Lunar surface positioning method based on cross-view-angle image falling pit space relation matching

Abstract

The invention relates to a lunar surface positioning method based on cross-view-angle image trocar space relation matching, and the method comprises the following steps: obtaining a lunar surface image of a front view angle, extracting trocar features, and calculating a space relation; comprise the horizontal view angle of each troille, the horizontal view angle of the center of the adjacent troille and the distance relation between the troille and the position of the camera; the method comprises the following steps: acquiring a track image of an overlooking view angle, marking a trocar, and constructing a trocar matching landmark library; traversing multi-stage matching is carried out on the spatial relationship and the troille matching landmark library to obtain a preliminary positioning result, and the multi-stage matching comprises tangent included angle matching of a single troille, feature included angle matching of adjacent troilles and distance relationship matching of multiple troilles; and based on the cosine theorem, optimizing the preliminary positioning result by using least square adjustment. Compared with the prior art, the lunar surface positioning precision can be further improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of deep space exploration, and in particular relates to a lunar surface positioning method based on the matching of crater spatial relationships in cross-perspective images. Background Art

[0002] For future lunar surface exploration missions, remote and continuous positioning is a basic requirement. At present, lunar surface positioning research mainly includes relative positioning and absolute positioning methods. The relative positioning dead reckoning method calculates the position relative to the lunar landing point by combining visual odometry (VO) and inertial measurement unit (IMU), and uses the images transmitted to the earth with the help of ground personnel to achieve global position update. However, there are various cumulative errors in the calculation process of this method, and the errors will increase with the increase of distance, resulting in low accuracy of heading estimation. Absolute positioning mainly includes methods such as astronomical navigation technology positioning, digital elevation model (DEM) matching positioning, and digital orthophoto model (DOM) matching positioning. Among them, astronomical navigation technology positioning mainly obtains the information position of the target by relying on the observation information of the earth, the sun, and other celestial bodies. However, the current lunar surface positioning accuracy of astronomical navigation technology positioning is relatively low; the digital elevation model matching positioning method uses the local DEM created by lunar surface images to match the DEM created by orbital images for positioning. However, for the lunar environment, the change of the sun angle will cause drastic changes in the terrain appearance, increasing the difficulty of DEM matching; the digital orthophoto model matching positioning method uses the local DOM created by lunar surface images to match the landing image DOM for positioning. However, this method has a complex process, high requirements for image data, and lacks sufficient high-resolution DOM for its use. Therefore, it is necessary to design a new lunar surface positioning method to further improve the lunar surface positioning accuracy. Summary of the Invention

[0003] The purpose of the present invention is to overcome the above-mentioned defects existing in the prior art and provide a lunar surface positioning method based on the matching of crater spatial relationships in cross-perspective images, so as to solve the problems of insufficient positioning accuracy and high requirements for lunar surface image data in the existing lunar surface positioning methods.

[0004] The purpose of the present invention can be achieved by the following technical solutions:

[0005] The present invention provides a lunar surface positioning method based on the matching of crater spatial relationships in cross-perspective images, including the following steps:

[0006] Obtain the lunar surface image of the front view angle, extract the crater features and calculate the spatial relationships, where the spatial relationships include the horizontal view angle of each crater, the horizontal view angle between the centers of adjacent craters, and the distance relationship between each crater relative to the camera position;

[0007] Obtain the orbital image from the top-down perspective, perform crater marking, and construct a crater matching landmark library;

[0008] Traverse and perform multi-level matching on the spatial relationship and the crater matching landmark library to obtain a preliminary positioning result. The multi-level matching includes the tangent angle matching of a single crater, the feature angle matching of adjacent craters, and the distance relationship matching of multiple craters;

[0009] Based on the cosine theorem, use the least squares adjustment to optimize the preliminary positioning result.

[0010] Furthermore, use the YOLOv8 model to extract crater features.

[0011] Furthermore, when the number of crater features extracted from a certain lunar surface image is less than the set threshold or the display is incomplete, perform matching and stitching on this lunar surface image and then re-extract the crater features.

[0012] Furthermore, the specific process of matching and stitching the lunar surface images is as follows:

[0013] Use Scale-Invariant Feature Transform (SIFT) to perform matching and stitching of adjacent lunar surface images;

[0014] Use distance-weighted fusion to eliminate the image stitching line.

