Star sensor error correction method based on star point slice image

Abstract

The invention discloses a star point slice image-based star sensor error correction method. The method comprises a vacuum partition calibration method, a star point slice image data structure design method, a predicted star point distortion correction method and a star point coordinate trajectory fitting method. The vacuum partition calibration comprises the steps of carrying out vacuum preset calibration on the ground and dividing view field partition calibration according to an imaging area; the data structure of the star point slice image comprises coordinates, gray scale information, a star catalogue sequence number and telemetry data packet information of an original slice image of a detection star; the predicted star point distortion correction method comprises the following steps: the star sensor corrects navigation star point coordinates on the basis of current attitude information; the star point coordinate trajectory fitting method comprises the steps of solving a star point slice image fixed star target mass center, carrying out uniform distribution processing on star point target image gray scale data through a bilinear difference value method, and filtering the fixed star target mass center according to multi-frame star point slice image data. And carrying out prediction and filling on the intermittent jump star point or dark weak star point target according to the star point drawing track.

Description

Technical Field

[0001] This invention relates to the field of satellite star sensor attitude measurement technology, and specifically to a star sensor error correction method based on star point slice images. Background Technology

[0002] Star sensors are the most precise attitude measurement sensors. They image the night sky using image sensors, measure the components of stellar vectors in the star sensor's coordinate system, and use the precise positions of known stars to determine the satellite's three-axis attitude relative to the inertial coordinate system. They have become indispensable attitude sensors on satellites. Existing satellites typically acquire attitude measurement data from star sensors in orbit, providing attitude measurement and determination information for the attitude and orbit control subsystem and a measurement reference for the payload. Because the optical system of a star sensor has a large field of view, the same star point will exhibit aberrations at the edge and center of the field of view. Factors such as rocket launch vibration and space radiation temperature cause differences between ground calibration parameters and in-orbit operating conditions. The measurement error of the star sensor plays a crucial role in satellite attitude determination; the accuracy of satellite nadir positioning mainly depends on the attitude measurement error of the star sensor.

[0003] Current methods for improving the measurement accuracy of star sensors are all based on star sensor products, utilizing their optical systems, hardware circuits, structures, software, and algorithms. However, because on-board products are designed for in-orbit applications, they are generally limited by factors such as space, size, cost, and reliability. Summary of the Invention

[0004] The technical problem solved by this invention is to overcome the shortcomings of the prior art and provide a star sensor error correction method based on star point slice images, which solves the problem that the attitude measurement accuracy of existing satellite star sensors does not meet the requirements of high-precision imaging and ground processing correction.

[0005] The technical solution of this invention is: a star sensor error correction method based on star point slice images, comprising:

[0006] When the star sensor is working in orbit, it acquires star point slice image data, obtains the star sensor optical system parameters under vacuum conditions through the vacuum partition calibration method, and loads the optical system parameters into the star sensor default calibration parameters through a preset method.

[0007] The star sensor uses default calibration parameters to perform on-orbit imaging and attitude measurement, obtains star sensor attitude measurement information, and outputs star point slice image data and transmits it to the ground.

[0008] The ground server corrects the navigation star coordinates using a predictive star distortion correction method based on the star sensor attitude measurement information and star point slice image data.

[0009] The star sensor error correction is completed by processing the star point coordinates through bilinear interpolation and star point centroid filtering based on the corrected navigation star point coordinates.

[0010] The method of reducing the difference between the center and periphery of the star point slice image data through vacuum partition calibration to obtain the optical system parameters of the star sensor under vacuum conditions includes:

[0011] The parameters of the star sensor optical system under vacuum conditions are obtained through a vacuum partition calibration method. The imaging area is divided into multiple regions at fixed intervals according to a preset number of partitions. The calibration points of each region are calibrated individually, and the parameters of the star sensor optical system in each region are calculated. The specific steps are as follows:

[0012] Mount the star sensor on the two-axis rate position turntable, and turn on the collimator of the star simulation light source located opposite the star sensor after powering on; set the magnitude and adjust the gain until the centroid positioning accuracy of the star point is less than 0.05 pixels;

[0013] The collimator and turntable of the stellar simulation light source are both placed inside a vacuum chamber; the vacuum chamber is closed, and a vacuum is drawn to maintain a vacuum level of ≤1×10⁻⁶. -3 Under a pressure of Pa, the vacuum tank temperature is set and stabilized. When the temperature change rate of the lens, detector, and housing of the star sensor is less than 1℃ / hour, the star sensor is deemed to be thermally stable, and parameter calibration begins.

