CN102840861B - Navigational star screening method for star sensors - Google Patents

Abstract

The present invention relates to a kind of method that is used for the screening navigation star of star sensor, comprises: 1, according to the limit magnitude of star sensor, carry out star filtering process to the original star list of whole celestial sphere, and determine the star number threshold _Nth ; Two, the number of remaining stars of the star sensor in the field of view of the current sky area is set to N, if N≤N _th , then the remaining stars are all selected as navigation stars, and step 3 is performed; if N>N _th , the navigation stars in the field of view of the current sky area are screened through multi-scale image plane segmentation. 3. After the navigation stars in the field of view of the current sky area are screened, the star sensor turns to the next position and repeats the steps ( 2) Screening navigation stars until traversing the whole celestial sphere; the multi-scale image plane segmentation and screening method in the present invention can adapt to changes in the number of stars in different sky areas, delete redundant stars in high-density sky areas of star distribution, and retain low-density sky areas All the stars in the area, and the screened navigation stars are evenly distributed.

Description

A method for screening navigation stars for star sensors

technical field

The invention belongs to the technical field of celestial navigation, and relates to a method for screening navigation stars for a star sensor.

Background technique

The star sensor recognizes the star map, compares the characteristics of the observation star group and the navigation star group, identifies the observation star, determines their coordinates in the body coordinate system and the inertial coordinate system, and then measures the satellite attitude. It is a modern aerospace field. The most accurate satellite attitude measurement instrument. Star map recognition is the core technology of the star sensor. Establishing a navigation star library is an important prerequisite for star map recognition. Reasonable selection of navigation stars can reduce the similarity of navigation star group features, improve star map recognition speed and success rate, and enhance star map recognition. It is of great significance to improve the anti-false star interference ability of the sensor and improve the accuracy of attitude measurement.

When the navigation stars are evenly distributed on the whole celestial sphere, the redundancy of the star group features of the navigation stars is small, and the star map recognition stability is high. Usually, the optimization (screening) algorithm is evaluated by the distribution uniformity of the navigation stars. The current navigation star selection (screening) ) algorithms can be roughly divided into two categories.

The first type of algorithm takes the uniform distribution of navigation stars in the whole celestial sphere as the starting point. The orthogonal grid method proposed by Lin Tao et al. in 1998 projects the unit celestial sphere onto a plane, divides the projection plane orthogonally, and divides the whole celestial sphere into many non-intersecting equal-area sky areas. The stars are the navigation stars. Because the aspect ratio of the sky varies with latitude, the density of navigation stars is not uniform. Algorithms such as The Spherical Patches method, The Fixed-Slope Spiral method and The Charged Particles method proposed by Samaan and Malak A in 2004 divide the celestial sphere equally, and each The aspect ratio of a sky area has little relationship with the location, and the distribution of the obtained navigation stars is also more uniform. The navigation star optimization algorithm based on Boltzmann entropy published in ELECTRONICS LETTERS Volume 40 No. 2 in 2004 starts from the selected two navigation stars and selects other navigation stars one by one, so that the glass of all selected navigation stars The Boltzmann entropy is the smallest, and this algorithm can effectively delete redundant stars and obtain a uniform distribution of global navigation stars. This type of algorithm seldom considers the field of view of the star sensor and the number of navigation stars in the field of view of each sky area. Although the uniform distribution of navigation stars can be achieved, when the field of view is large, the number of navigation stars that can be observed each time is still redundancy.

The second type of algorithm starts from the uniform distribution of navigation stars on the local celestial sphere, and realizes the uniform distribution on the whole celestial sphere. In 2000, Li Lihong and others proposed a magnitude weighting method, which assigns different weights to each star according to the magnitude. Stars with low magnitudes have high weights, and stars with high magnitudes have low weights. According to the weights, guide stars are selected. The algorithm is superior to the orthogonal grid method, but the algorithm less considers the position of the stars, and the uniformity of the distribution of the navigation stars needs to be improved. In 2002, Hye-Young Kim of Texas A&M University and others proposed a self-organizing navigation star selection algorithm. On the premise that a certain number of navigation stars is reached in the field of view that satisfies any axis pointing, the navigation stars are selected one by one according to the positional relationship of the stars. The distribution is relatively uniform both locally and globally. The regression selection algorithm proposed by Zheng Sheng et al. in 2004 is based on the number of observable stars in the field of view, based on the method of support vector machine, to generate a dynamic magnitude threshold, and according to this threshold, observe stars in the field of view in different sky areas to obtain navigation stars , this method can obtain a relatively uniform distribution of navigation stars, but for star sensors with fixed limit magnitudes, the distribution of navigation stars obtained by the regression selection algorithm is still not uniform enough.

