A high-reliability star image recognition method based on improved triangle matching

Abstract

A high-reliability star map recognition method based on improved triangle matching belongs to the field of spacecraft navigation, guidance and control, and includes the following steps: according to the camera line of sight guidance quaternion and the camera field of view angle, the coverage star area is calculated, and the star-to-star angular distance between two stars is used as the feature to construct a navigation star database; extract the star points in the image, sort them according to the Euclidean distance from the center of the field of view, and use a specific star point selection order to construct an observation triangle; find a characteristic triangle matching the observation triangle in the navigation star database; add a test star for verification, and if successful, calculate the camera line of sight pointing quaternion, and if failed, enter the next round of matching process. The improved triangle star map recognition method proposed by the present invention improves the correctness of triangle matching and solves the problem of too long matching time when non-star target points are concentrated in the center of the field of view.

Description

Technical Field

[0001] The invention relates to a high-reliability star map recognition method based on improved triangle matching, and belongs to the technical field of spacecraft navigation, guidance and control. Background Art

[0002] Space situational awareness, as the basis for mastering the space situation and ensuring the safety of the space environment and the operation of spacecraft, has become a key area of ​​development for various space-faring countries. Space target detection and identification is the data source for space situational awareness and an important guarantee for obtaining future space advantages. It has become a priority development area. The U.S. Space Surveillance Network (SSN) is the world's most complete space target monitoring system. It can continuously discover, track, measure, identify, calculate orbits and catalog various space targets flying around the earth, including various spacecraft, terminal rockets, missiles, debris, meteorites, etc. of its own, friendly and enemy sides; monitor and extract environmental conditions related to space systems and missions, evaluate and predict the impact of the space environment on space systems and missions, and has the ability to distinguish attributes and intentions.

[0003] Space targets mainly include satellites, space stations and space debris. The core problem of space target capture is to use the characteristic differences between targets and natural celestial bodies to quickly and accurately identify space targets in the background of a large number of natural celestial bodies. The current mainstream space target detection and recognition methods mainly include recognition based on triangle matching and recognition based on target motion characteristics. For the triangle matching method, Guo Lei et al. changed the star point feature dimension from three dimensions to four dimensions, improved the star table search algorithm, and improved the target recognition success rate and operation efficiency. Liu Xianyi et al. constructed the radial feature quantity of star points through the geometric distribution of star points, and initially recognized the star map based on the radial feature quantity. The recognition results of the photographed star points were limited to several navigation stars, and the triangle star map recognition algorithm was used for recognition again, making the triangle star map recognition more targeted and improving the accuracy of recognition. For the motion feature method, Daqi Liu et al. proposed a new multi-target detection technology based on topological scanning to find GEO objects from a short sequence of optical images. It has high computational efficiency and is suitable for online processing. Taking advantage of the large field of view star map recognition, Yang Yuanzhao and others proposed a rapid capture algorithm for space targets under a complex starry sky background, achieving rapid and accurate capture of targets.

[0004] When the satellite is in the GEO orbit and the camera's visual axis is facing due east or due west, there will be more artificial satellites concentrated in the center of the field of view, and the image will be a non-stellar star point. There may be non-stellar star points in the observation triangle for triangle matching. If the existing triangle matching method is used to select three stars in sequence, the non-stellar target will always occupy one or more stars in the observation triangle, and the matching will still fail after multiple cycles; or the matching will be successful, but it will be an incorrect match. Summary of the invention

[0005] The technical problem solved by the present invention is: to overcome the deficiencies of the prior art, to provide a high-reliability star map recognition method based on improved triangle matching, and to solve the problem of long matching time when non-stellar target points are concentrated in the center of the field of view.

[0006] The technical solution of the present invention is: a high-reliability star map recognition method based on improved triangle matching, comprising:

[0007] (1) Calculate the coverage star area based on the camera's visual axis guidance quaternion and the camera's field of view angle, and build a navigation star database based on the star-to-star angular distance between two stars;

[0008] (2) Extract star points from the image, sort them by their Euclidean distance from the center of the field of view, and construct an observation triangle using a specific star point selection sequence;

[0009] (3) Find a characteristic triangle matching the observed triangle in the navigation star database;

[0010] (4) Add verification star verification. If successful, calculate the camera's visual axis pointing quaternion. If failed, enter the next round of matching process.

