precision, which can timely find the geometric misplacement problem at the image edge joint, avoid the flow of precision defects into the subsequent application link, and effectively guarantee the geometric accuracy and reliability of the satellite image mosaic results.

[0044] 3.The method can automatically perform secondary registration correction based on the reference image for the edge joint precision exceeding problem, and complete image mosaic after ensuring the precision meets the standard through iterative detection, and construct a full-process automatic closed loop of precision detection, exceeding warning, automatic correction and standard mosaic. The process does not need manual intervention for correction and mosaic operation, solves the pain points of traditional algorithms staying at the theoretical level and engineering application relying on manual intervention, and provides a practical engineering solution for automatic mosaic processing of large-scale and heterogeneous satellite images. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 is a flowchart of the method of the present application;

[0046] Figure 2 is a principle flowchart of the method of the present application;

[0047] Figure 3 is a total mosaic line region C of the image in the embodiment of the present application;

[0048] Figure 4 is a total effective region A of the image in the embodiment of the present application;

[0049] Figure 5 is a difference region schematic diagram in the embodiment of the present application;

[0050] Figure 6 is a blank region generated by the mosaic line file schematic diagram in the embodiment of the present application;

[0051] Figure 7 is an edge joint region positioning precision detection schematic diagram in the embodiment of the present application;

[0052] Figure 8 is a polynomial correction schematic diagram based on the reference base map in the embodiment of the present application;

[0053] Figure 9 Fig. 1 is a schematic diagram of the structure of an electronic device according to the present application. DETAILED DESCRIPTION

[0054] Embodiments of the present application are described in detail below with reference to the attached drawings, which are to be considered in conjunction with the description, wherein like or similar symbols designate like or similar elements throughout the several views. The embodiments described below are examples of the present application, which are only for the purpose of explanation and are not to be considered as limiting the present application.

[0055] The existing mosaic line edge precision detection method has the following problems: the topological position relationship of the image mosaic line is not correctly verified, which often leads to serious misplacement of the image edge.

[0056] Embodiment 1

[0057] In view of the above problems, the embodiment discloses a mosaic line edge precision detection method based on spatial topological relationship, Figure 1 Fig. 1 is a schematic diagram of the structure of an electronic device according to the present application. Figure 2 Fig. 1 is a schematic diagram of the structure of an electronic device according to the present application.

[0058] S1, receiving a plurality of remote sensing images and performing standardization processing.

[0059] Receiving n (n≥2) remote sensing images to be processed, denoted as I1, I2,..., In. n The image is preliminarily standardized, which specifically includes:

[0060] Format unification, converting images of different formats into a unified format (such as GeoTIFF) supported by the process to eliminate processing obstacles caused by format differences.

[0061] Metadata extraction, parsing the metadata information of each image, especially the geometric model parameters (Rational Polynomial Coefficient RPC), providing a basis for subsequent object coordinate calculation, and organizing and storing the extracted metadata information for subsequent use.

[0062] Orthorectification, according to RPC or other metadata information, orthorectification processing is performed on the image to output orthographic image results with the same resolution, so as to eliminate the geometric distortion of the image caused by factors such as terrain undulation and sensor attitude, and make the position of the ground object in the image correspond to the actual geographical position. Through orthorectification, the geographical reference of different images can be unified to provide accurate geometric basis for subsequent mosaic line edge precision detection.

[0063] S2. Automatically generate a mosaic line vector file of remote sensing imagery, and find the union of all mosaic vector surfaces in the mosaic line to obtain the total mosaic line region C.

[0064] Specifically, based on graph theory's shortest path algorithm (such as Dijkstra's algorithm), the difference values ​​of pixels in overlapping areas (such as spectral and texture differences) can be regarded as costs, effectively avoiding obvious features such as houses and vehicles, and finding a path with the minimum crossing cost as the edge line. Based on the edge line, a mosaic line vector file describing the positional relationship of the image edge is generated.

[0065] From the mosaic line vector file, such as Figure 3 As shown, extract M polygon tessellation vector faces, denoted as C1, C2, ..., C6. M tessellation vector planes C1-C of all overlapping regions M

[0066] Perform a union operation to generate a total mosaic line vector region C, which represents the area covered by all images in the entire survey area.

[0067] S3. Calculate the effective vector surface of the effective area of ​​each remote sensing image, and find the union of all effective vector surfaces to obtain the total effective area A.

[0068] For each image Iᵢ, invalid pixels (such as black border areas) are removed, and a vector image Aᵢ representing its valid area is generated. For example... Figure 4 As shown, the vector diagram of the effective area of ​​all images A1-A n Perform a union operation to obtain the largest area vector map A that can theoretically be covered by all images in the entire survey area. This total effective area vector map A reflects the theoretical range that all images can cover in the entire survey area without considering factors such as mosaic lines.