Contents of the invention

The technical problem to be solved by the present invention is to provide a method suitable for evenly screening out navigation stars for a star sensor.

The basic idea of the present invention is that since the sky area in the field of view of the star sensor only occupies a very small part of the whole celestial sphere, the sky area in the field of view can be regarded as a plane area, if any navigation in the field of view If the image plane of star imaging is uniformly distributed, then the navigation stars are also approximately uniformly distributed on the whole celestial sphere. In this way, the guide star can be screened according to the star image density on the image plane, and the distribution problem of the guide star on the whole celestial sphere can be converted into the distribution problem of its star image on the image plane.

Under the basic idea, the present invention provides a method for screening navigation stars for star sensors, comprising:

Step 1. According to the limit magnitude of the star sensor, filter the original star catalog of the whole celestial sphere, that is, delete double stars, variable stars and stars whose magnitude is higher than the limit magnitude; and determine the number of stars according to the star map recognition algorithm threshold N _th ;

Step 2. The number of remaining stars in the field of view of the star sensor in the current sky area is set to N. If N≤N _th , the remaining stars are all selected as navigation stars, and step 3 is performed;

If N>N _th , then filter the navigation stars in the field of view of the current sky area through multi-scale image plane segmentation, the steps are as follows:

Step (1) Image the remaining stars to the image plane, and divide the image plane into an orthogonal grid with p rows and q columns; each grid in the orthogonal grid is a sub-district ;

Step (2) Traversing each cell in turn, checking the number of remaining stars in it, among them, if there are many remaining stars in a cell, keep the brightest star among them and delete the rest; If N≤N _th , then set the current remaining star as the navigation star, the traversal ends, and execute step 3; if N>N _th , continue traversing; if after traversing all the cells, N is still greater than N _th , set the cell As a pixel, if there is a star in the cell, the gray value of the pixel is not 0, if there is no star in the cell, the gray value of the pixel is 0, and the adjacent cells with remaining stars after traversal Divide into connected domains, and calculate the centroid coordinates and the number of cells of each connected domain;

Step (3) Select the connected domain with the largest number of cells, and set the star closest to the centroid coordinates of the connected domain as a redundant star; if there are multiple stars in the connected domain closest to the centroid coordinates , the darkest star among them is a redundant star; if there are multiple connected domains with the largest number of cells, then select the darkest star in these connected domains as a redundant star; delete the redundant star; judge The number of remaining stars at this time, if N≤N _th , set the current remaining stars as navigation stars, and perform step 3; if N>N _th , repeat step (3);

Step (4) If there are no more connected domains; N is still greater than N _th , then the values of p and q are both reduced by 1, and steps (1) to (4) are repeated until N≤N _th ;

Step 3: After the navigation star screening of the current field of view in the sky area is completed, the star sensor turns to the next position and repeats step 2 to screen the navigation stars until the entire celestial sphere is traversed.

Further, in the step (2), if there are many remaining stars in a cell, the method of retaining the brightest star and deleting the remaining stars includes: pre-defining three stars in the orthogonal grid Two-dimensional arrays Marray, Idarray and MAGarray; if there is at least one star in the cell in the mth row and n column in the orthogonal grid, then Marray[m][n]=1, otherwise it is zero; if the If there are many stars in the cell, use IDarray[m][n] and MAGarray[m][n] to record the star number and magnitude of the brightest star respectively, and delete the rest.

Further, in order to quickly calculate the centroid coordinates of the connected domain, the centroid coordinate calculation method in the step (3) is:

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>$ ${\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>$

Where x _c , y _c are the gray-scale centroid coordinates of each connected domain, x _i , y _i represent the row and column numbers of the i-th cell in the connected domain, and k represents the total number of cells in the connected domain.

Further, in order to better segment the image plane; the ratio of the initial values of p and q is the same as the row and column size ratio of the image plane.