[0015] Furthermore, the specific process of constructing the crater matching landmark library is as follows:

[0016] Obtain the LROC NAC image of the orbital area and perform orthorectification processing to obtain the first DOM image;

[0017] According to the range of the first DOM image, obtain the low-altitude landing image and perform orthorectification processing to obtain a second DOM image with higher resolution;

[0018] Perform crater marking on the second DOM image with a vector surface to complete the construction of the crater matching landmark library.

[0019] Furthermore, the specific process of traversing and performing multi-level matching on the spatial relationship and the crater matching landmark library is as follows:

[0020] Construct a square approximate range centered on the first dead reckoning position, divide this approximate range into several small blocks, take the center of each small block as the approximate camera position for multi-level matching, and calculate the score of each small block, and accordingly determine the preliminary matching result;

[0021] Take the preliminary matching result as a new starting point, superimpose the driving distances in the x and y directions of the lunar rover to form new coordinates, construct a new approximate range centered on these coordinates, and loop the above steps until the lunar surface mission is completed.

[0022] Further, the specific process of the multi-level matching is as follows:

[0023] Tangent angle matching of a single crater: Calculate the tangent angle of each crater in the crater matching landmark library relative to the center of the small block, and compare it with the horizontal viewing angle of each crater in the spatial relationship respectively, and retain the crater landmarks whose difference satisfies the preset threshold;

[0024] Characteristic angle matching of adjacent craters: Calculate the characteristic angle between the centers of adjacent craters in the crater matching landmark library and the center of the small block, compare it with the horizontal viewing angle of the centers of adjacent craters in the spatial relationship, retain the crater landmarks whose difference satisfies the preset threshold, and screen out the series of landmarks that satisfy the end-to-end connection relationship from them;

[0025] Distance relationship matching of multiple craters: Calculate the distance relationship of each crater in the crater matching landmark library relative to the center of the small block, and compare it with the distance relationship between each crater relative to the camera position in the spatial relationship respectively, and retain the series of landmarks whose difference satisfies the preset threshold; Take the center of the small block corresponding to the series of landmarks with the longest distance relationship as the preliminary positioning result. If there are multiple, take the center of the small block corresponding to the series of landmarks with the smallest difference between the tangent angle and the characteristic angle as the preliminary positioning result.

[0026] Further, the horizontal viewing angle of each crater is calculated by the following formula:

[0027] A i =W i n

[0028] The horizontal viewing angle of the centers of adjacent craters is calculated by the following formula:

[0029] B j =W j n

[0030] Where, A i is the horizontal viewing angle of the i-th crater, n is the horizontal viewing angle of a single pixel of the lunar image, and W i is the width of the i-th crater in the lunar image; B j is the horizontal viewing angle of the centers of adjacent craters, and W j is the width between the centers of adjacent craters.

[0031] Further, the specific process of optimizing the preliminary positioning result is as follows:

[0032] Based on the triangle formed by the preliminary positioning result and the centers of two adjacent craters, construct an observation equation based on the cosine theorem;

[0033] Taking the feature angle as the observed value, where the feature angle is the interior angle of the aforementioned triangle with the vertex being the preliminary positioning result, and taking the crater coordinates as unknowns and as additional observed values, substituting the two into the aforementioned observation equation for least squares adjustment to optimize the preliminary positioning result.

[0034] Further, the expression of the observation equation is specifically as follows:

[0035]

[0036] Among them, B j is the feature angle, θ nm is the residual of B j r is the distance between the camera position p determined by the preliminary positioning result and the crater n, r np is the distance between the camera position p and the crater m, r mp is the distance between the camera position p and the crater m, r nm is the distance between the centers of adjacent craters n and m.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. The present invention proposes a lunar surface positioning method based on the matching of crater spatial relationships in cross-perspective images. First, a lunar surface image in the frontal view is obtained, crater features are extracted and spatial relationships are calculated, including the horizontal viewing angle of each crater, the horizontal viewing angle between the centers of adjacent craters, and the distance relationship between each crater relative to the camera position. At the same time, an orbital image in the top view is obtained, crater marking is performed, and a crater matching landmark library is constructed; then, the spatial relationships and the crater matching landmark library are traversed for multi-level matching to obtain a preliminary positioning result. The multi-level matching specifically includes the tangent angle matching of a single crater, the feature angle matching of adjacent craters, and the distance relationship matching of multiple craters; finally, based on the cosine theorem, the least squares adjustment is used to further optimize the preliminary positioning result; the above method realizes lunar surface positioning by constructing the corresponding matching between the angles and distance relationships of crater features in the cross-perspective lunar surface and orbital images relative to the camera position, and can effectively improve the lunar surface positioning accuracy.