[0014] Adjust the two-axis rate position turntable so that the star point of the simulated star light source collected by the star sensor reaches the center point of the pixel plane of the star sensor detector.

[0015] The star sensor's field of view is divided into n imaging regions. Each imaging region is further divided into multiple sub-imaging regions at fixed intervals, with the star sensor's main point as the center. Each sub-imaging region is then further divided into multiple unit regions. The calibration points of each unit region are calibrated individually. The rotation and flip axes of the two-axis rate-position turntable are set to move within each unit region, and the centroid data of the star points at the calibration points are collected. Each calibration point is collected 15 times, and the average value is calculated as the measured calibration value for each calibration point. Based on the position of the two-axis rate-position turntable and the measured calibration values ​​of each calibration point, the parameters of the star sensor's optical system are calculated.

[0016] The star point slice image data includes the coordinates of the original slice image of the probe star, the grayscale of the diffuse spot pixels of the original slice image of the probe star, the centroid coordinates and grayscale values ​​of the original slice image of the probe star, the centroid coordinates, grayscale values ​​and star catalog numbers of the attitude-determining star and the tracking star, the number of tracking stars and the number of processing gates, the status word, the attitude quaternion and the UTC time.

[0017] When the star point slice image data is transmitted to the ground, an asynchronous communication method is used, namely 1 start bit, 8 data bits, 1 odd parity bit, and 1 stop bit.

[0018] When transmitting the star point slice image data to the ground, a multi-byte data transmission method is used, that is, the highest byte data is transmitted first, then the second highest byte data is transmitted, and finally the lowest byte data is transmitted; in the transmitted bit stream, the least significant byte comes first and the most significant byte comes last.

[0019] The star slice image data are arranged in ascending order of apparent star magnitude. Each star slice image data is arranged from left to right along the X-axis of the detector image plane, and then from bottom to top along the Y-axis of the detector image plane. The positions of the first three pixels of each star slice image data are used to store the x-coordinate of the lower left corner of the star slice image, the y-coordinate of the lower left corner of the star slice image, and the tracking star index ID. The X-axis points to the horizontal axis of the detector image plane, and the Y-axis points to the vertical axis of the detector image plane.

[0020] The ground server corrects the navigation star point coordinates based on star sensor attitude measurement information and star point slice image data using a predicted star point distortion correction method, including:

[0021] Based on the attitude quaternion q(t) of the kth meteor sensor k ), calculate the predicted value of the next quaternion. Based on predictive quaternions The attitude transfer matrix A is calculated as follows:

[0022]

[0023] In the formula, the attitude transfer matrix A has 9 elements, which are a, ... 11 ,a 12 ,a 13 ,a 21 ,a 22 ,a 23 ,a 31 ,a 32 ,a 33 ;

[0024] Based on the obtained attitude transfer matrix A, calculate the coordinates r of the star sensor optical axis vector in the inertial frame;

[0025] Find the coordinates of the star sensor's optical axis vector in the inertial frame within the navigation star catalog. The set of navigation stars S with an included angle less than half the field of view I :

[0026]

[0027] In the formula, N is the total number of navigation satellites within the field of view, and v l Let v be the coordinates of the i-th navigation satellite in the inertial frame. l1 v l2 v l3 For v l The three-axis components in the inertial frame;

[0028] Let the coordinates of the l-th navigation satellite in the inertial frame be... If the focal length of the lens is f, then the mapped coordinates (x, y) of the navigation satellite on the detector target surface are:

[0029]

[0030] Let the principal point coordinates of the star sensor's optical lens on the detector target surface be (x0, y0). Based on this, distortion correction is performed, with radial distortion p1, p2 and tangential distortion q1, q2. Then the corrected navigation star point coordinates (x′, y′) are:

[0031]

[0032] The process of processing the star point coordinates based on the corrected navigation star point coordinates using bilinear interpolation and star point centroid filtering includes:

[0033] The centroid coordinates of the star target are obtained based on the star point slice image data of each frame. The grayscale data of the star target image is uniformly distributed by bilinear interpolation to correct the grayscale and centroid of the non-photosensitive area of ​​the pixel. Then, the grayscale mean and threshold segmentation methods are used to segment the star target and the background image.