Compared with the prior art, the present invention has the following advantages: (1) the method of adopting multi-scale image plane segmentation and screening in the present invention can adapt to changes in the number of stars in different sky areas and delete redundant stars in star distribution high-density sky areas, Keep all the stars in the low-density sky area, and the selected navigation stars are evenly distributed; (2) The present invention regards the cell as a pixel, if there is a star in the cell, the gray value of the pixel is not 0, and if there is no star in the cell, then The gray value of this pixel is 0, and the adjacent cells with remaining stars after traversal are divided into connected domains, and the centroid coordinates of the connected domains can be quickly calculated through the ranks and columns of the connected domains, and the centroid coordinates can be effectively Delete the redundant stars in the high-density area, so that the distribution of the remaining stars tends to be uniform, that is, finally obtain a uniform navigation star; (3) complete the navigation star screening of the whole sky through the star sensor, and the screened navigation star of the whole sky The distribution is also uniform; (4) The ratio of the initial value of p and q is the same as the ratio of the row and column sizes of the image plane, which makes the segmentation of the cell more reasonable, and facilitates the decrement of p and q to complete the multi-scale image plane segmentation .

Description of drawings

In order to make the content of the present invention more easily understood, the present invention will be described in further detail below in conjunction with the specific embodiments according to the accompanying drawings, wherein

Fig. 1 is a schematic diagram of the body coordinate system based on the optical system of the present invention;

Fig. 2 is a schematic diagram of the rotation relationship between the inertial coordinate system and the body coordinate system;

Fig. 3 is the flowchart of the method for screening navigation star of the present invention;

Figure 4 is an example of the original star map;

Figure 5 is a distribution diagram of stars in the current field of view after dividing the image plane according to 5×5;

Figure 6 is a distribution diagram of the remaining stars in the current field of view after deleting dark stars in each cell;

Figure 7 is a distribution diagram of the remaining stars in the current field of view after deleting the star closest to the centroid of the largest connected domain;

Figure 8 is the distribution diagram of the remaining stars in the current field of view after deleting the darker stars in the two largest connected domains;

Fig. 9 is a distribution diagram of navigation stars after screening stars in the current field of view;

Figure 10 is the distribution diagram of stars in the current field of view after dividing the image plane according to 8×8;

Figure 11 is a distribution diagram of the remaining stars in the current field of view after deleting dark stars in each cell;

Figure 12 is a distribution diagram of the remaining stars in the current field of view after deleting the star closest to the centroid of the connected domain;

Figure 13 is a distribution diagram of the remaining stars in the current field of view after dividing the image plane according to 7×7;

Figure 14 is a distribution diagram of the remaining stars in the current field of view after dividing the image plane according to 6×6;

Figure 15 is a distribution diagram of the remaining stars in the current field of view after dividing the image plane according to 5×5;

Fig. 16 is a distribution diagram of navigation stars in the current field of view after processing according to the 5×5 divided image plane;

Figure 17 is the distribution diagram of navigation stars after star filtering and before filtering when the limit star magnitude is 5.2;

Fig. 18 is when the limit star magnitude is 5.2 and the field of view is 21.91°×16.47°, the distribution diagram of navigation stars on the whole celestial sphere after using the present invention to filter;

Figure 19 is a probability distribution diagram of the number of navigation stars in the field of view before screening when the limiting magnitude is 5.2 and the field of view is 21.91°×16.47°;

Figure 20 is a probability distribution diagram of the number of navigation stars in the field of view after screening when the limiting magnitude is 5.2 and the field of view is 21.91°×16.47°;

Fig. 21 is a cumulative probability distribution diagram of the number of navigation stars in the field of view before and after screening when the limiting magnitude is 5.2 and the field of view is 21.91°×16.47°.

Detailed ways

Below in conjunction with accompanying drawing and embodiment the present invention is described in detail:

Example 1

There are two characteristics of the star-image high-density area, one is that there are many star images in the same area, and the other is that the distance between the star images is short. According to these two characteristics, the present invention designs a method for screening navigation stars for star sensors.

First, according to the first feature, the image plane is divided into multiple rectangular areas of equal area along the row and column directions of the focal plane of the image plane detector to form an orthogonal grid. To simplify the description, each rectangular area is called a cell (that is, each grid in the orthogonal grid is a cell), and the equally divided interval in the row and column directions is called a scale. Select a certain scale to segment the image plane (also known as the image plane), and in actual operation, set the number of rows p and the number of columns q to realize the division of the image plane; if the high-density area of the star image contains many According to the characteristics of the brighter star, which has a high probability of being detected and a high signal-to-noise ratio of the star image, the brightest star is reserved, and all cells are processed by this method.

Then, according to the second feature, those remaining stars that are closer to each other must be in the neighborhood of each other, that is, the cells where they are located are connected to form a connected domain, and the higher the star density, the larger the range of the connected domain. If the star closest to the centroid of the largest connected domain is deleted, the connected domain can be split into several small connected domains, and the star density in this area will decrease. Then select the next largest connected domain from the processing results, and process it in a similar way until there are no more connected domains.