[0039] 2. When the present invention traverses and multi-level matches the spatial relationships and the crater matching landmark library, first, a square approximate range is constructed with the first dead reckoning position as the center, the approximate range is divided into several small blocks, the center of each small block is taken as the approximate camera position for multi-level matching, and the score of each small block is calculated, and the preliminary matching result is determined accordingly; then, the preliminary matching result is used as a new starting point, the driving distances in the x and y directions of the lunar rover are superimposed to form a new coordinate, and a new approximate range is constructed with this coordinate as the center to repeat the above steps, which can eliminate the cumulative error caused by integrated navigation and ensure the lunar surface positioning accuracy.

[0040] 3. Based on the triangle formed by the preliminary positioning result and two adjacent meteorite craters, the present invention constructs an observation equation based on the cosine theorem, uses the characteristic angle as the observed value, takes the coordinates of the meteorite craters as unknowns and additional observed values, substitutes the two into the observation equation for least squares adjustment, and optimizes the preliminary positioning result, which enables the detection of possible gross errors and provides corrections for all coordinates. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a flowchart of the method of the present invention.

[0042] Among them, "crater" in the enlarged view of meteorite crater extraction represents a meteorite crater, and the number behind it represents the confidence score of meteorite crater detection and extraction; "A" in the enlarged view of calculating spatial relationship represents the horizontal viewing angle of the meteorite crater, and "B" represents the horizontal viewing angle between the centers of adjacent meteorite craters.

[0043] Figure 2 is a schematic diagram of the construction process of the meteorite crater matching landmark library.

[0044] Among them, (2a) is the DOM obtained after processing the LROC NAC image, (2b) is the DOM with a higher resolution of 0.11 m obtained using a series of Chang'e landing images, and (2c) is the finally constructed meteorite crater matching landmark library.

[0045] Figure 3 is a diagram of the cosine theorem model in the embodiment of the present invention.

[0046] Figure 4 is the meteorite crater feature matching and positioning result in the embodiment of the present invention.

[0047] Among them, the blue dots represent the dead reckoning method, the red dots represent the cross-site visual positioning method, the yellow dots represent the DOM matching positioning method, and the green dots represent the meteorite crater matching positioning method (i.e., the method proposed in the present invention). DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and gives detailed implementation manners and specific operation processes, but the protection scope of the present invention is not limited to the following embodiments.

[0049] Embodiment:

[0050] Taking the matching and positioning experiment of each position point of the Chang'e-3 lander and the Yutu rover as an example, this embodiment provides a lunar surface positioning method based on the matching of meteorite crater spatial relationships from cross-perspective images, as Figure 1 shown, including the following steps:

[0051] Step 1: Obtain the lunar surface image from the front view, use the YOLOv8 model to extract meteorite crater features, and calculate spatial relationships.

[0052] The goal of feature extraction is to extract three or more crater features from lunar surface images. If the number of available craters extracted from some single images is less than 3, or the craters are located at the image edge and are incompletely shown, then image stitching is required to obtain sufficient crater features. The specific process is as follows: Scale-Invariant Feature Transform (SIFT) is used for matching and stitching adjacent lunar surface images. Then, distance-weighted fusion is used to eliminate the image stitching line, and in the overlapping part, it gradually transitions from the previous image to the second image, and the pixel values in the overlapping area of the images are added together according to certain weights to synthesize a new image.

[0053] Through the horizontal viewing angle r and width W of the image, the horizontal viewing angle of a single pixel of the image can be calculated as Spatial relationships include the horizontal viewing angle of each crater, the horizontal viewing angle between the centers of adjacent craters, and the distance relationship between each crater relative to the camera position. Among them, the horizontal viewing angle A of the i-th crater i =W i n, W i is the width of the i-th crater in the lunar surface image; the horizontal viewing angle B between the centers of adjacent craters j =W j n, W j is the width between the centers of adjacent craters. The distance relationship range between each crater relative to the camera position is: p1>p2, where p1 and p2 are the distances from the two craters to the camera position respectively. Through A i 、B j and range, subsequent accurate matching can be achieved, thereby positioning the camera shooting position.