[0011] Furthermore, in step (2), the method for selecting an arrangement of observation triangles in order includes:

[0012] (a) In the first cycle, the first star number increases in sequence, the interval between the second star and the first star number is Gap12, and the interval between the third star and the second star number is Gap23, which is 1;

[0013] (b) In the second cycle, the first star number increases successively, Gap12 is 1, Gap23 is 2, that is, the gap between the third star and the second star number Gap23 increases by 1 after each cycle;

[0014] (c) When Gap23 increases to its maximum value, Gap12 increases to 2, and Gap23 increases from 1 again, and the process in (b) is repeated until all star point combinations are traversed.

[0015] Furthermore, in step (1), the navigation star database is constructed based on the SAO star catalog, and the number, right ascension, and declination information of the stars in the SAO star catalog are extracted according to the calculated coverage star area, and the projected star point coordinates of the stars in the image and the angular distances of all the projected star points in pairs are calculated.

[0016] Furthermore, in step (1), in order to avoid the influence of the edge distortion of the camera field of view, the field of view angle is selected according to the camera characteristics so that the distortion of all image points is within 1 pixel; or an internal orientation element test is performed to correct the distortion of the image points.

[0017] Further, in step (2), when calculating the coordinates of the star points, the centroid method with a threshold is adopted. The original image is subtracted from the background threshold, and the centroid of the pixel points in the original image that are greater than the threshold is calculated.

[0018] Further, in step (2), when constructing the observation triangle, the value of the diagonal distance is restricted, and only the star pair information is constructed for which the Euclidean distance between the star pairs corresponding to the angular distances on the image is less than 1 pixel.

[0019] Further, in step (3), for the three angular distances of the observation triangle, three star pairs are searched in the navigation star pair database whose differences from the angular distance values are less than a certain threshold. If the stars that make up these three star pairs are pairwise connected, that is, the three star pairs form a triangle, it is considered that the triangle matching is successful.

[0020] Further, in step (3), if multiple triangles are matched in step (3) and the verification star in step (4) is verified, as long as one triangle is verified correctly, it is considered that the current star map matching is correct, and the subsequent triangles are no longer verified.

[0021] Further, in step (4), the star points are numbered 1, 2, ……, n in the order of the Euclidean distance from the center of the field of view; n is the number of star points; if the numbers of the triangles successfully matched in step (3) are i, j, k, and i < j < k, then the number of the selected verification star is k + 1.

[0022] Further, in step (4), if the verification of the verification star fails, it is considered that the triangle Δ1 in step (3) is a mis-matched triangle, and thus three other stars are selected to continue the triangle matching.

[0023] The advantages of the present invention compared with the prior art are as follows:

[0024] (1) After the triangle matching is successful, a fourth star is added for hexagon verification, effectively eliminating the mis-matched triangles and ensuring the correctness of the star matching.

[0025] (2) When selecting the three stars of the observation triangle, instead of following the conventional method of increasing in sequence, the star points are first arranged in the order of the Euclidean distance from the center of the field of view from near to far, and then all the star points in the field of view are traversed as quickly as possible in the first round of loop, and the interval between two stars is increased by two in the subsequent loops; when there is a non-star target at the center of the field of view, this method can avoid the situation where the matching is still unsuccessful after multiple loops. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flow schematic diagram of the high-reliability star map recognition method based on improved triangle matching of the present invention;

[0027] Figure 2 Schematic diagram of the observation triangle with the test star added. DETAILED DESCRIPTION

[0028] In order to better understand the above technical scheme, the technical scheme of the present application is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical scheme of the present application, rather than limitations on the technical scheme of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0029] The following is a further detailed description of a high-reliability star map recognition method based on improved triangle matching provided by an embodiment of the present application in conjunction with the accompanying drawings. The specific implementation method may include (eg Figures 1-2 The steps are as follows:

[0030] (1) The coverage star area is calculated based on the camera line of sight guidance quaternion and the camera field of view angle, and the navigation star database is constructed using the star pair angular distance between two stars as the feature.

[0031] Furthermore, in one possible implementation, the navigation star database is constructed based on the SAO star catalog. The number, right ascension, declination and other information of the stars in the SAO star catalog are extracted according to the calculated coverage star area, and the projected star point coordinates of the stars in the image and the angular distances of all pairwise combinations of the projected star points are calculated.

[0032] In one possible implementation, in order to avoid the influence of the edge distortion of the camera field of view, it is necessary to select the field of view angle according to the camera characteristics so that the distortion of all image points is within 1 pixel; or perform an internal orientation element test to correct the distortion of the image points.

[0033] (2) Extract the star points in the image, sort them according to the Euclidean distance from the center of the field of view, and construct the observation triangle using a specific star point selection order.