It is no longer possible to determine which star can be deleted by using this scale, so increase the equal interval and decrease the equal fraction, that is, the values of p and q are both reduced by 1.

To facilitate the description of the principle of the present invention, it is now assumed that the star sensor body coordinate system is established on the optical system, as shown in FIG. 1 . Let the optical system be equivalent to an ideal imaging system, H and H′ are the principal points of the object and image respectively, f is the focal length of the optical system, the origin of the star sensor body coordinate system is at the principal point H′ of the image, X _b Axis and _Yb axis are in the main plane of the image side, parallel to the row and column of the focal plane of the image plane detector respectively, _Zb axis is along the optical axis, and its positive direction is shown in Figure 1, and the three axes form a right-handed coordinate system. The direction cosine vector of the star S in this coordinate system is V _b , the field angles XFLD and YFLD in the X _b and Y _b directions, and the coordinates on the image plane are (x _b , y _b ).

Assuming that in the inertial coordinate system, the optical axis points to (α _c , δ _c ), the body coordinate system can be obtained by rotating the inertial coordinate system in a certain way. As shown in Figure 2, the inertial coordinate system first rotates α _c around the Z axis from the +X axis to the +Y axis to obtain the X′Y′Z′ coordinate system, and then the new coordinate system revolves around the Y′ axis from the +Z′ axis to + The X′ axis is rotated by 90°-δ _c to obtain the X″Y″Z″ coordinate system, and the coordinate system is rotated around the Z″ axis by φ to obtain the body coordinate system X _b Y _b Z _b .

According to the above-mentioned star sensor setting and the conversion of inertial coordinates and body coordinates, the method for screening navigation stars for star sensors of the present invention includes:

Step 1. According to the limit magnitude of the star sensor, filter the original star catalog of the whole celestial sphere, that is, delete double stars, variable stars and stars whose magnitude is higher than the limit magnitude; and determine the number of stars according to the star map recognition algorithm threshold N _th ;

Step 2. The number of remaining stars in the field of view of the star sensor in the current sky area is set to N. If N≤N _th , the remaining stars are all selected as navigation stars, and step 3 is performed;

If N>N _th , then filter the navigation stars in the field of view of the current sky area through multi-scale image plane segmentation, the steps are as follows:

Step (1) Image the remaining stars to the image plane, and divide the image plane into an orthogonal grid with p rows and q columns; each grid in the orthogonal grid is a sub-district ;

Step (2) Traversing each cell in turn, checking the number of remaining stars in it, among them, if there are many remaining stars in a cell, keep the brightest star among them and delete the rest; If N≤N _th , then set the current remaining star as the navigation star, the traversal ends, and execute step 3; if N>N _th , continue traversing; if after traversing all the cells, N is still greater than N _th , set the cell As a pixel, if there is a star in the cell, the gray value of the pixel is not 0, if there is no star in the cell, the gray value of the pixel is 0, and the adjacent cells with remaining stars after traversal Divide into connected domains, and calculate the centroid coordinates and the number of cells of each connected domain;

Step (3) Select the connected domain with the largest number of cells, and set the star closest to the centroid coordinates of the connected domain as a redundant star; if there are multiple stars in the connected domain closest to the centroid coordinates , the darkest star among them is a redundant star; if there are multiple connected domains with the largest number of cells, then select the darkest star in these connected domains as a redundant star; delete the redundant star; judge The number of remaining stars at this time, if N≤N _th , set the current remaining stars as navigation stars, and perform step 3; if N>N _th , repeat step (3);

Step (4) If there are no more connected domains; N is still greater than N _th , then the values of p and q are both reduced by 1, and steps (1) to (4) are repeated until N≤N _th ;

Step 3: After the navigation star screening of the current field of view in the sky area is completed, the star sensor turns to the next position and repeats step 2 to screen the navigation stars until the entire celestial sphere is traversed.

The centroid coordinate calculation method in the described step (3) is:

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>$ ${\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>$

Where x _c , y _c are the gray-scale centroid coordinates of each connected domain, x _i , y _i represent the row and column numbers of the i-th cell in the connected domain, and k represents the total number of cells in the connected domain.

The ratio of the initial values of p and q is the same as the row and column size ratio of the image plane.

Example 2

Realize on the basis of embodiment 1 the method for the screening navigation star of star sensor, its specific implementation process is as follows:

The first step is to filter the original star catalog. Delete dark stars according to the limit star magnitude, and delete variable stars and double _stars at the same time. It is necessary to select a corresponding star map recognition algorithm to determine the star number threshold N _th , or it can be set by itself according to needs.