[0054] Step 2: Obtain the orbital image from the top-down view, perform crater marking, and construct a crater matching landmark library. The specific process is as follows:

[0055] First, obtain the LROC NAC image of the orbital area and perform orthorectification processing to obtain the first DOM image, as shown in (2a) of Figure 2 . Then, according to the range of the first DOM image, obtain the low-altitude landing images of Chang'e-3 and perform orthorectification processing to obtain a second DOM image with higher resolution, which covers the landing area and the driving area of Yutu-1, as shown in (2b) of Figure 2 . Finally, perform crater marking on the second DOM image with a vector surface to complete the construction of the crater matching landmark library, as shown in (2c) of Figure 2 .

[0056] Step 3: Traverse and perform multi-level matching on the spatial relationship and the crater matching landmark library to obtain a preliminary positioning result.

[0057] Dead reckoning integrated navigation can track the position relative to the starting point of each trip. Each relative position can be used as the prior information of the current position of the lunar rover. However, there will be cumulative errors in the integrated navigation, which are positively correlated with the distance of each trip. Therefore, in this embodiment, a square approximation range is constructed centered on the first dead reckoning position, as shown in Figure 1 As shown, the approximation range is divided into several small blocks. The center of each small block is taken as the approximate camera position for multi-level matching, and the score of each small block is calculated to determine the preliminary matching result. Then, the preliminary matching result is used as the new starting point, and the driving distances in the x and y directions of the lunar rover are superimposed to form new coordinates. A new approximation range is constructed centered on these coordinates, and the above steps are repeated until the lunar surface mission is completed. The above process can eliminate the cumulative errors caused by the integrated navigation and ensure the accuracy of the lunar surface positioning result.

[0058] The multi-level matching includes the tangent angle matching of a single crater, the characteristic angle matching of adjacent craters, and the distance relationship matching of multiple craters. The specific process is as follows:

[0059] Step 31, tangent angle matching of a single crater: Calculate the tangent angle of each crater in the crater matching landmark library relative to the center of the small block, and compare it with the horizontal viewing angle A of each crater in the spatial relationship i respectively, and retain the crater landmarks whose difference satisfies the preset threshold.

[0060] Step 32, characteristic angle matching of adjacent craters: Calculate the characteristic angle between the centers of adjacent craters in the crater matching landmark library and the center of the small block, and compare it with the horizontal viewing angle B between the centers of adjacent craters in the spatial relationship j respectively, retain the crater landmarks whose difference satisfies the preset threshold, and screen out the series of landmarks that satisfy the head-to-tail connection relationship from them.

[0061] Step 33, distance relationship matching of multiple craters: Calculate the distance relationship of each crater in the crater matching landmark library relative to the center of the small block, and compare it with the distance relationship range between each crater relative to the camera position in the spatial relationship respectively. Retain the series of landmarks whose difference satisfies the preset threshold. The center of the small block corresponding to the series of landmarks with the longest distance relationship is used as the preliminary positioning result; if there are multiple, the center of the small block corresponding to the series of landmarks with the smallest difference between the tangent angle and the characteristic angle is used as the preliminary positioning result.

[0062] Step 4, based on the cosine theorem, use the least squares adjustment to optimize the preliminary positioning result.

[0063] As shown in Figure 3 As shown, the preliminary positioning result (i.e., the camera position p) and the centers of two adjacent craters n and m can form a triangle. Based on the cosine theorem model, the following observation equation is constructed:

[0064]

[0065] Among them, B j is the feature included angle, and the feature included angle is the interior angle with the vertex of the preliminary positioning result in the aforementioned triangle, and θ nm is the residual of B j , r np is the distance between the camera position p determined by the preliminary positioning result and the crater n, and r mp is the distance between the camera position p and the crater m, and r nm is the distance between the centers of adjacent craters n and m.