[0034] In one possible implementation, the order of star point selection is arranged as follows:

[0035] (a) In the first cycle, the first star number increases in sequence, the interval between the second star and the first star number is Gap12, and the interval between the third star and the second star number is Gap23, which is 1;

[0036] (b) In the second cycle, the first star number increases successively, Gap12 is 1, Gap23 is 2, that is, the gap between the third star and the second star number Gap23 increases by 1 after each cycle;

[0037] (c) When Gap23 increases to the maximum value (depending on the total number of star points), Gap12 increases to 2, Gap23 then starts increasing from 1, and the process in (b) is repeated until all star point combinations are traversed.

[0038] Furthermore, the centroid method with a threshold is used to calculate the coordinates of the star points. The original image is subtracted from the background threshold, and the centroid of the pixel points in the original image greater than the threshold is calculated. The error in calculating the star point coordinates by this method can reach the sub-pixel level.

[0039] In a possible implementation, when constructing the observation triangle, since the angular distance needs to be constrained, only the star pair information with the angular distance within a certain threshold range is constructed.

[0040] (3) Search for the characteristic triangle that matches the observation triangle in the navigation star database;

[0041] Furthermore, for the three angular distances of the observation triangle, search for three star pairs in the navigation star pair database whose difference from the angular distance values is less than a certain threshold. If the stars that make up these three star pairs are pairwise connected, that is, the three star pairs form a triangle, it is considered that the triangle matching is successful.

[0042] In a possible implementation, if multiple triangles are matched in step (3) and the verification of the verification star in step (4) is performed, as long as one triangle is verified correctly, it is considered that the current star map matching is correct, and the subsequent triangles are no longer verified.

[0043] (4) Add the verification of the verification star. If successful, calculate the quaternion of the camera optical axis pointing; if failed, enter the next round of matching process.

[0044] Furthermore, the numbers of the star points sorted by the Euclidean distance from the center of the field of view are 1, 2, ……, n (n is the number of star points). If the numbers of the triangles successfully matched in step (3) are i, j, k, and i < j < k, then the number of the selected verification star is k + 1.

[0045] In a possible implementation, if the verification of the verification star fails, it is considered that the triangle Δ1 in step (3) is a mis-matched triangle, so other three stars are selected to continue the triangle matching, rather than selecting other verification stars to continue verifying the triangle Δ1.

[0046] In the solution provided in the embodiments of the present application, it specifically includes the following steps:

[0047] A. Convert the camera's visual axis guidance quaternion into its rotation matrix in the inertial space coordinate system, and calculate the right ascension and declination in the inertial system pointed by the visual axis; combine the camera's field of view and imaging capabilities to screen out stars in the sao star catalog with magnitudes below a specific value and within the camera's field of view; project these stars to the camera's focal plane and calculate the coordinates of the stars on the camera image; among these stars, use the star pair angular distance between two stars as a feature to build a navigation star pair database.

[0048] B. Extract star points from the star map captured by the camera and sort them from near to far according to the Euclidean distance from the center of the field of view. Select three stars in a specific order and construct an observation triangle for all selected triangles. The for loop pseudocode for the order of selecting three stars is as follows, where N is the total number of star points extracted from the star map, Gap12 is the number interval between the first and second stars, Gap23 is the number interval between the second and third stars, imagepoint is the star point array, indexStart is the number of the first star, indexStart+Gap12 is the number of the second star, and indexStart+Gap12+Gap23 is the number of the third star:

[0049]

[0050] C. For each of the three sides of the observed triangle, the angular distance C 12 , C 23 , C 13 First, search for the satellite pair corresponding to C in the navigation star pair database. 12 , C 23 Corresponding to equal angular distances D 12 =C 12 , D 23 =

[0051] C 23 , find D with common star points 12 , D 23 , which now constitutes D 12 The two stars are Index1 and Index2, which constitute D 23 The two stars are Index2 and Index3; then search for the corresponding stars in the navigation star database. 13 Equal angular distance D 13 =C 13 If D 13 The two stars of D are Index1 and Index3, then 12 , D 23 , D 13 is the angular distance of the three sides of the characteristic triangle corresponding to the observation triangle, and D 12 =C 12 , D 23=C 23 , D 13 =C 13 .