In the second step, the optical axis of the star sensor points to the position of coordinates (α _i , δ _i ) on the whole celestial sphere, and α _i or δ _i changes by 1° each time, traversing the whole celestial sphere. Extract each remaining star pointing to the field of view, and calculate the total number N of remaining stars. If N≤N _th , the remaining stars are all selected as navigation stars; the optical axis of the star sensor is turned to the next position, and then judge until the number of stars is greater than the threshold, and start the following steps.

The method of extracting the remaining stars in the field of view is to first select the coordinates (α, δ) satisfying

|δ-δ _c |≤w _m

The star, where w _m represents the field angle corresponding to the diagonal of the star sensor image surface detector.

The above formula defines the upper and lower limits of the declination of the remaining stars in the current field of view. Since the value range of declination δ is -90°-90°, when δ _i -w _m is less than -90°, the lower limit should be set to -90°; similarly, when δ _i -w _m is greater than 90°, the upper limit should be set to 90°, giving

The remaining star data is sorted by declination, and the position of the star whose declination value is just greater than δ _bot is determined by the dichotomy method, and then the subsequent data is read to extract the remaining stars until the declination value is greater than δ _top .

Next, calculate the azimuths of the remaining stars that have been extracted in the body coordinate system. For the stars whose right ascension and declination are (α, δ), we have

$\left[\begin{array}{c}{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; bx</span>}}\\ {\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; by</span>}}\\ {\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; bz</span>}}\end{array}\right]<span\; class="notranslate">\; =</span>\left[\begin{array}{ccc}\mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&phi;</span>}& \mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&phi;</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&phi;</span>}& \mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&phi;</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]\left[\begin{array}{ccc}\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 90</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&delta;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>& \mathrm{<span\; class="notranslate">\; 0</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; sin</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 90</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&delta;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; sin</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 90</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&delta;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 90</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&delta;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>\end{array}\right]<span\; class="notranslate">\; \&times;</span>$

$\left[\begin{array}{ccc}{\mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}& \mathrm{<span\; class="notranslate">\; sin</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; sin</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}& {\mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]\left[\begin{array}{c}\mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&alpha;</span>}\mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&delta;</span>}\\ \mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&alpha;</span>}\mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&delta;</span>}\\ \mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&delta;</span>}\end{array}\right]$

Its field of view XFLD and YFLD in the directions of X _b and Y _b are

$\mathrm{<span\; class="notranslate">\; XFLD</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; tg</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; bx</span>}}}{{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; bz</span>}}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; YFLD</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; tg</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; by</span>}}}{{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; bz</span>}}}<span\; class="notranslate">\; )</span>$

If it is assumed that the maximum viewing angles of the optical system in the directions of X _b and Y _b are _WA and W _B . , only if the

|XFLD|≤W _A /2, |YFLD|≤W _B /2

stars can be observed. Filter through the above formula to get the remaining stars in the current field of view, and also get their field angles XFLD, YFLD, and their total number N.

The third step is to image all the stars in the field of view to the image plane, press

x _b =ftan(XFLD), _yb =ftan(YFLD)

Calculate and record the position of each star image.

The fourth step is to divide the image plane into a p×q grid along the row and column directions of the focal plane, and establish and initialize three two-dimensional arrays Marray, Idarray and MAGarray of p rows and q columns corresponding to the grid, p and q The ratio of should be consistent with the size ratio of the row and column directions of the focal plane as far as possible, so as to ensure that the scales of the two directions are the same. At the beginning, p and q should take slightly larger values in order to examine the star distribution density in detail. The arrays Marray and Idarray are initialized to 0, and MAGarray is initialized to -99.99.

The fifth step is to traverse each cell to calculate the cell where the extracted star is located. For a star image with coordinates (x _b , y _b ), it is located in the mth row and n column of the image surface, and there is a star image, then

$\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; floor</span>}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; f</span>}\mathrm{<span\; class="notranslate">\; the\; tan</span>}<span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; w</span>}}_{\mathrm{<span\; class="notranslate">\; A</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

$\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; floor</span>}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; q</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; f</span>}\mathrm{<span\; class="notranslate">\; the\; tan</span>}<span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; w</span>}}_{\mathrm{<span\; class="notranslate">\; B</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; q</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

Among them, floor(x) means to take the nearest integer smaller than x, and Marray[m][n]=1. If there are many star images in the cell, only the brightest star will be kept, IDarray[m][n] will record the number of the preserved star, MAGarray[m][n] will record its magnitude, and update the current field of view The total number of remaining stars N is the star selection for multi-star cells.