[0066] Taking the feature included angle B j as the observed value, taking the crater coordinates as unknowns and as additional observed values, substituting the two into the aforementioned observation equation for least squares adjustment, realizing the best fit between the two, and further optimizing the preliminary positioning result, which can further improve the accuracy of the lunar surface positioning result.

[0067] To verify the effectiveness of the above method, this embodiment has completed the matching positioning experiment of each shooting position of the Yutu rover camera and the landing position of the lander. Figure 4 are the positioning results of four methods. Among them, the blue dots represent the dead reckoning method, the red dots represent the cross-site visual positioning method, the yellow dots represent the DOM matching positioning method, and the green dots represent the crater matching positioning method (i.e., the method proposed in the present invention). As Figure 4 shown in the enlarged partial view of, under the crater matching positioning method, the lander positioning result (green dot) is located at the prominent white part on the LROC NAC image. This is because the lander will cause disturbances during landing, including the displacement, compaction, and dust raising of lunar soil caused by the rocket jets ejected by the lander, making the surface features of the landing area become prominent in the image, verifying the correctness of the positioning result.

[0068] Both dead reckoning and cross-site visual positioning have cumulative errors. The nominal positioning accuracy of dead reckoning is 10% of the driving distance, and the positioning accuracy of cross-site visual positioning is about 4% of the driving distance. The DOM matching positioning result is better than the previous two. Table 1 shows the x, y coordinates and distance difference data of the positioning results of the five positions of the method of the present invention and the positioning results of the other three methods. Compared with dead reckoning, the average distance difference is 2.575 m, and as the driving distance increases, the distance difference will increase significantly, meeting the expectation of a large cumulative error; compared with cross-site visual positioning, the average distance difference is 0.653 m, and as the driving distance increases, the increasing trend of the distance difference is not obvious. This is because the driving distance of the Yutu rover is not long, about 100 m, and the cross-visual positioning method itself has a small cumulative error; compared with the DOM matching absolute positioning distance, the average distance difference is 0.813 m, reaching the sub-meter level.

[0069] Comparison of the Difference in the Positioning Results between the Method Proposed in Table 1 and the Existing Methods

[0070]

[0071] The present invention proposes a lunar surface matching and positioning method based on the angular and distance spatial relationships of lunar craters in lunar surface and orbital images from different perspectives. The effectiveness of this matching and positioning method is verified through the crater matching and positioning experiments of the positions of the Chang'e-3 lander and the shooting positions of the Yutu rover. The positioning result of the lander can be verified through the LROC NAC images. The positioning result of the Yutu rover has an average difference of 0.813 m from the currently known best DOM positioning result of the Chang'e-3 mission. Thus, it can be seen that the present invention can achieve high-precision lunar surface positioning, and it can be achieved through monocular lunar surface images, and is expected to be applied to future lunar surface remote navigation and positioning.

[0072] If the above method is implemented in the form of software functional units and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. And the aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0073] The above description of the embodiments is to facilitate the understanding and use of the invention by those of ordinary skill in the art. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative labor. Therefore, the present invention is not limited to the above embodiments, and the improvements and modifications made by those skilled in the art without departing from the scope of the present invention according to the disclosure of the present invention should be within the protection scope of the present invention.

Claims

1. A lunar surface positioning method based on spatial relationship matching of craters in cross-view images, characterized in that: The following steps are involved: Acquire a lunar surface image of a normal viewing angle, extract crater features and calculate spatial relationships, wherein the spatial relationships include a horizontal viewing angle of each crater, a horizontal viewing angle of the center of adjacent craters, and a distance relationship between craters relative to a camera position; Acquire orbital images from a bird's-eye view, mark craters, and build a crater matching landmark library; Performing multi-level matching on the spatial relationship and the crater matching landmark library to obtain a preliminary positioning result, wherein the multi-level matching includes tangent angle matching of a single crater, characteristic angle matching of adjacent craters, and distance relationship matching of multiple craters; Based on the law of cosines, the least squares adjustment is used to optimize the preliminary positioning result.

2. The lunar surface positioning method based on cross-view image crater spatial relationship matching according to claim 1 is characterized in that: The YOLOv8 model is used to extract crater features.

3. The lunar surface positioning method based on cross-view image crater spatial relationship matching according to claim 1, characterized in that: When the number of crater features extracted from a lunar image is less than a set threshold or is incomplete, the lunar image is matched and spliced ​​and then the crater features are extracted again.

