A multi-endpoint star map recognition method based on optimal path

Abstract

The present invention discloses a multi-endpoint star map recognition method based on an optimal path. The method comprises the following steps: S1. constructing a navigation star library and, based on the concept of a minimum weight spanning tree, constructing a navigation feature library containing navigation star characteristic patterns; S2. processing an observed star map and, based on the concept of a minimum weight spanning tree, generating characteristic patterns of observed stars; and S3. matching and identifying the navigation star characteristic patterns with the observed star characteristic patterns based on the bidirectional Hausdorff distance. The present invention integrates graph theory knowledge and the optimal path problem with star map recognition methods, utilizing the concept of a minimum weight spanning tree to extract star point characteristic patterns. Due to the unique distribution of star points, the characteristic pattern of each star point is ensured to be unique, thereby increasing the accuracy of star map recognition.

Description

Technical Field

[0001] The present invention relates to the technical field of celestial navigation, and in particular to a multi-endpoint star map recognition method based on an optimal path. Background Art

[0002] Celestial navigation primarily determines the attitude and position of an aircraft by measuring the position vectors of natural celestial bodies relative to the aircraft, thereby achieving positioning and navigation. In navigation systems, the accuracy of attitude measurement is a crucial indicator of system performance. Star sensors, as high-precision attitude measurement components, are widely used in celestial navigation.

[0003] A star sensor uses an image sensor to capture a real star map. It then compares the characteristics of the captured star map with those of navigation stars for matching and identification. It then uses the image information and information about the identified stars to calculate the three-axis attitude. The star sensor's attitude measurement process shows that the quality of the star map recognition algorithm directly determines the star sensor's attitude performance, making star map recognition a key component of attitude determination.

[0004] Currently, star pattern recognition algorithms are broadly categorized into three types based on the method used to extract observed star features: subgraph isomorphism algorithms, pattern recognition algorithms, and artificial intelligence algorithms. Subgraph isomorphism algorithms treat the star pattern recognition problem as a special case of the subgraph isomorphism problem. These algorithms primarily use the positional relationships between star points to construct triangular or polygonal feature patterns, which are then matched against a database of navigation star features for identification. Currently, the most widely used, mature, and successfully operational star pattern recognition algorithm is the triangle algorithm. However, due to the low dimensionality of the features used in this algorithm, redundant matches and mismatches are prone to occur, resulting in low recognition rates and a slow recognition rate. Pattern recognition algorithms construct a unique feature for each star, known as a "star pattern." This feature is the geometric distribution characteristics of other stars within a certain range of the target star, serving as the star's characteristic pattern to distinguish it from other stars. Typical pattern recognition algorithms include grid algorithms, singular value decomposition, and star pattern recognition based on radial and circumferential distribution features. Artificial intelligence algorithms primarily utilize neural networks and intelligent algorithms to identify the geometric distribution characteristics surrounding star points. However, it usually requires the use of a large-scale parallel architecture to implement, and the recognition accuracy is related to the training process. Therefore, it has high hardware requirements and is not easy to implement. Summary of the Invention

[0005] The present invention aims to propose a multi-endpoint star pattern recognition method based on an optimal path. By solving the optimal path problem, a characteristic pattern of stars is constructed and matched using the Hausdorff distance based on this characteristic pattern. Once the primary star is identified, the path endpoint information is used to identify other stars within the primary star's field, thereby improving the recognition rate. Furthermore, this method requires less space in the navigation star library.

[0006] Technical solution: A multi-endpoint star map recognition method based on an optimal path, which includes the following steps:

[0007] S1. Construct a navigation star library, and construct a navigation feature library containing navigation star feature patterns based on the idea of ​​minimum weight spanning tree;

[0008] S2. Process the observed star map and generate characteristic patterns of the observed stars based on the idea of ​​minimum weight spanning tree;

[0009] S3. Based on the bidirectional Hausdorff distance, the characteristic pattern of the navigation star and the characteristic pattern of the observation star are matched and identified.

