Multi-star interference discrimination method for high-sensitivity star sensor

Abstract

The invention discloses a multi-star interference discrimination method for a high-sensitivity star sensor. The method comprises the following steps: acquiring a window graph by using the star sensor; judging whether a plurality of targets exist in the window graph or not, if so, entering the next step, otherwise, ending; judging whether a plurality of targets in the window image are star points or not, removing non-star points, if the number of the star points in the window image after removal is greater than 1, entering the next step, and otherwise, ending; and identifying a navigation star from the remaining star points. According to the method, the problems that a multi-target window is misjudged as a single-target window and an interference target is misjudged as a navigation star point in the prior art can be avoided, and the centroid positioning precision and the attitude measurement precision are effectively improved.

Description

Technical Field

[0001] The invention belongs to the field of spacecraft control and relates to an interference identification method for a star sensor. Background Art

[0002] Star sensors are optical sensors that use stars as attitude reference sources. They have the advantages of high accuracy, almost no drift, and high reliability, and have become key attitude measurement components for satellites. With the increasing requirements for satellite attitude control accuracy, higher and higher requirements are placed on the technical indicators of star sensors. As the sensitivity of star sensors continues to increase, multiple star points or non-cooperative targets will appear in some window images in the window tracking mode, resulting in misidentification of navigation stars and attitude jumps.

[0003] When performing star identification in the traditional window tracking mode, first calculate the window mean, threshold, and standard deviation, traverse the valid pixels in the window, extract the number of star points and star point information in the output window, and remove pseudo stars if the number of extracted stars is greater than 1. After pseudo star removal, if there is only one star point in the window image, calculate the centroid coordinates of the star point, otherwise the star point with the largest energy is assumed to be the navigation star, and calculate its centroid coordinates. Otherwise, the star point with the largest energy is assumed to be the navigation star, and calculate the centroid coordinates. In this algorithm flow, the first-order centroid method is generally used for window star point traversal, and the pseudo star point removal criteria include the number of pixels in the star point column direction, the number of pixels in the star point row direction, the relationship between the total number of pseudo star point pixels and the minimum and maximum number of star point pixels when distinguishing pseudo stars, and the ratio of the total number of rows and columns to the total number of pixels.

[0004] Based on the above algorithm flow, two types of abnormal window images cannot be identified and processed: 1) If there are two targets in the window, but the two targets are too close to each other and are stuck together, the first-order centroid method will mistakenly think that there is only one target when extracting star points, and regard the target as the correct navigation star point, which will increase the centroid extraction error; 2) If there are two targets in the window, but the energy of the two targets is close or the energy of the interference target is greater than the navigation star point, according to the pseudo-star removal conditions, the navigation star point will be removed, leaving the interference target, and the centroid extraction error will increase. The judgment and processing methods of this part of the multi-star window need to be solved. Summary of the invention

[0005] The technical problem solved by the present invention is: to overcome the shortcomings of the prior art and provide a multi-star interference identification method for a high-sensitivity star sensor, which can determine the number of targets in the window, determine whether the target is a star point and obtain its energy and position information on the basis of obtaining a window image, and identify the correct navigation star on this basis, thereby avoiding misidentification and attitude deviation caused by multi-star interference.

[0006] The technical solution of the present invention is: a method for identifying multi-star interference of a high-sensitivity star sensor, comprising the following steps:

[0007] (1) Obtain a window image using a star sensor;

[0008] (2) Determine whether there are multiple targets in the window image. If there are multiple targets, go to step (3); if there are no multiple targets, end;

[0009] (3) Determine whether the multiple targets in the window image are star points, and eliminate non-star points. If the number of star points in the window image is greater than 1 after elimination, go to the next step; if the number of star points in the window image is equal to 1 after elimination, take the only star point as the navigation star and end;

[0010] (4) Identify the navigation star from the remaining star points.

[0011] Further, the determination of whether there are multiple targets in the window image is specifically as follows:

[0012] (21) Calculate the average value mean and standard deviation std of the background gray level using a square area with a window image edge width of K pixels, and construct a background threshold threshold = mean + X * std, where X is a threshold coefficient;

[0013] (22) Traverse the pixels in the entire window image. If the gray level value of a certain pixel is greater than the background threshold threshold and greater than the gray level values of all its adjacent pixels around it, then determine that the pixel is an energy peak, record its gray level value and coordinate position [DN, x, y], and at the same time, the energy peak number peak_number in the window image is cumulatively incremented by 1, and the initial value of peak_number is 0;

[0014] (23) After the traversal is completed, if peak_number > 1, it is determined that there are multiple targets in the window image.

[0015] Preferably, the value range of K is 2.3 to 5.