[0052] D. When there is a target among the three stars of the observation triangle, if there is still a feature triangle with equal angular distances to the observation triangle, this is an incorrect matching result, so a fourth star needs to be added for verification. The angular distances between the fourth star and the three stars of the observation triangle are C 14 , C 24 , C 34 First, search for the star index Index1 in the navigation star pair database, which is the same as C 14 Equal angular distance D 14 =C 14 , which now constitutes D 14 The two stars are Index1 and Index4; calculate the angular distance D between Index2 and Index4 24 , the angular distance D between Index3 and Index4 34 , if D 24 =C 24 , D 34 =C 34 , then the three stars of the observation triangle and the test stars are combined into a hexagon (such as Figure 1 As shown in Figure 1, the hexagon formed by the navigation stars Index1 to Index4 is correctly matched. At this point, the triangle is successfully identified and the subsequent camera visual axis calculation process is entered. If no triangle that satisfies D is found, 14 =C 14 angular distance, or D 24 ≠C 24 , D 34 ≠C 34 , then the current round of triangle matching fails, and the feature triangle is reselected to enter the next round of matching process.

[0053] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

[0054] The contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.

Claims

1. A high-reliability star map recognition method based on improved triangle matching, characterized in that: Including: (1) Calculate the covered star area according to the camera optical axis guidance quaternion and the camera field of view angle, construct a navigation star database with the star diagonal distance between two stars as the feature; (2) Extract the star points in the image, sort them according to the Euclidean distance from the center of the field of view, and construct an observation triangle using a specific star point selection order; (3) Search for a feature triangle in the navigation star database that matches the observation triangle; (4) Add a verification star for verification. If successful, calculate the camera optical axis pointing quaternion. If failed, enter the next round of matching process; In (2), the arrangement method for constructing the observation triangle with the selection order is: (a) In the first round of loop, the numbers of the first star increase sequentially, the number interval Gap12 between the second star and the first star is 1, and the number interval Gap23 between the third star and the second star is 1; (b) In the second round of loop, the numbers of the first star increase sequentially, Gap12 is 1, Gap23 is 2, that is, the number interval Gap23 between the third star and the second star increases by 1 after each round of loop; (c) When Gap23 increases to the maximum value, Gap12 increases to 2, Gap23 starts to increase from 1 again, and repeat the process of (b) until all star point combinations are traversed.

2. A high-reliability star map recognition method based on improved triangle matching according to claim 1, characterized in that: In step (1), the navigation star database is constructed based on the SAO star catalog. Extract the number, right ascension, and declination information of the stars in the SAO star catalog according to the calculated covered star area, calculate the projected star point coordinates of the stars in the image, and the angular distance between every two combinations of all projected star points.

3. The high-reliability star map recognition method based on improved triangle matching according to claim 1, characterized in that: In step (1), in order to avoid the influence of the distortion at the edge of the camera field of view, select the field of view angle according to the camera characteristics so that the distortion of all image points is within 1 pixel; or perform an interior orientation element test and correct the distortion of the image points.

4. The high-reliability star map recognition method based on improved triangle matching according to claim 1, characterized in that: In step (2), the centroid method with a threshold is used to calculate the coordinates of the extracted star points. Subtract the background threshold from the original image, and calculate the centroid of the pixel points in the original image that are greater than the threshold.

5. The high-reliability star map recognition method based on improved triangle matching according to claim 1, characterized in that: In step (2), when constructing the observation triangle, the value of the angular distance is constrained, and only construct the star pair information where the Euclidean distance between the star pairs corresponding to the angular distance on the image is less than 1 pixel.

6. The high-reliability star map recognition method based on improved triangle matching according to claim 1, characterized in that: In step (3), for the three angular distances of the observation triangle, search for three star pairs in the navigation star pair database whose difference from the angular distance value is less than a certain threshold. If the stars that make up these three star pairs are pairwise connected, that is, the three star pairs form a triangle, it is considered that the triangle matching is successful.

7. The high-reliability star map recognition method based on improved triangle matching according to claim 1, characterized in that: In step (3), if multiple triangles are matched in step (3) and the verification star verification in step (4) is performed, as long as one triangle is verified correctly, it is considered that the current star map matching is correct, and the subsequent triangles will no longer be verified.

8. The high-reliability star map recognition method based on improved triangle matching according to claim 1, characterized in that: In step (4), the numbers of the star points after being sorted according to the Euclidean distance from the center of the field of view are 1, 2,..., n; n is the number of star points; if the numbers of the triangles successfully matched in step (3) are i, j, k, and i < j < k, then the number of the selected verification star is k + 1.

9. The high-reliability star map recognition method based on improved triangle matching according to claim 1, characterized in that: In step (4), if the verification star verification fails, it is considered that the triangle Δ1 in step (3) is a mis-matched triangle, and thus select other three stars to continue the triangle matching.

Images