[0010] Preferably, the navigation star library in S1 includes the navigation star number, navigation star, right ascension, declination and direction vector of the navigation star; the direction vector V_star of the navigation star in the celestial coordinate system is:

[0011] Among them, α is the right ascension of the navigation star, and δ is the declination of the navigation star.

[0012] Preferably, the construction of the navigation feature library in S1 includes the following steps:

[0013] S1-2-1. Select any navigation star S as the primary star, record the numbers of the other companion stars within its neighborhood radius R, and take the neighborhood radius R as one-third of the camera field of view, that is: R = FOV / 3, where FOV represents the camera field of view; and calculate the angular distance d between the primary star and the remaining companion stars. si Angular distance d between the preceding and following companion stars ij :

[0014]

[0015]

[0016] Among them, V s represents the direction vector of the primary star S, V i represents the direction vector of the companion star i, V j represents the direction vector of the companion star j

[0017] S1-2-2. Based on the positional relationships between the primary and companion stars, and between the companion stars, construct an undirected graph G = (V, E): the numbers of the primary and companion stars serve as the vertices of the graph; the connections between the primary and companion stars, and the preceding and following companion stars serve as the edges (u, v); the angular distances between the primary and companion stars, and the preceding and following companion stars serve as the weights w(u, v);

[0018] S1-2-3. Use the idea of ​​minimum weight spanning tree to construct the characteristic pattern of the main star:

[0019] ① Select the main star as the starting point of the minimum weight spanning tree and add it to the visited node set. The remaining nodes belong to the unvisited node set; select the edges connected to the main star and add them to the possible minimum weight edge set;

[0020] ② Select the edge with the minimum weight connected to the main star, add the other endpoint of the edge to the visited node set; and update the possible minimum weight edge set: the set contains all edges whose starting point is a visited node and the end point is an unvisited node;

[0021] ③ Select the edge with the smallest weight in the minimum possible weight edge set and add the new endpoint to the visited node set; at the same time, update the possible minimum weight edge set;

[0022] ④ Repeat the above steps until all endpoints in the undirected graph are in the visited node set, that is, the primary star and all companion stars are traversed;

[0023] The asterisks in the visited node set are added in the order of their addition to form feature pattern A; the minimum weights added before and after are added in the order of their addition to form feature pattern B; and the sum of the minimum weights is used as feature pattern C;

[0024] At this point, the characteristic pattern of the main star is obtained;

[0025] S1-2-4: Use any navigation star S as the primary star and repeat S1-1 to S1-3 to obtain the navigation feature library.

[0026] Preferably, the processing steps of the observed star map in S2 are:

[0027] S2-1. After the observed star map is segmented, the star point coordinates (X, Y) are extracted using the weighted centroid method with a threshold.

[0028] S2-2. Calculate the direction vector V_pic of the star point in the camera coordinate system:

[0029]

[0030] Where X and Y represent the horizontal and vertical coordinates of the star point on the image plane, respectively; xo and yo represent the horizontal and vertical coordinates of the principal point on the image plane; f represents the focal length of the camera; dx and dy represent the pixel size of the camera in the x and y directions;

[0031] S2-3. Calculate the distance d between the star point pic on the image plane and the principal point o on the image plane pic-o :

[0032]

[0033] They are numbered in sequence according to their distance, with the closest star being designated as star number one;

[0034] S2-4, using the method and steps for constructing a characteristic pattern of the navigation star, construct a corresponding characteristic pattern for the observation star;

[0035] S2-5. Take any observed star pic as the primary star and repeat S2-1 to S2-4 to obtain the characteristic pattern of each observed star.