[0034] Based on multi-frame star point slice image data, the centroid of the star points is filtered to realize the continuous motion of the star target on the image plane of the star sensor detector.

[0035] The process involves calculating the centroid coordinates of the star target based on the star point slice image data of each frame, uniformly distributing the grayscale data of the star target image using bilinear interpolation to correct the grayscale and centroid effects of the non-photosensitive area of ​​the pixels, and then segmenting the star target and background image using grayscale mean and threshold segmentation methods, including:

[0036] The image grayscale of each star point slice is obtained, and the grayscale mean m is calculated. Then, the star point extraction threshold T = m + offset is determined according to the preset threshold offset. Each pixel in the star point slice image is binarized according to the star point extraction threshold, with grayscale values ​​greater than the star point extraction threshold being 1 and those less than the threshold being 0. The gate image is labeled using the four-connected domain criterion to obtain the star point map slice image, and the centroid coordinates of the star point map slice image are calculated using the following formula:

[0037]

[0038] Where x i y i g i These represent the x-coordinate, y-coordinate, and grayscale value of each pixel that makes up the star point;

[0039] Assume the source image of the star point slice of the star sensor is m×n in size, and the interpolated image is a×b. The side length ratios of the two images are m / a and n / b, respectively. The (i,j)th pixel of the interpolated image can be mapped back to the source image through the side length ratio, and its corresponding coordinates are (i×m / a,j×n / b).

[0040] The grayscale value of the stellar target at image coordinates (x,y) is f(x,y);

[0041]

[0042]

[0043] In the formula, R1 = (x, y1), R2 = (x, y2); Q 11 = (x1, y1), Q 12 = (x1, y2); Q 21 = (x2, y1), Q 22 = (x2, y2); P = (x, y);

[0044]

[0045] Then f(x,y) is:

[0046]

[0047] The step of filtering the centroid of star points based on multi-frame star point slice image data to achieve continuous motion of the stellar target on the image plane of the star sensor detector includes:

[0048] Let the centroid coordinates of the star point in the k-th frame be (x... k y k The predicted position coordinates of the target within the k+1 frame image are (x... k+1 ,y k+1 ); in (t1,t2,…,t k During time, the centroid coordinate x has k measured values ​​(x1, x2, ..., xk). k );

[0049] Let the theoretical coordinates of the centroid of the star be...

[0050]

[0051] The deviation between the measured and theoretical values ​​of the centroid of the star is:

[0052]

[0053] The sum of squares of the deviations between the measured and theoretical values ​​of the centroids of k star points is:

[0054]

[0055] By fitting a quadratic function to the centroid data of the star points, and taking the partial derivatives of E with respect to a0, a1, and a2 respectively, and setting each partial derivative equal to 0, we obtain three equations about a0, a1, and a2:

[0056]

[0057] This system of linear equations can be written in matrix form as Ma = b, where the coefficient matrix M, the unknown vector a, and the constant vector b are respectively:

[0058]

[0059] According to the method for solving linear equation systems, when the coefficient matrix M is non-singular, the solution is a = M. -1 b; By calculating the product of the inverse of matrix M and vector b, we can obtain the expressions for a0, a1, and a2; the specific expressions are obtained by Cramer's rule and then simplified through the expansion of the determinant:

[0060]

[0061] The above equation is the general solution for fitting a quadratic function of x in the sense of minimum deviation, where Δ=k(∑t i 2 ∑t i 4 -(∑t i 3 ) 2 )+∑t i (∑t i 3 ∑t i -∑t i 2 ∑t i 2 )+∑t i 2 (∑t i ∑t i 3 -∑t i 2 ∑t i 2Substituting the obtained values ​​of a0, a1, and a2 into the equation, the target value at time t is calculated. k+1 Predicted value at time :

[0062]

[0063] The beneficial effects of this invention are:

[0064] (1) This invention utilizes the acquisition of star sensor to detect the original slice image of star point area, re-determines the centroid coordinates of the image, performs trajectory fitting on star point targets within the field of view, fills the flickering area of ​​faint star targets, and performs real-time correction of optical parameters such as focal length and distortion. This can effectively reduce the low-frequency error of field of view period caused by the distortion of the optical lens at the edge of the image plane, the high-frequency error at the pixel level caused by the pixel fill factor of the APS detector, and the attitude calculation error caused by the instability of star point extraction.