[0016] Further, the determination of whether the multiple targets in the window image are star points is specifically as follows: According to the recorded information [DN, x, y] of each energy peak, sum the gray level values of the 4 pixel points above, below, left, and right at the peak, that is, sum = DN1 + DN2 + DN3 + DN4. When sum >= 4 * threshold, it is determined that the position corresponding to the energy peak information is a star point.

[0017] Further, to determine whether multiple targets in the judgment window graph are star points specifically includes: according to each piece of recorded energy peak information [DN, x, y], a window of m*m is opened with the peak coordinates (x, y) as the center, and the pixels within the window are traversed. If the gray value of a certain pixel is greater than the background threshold threshold, the number of valid pixels pixel_number within the window is incremented by 1, and the initial value of pixel_number is 0. If pixel_number >= m*m / 2, it is determined that the peak is a star point, and m is the window width represented by the number of pixels.

[0018] Preferably, the value range of m is 3 to 5.

[0019] Further, to identify navigation stars from the remaining star points specifically includes: according to each piece of information [DN, x, y] of each recorded star point, this window graph is compared with other window graphs containing only a single star at the same moment. Each star point in this window graph is individually identified, and the identification method is the same, and the following process is executed for all: calculate the angular distance value between the star point in this window graph and the star points in other window graphs containing only a single star. If the difference between each calculated angular distance value and the theoretical angular distance value of the two corresponding stars involved in the calculation in the star catalog is less than the angular distance measurement error of the star sensor, it is considered that the star point in this window graph is a navigation star.

[0020] Further, to identify navigation stars from the remaining star points specifically includes: according to each piece of information [DN, x, y] of each recorded star point, this window graph is compared with the previous frame window graph. If the difference between the star point positions in the previous frame window graph and the current window graph is less than the position threshold, and the difference between the gray value of the star point in the previous frame window graph and the current star point gray value is less than the gray threshold, it is considered that the star point is a navigation star.

[0021] Preferably, the position threshold is the angular velocity of the aircraft where the star sensor is located * dt + the angular resolution of the star sensor, where dt is the time interval between the acquisition times of two frame window graphs.

[0022] Preferably, the gray threshold is the standard deviation std of the background gray value.

[0023] The advantages of the present invention compared with the prior art are as follows: First, a recognition method for a multi-target window graph, namely an energy peak search algorithm, is proposed. After obtaining the window graph, the background threshold of the window graph is calculated, and the window graph is traversed to find pixels with gray values greater than the threshold and greater than other surrounding pixels, and the pixel is marked as an energy peak. After the full graph traversal is completed, multiple such energy peaks can be recorded, including their positions and gray levels. Then, a true and false multi-star judgment method is proposed, which includes two criteria: the total energy criterion and the number of effective pixels criterion. That is, on the basis of obtaining the energy peak information, a small window of M*M is selected with the peak as the center, and it is judged whether the peak is a star point according to the total energy of the four pixels above, below, left, and right at the peak and the number of pixel points with gray values greater than the threshold in the small window, and then it is selected to be removed or retained. If the number of star points retained in the window is greater than 1, a navigation star judgment will be performed to identify the correct navigation star for attitude calculation. Based on the method of the present invention, the problems of misjudging a multi-target window as a single-target window and misjudging an interference target as a navigation star point in the prior art can be avoided, and the centroid positioning accuracy and attitude measurement accuracy can be effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a flow chart of the method of the present invention;

[0025] Figure 2 is a schematic diagram of the selection of a square area when calculating the background gray level of the present invention, where K = 3 in the figure;

[0026] Figure 3 is an example of the energy peak search of the present invention;

[0027] Figure 4 is a schematic diagram of the comparison and judgment between the previous and subsequent frames of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0028] The main information source of the method of the present invention is the window graph, the main recognition object is the number of star points, and the main content is the star point criterion and recognition.

[0029] Specifically, as Figure 1 shown, is a flow chart of the method of the present invention, which mainly includes:

[0030] 1. Obtain a window graph image containing multiple targets

[0031] After the star sensor is powered on, it images the spatial star background. The obtained original star map (full map) is processed by FPGA or software. After image filtering, the effective point targets in the original star map are extracted, and a window is opened with the point target as the center, thereby generating a large number of window graphs.

[0032] During subsequent operations, the star sensor can predict the target position at the next moment based on the target position and angular velocity in the window image generated at the previous moment, and then open a window at the predicted position to directly obtain the window image.

[0033] 2. Multi-target window image recognition

[0034] The present invention proposes an energy peak search algorithm, which can determine whether there are multiple targets in the input window image.