[0036] Preferably, the specific steps of matching and identifying in S3 are:

[0037] S3-1. Select the satellite No. 1 closest to the principal point of the image plane as the main observation star, and perform matching and identification based on characteristic patterns B and C:

[0038] ① Use the characteristic pattern C, i.e. the weight sum, for preliminary screening: retain the characteristic pattern C of the navigation star S s In the observed star pic characteristic mode C pic The navigation star within the range of radius R belongs to (C pic -R,C pic +R) range of the navigation star;

[0039] ②Use the characteristic pattern B, that is, add the edge of the minimum weight spanning tree before and after to further screen the observed main star: pic , and the characteristic mode B of the navigation star S s Considered as two ordered point sets, the two-way Hausdorff distance between each navigation star and the observed primary star is constructed, and the navigation star with a distance less than a certain threshold is considered to be the observed primary star;

[0040] S3-2. Determine the primary star to be observed based on characteristic patterns A and B, and determine the star numbers of the remaining companion stars in the neighborhood of the primary star to be observed:

[0041] ① The characteristic mode B of the navigation star S determined above s The characteristic pattern B of the observed primary star pic picCompare: If the lengths of the two point sets are consistent and the difference between the corresponding elements of the two point sets is less than a certain threshold, then the observed main star is considered to be the navigation star;

[0042] ② Further determine the serial numbers of the remaining companion stars based on the characteristic pattern A: s and the characteristic mode A of the observed primary star pic pic The elements in correspond one to one, and the serial numbers of the remaining companion stars are obtained;

[0043] S3-3. If the identification of the first star fails, the identification of the nth star is performed, and steps S3-1 and S3-2 are repeated.

[0044] Preferably, the bidirectional Hausdorff distance formula in S3 is:

[0045] H(A,B)=max(h(A,B),h(B,A))

[0046]

[0047] Where A is a finite point set constructed based on the navigation stars, B is a finite point set constructed based on the observation stars, a is an element in point set A, b is an element in point set B, h(A, B) is the one-way Hausdorff distance from A to B, h(B, A) is the one-way Hausdorff distance from B to A, and H(A, B) is the two-way Hausdorff distance between A and B.

[0048] Preferably, it further includes step S4, verifying the recognition result using unused angular distance information.

[0049] Preferably, step S4 is specifically as follows: in the recognition process, N stars are successfully recognized using M groups of angular distance information, where M≤2N; the star numbers num of the N stars obtained by recognition are i (i=1~N), find the corresponding direction vector in the navigation star library based on the star number Since not all angular distance information is used in the recognition process, the unused angular distance d can be used pq-pic The corresponding pair of asterisks: x, y (x = num p , y = num q ), find the direction vector V corresponding to x, y x , V y , calculate the angular distance d of the star pair in the celestial coordinate system xy-s ,in Combine this value with the unused angular distance d in the camera coordinate system. pq-pic For comparison, if |d xy-s -d pq-pic|<ε, it means that the star pair is correctly identified, and ε is the set threshold.

[0050] Beneficial effects of the present invention

[0051] 1. We integrate graph theory and optimal path problems with star map recognition methods, using the concept of minimum weight spanning trees to extract star point feature patterns. Due to the unique distribution of star points, we ensure that the feature pattern of each star point is unique, thereby increasing the accuracy of star map recognition.

[0052] 2. The storage capacity of the navigation star library and navigation feature library is small.

[0053] 3. Due to the orderliness of the characteristic patterns, the bidirectional Hausdorff distance is used to match and identify the observed stars and the navigation stars.

[0054] 4. While identifying the main star, the companion stars in the main star's neighborhood can also be identified, which improves the speed of star map recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 Flow chart of the method of the present invention.

[0056] Figure 2 Schematic diagram of the steps to construct a minimum weight spanning tree for the navigation star.

[0057] Figure 3 For observing star maps.

[0058] Figure 4 for Figure 3 Schematic diagram in reverse color and star point numbers arranged in sequence.

[0059] Figure 5 is the undirected graph of satellite No. 1.

[0060] Figure 6 Spanning tree for minimum weight of star number one.

[0061] Figure 7 This is a schematic diagram of the multi-endpoint identification of the companion star of Star No. 1. DETAILED DESCRIPTION

[0062] In order to more clearly describe the principles, technical solutions and advantages of the present application, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0063] Specifically, if Figure 1 As shown, a multi-endpoint star map recognition method based on an optimal path includes the following technical steps:

[0064] S1. Build navigation star library and navigation feature library

[0065] S1-1. First, construct a navigation star library, which includes the navigation star number, navigation star class, navigation star right ascension, declination and direction vector.

