A fast star map recognition method based on refracting star / non-refracting star information fusion and prediction

Abstract

A fast star map recognition method based on the fusion and prediction of refraction star / non-refraction star information. In the star sensor window tracking mode, according to the predicted attitude, the navigation stars appearing in the field of view can be searched, and the refraction stars with large refraction angles are retained first, and the remaining stars are retained first. A total of N navigation stars that meet the accuracy requirements are retained. According to the projection position of the navigation star on the image plane, if it is a refraction star, it is projected according to the visual position after refraction. After the FPGA window is opened, star point extraction is performed. After the star point extraction is completed, the navigation star marked as the non-refraction star window is used for star point identification and matching, and the inertial attitude of the star sensor is calculated to calculate the angular velocity of the star sensor and predict the attitude of the next frame. After the star point extraction is successful in the window marked as the refraction star, the image plane position is converted into an inertial vector according to the attitude calculated by the non-refraction star, and it is matched with the corresponding inertial vector in the navigation star catalog to obtain the refraction angle of the refraction star.

Description

Technical Field

[0001] The invention proposes a fast star map recognition method for refracting star / non-refracting star information fusion and prediction, and the method belongs to the field of astronomical navigation or satellite autonomous navigation. Background Art

[0002] When starlight passes through the atmosphere, due to the uneven density of the atmosphere, the light is refracted and bent toward the center of the earth. The angle between the refracted light observed from the satellite and the incident light from the star is the starlight refraction angle. There is a definite relationship between the starlight refraction angle and the atmospheric density, and there is also a relatively accurate model for how the atmospheric density changes with altitude. Therefore, the starlight refraction angle information measured by the star sensor can be used to calculate the altitude of the starlight in the atmosphere. This altitude observation is a function of the satellite position. Using this measurement data and combining it with the orbital dynamics equation for filtering, the satellite's position, velocity and other navigation information can be obtained.

[0003] Star sensors are the main working parts of spacecraft to complete space missions. They can provide spacecraft with high-precision attitude information. The steps of attitude measurement using star sensors can generally be divided into shooting star maps, centroid extraction, star map recognition and attitude determination. Star map recognition is a key step in the attitude measurement process of star sensors. Starlight refraction autonomous navigation technology requires the simultaneous identification of navigation stars and refraction stars. Star map recognition technology is the key technology in the process of spacecraft attitude determination. Although the United States and other countries have successfully conducted flight tests on starlight refraction autonomous navigation technology, due to the confidentiality of the technology, the relevant information that can be obtained is very limited.

[0004] In autonomous navigation based on starlight refraction, accurate identification of refracted stars and precise acquisition of refraction angles are the basis for precision navigation. Since the image of starlight formed on the sensor after atmospheric refraction is different from the standard star map, the traditional star map recognition method cannot identify refracted stars, and thus cannot obtain refraction angle information. There are currently two solutions. One solution is to use multiple star sensors, one facing the zenith to obtain non-refracted star information, and the other using the installation array relationship to observe refracted stars. Another solution is to use one product to simultaneously observe refracted and non-refracted stars.

[0005] Both Scheme 1 and Scheme 2 simulate the refracted star map based on the attitude, and obtain the refracted stars by comparing the refracted star map with the photographed star map. This leads to the following problems: 1) There are errors in the installation array of the two star sensors in Scheme 1, which will cause deviations in the star point positions; 2) In Scheme 2, after the refracted stars are identified, the attitude is calculated and then projected for verification. This requires the entire map to operate, which will inevitably limit the image update rate. At the same time, due to the constraints of storage space and time complexity, the number of identified star points cannot be too many, which will restrict the accuracy of the product and cannot meet the growing demand for high-precision and high-update rate astronomical navigation; 3) Both Scheme 1 and Scheme 2 have error sources such as installation matrix, attitude calculation, and centroid extraction. How to determine the appropriate matching threshold when aligning the simulated star map with the photographed star map is a difficult problem in its surface connection. Summary of the invention

[0006] The technical problem solved by the present invention is: to overcome the deficiencies of the prior art and provide a fast star map recognition method for refracting star / non-refracting star information fusion and prediction, which is easy to implement, is not affected by parameter settings, and has no random uncertainty process.

[0007] The technical solution of the present invention is: a fast star map recognition method based on refracting star / non-refracting star information fusion and prediction, comprising:

[0008] Classify the navigation stars appearing in the field of view into non-refracting stars and candidate refracting stars;

[0009] The candidate refraction stars are retained first for star point extraction, and non-refraction stars are retained according to the star point energy, with bright stars being given priority;

[0010] The attitude matrix is ​​obtained by performing attitude calculation using the above-retained non-refractive stars;

[0011] Extract candidate refraction stars, and mark the extracted candidate refraction stars as refraction stars;

[0012] Calculate the coordinates of the refraction star in the star sensor system based on the image plane coordinates of the refraction star combined with the calculated attitude matrix;

[0013] According to the inertial coordinates of the refracting star, calculate the coordinates of the refracting star in the star sensor system before refraction;

[0014] The refraction angle of the output refracted star is calculated based on the coordinates of the refracted star in the star sensor system and its coordinates in the star sensor system before refraction.