The sixth step, if N≤N _th , return to the second step, otherwise use the pixel clustering algorithm (for the pixel clustering algorithm, refer to the author Yang Fan, "Digital Image Processing and Analysis", ISBN 978-7-5124-0188-4 ) to connect the domain to calculate the coordinates of the center of mass, and make further screening. That is to say, the small area is regarded as a pixel, and the number of stars in the small area is regarded as a gray value, and the image surface grid is divided into multiple connected domains by eight connections, and the centroid coordinates and the number of small areas of each connected domain are calculated. Select the connected domain with the largest number of cells, and delete the star closest to the centroid coordinates. If multiple stars are closest to the centroid coordinates, delete the faintest star. If there are multiple connected domains with the largest number of cells, delete the dimmest stars in these connected domains. At the same time, update the values of N, Marray, and IDarray, MAGarray. If N>N _th , repeat this step, and then search for the next largest connected domain in the range. When N≤N _th , the navigation stars in the current field of view are screened. If N>N _th , and the cells where any two stars are located are no longer connected (that is, there is no connected domain), go to the next step.

In the seventh step, both p and q are reduced by 1, and the average division scale is increased, the area of the image area is slightly increased, and several star images at a distance may be connected again, and then execute from the fourth step until N ≤ N _th .

The eighth step, when the whole celestial traversal is completed, the navigation star screening ends.

The above steps are simplified into a flowchart, as shown in FIG. 3 .

Example 3

On the basis of Example 1 and Example 2, the content of the present invention is further illustrated by implementing two kinds of segmentation scales (ie, different values of q and p) and comparing with the prior art.

The following embodiments use the SAO star catalog as the original star catalog; the SAO star catalog (The SmithsonianAstrophysical Observatory Star Catalog/Smithsonian Astrophysical Observatory Star Catalog) is an astrometric star catalog published by the Smithsonian Astrophysical Observatory in 1966. Contains a total of 258,997 stars. This catalog is compiled from some previous catalogs, but only includes stars above magnitude 9.0 and whose proper motions have been precisely measured. The star names in the SAO star catalog start with the letter SAO followed by a serial number. The stars are divided into divisions by declination, and every 10 degrees is a division, which is divided into 18 divisions. The stars in each division are sorted according to the position of right ascension.

Implementation Mode 1

Select the SAO star catalog as the original star catalog, the limit star magnitude is 5.5, and the field of view is 20°×20°. After the original star catalog is filtered, a certain optical axis points to the remaining stars in the field of view on the image plane The distribution of star images on is shown in Figure 4. At this time, there are 18 remaining stars, and the distribution of these stars is not uniform enough. Take N _th =7, the initial values of p and q are both taken as 5, and after the image plane is divided by 5×5, as shown in Figure 5, there are multiple stars in multiple sub-districts, and each sub-district deletes the dark star and keeps the brightest After the star, get Figure 6, and there are 10 stars left. In Figure 6, rows 1 to 5, columns 1 and 2, there are five small areas forming a connected domain, which contains the largest number of star images, and the star density in this area is high. The star closest to the centroid of the connected domain is deleted, and the star density in this region decreases. The coordinates of these five stars are (1, 1), (2, 1), (3, 2), (4, 1), (5, 1), and their centroids are (3, 1.25), and they reach ( 3, 2) is the closest, then delete the star of (3, 2) to get Figure 7. At this time, there are 3 connected domains, each of which contains 2 cells, and the range of their connected domains is the largest. If the star in row 2 and column 1 is the darkest, delete it. The remaining 2 connected domains contain 2 small areas with the largest range. The star in the 1st row and 3rd column is the darkest star in the 2 connected domains, and then delete it. In this way, the number of remaining stars is not greater than the threshold N _th , and the screening results are shown in FIG. 8 and FIG. 9 .

Implementation mode two

For the screening of the remaining stars in the field of view corresponding to Figure 4, N _th is still taken as 7, and the initial values of p and q are both taken as 8. After the image plane is divided by 8×8, as shown in Figure 10, there are multiple small areas There are multiple stars in , and after deleting the dark stars and keeping the brightest stars in each cell, Figure 11 is obtained.

As shown in Figure 11, in rows 1 to 4 and columns 1 and 2, there are four cells forming a connected domain, which contains the largest number of star images, and the star density in this area is high. The star closest to the centroid of the connected domain is deleted, and the star density in this region decreases. The coordinates of these four stars are (1, 1), (2, 2), (3, 1), (4, 1), their center of mass is (2.5, 1.25), and the distance from it to (3, 1) Most recently, then delete the star at (3, 1). Then select the largest connected domain from the processing results, and then process it in a similar way until the cells where any two remaining stars are located are no longer connected. The result after processing Figure 11 is shown in Figure 12, and the remaining stars are more evenly distributed within this scale.