[0066] The Tycho II star catalog was used as the original star catalog to extract the magnitude, right ascension, and declination information of the navigation stars. Based on the original star catalog, navigation stars with a magnitude below 6 were selected according to the lens observable magnitude threshold to reduce the storage capacity of the navigation star library. The selected navigation stars were numbered in sequence. The direction vector V_star of the navigation star in the celestial coordinate system was generated based on the right ascension α and declination δ of the navigation star:

[0067]

[0068] S1-2. Secondly, construct a navigation feature library. The navigation feature library contains the feature pattern A, feature pattern B, and feature pattern C of each star.

[0069] S1-2-1. Based on the camera's field of view (FOV), select one-third of the field of view as the neighborhood range R of the navigation star, i.e., R = FOV / 3. This selection principle ensures that the companion star of the navigation star can appear in the field of view, which improves the recognition accuracy to a certain extent.

[0070] Select any navigation star S as the primary star and record the numbers of the other companion stars in its neighborhood R. Calculate the angular distance d between the primary star and the other companion stars. si Angular distance d between the preceding and following companion stars ij .

[0071]

[0072]

[0073] Among them, V s represents the direction vector of the primary star S, V i represents the direction vector of the companion star i, V j represents the direction vector of the companion star j.

[0074] S1-2-2. Then, according to the positional relationship between the primary star and the companion star, and between the companion stars, a characteristic pattern of the primary star based on the minimum weight spanning tree concept is constructed.

[0075] Based on graph theory, and the positional relationships between the primary and companion stars, and between companion stars, an undirected graph G = (V, E) is constructed. The elements in set V are the vertices of the undirected graph, with the numbers of the primary and companion stars serving as vertices. The elements in set E are the edges of the undirected graph, with the edges connecting the primary and companion stars, and the preceding and following companion stars, serving as edges (u, v), and the angular distances between the primary and companion stars, and the preceding and following companion stars, serving as weights w(u, v). This constructs an undirected graph of the primary star.

[0076] S1-2-3, see Figure 2 The six steps to construct a minimum weight spanning tree for a navigation star are as follows: Based on the optimal path problem and the characteristics of the undirected graph of the primary star, the concept of a minimum weight spanning tree is used to construct the characteristic pattern of the primary star. The minimum weight spanning tree contains all the nodes of the original undirected graph, and the sum of the weights of the connecting paths is minimized. The steps are as follows:

[0077] ① Select the main star as the starting point of the minimum weight spanning tree and add it to the visited node set. The remaining nodes belong to the unvisited node set. Select the edges connected to the main star and add them to the possible minimum weight edge set.

[0078] ② Select the edge with the smallest weight connected to the main star, add the other endpoint of the edge to the visited node set; and update the possible minimum weight edge set: the set contains all edges whose starting points are visited nodes and whose end points are unvisited nodes.

[0079] ③ Select the edge with the smallest weight in the minimum possible weight edge set and add the new endpoint to the visited node set; at the same time, update the possible minimum weight edge set.

[0080] ④ Repeat the above steps until all endpoints in the undirected graph are in the visited node set, that is, the primary star and all companion stars are traversed.

[0081] The endpoint numbers of the visited nodes except the main star (the main star is the default starting point) are added in the order of addition to form feature pattern A; the minimum weights of the previous and next nodes are added in the order of addition to form feature pattern B, and the sum of the minimum weights is used as feature pattern C. Feature patterns A and B are both ordered sets.

[0082] Since the distribution of companion stars of each navigation star is unique, and the weight of each edge in the undirected graph of the primary star is different, it can be guaranteed that the characteristic pattern of each navigation star is unique.

[0083] S2. Process the observed star map and generate characteristic patterns of the observed stars based on the minimum weight spanning tree concept.