[0015] Preferably, assuming that the direction vector of a star's starlight is s, the star that meets the following equation can be selected as a candidate refraction star:

[0016]

[0017] where θ R The refraction height is ha The corresponding starlight refraction angle at ; r = [xyz] is the position vector of the carrier in the geocentric equatorial inertial coordinate system, is its modulus, representing the distance from the carrier to the center of the earth; Thickness of the stratosphere Refraction height range (h a ,h b ), R e is the radius of the Earth.

[0018] Preferably, the thickness refraction height of the stratosphere is selected to be 20 to 50 km.

[0019] Preferably, the coordinates (u, v) of the star point marked as the refracted star are extracted, and the coordinates of the incident starlight vector, i.e., the observation vector, in the star sensor body coordinate system are According to the coordinates (u, v) of the star point on the image plane and the pinhole imaging model, we can get:

[0020]

[0021] Where: u is the column coordinate of the star point on the image plane, v is the row coordinate, (u0, v0) is the coordinate of the center point of the image plane, and f is the focal length of the star sensor.

[0022] Preferably, all navigation stars appearing in the field of view are sorted according to the size of the refraction angle, and navigation stars with large refraction angles are retained first, with the number not exceeding 34.

[0023] Preferably, the inertial direction vector of the refraction star before refraction is obtained according to the navigation star catalog Direction vector of the refracting star in the star sensitive system A is a non-refractive star, and the attitude matrix is ​​obtained by performing attitude calculation on it.

[0024] Preferably, in the star-sensitive system, according to the vector of the refracting star before refraction The refracted vector Find the refraction angle corresponding to the refraction star

[0025] Preferably, the field of view is the field of view of a single star sensor.

[0026] Preferably, the extraction is window image extraction.

[0027] Preferably, no less than 4 non-refractive stars are extracted.

[0028] The beneficial effects of the present invention compared with the prior art are:

[0029] (1) The present invention does not need to set a threshold, and can get rid of the difficulty of selecting a matching threshold when aligning a simulated star map with a photographed star map in the existing solution. The present invention can predict and retain the refraction stars that appear in the field of view by fusing and predicting the information of refraction stars and non-refraction stars. As long as the refraction stars can be extracted, the refraction angle can be calculated, and the refraction stars with smaller refraction angles will not be unable to be identified or mistakenly classified as non-refraction stars due to inappropriate setting of the matching threshold. In starlight refraction navigation, each refraction star is very precious and valuable. The present invention uses non-refraction stars in the field of view to calculate the attitude of the star sensor, thereby improving the accuracy of attitude calculation.

[0030] (2) The present invention has a fast processing speed, is easy to implement and can be applied to window images. Existing solutions are suitable for full-image operations and cannot meet the growing demand for high-precision and high-update rate astronomical navigation technology. Currently, star sensors are generally in tracking mode, and the object processed by the application software is the window image or star point coordinates, so the previous solutions cannot be used.

[0031] (3) The present invention is simple to implement, requiring only one star sensor and no threshold setting, thereby eliminating the difficulty of selecting a matching threshold when aligning simulated star maps with photographed star maps in existing solutions.

[0032] (4) The present invention has a fast processing speed, is easy to implement and can be applied to window images, meeting the current requirements of star sensor star map processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a schematic diagram of the principle of determining a refraction star according to the present invention;

[0034] Figure 2 This is a flow chart of star map recognition of the present invention; DETAILED DESCRIPTION

[0035] In the autonomous navigation of starlight refraction, the accurate identification of refracting stars and the precise acquisition of refraction angles are the basis for achieving precision navigation. In view of the current high-precision and high-update rate astronomical navigation requirements, the method of the present invention is suitable for the window tracking mode, and does not require star map simulation and star map registration based on the pre-obtained attitude. In the window tracking mode of the star sensor, the present invention can search for navigation stars that appear in the field of view according to the predicted attitude, and give priority to retaining refraction stars with large refraction angles. The remaining stars give priority to retaining bright stars, and a total of N bright navigation stars that meet the accuracy requirements (N≥4) are retained. According to the projection position of the navigation star on the image plane, if it is a refraction star, it is projected according to the visual position after refraction. After the FPGA opens the window, star point extraction is performed. After the star point extraction is completed, the navigation star marked as the non-refracting star window is used to identify and match the star points, calculate the inertial attitude of the star sensor, calculate the angular velocity of the star sensor, and predict the attitude of the next frame. After the star point of the window marked as a refraction star is extracted successfully, the image plane position is converted into an inertial vector according to the posture calculated by the non-refracting star, and it is matched with the corresponding inertial vector in the navigation star catalog to obtain the refraction angle of the refraction star. The present invention does not need to set a threshold, and can get rid of the difficulty of selecting the matching threshold when aligning the simulated star map and the photographed star map in the existing scheme. At the same time, the processing speed is fast, easy to implement and can be applied to the window image. The specific implementation is as follows:

[0036] (1) According to the thickness of the stratosphere, the refraction height is generally selected from 20 to 50 km, that is, Figure 2 Where ha = 20 km, hb = 50 km; assuming that the direction vector of a star's light is s, then Figure 2 Stars that satisfy the following equation can be selected as candidate refraction stars.