[0035] When a target or noise appears in the window, it will cause the gray value of the corresponding position coordinate to be higher than the background. Therefore, by traversing the pixels in the window image, searching for the gray peak, and judging whether the peak can become a star point, the effective discrimination of the number of targets in the window can be realized. The specific steps are as follows:

[0036] 2.1 Background threshold setting

[0037] Use a square area with an edge width of K pixels in the window image to calculate the average value mean and standard deviation std of the background gray level. For example, if K = 3, then take the area of 3 rows and 3 columns at the edge of the window image for calculation, as Figure 2 shown. The selection of the K value is related to the size of the window image and the image uniformity.

[0038] The background threshold threshold = mean + X * std, where X is the threshold coefficient, which is adjusted according to the background noise size of the window image and the algorithm accuracy rate. The general value range is 2.3 to 5.

[0039] 2.2 Energy peak search

[0040] Traverse all the pixels in the window image. If the gray value of a pixel is greater than the threshold threshold and greater than the gray values of n * n - 1 (open a small square window centered on this pixel, the side length of this small window is n pixels, and there are n * n - 1 pixels except this pixel in the small window) pixels around it, then this pixel is considered an energy peak, record its gray value and coordinate position [DN, x, y], and at the same time, the energy peak number peak_number in this window image is cumulatively incremented by 1 (the initial value of peak_number = 0).

[0041] Figure 3 In the energy peak search example shown, DN > threshold, and DN > DN12, DN1, DN13, DN2, DN3, DN4, DN24, DN34, with a total of 8 pixels around.

[0042] 2.3 Multi-target window image judgment

[0043] If the number of energy peaks peak_number > 1 in the window image, it is considered that there are multiple targets in the window image and further processing is performed; otherwise, the window image is considered normal, that is, there is only a single target.

[0044] 3. True and False Multi-star Judgment

[0045] It includes the total energy criterion method and the effective pixel number criterion method, and either of the two algorithms can be selected. It can be judged whether the multiple targets in the window image are star points or interferences.

[0046] 3.1 Total Energy Criterion

[0047] According to each energy peak information [DN, x, y] recorded in step 2.2, sum the gray values of the 4 pixel points above, below, left, and right of the peak, that is, sum = DN1 + DN2 + DN3 + DN4. When sum >= 4 * threshold, it is determined that the peak is a star point.

[0048] 3.2 Effective Pixel Number Criterion

[0049] According to each energy peak information [DN, x, y] recorded in step 2.2, open a small window of m * m (the value range of m is 3 to 5) centered on the peak coordinates (x, y), traverse the pixels in the small window. If the gray value of a certain pixel is greater than the threshold (threshold), the number of effective pixels pixel_number in the window is cumulatively incremented by 1 (the initial value is 0). If pixel_number >= m * m / 2, it is determined that the peak is a star point.

[0050] 3.3 Retain Star Points

[0051] Eliminate non-star points. If the number of star points star_number > 1 in the window image, proceed to the next step. If the number of star points star_number = 1 in the window image, it is determined that there is only a single star point.

[0052] 4. Navigation Star Judgment

[0053] It includes the front and back frame comparison criterion method and the angular distance comparison criterion method, and either of the two algorithms can be selected. It can be judged the correct navigation star among the multiple star points in the window image.

[0054] 4.1 Front and Back Frame Comparison Criterion

[0055] According to the information [DN, x, y] of each star point recorded in step 3, compare the current window image with the previous frame window image. If the difference between the star point position in the previous frame window image and the current star point position is less than the threshold, and the difference between the star point gray value in the previous frame window image and the current star point gray value is less than the threshold, then it is considered that the star point is the correct navigation star.

[0056] As shown Figure 4 Let the exposure time of the current frame be \(t\), and the exposure time of the previous frame be \(t - dt\). Then the criterion is:

[0057] [x(t)-x(t - dt)]^2+[y(t)-y(t - dt)]^2<(threshold_distance)^2

[0058] DN(t)-DN(t - dt)<threshold_DN

[0059] where, threshold_distance = angular velocity of the aircraft * dt + angular resolution of the star sensor, threshold_DN = standard deviation std.

[0060] If it is the first frame currently, the following criterion of angular distance comparison in 4.2 is used for judgment.

[0061] 4.2 Criterion of Angular Distance Comparison

[0062] According to the information [DN, x, y] of each star point recorded in step 3, this window image is compared with other window images with only a single star at the same time. When comparing, for each star point in this window image, the following judgment process is executed separately:

[0063] Calculate the angular distance value (simply referred to as the calculated value) between the star point in this window image and the star points in other window images with only a single star at the same time. If the difference between the calculated value and the angular distance value (theoretical value) of the corresponding two stars in the star catalog is less than the angular distance measurement error of the star sensor, it is considered that the star point in this window image is the correct navigation star.