[0084] S2-1. After the observed star map is segmented, the star point coordinates (X, Y) are extracted using the weighted centroid method with a threshold.

[0085] S2-2. According to the relationship between the image plane coordinate system and the camera coordinate system, the direction vector V_pic of the star point in the camera coordinate system can be calculated:

[0086]

[0087] Where X and Y represent the horizontal and vertical coordinates of the star point on the image plane, respectively; xo and yo represent the horizontal and vertical coordinates of the principal point on the image plane; f represents the focal length of the camera; dx and dy represent the pixel size of the camera in the x and y directions;

[0088] Calculate the distances between N star points on the image plane and the principal point of the image plane, and number them in sequence according to the distance, with the closest star point being designated as star number one.

[0089] See also Figure 4 for Figure 3 A schematic diagram showing the inverted color of the observed star map, with each star point numbered according to its distance from the principal point.

[0090] S2-3, select the nth star as the main star, construct a characteristic pattern for it, calculate the angular distance between the main star and other star points, and select the companion star within the range of the main star's field radius R, and record the star point number i and the corresponding angular distance d within this range si_pic Next, calculate the angular distance d between the leading and trailing companion stars. ij_pic Based on the positional relationships between the primary star and its companions, and between the preceding and succeeding companions, an undirected graph of the primary star is constructed. Furthermore, the characteristic patterns A, B, and C of the primary star are constructed using the concept of a minimum weight spanning tree.

[0091] See also Figure 5 For the Star Figure 1 The undirected graph constructed by the star, Figure 6 Schematic diagram of generating the minimum spanning tree for satellite No. 1.

[0092] S3. Match and identify the navigation star feature pattern and the image plane star point feature pattern based on the bidirectional Hausdorff distance.

[0093] Bidirectional Hausdorff distance is a method to measure the similarity between two point sets. The Hausdorff distance can be understood as the maximum value of the shortest distance from a point in one point set to another point set. The specific formula is as follows:

[0094] H(A,B)=max(h(a,b),h(b,a))

[0095]

[0096] Where A is a finite point set constructed based on the navigation stars, B is a finite point set constructed based on the observation stars, a is an element in point set A, b is an element in point set B, h(A,B) is the one-way Hausdorff distance from A to B, h(B,A) is the one-way Hausdorff distance from B to A, and H(A,B) is the two-way Hausdorff distance between A and B.

[0097] S3-1. First, select the No. 1 star closest to the principal point of the image plane as the main star for observation. Match and identify it based on the characteristic pattern B and characteristic pattern C:

[0098] ① Use the characteristic pattern C, i.e. the weight sum, to perform preliminary screening and retain the characteristic pattern C of the navigation star s In (C pic -R,C pic +R) of the navigation star.

[0099] ②Use the characteristic pattern B, that is, the edge of the minimum weight spanning tree added before and after, to further screen the main star to be observed. pic , and the characteristic mode B of the navigation star s The two-way Hausdorff distance between each navigation star and the observed primary star is constructed as two ordered point sets, and the navigation star with a distance less than a certain threshold is considered to be the observed primary star.

[0100] S3-2. Further, the observed primary star is determined based on the characteristic pattern A and the characteristic pattern B, and the star numbers of the remaining companion stars in the neighborhood of the observed primary star are determined.

[0101] ① Since the characteristic pattern B is an ordered point set, the characteristic pattern B of the navigation star determined above is s The characteristic pattern B of the observed primary star pic Compare: If the lengths of the two point sets are consistent and the difference between the corresponding elements of the two point sets is less than a certain threshold, then the observed main star is considered to be the navigation star.

[0102] ②At the same time, the serial number of the observed main star can be further determined according to the characteristic pattern A: the characteristic pattern A of the navigation star s and the characteristic mode A of the observed primary star pic The elements in correspond one to one, and the serial numbers of the remaining companion stars are obtained.

[0103] See also Figure 7 Schematic diagram of multi-endpoint identification for observing the primary star's companion star.