It is no longer possible to decide which star can be deleted by using this scale, so the equal division interval is increased and the equal fraction is decreased. Divide the image plane into 49 equal-area areas of 7×7 to obtain Figure 13, and perform similar processing according to the method of the previous scale. As shown in Figure 14, the scale is increased again, and the image plane is divided into 6×6 for processing, and Figures 15 and 16 are obtained. It can be seen that the distribution of the remaining stars is very uniform, and these remaining stars can be used as navigation stars.

Implementation Mode Three

In order to compare the method for screening navigation stars of the present invention with the orthogonal grid method and the Boltzmann entropy algorithm, Table 1 shows that the limit magnitude Take 6 and 7.5 mag, respectively, and compare the data of the navigation star results. The data of the orthogonal grid method and the Boltzmann entropy algorithm come from the paper "A General Method of the automatically selection of guide star" published at the Proceedings of ICSP'98 conference and the 40th issue of "ELECTRONICS LETTERS" published in 2004 Volume 2 paper "Boltzmannentropy-based guide star selection algorithm for star tracker". With the present invention, the star number threshold _Nth is taken as 6, and the initial values of p and q, which are equal fractions in the row and column directions, are both taken as 8, and the navigation star library Boltzmann entropy thus established is the smallest, and the uniformity of the whole celestial globe is the best . From the perspective of local celestial uniformity, the maximum number of navigation stars obtained by the orthogonal grid method and the Boltzmann entropy algorithm is larger, the minimum value is smaller, and the uniformity is slightly worse.

Table 1 The present invention and the comparison of orthogonal grid method and Boltzmann entropy algorithm

In order to compare with the regression selection algorithm, magnitude weighting algorithm and self-organization algorithm, when the field of view is 8 ° × 8 °, and the limit magnitude is 7 values between 6.5 and 7.9, use the present invention to filter the navigation star. The results are shown in Table 2. Regression selection algorithm, magnitude weighting algorithm and self-organization algorithm The data are respectively from "Research on a New Navigation Star Selection Algorithm" published by Zheng Sheng et al. in 2004, Volume 25, No. 1 of "Acta "An Improved Algorithm for All-Sky Autonomous Triangular Star Pattern Recognition" published in "Optical Technology" Volume 26, Issue 4, Hye-Young Kim et al. published the paper "Self-organizing at the IEEE on aerospace conference proceedings in 2002 Guide Star Selection Algorithm for Star Trackers: Thinning Method".

Table 2 shows that the present invention can effectively reduce the number of stars in the high-density sky area, but has little effect on the number of stars in the low-density sky area, and the number of navigation stars in the field of view of more than 95% of the sky area is between 5 and 12. When the limit magnitude is low, the number of navigation stars is a little more, mainly because the number of stars in the field of view in some sky areas is less than N _th , even if there are redundant stars in the surrounding sky areas, they cannot be deleted.

When the limit star magnitude is 6.5 and 7.3, the regression selection algorithm deletes too many stars in some sky areas, and the number of navigation stars in more than 10% of the sky areas is less than 5. When the limit star magnitude is greater than 7.5, the navigation established by this algorithm In the star library, there are still too many stars in the field of view in many sky areas, and the maximum number of stars is relatively large. Using magnitude weighting algorithm and self-organizing algorithm to screen navigation stars, the proportion of low-density sky area is relatively large. With the increase of the limit star magnitude, the redundancy of the navigation star library established by the self-organizing algorithm is getting bigger and bigger. The navigation star library obtained by the invention has the smallest standard deviation and the most uniform distribution, and is superior to magnitude weighting algorithms, self-organizing algorithms and regression algorithms.

Table 2 Comparison of navigation star screening algorithms in 8°×8° field of view

Implementation Mode Four

Take the SAO star catalog as the original star catalog, the limiting magnitude is 5.2, the aspect ratio of the star sensor detector is 4:3, the field of view is 21.91°×16.47°, and the initial values of the equal fractions p and q in the row and column directions Take 12 and 9, and select the star number threshold N _th to be 6. When using the present invention to build the navigation star library, 529 stars were deleted and 1078 stars were kept. Figure 17 and Figure 18 show the distribution of navigation stars on the whole sphere before and after selection, respectively. The Boltzmann entropy dropped from 0.0119 to 1.3643×10 ^-4 , and the distribution of navigation stars is more uniform.