[0069] S4. Calculate the difference between the total effective region A and the total mosaic line region C, and use the difference region to fill in the empty region of the total mosaic line region C.

[0070] like Figure 5 As shown, a difference operation is performed between the total effective region A and the total mosaic line region C. The formula is as follows:

[0071] W1=A–C

[0072] Among them, such as Figure 6 As shown, the difference region W1 is the blank area in the total mosaic line region C that is not covered by the overlapping area of ​​any adjacent image pair. This is usually caused by incomplete mosaic line vector file generation or gaps between images.

[0073] Further, the detected difference area W1 (blank area) is subjected to a Union operation with the adjacent tessellation line vector graph Cᵢ, and the original overlapping area definition is dynamically expanded to ensure that the total tessellation line area C covers the entire total effective area A.

[0074] Further, it is judged whether the area of the difference area W1 is zero or not greater than a set tolerance threshold. If the difference area W1 (blank area) is within an acceptable range, it is considered that the spatial layout of the total tessellation line area C is reasonable, and the process continues to the geometric accuracy check.

[0075] If it is judged that the area of the difference area W1 is greater than the set tolerance threshold, it is considered that the blank area is not within the acceptable range, the difference area W1 (blank area) is subjected to a Union operation with the adjacent tessellation line vector graph Cᵢ, and then the tessellation line is iteratively optimized. If it still cannot meet the requirement after multiple iterations, a warning is needed to prompt the user for manual intervention or to check the coverage range of the input image.

[0076] S5, the same name point pairs of the adjacent overlapping area of the completed total tessellation line area C' tessellation line are subjected to object coordinate positioning accuracy difference calculation.

[0077] First, as shown in Figure 7 , for the completed total tessellation line area C', adjacent image pairs (I1, I2) are sequentially selected, and a matching algorithm based on image features (such as SIFT, ORB, SURF, etc.) is used in the overlapping area of the image pairs I1 and I2 to automatically establish the connection relationship between the images and extract a large number of uniformly distributed same name point pairs.

[0078] Then, using the geometric model parameters (such as RPC) provided by each image, the image coordinates (x, y) of the same name points are converted into object coordinates (Lat, Lon, Height).

[0079] Then, for each same name point pair, the object coordinate difference ΔX, ΔY, ΔZ of the same name point pair is calculated, and the statistical quantity of the difference value of all same name point pairs of the image pair is calculated, such as the root mean square error (RMSE), which is denoted as the positioning accuracy difference W2.

[0080] S6, according to the positioning accuracy difference result, the completed total tessellation line area C' is subjected to geometric correction to obtain a seamless image graph product.

[0081] The calculated accuracy difference W2 is compared with the accuracy threshold required by the project (for example, 1 pixel size corresponds to the ground distance).

[0082] If it meets the accuracy threshold requirement, it is considered that the relative geometric accuracy of the image pair meets the requirement, and the process continues to the next pair of checks or enters the final output stage.

[0083] If not, it is determined that the geometry of the image pair has significant systematic errors, and the geometric correction needs to be performed again.

[0084] Specifically, as shown in Figure 8 The geometric correction includes:

[0085] First, select a high-precision and high-quality image from the reference as the reference image. The same-named points between the image to be corrected and the reference image are used as control points to construct a quadratic polynomial correction model.

[0086] Then, perform correction and apply the correction model to resample the image to be corrected to eliminate the systematic geometric deviation between the image to be corrected and the reference image.

[0087] Then, output the corrected image, and use the corrected image to perform matching and precision checking again until the precision difference W2 of the image to be corrected meets the requirements.

[0088] Finally, when the linking relationship of all image pairs passes the void check and the precision check, the seamless linked image map product is finally generated according to the optimized and corrected image mosaic line vector file.

[0089] The application also provides an electronic device, Figure 9 The structure diagram of the electronic device provided by the embodiment of the application is shown in Figure 9 The electronic device can include a processor, a communications interface, a memory, and a communications bus, wherein the processor, the communications interface, and the memory complete mutual communication through the communications bus. The processor can call logical instructions in the memory, for example, to execute the following method:

[0090] S1, receiving multiple remote sensing images and performing standardization processing;

[0091] S2, automatically generating a remote sensing image mosaic line vector file, and obtaining a total mosaic line area C by performing a union set on all mosaic vector faces in the mosaic line;

[0092] S3, calculating effective vector faces of effective areas of each remote sensing image, and obtaining a total effective area A by performing a union set on all effective vector faces;

[0093] S4, calculating a difference area between the total effective area A and the total mosaic line area C, and supplementing the void area of the total mosaic line area C by using the difference area;

[0094] S5, calculating the positional precision difference of the same-named point pairs in the adjacent overlapping area of the supplemented total mosaic line area C';

[0095] S6、According to the positioning accuracy difference result, the total mosaic line area C' is geometrically corrected to obtain a seamless linked image map product.