[0104] S3-3. If the identification of the first star fails, the identification of the nth star is performed, and steps S3-1 and S3-2 are repeated.

[0105] S4. Use the unused angular distance information to verify the recognition result.

[0106] In the recognition process, N stars are successfully identified using M sets of angular distance information, where M≤2N; the star numbers of the N stars obtained by recognition are i (i=1~N), find the corresponding direction vector in the navigation star library based on the star number Since not all angular distance information is used in the recognition process, the unused angular distance d can be used pq-pic The corresponding pair of asterisks: x, y (x = num p , y = num q ), find the direction vector V corresponding to x, y x , V y , calculate the angular distance d of the star pair in the celestial coordinate system xy-s ,in Combine this value with the unused angular distance d in the camera coordinate system. pq-pic For comparison, if |d xy-s -d pq-pic |<ε, it means that the star pair is correctly identified, and ε is the set threshold.

[0107] Those skilled in the art should understand that the present invention is not limited to the above-mentioned embodiments. The above-mentioned embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, and these changes and improvements shall fall within the scope of the present invention to be protected.

Claims

1. A multi-endpoint star map recognition method based on optimal path, characterized by It includes the following steps: S1. Constructing a navigation star library and constructing a navigation feature library containing navigation star feature patterns based on the idea of ​​minimum weight spanning tree. The construction of the navigation feature library specifically includes the following steps: S1-2-1. Select any navigation star S as the primary star, record the numbers of the other companion stars within its neighborhood radius R, and take the neighborhood radius R as one-third of the camera field of view, that is: R = FOV / 3, where FOV represents the camera field of view; and calculate the angular distance d between the primary star and the remaining companion stars. si Angular distance d between the preceding and following companion stars ij ; S1-2-2. Based on the positional relationships between the primary and companion stars, and between the companion stars, construct an undirected graph G = (V, E): the numbers of the primary and companion stars serve as the vertices of the undirected graph; the connections between the primary and companion stars, and the preceding and following companion stars serve as the edges (u, v); the angular distances between the primary and companion stars, and the preceding and following companion stars serve as the weights w(u, v); S1-2-3. Use the idea of ​​minimum weight spanning tree to construct the characteristic pattern of the main star: ① Select the main star as the starting point of the minimum weight spanning tree and add it to the visited node set. The remaining nodes belong to the unvisited node set; select the edges connected to the main star and add them to the possible minimum weight edge set; ② Select the edge with the minimum weight connected to the main star, add the other endpoint of the edge to the visited node set; and update the possible minimum weight edge set: the set contains all edges whose starting point is a visited node and the end point is an unvisited node; ③ Select the edge with the smallest weight in the minimum possible weight edge set and add the new endpoint to the visited node set; at the same time, update the possible minimum weight edge set; ④ Repeat the above steps until all endpoints in the undirected graph are in the visited node set, that is, the primary star and all companion stars are traversed; The asterisks in the visited node set are added in the order of their addition to form feature pattern A; the minimum weights added before and after are added in the order of their addition to form feature pattern B; and the sum of the minimum weights is used as feature pattern C; At this point, the characteristic pattern of the main star is obtained; S1-2-4, take any navigation star S as the primary star, repeat S1-1 to S1-3 to obtain the navigation feature library; S2. Process the observed star map and generate characteristic patterns of the observed stars based on the idea of ​​minimum weight spanning tree; S3. Based on the bidirectional Hausdorff distance, the characteristic pattern of the navigation star and the characteristic pattern of the observation star are matched and identified.

2. The method according to claim 1, characterized in that The navigation star library described in S1 includes the navigation star number, navigation star, right ascension, declination and direction vector of the navigation star; the direction vector V_star of the navigation star in the celestial coordinate system is: Among them, α is the right ascension of the navigation star, and δ is the declination of the navigation star.

3. The method according to claim 1, characterized in that In S1, the angular distance d between the primary star and the remaining companion stars is calculated by the following formula: si Angular distance d between the preceding and following companion stars ij : Among them, V s represents the direction vector of the primary star S, V i represents the direction vector of the companion star i, V j represents the direction vector of the companion star j.