Through the whole celestial traversal, the distribution of the number of navigation stars before and after the screening is counted, and the results are shown in Figure 19 and Figure 20. The maximum number of navigation stars in the field of view is reduced from the original 47 to 18, while the minimum value remains unchanged at 2. The standard deviation of the calculated number of stars is reduced from the original 6.15 to 1.87, and the average number of stars is reduced from 13.75 to 9.37 stars, the uniformity of the navigation star on the local celestial sphere is improved. Figure 21 is the cumulative probability distribution of the number of navigation stars in the field of view. When the number of stars is less than 4, the two curves before and after the selection of navigation stars coincide, and the probability of more than 4 navigation stars appearing in the field of view is 99.94%, which shows that the present invention can Effectively reduce the number of stars in the high-density sky area with star distribution, and reduce the redundancy of navigation star features.

Apparently, the above-mentioned embodiments are only examples for clearly illustrating the present invention, rather than limiting the implementation of the present invention. For those of ordinary skill in the art, other changes or changes in different forms can be made on the basis of the above description. It is not necessary and impossible to exhaustively list all the implementation manners here. And these obvious changes or modifications derived from the spirit of the present invention are still within the protection scope of the present invention.

Claims (3)

1., for a method for the screening nautical star of star sensor, comprising:

Step one, limiting magnitude according to star sensor, make star filtration treatment to the original star catalogue of whole day ball, namely delete double star, variable and the magnitude fixed star higher than limiting magnitude; And according to star Pattern Recognition Algorithm determination star number threshold value N _th;

Step 2, star sensor optical axis point to coordinate (α on whole day ball _i, δ _i) position, α _ior δ _ieach change 1 °, traversal whole day ball; Described star sensor is set to N, if N≤N in the quantity when the residue star in visual field, district's day before yesterday _th, then described residue star all elects nautical star as, performs step 3;

If N > is N _th, then described when the nautical star in visual field, district's day before yesterday by multiple dimensioned image planes segmentation screening, its step is as follows:

Described residue star is imaged onto image planes by step (1), and concrete grammar is: all stars in visual field are imaged onto image planes, press

x _b＝f tan(XFLD)，y _b＝f tan(YFLD)

Calculate and record the position of each star image;

These image planes are divided into line number is p, columns is the orthogonal grid of q; Each grid in described orthogonal grid is a community;

Step (2) travels through each community successively, checks the quantity of residue star wherein, wherein, if the quantity of a community residue star has many, then retains a wherein the brightest star, deletes all the other stars; Judge the quantity now remaining star, if N≤N simultaneously _th, then set current residual star as nautical star, traversal terminates, and performs step 3; If N > is N _th, then traversal is continued; If after traveling through all communities, N is still greater than N _th, Ze Ba is used as pixel in community, if having star in community, then the gray value of this pixel is non-zero, if without star in community, then the gray value of this pixel is 0, the neighbor cell with residue star after traversal is divided into connected domain, calculates center-of-mass coordinate and the community number of each connected domain;

Step (3) chooses the maximum connected domain of its small area number, be located in this connected domain from the star that the center-of-mass coordinate of this connected domain is nearest be redundant star; If have many from the star that center-of-mass coordinate is nearest in this connected domain, then a wherein the darkest star is redundant star; If the connected domain that community number is maximum has multiple, then a star the darkest in these connected domains is selected to be redundant star; Delete described redundant star; Judge the quantity now remaining star, if N≤N _th, then set current residual star as nautical star, perform step 3; If N > is N _th, then this step (3) is repeated;

Step (4) is if after no longer including connected domain; N is still greater than N _th, then the value of described p and q all subtracts 1, repeats step (1) to (4); Until N≤N _th;

After the screening of step 3, the described nautical star when visual field, district's day before yesterday terminates, described star sensor forwards next orientation to and repeats step 2 to screen nautical star, until traversal whole day ball.

2. the method for screening nautical star according to claim 1, is characterized in that: the center-of-mass coordinate computational methods in described step (3) are:

$x_c=\frac{\mathrm{\&Sigma;}{x}_i}{k},y_c=\frac{\mathrm{\&Sigma;}{y}_i}{k}$

Wherein x _c, y _cfor the gray scale center-of-mass coordinate of each connected domain, x _i, y _irepresent the sequence number of the row and column at this place, connected domain ZhongiGe community, k represents the community sum in connected domain.

3. the method for screening nautical star according to claim 1 and 2, is characterized in that: the ratio of the initial value of described p with q and the row, column size of described image planes are than identical.