[0064] An example is as follows:

[0065] Let the serial number of this window image be \(a\), and there are two other window images with only a single star at the same time, and their serial numbers are \(b\) and \(c\) respectively. Then the criterion is:

[0066] [x(a)-x(b)]^2+[y(a)-y(b)]^2<(angular distance measurement error of the star sensor + angular distance theoretical value in the star catalog)^2[x(a)-x(c)]^2+[y(a)-y(c)]^2<(angular distance measurement error of the star sensor + angular distance theoretical value in the star catalog)^2

[0067] The above ^2 all represents the square operation.

[0068] The content not described in detail in the specification of the present invention belongs to the well-known technology in the art.

Claims

1. A method for discriminating multi-star interference of a high-sensitivity star sensor, characterized in that: It includes the following steps: (1) Use a star sensor to obtain a window image; (2) Determine whether there are multiple targets in the window image. If there are multiple targets, go to step (3); if there are no multiple targets, end; (3) Determine whether the multiple targets in the window image are star points, and eliminate non-star points; If the number of star points in the window image is greater than 1 after elimination, go to the next step; if the number of star points in the window image is equal to 1 after elimination, use the only star point as the navigation star and end; (4) Identify the navigation star from the remaining star points.

2. The multi-star interference discrimination method for a high-sensitivity star sensor according to claim 1, characterized in that: The determination of whether there are multiple targets in the window image is specifically as follows: (21) Calculate the average value mean and standard deviation std of the background gray level using a square area with a window image edge width of K pixels, and construct a background threshold threshold = mean + X * std, where X is a threshold coefficient; (22) Traverse the pixels in the entire window image. If the gray level value of a certain pixel is greater than the background threshold threshold and greater than the gray level values of all its adjacent pixels around it, determine that the pixel is an energy peak, record its gray level value and coordinate position [DN, x, y], and at the same time, the energy peak number peak_number in the window image is incremented by 1. The initial value of peak_number is 0; (23) After traversal, if peak_number > 1, determine that there are multiple targets in the window image.

3. A method for discriminating multi-star interference of a high-sensitivity star sensor according to claim 2, characterized in that: The value range of K is 2.3 to 5.

4. A method for discriminating multi-star interference of a high-sensitivity star sensor according to claim 2, characterized in that: The determination of whether the multiple targets in the window image are star points is specifically as follows: According to each recorded energy peak information [DN, x, y], sum the gray level values of the 4 pixel points above, below, left, and right at the peak, that is, sum = DN1 + DN2 + DN3 + DN4. When sum >= 4 * threshold, determine that the position corresponding to the energy peak information is a star point.

5. A method for discriminating multi-star interference of a high-sensitivity star sensor according to claim 2, characterized in that: The determination of whether the multiple targets in the window image are star points is specifically as follows: According to each recorded energy peak information [DN, x, y], open an m * m window centered on the peak coordinates (x, y), traverse the pixels in the window. If the gray level value of a certain pixel is greater than the background threshold threshold, the number of valid pixels pixel_number in the window is incremented by 1. The initial value of pixel_number is 0; if pixel_number >= m * m / 2, determine that the peak is a star point, and m is the window width represented by the number of pixels.

6. A method for discriminating multi-star interference of a high-sensitivity star sensor according to claim 5, characterized in that: The value range of m is 3 to 5.

7. A method for discriminating multi-star interference of a high-sensitivity star sensor according to claim 4 or 5, characterized in that: Identifying the navigation stars from the remaining star points specifically includes: according to the information [DN, x, y] of each recorded star point, comparing the current window image with other window images that contain only a single star at the same moment. Each star point in the current window image is identified individually, and the identification method is the same, and all perform the following process: calculate the angular distance value between the star point in the current window image and the star points in other window images that contain only a single star. If the difference between each calculated angular distance value and the theoretical angular distance value of the two corresponding stars participating in the calculation in the star catalog is less than the angular distance measurement error of the star sensor, then it is considered that the star point in the current window image is a navigation star.

8. A method for discriminating multi-star interference of a high-sensitivity star sensor according to claim 4 or 5, characterized in that: Identifying the navigation stars from the remaining star points specifically includes: according to the information [DN, x, y] of each recorded star point, comparing the current window image with the previous frame window image. If the difference between the star point position in the previous frame window image and the star point position in the current window image is less than the position threshold, and the difference between the star point gray value in the previous frame window image and the current star point gray value is less than the gray value threshold, then it is considered that the star point is a navigation star.

9. A method for discriminating multi-star interference of a high-sensitivity star sensor according to claim 8, characterized in that: The position threshold is the angular velocity of the aircraft where the star sensor is located * dt + the angular resolution of the star sensor, where dt is the time interval between the acquisition times of two frame window images.

10. A method for discriminating multi-star interference of a high-sensitivity star sensor according to claim 8, characterized in that: The gray value threshold is the standard deviation std of the background gray value.

Images