4. The method according to claim 1, characterized in that The processing steps of the observed star map described in S2 are: S2-1. After the observed star map is segmented, the star point coordinates (X, Y) are extracted using the weighted centroid method with a threshold. S2-2. Calculate the direction vector V_pic of the star point in the camera coordinate system: Where X and Y represent the horizontal and vertical coordinates of the star point on the image plane, respectively; xo and yo represent the horizontal and vertical coordinates of the principal point on the image plane; f represents the focal length of the camera; dx and dy represent the pixel size of the camera in the x and y directions; S2-3. Calculate the distance d between the star point pic on the image plane and the principal point o on the image plane pic-o : They are numbered in sequence according to their distance, with the closest star being designated as star number one; S2-4, using the method and steps for constructing a characteristic pattern of the navigation star, construct a corresponding characteristic pattern for the observation star; S2-5. Take any observed star pic as the primary star and repeat S2-1 to S2-4 to obtain the characteristic pattern of each observed star.

5. The method according to claim 3, characterized in that The specific steps of matching and identification in S3 are: S3-1. Select the satellite No. 1 closest to the principal point of the image plane as the main observation star, and perform matching and identification based on characteristic patterns B and C: ① Use the characteristic pattern C, i.e. the weight sum, for preliminary screening: retain the characteristic pattern C of the navigation star S s In the observed star pic characteristic mode C pic The navigation star within the range of radius R belongs to (C pic -R,C pic +R) range of the navigation star; ②Use the characteristic pattern B, that is, add the edge of the minimum weight spanning tree before and after to further screen the observed main star: pic , and the characteristic mode B of the navigation star S s Considered as two ordered point sets, the two-way Hausdorff distance between each navigation star and the observed primary star is constructed, and the navigation star with a distance less than a certain threshold is considered to be the observed primary star; S3-2. Determine the primary star to be observed based on characteristic patterns A and B, and determine the star numbers of the remaining companion stars in the neighborhood of the primary star to be observed: ① The characteristic mode B of the navigation star S determined above s The characteristic pattern B of the observed primary star pic pic Compare: If the lengths of the two point sets are consistent and the difference between the corresponding elements of the two point sets is less than a certain threshold, then the observed main star is considered to be the navigation star; ② Further determine the serial numbers of the remaining companion stars based on the characteristic pattern A: s and the characteristic mode A of the observed primary star pic pic The elements in correspond one to one, and the serial numbers of the remaining companion stars are obtained; S3-3. If the identification of the first star fails, the identification of the nth star is performed, and steps S3-1 and S3-2 are repeated.

6. The method according to claim 1, characterized in that The bidirectional Hausdorff distance formula described in S3 is: H(A,B)=max(h(A,B),h(B,A)) Where A is a finite point set constructed based on the navigation stars, B is a finite point set constructed based on the observation stars, a is an element in point set A, b is an element in point set B, h(A,B) is the one-way Hausdorff distance from A to B, h(B,A) is the one-way Hausdorff distance from B to A, and H(A,B) is the two-way Hausdorff distance between A and B.

7. The method according to claim 1, characterized in that It also includes step S4, using the unused angular distance information to verify the recognition result.

8. The method according to claim 7, characterized in that Step S4 is specifically as follows: in the recognition process, N stars are successfully recognized using M sets of angular distance information, where M≤2N; the star numbers num of the N stars obtained by recognition are i , find the corresponding direction vector in the navigation star library based on the star i=1~N;Use unused angular distance d pq-pic The corresponding pair of asterisks: x, y, x = num p ,y=num q , find the direction vector V corresponding to x,y x ,V y , calculate the angular distance d of the star pair in the celestial coordinate system xy-s ,in Combine this value with the unused angular distance d in the camera coordinate system. pq-pic For comparison, if |d xy-s -d pq-pic |<ε, it means that the star pair is correctly identified, and ε is the set threshold.

Images