4. The lunar surface positioning method based on cross-view image crater spatial relationship matching according to claim 3 is characterized in that: The specific process of matching and stitching lunar images is as follows: Adopt scale-invariant feature transformation to match and stitch adjacent lunar images; Eliminate image stitching lines using distance-weighted fusion.

5. The lunar surface positioning method based on cross-view image crater spatial relationship matching according to claim 1, characterized in that: The construction process of the crater matching landmark library is as follows: Acquire the LROC NAC image of the orbital area and perform orthorectification processing to obtain the first DOM image; Acquire a low-altitude landing image according to the range of the first DOM image and perform orthorectification processing to obtain a second DOM image with higher resolution; The craters are marked on the second DOM image with vector surfaces to complete the construction of the crater matching landmark library.

6. The lunar surface positioning method based on cross-view image crater spatial relationship matching according to claim 1, characterized in that: The specific process of traversing the spatial relationship and the crater matching landmark library for multi-level matching is as follows: A square approximate range is constructed with the first dead reckoning position as the center, and the approximate range is divided into several small blocks. The center of each small block is taken as the approximate camera position for multi-level matching, and the score of each small block is calculated to determine the preliminary matching result; Use the preliminary matching result as a new starting point, superimpose the travel distance of the lunar rover in the x and y directions to form a new coordinate, build a new approximate range with this coordinate as the center, and repeat the above steps until the lunar mission is completed.

7. The lunar surface positioning method based on cross-view image crater spatial relationship matching according to claim 6, characterized in that: The specific process of the multi-level matching is as follows: Tangent angle matching of a single crater: Calculate the tangent angle of each crater in the crater matching landmark library relative to the center of the small block, compare them with the horizontal viewing angle of each crater in the spatial relationship, and retain the crater landmarks whose difference meets the preset threshold; Matching characteristic angles of adjacent craters: Calculate characteristic angles between the centers of adjacent craters and the centers of small blocks in the crater matching landmark library, compare them with the horizontal viewing angles of the centers of adjacent craters in the spatial relationship, retain crater landmarks whose differences meet a preset threshold, and select a series of landmarks that meet the head-to-tail connection relationship; Distance relationship matching of multiple craters: Calculate the distance relationship between each crater in the crater matching landmark library and the center of the small block, and compare them with the distance relationship between each crater in the spatial relationship and the camera position, and retain the series of landmarks whose difference meets the preset threshold; take the center of the small block corresponding to the series of landmarks with the longest distance relationship as the preliminary positioning result. If there are multiple ones, take the center of the small block corresponding to the series of landmarks with the smallest difference between the tangent angle and the feature angle as the preliminary positioning result.

8. The lunar surface positioning method based on cross-view image crater spatial relationship matching according to claim 1, characterized in that: The horizontal viewing angle of each crater is calculated using the following formula: A i =W i n The horizontal viewing angles of adjacent crater centers are calculated using the following formula: B j =E j n Among them, A i is the horizontal viewing angle of the i-th crater, n is the horizontal viewing angle of a single pixel in the lunar image, and W i is the width of the i-th crater in the lunar image; B j is the horizontal viewing angle of the adjacent crater centers, W j is the width between the centers of adjacent craters.

9. The lunar surface positioning method based on cross-view image crater spatial relationship matching according to claim 1, characterized in that: The specific process of optimizing the preliminary positioning result is as follows: Based on the triangle formed by the preliminary positioning result and the centers of two adjacent craters, the observation equation is constructed based on the law of cosines. The characteristic angle is taken as the observation value, the characteristic angle is the inner angle of the aforementioned triangle whose vertex is the preliminary positioning result, and the crater coordinates are taken as unknown quantities and as additional observation values. The two are substituted into the observation equation for least squares adjustment to optimize the preliminary positioning result.

10. The lunar surface positioning method based on cross-view image crater spatial relationship matching according to claim 9, characterized in that: The expression of the observation equation is as follows: Among them, B j is the characteristic angle, θ nm For B j The residual, r np is the distance between the camera position p and the crater n determined by the preliminary positioning result, r mp is the distance between the camera position p and the crater m, r nm is the distance between the centers of adjacent craters n and m.

Images