[0096] In addition, the logic instructions in the above-mentioned memory can be realized in the form of a software function unit and sold or used as an independent product, and can be stored in a computer-readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0097] The embodiments of the present application also provide a non-transitory computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the method provided by the above-mentioned embodiments, for example, including:

[0098] S1, receiving a plurality of remote sensing images and performing standardization processing;

[0099] S2, automatically generating a remote sensing image mosaic line vector file, and obtaining a total mosaic line area C by taking the union of all mosaic vector faces in the mosaic line;

[0100] S3, calculating the effective vector faces of the effective area of each remote sensing image, taking the union of all effective vector faces, and obtaining a total effective area A;

[0101] S4, calculating the difference area of the total effective area A and the total mosaic line area C, and supplementing the hollow area of the total mosaic line area C by using the difference area;

[0102] S5, calculating the positioning accuracy difference of the same name point pairs of the adjacent overlapping area of the total mosaic line area C' mosaic line;

[0103] S6, according to the positioning accuracy difference result, the total mosaic line area C' is geometrically corrected to obtain a seamless linked image map product.

[0104] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0105] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and the necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0106] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for detecting the edge accuracy of tessellation lines based on spatial topological relationships, characterized in that, include: S1. Receive multiple remote sensing images and perform standardization processing; S2. Automatically generate a mosaic vector file of remote sensing imagery, and find the union of all mosaic vector surfaces in the mosaic to obtain the total mosaic region C. S3. Calculate the effective vector surface of the effective region of each remote sensing image, and find the union of all effective vector surfaces to obtain the total effective region A. S4. Calculate the difference between the total effective region A and the total mosaic line region C, and use the difference region to fill in the empty areas of the total mosaic line region C. S5. For the pairs of corresponding points in the adjacent overlapping areas of the completed mosaic line region C', calculate the difference in object coordinate positioning accuracy. S6. Based on the positioning accuracy difference result, perform geometric correction on the complete mosaic line area C' to obtain a seamless link image product.

2. The method according to claim 1, characterized in that, The standardization process in S1 includes: The system converts images of different formats into a unified GeoTIFF format, parses the metadata information of each image, performs orthorectification on each image, and outputs orthophotos with the same resolution.

3. The method according to claim 2, characterized in that, The metadata information includes the RPC geometric model parameters that come with the image.

4. The method according to claim 1, characterized in that, The automatic generation of remote sensing image mosaic lines in S2 includes: The graph theory-based shortest path algorithm treats the difference in pixel values ​​in overlapping regions as a cost, finds a path with the minimum cost to cross the edge as the edge line, describes the positional relationship of the image edge based on the edge line, and obtains the mosaic line vector file.

5. The method according to claim 1, characterized in that, The process of obtaining the total effective region A in S3 includes: Remove invalid pixels from each image and generate its valid region vector map. Then, combine the valid region vector maps of all images to obtain the total valid region A.

6. The method according to claim 5, characterized in that, The process of filling in the voids in the total mosaic line region C using the difference region in S4 includes: The difference between the total effective region A and the total mosaic line region C is calculated using the following formula: W1=A–C Wherein, W1 represents the difference region; If the difference region W1 is greater than the set tolerance threshold, detect the tessellation vector surface adjacent to the difference region W1, and find the union of the difference region W1 and the adjacent tessellation vector surface.

7. The method according to claim 3, characterized in that, S5 includes the following steps: S51. For the completed mosaic region C', select adjacent image pairs in sequence, and use a matching algorithm based on image features in the overlapping area of ​​the image pairs to establish the connection relationship between the images and extract a large number of evenly distributed pairs of the same name. S52. Based on the geometric model parameters of each image, convert the image coordinates of the corresponding points into object coordinates; S53. For each pair of points with the same name, calculate the difference in the object coordinates of the pair of points with the same name, ΔX, ΔY, ΔZ. S54. The statistical measure of the coordinate difference of all corresponding point pairs in the statistical image is denoted as the positioning accuracy difference W2.

8. The method according to claim 7, characterized in that, The conditions for geometric correction in S6 include: The accuracy difference W2 is compared with the accuracy threshold. If the accuracy difference W2 exceeds the accuracy threshold, geometric correction is performed.

9. The method according to claim 8, characterized in that, The geometric correction steps include: S61. Select the reference image; S62. Use the corresponding points between the image to be corrected and the reference image as control points to construct a correction model; S63. Apply the correction model to resample the image to be corrected, eliminate the systematic geometric deviation between the image to be corrected and the reference image, and output the corrected image. S64. Take the corrected image as the image to be corrected, and repeat steps S62 and S63 until the accuracy difference W2 of the image to be corrected meets the requirements.

Images