[0037]

[0038] Among them, θ R is the starlight refraction angle corresponding to the refraction height of 20 km, which can be calculated by the atmospheric refraction model; r = [xyz] is the position vector of the carrier in the geocentric equatorial inertial coordinate system, is its modulus, representing the distance from the carrier to the center of the earth; α and β are Figure 1 As shown, it can be expressed as and R e is the radius of the Earth.

[0039] (2) Due to the influence of space complexity and time complexity, only a limited number of star points may be retained for star point extraction. Therefore, stars with large refraction angles are preferentially retained for star point extraction (the number is generally not more than 34). The rest are sorted according to star point energy, and bright navigation stars are retained and marked as non-refracting stars. In order to ensure the accuracy of attitude calculation, it is generally required to extract no less than 4 non-refracting stars.

[0040] (3) Perform attitude calculation on the extracted non-refractive stars, assuming that the calculated attitude matrix is ​​A.

[0041] (4) Based on the navigation star catalog, we can obtain the inertial direction vector of the refracting star before refraction Its direction vector in the star sensitive system

[0042] (5) Extract the coordinates (u, v) of the star point marked as a refracted star and the coordinates of the incident starlight vector (i.e., the observation vector) in the star sensor body coordinate system According to the coordinates (u, v) of the star point on the image plane, according to the pinhole imaging model, it can be obtained by the following formula:

[0043]

[0044] Where: u is the column coordinate of the star point on the image plane, v is the row coordinate, (u0, v0) is the coordinate of the center point of the image plane, and f is the focal length of the star sensor.

[0045] (6) In the star-sensing system, according to the vector of the refracting star before refraction The refracted vector The refraction angle corresponding to the refraction star can be obtained

[0046] (7) Using the attitude calculated by non-refractive stars to predict the star points in the next frame of the star map

[0047] Parts of the present invention that are not described in detail belong to common knowledge of those skilled in the art.

Claims

1. A fast star map recognition method based on refracting star / non-refracting star information fusion and prediction, characterized in that Applicable to window tracking mode, no star map simulation and star map registration are required, the steps include: Classify the navigation stars appearing in the field of view into non-refracting stars and candidate refracting stars; The candidate refraction stars are preferentially retained for star point extraction, and non-refractive stars are extracted according to the star point energy, with bright stars being given priority; the extraction is window image extraction; and no less than 4 non-refractive stars are extracted; The attitude matrix is ​​obtained by performing attitude calculation using the non-refractive stars extracted above; Extract candidate refraction stars, and mark the extracted candidate refraction stars as refraction stars; According to the image plane coordinates of the refracted star, calculate the coordinates of the refracted star in the star sensor system; Based on the inertial coordinates of the refracting star combined with the calculated attitude matrix, the coordinates of the refracting star in the star sensor system before refraction are calculated; The refraction angle of the output refraction star is calculated based on the coordinates of the refraction star in the star sensor system and its coordinates before refraction in the star sensor system; Assuming that the direction vector of a star's light is s, the star that meets the following equation can be selected as a candidate refraction star: where θ R The refraction height is h a The corresponding starlight refraction angle at ; r = [xyz] is the position vector of the carrier in the geocentric equatorial inertial coordinate system, is its modulus, representing the distance from the carrier to the center of the earth; Thickness of the stratosphere Refraction height range (h a ,h b ), R e is the radius of the Earth.

2. The method according to claim 1, characterized in that: The thickness refraction height of the stratosphere is selected to be 20 to 50 km.

3. The method according to claim 1, characterized in that Extract the coordinates (u, v) of the star point marked as a refracted star, the coordinates of the incident starlight vector, i.e. the observation vector, in the star sensor body coordinate system According to the coordinates (u, v) of the star point on the image plane and the pinhole imaging model, we can get: Where: u is the column coordinate of the star point on the image plane, v is the row coordinate, (u0, v0) is the coordinate of the center point of the image plane, and f is the focal length of the star sensor.

4. The method according to claim 1, characterized in that All navigation stars that appear in the field of view are sorted according to the size of the refraction angle, and navigation stars with large refraction angles are retained first, with the number not exceeding 34.

5. The method according to claim 1 or 3, characterized in that: Obtain the inertial direction vector of the refracting star before refraction based on the navigation star catalog Direction vector of the refracting star in the star sensitive system A is a non-refractive star, and the attitude matrix is ​​obtained by performing attitude calculation on it.

6. The method according to claim 5, characterized in that: In the star-sensitive system, according to the vector of the refracting star before refraction The refracted vector Find the refraction angle corresponding to the refraction star 7. The method according to claim 1, characterized in that: The field of view is the field of view of a single star sensor.

Images