[0065] (2) This invention is verified by ground-based star observation and on-orbit data, which can effectively improve the attitude measurement accuracy of star sensors and provide a reference for the design of new space photoelectric sensors.

[0066] (3) The present invention can reduce the error of the optical axis angle between the star sensor and the payload, which can be used for ground processing and image correction by satellite users. Attached Figure Description

[0067] Figure 1 This is a general block diagram of a star sensor error correction method based on star point slice images according to the present invention;

[0068] Figure 2 This is a serial data bus timing diagram of a star sensor error correction method based on star point slice images according to the present invention.

[0069] Figure 3 This is a schematic diagram of a single star target on an image sensor based on star slice images, according to the present invention.

[0070] Figure 4 This is a schematic diagram of the positions of the turntable and single-star simulator inside the vacuum tank in the present invention, which is a method for correcting the error of a star sensor based on star point slice images.

[0071] Figure 5 This is a schematic diagram of the product relative to the turntable installation direction of the star sensor error correction method based on star point slice images according to the present invention.

[0072] Figure 6 This is a schematic diagram of bilinear interpolation coordinates of star points based on star point slice images according to the present invention;

[0073] Figure 7This is a schematic diagram of the trajectory fitting of the centroid of a star point in the image sensor imaging of a star sensor based on a star point slice image error correction method according to the present invention. Detailed Implementation

[0074] like Figure 1 As shown, this invention provides a star sensor error correction method based on star point slice images. It obtains information such as satellite designation, orbit number, stripe number, original star sensor image data, output data storage path, and local working path by reading "*.xml" files, and then performs star sensor attitude determination processing.

[0075] The API for processing star point and star map slice data transmits relevant information via an XML file. Specific fields and descriptions are shown in Table 1.

[0076] Table 1 StarMiner Input XML Information Table

[0077]

[0078]

[0079] The input file for the star point and star map slice data processing program is the original star map data (star point and star map slice data package), in .dat format, with a data length of approximately 24M bytes (10 minutes of data). The data in this input file has already been filtered according to the frame header.

[0080] Telemetry data from star sensors typically includes quaternions, time, valid flags, image grayscale, number of stars, temperature, and other information. The data packets range from tens to hundreds of bytes. They usually communicate with the attitude and orbit control computer using an RS422 asynchronous communication interface. The electrical signals conform to the ANSI / TIA / EIA-422 standard, with a data transmission rate of 115200 bps and a request / response frequency of approximately 4 Hz to 8 Hz.

[0081] The star-slice image data packet is 4096 bytes long and transmitted at 10Hz, approximately 128 times the original data volume, affecting serial port transmission capabilities. The communication rate is 921.6kbps. Asynchronous communication is used: 1 start bit, 8 data bits, 1 parity bit, and 1 stop bit; multi-byte data transmission: the most significant byte is transmitted first, followed by the second most significant byte, and finally the least significant byte. In the transmitted bit stream, the least significant byte comes first, followed by the most significant byte. The timing of the star sensor serial data bus is shown below. Figure 2 As shown.

[0082] The star point and star map slice data packets are sent according to the star sensor frame period. One packet is sent every frame period, and the length of each packet is 4k bytes. The frame header of each frame is 8AA2 53 53. The detailed data structure of each frame is shown in Table 2.

[0083] Table 2. Data packet format for star points and star map slices

[0084]

[0085] After the star point and star map slice data are processed, the relevant information is transmitted via an XML file. Specific fields and descriptions are shown in Table 3. The output XML file is stored in the "Star Affinity Attitude Determination Processing Result File Path".

[0086] Table 3 StarMiner Output XML Information Table

[0087]

[0088] Based on multi-frame data and following the algorithm function module entry format, the star point and star map slice data packets are parsed and converted into floating-point numbers stored in columns. The output file contains four parts: UTC time, ETR second pulse time, quaternion attitude data, and valid identifiers. The source code needs to parse the data into floating-point numbers, store them in columns, and store them in a txt text format, as shown in Table 4.

[0089] Table 4. Definition of Data Packet Output Format for StarMiner Ground Program

[0090]

[0091]

[0092] UTC time represents the midpoint of the star map exposure corresponding to the current frame's quaternion, synchronized via command, and is in milliseconds (ms). ETR second pulse time represents the midpoint of the star map exposure corresponding to the current frame's quaternion, synchronized via an external second pulse signal, and is in milliseconds (ms). Valid identifiers: 1 for valid, 0 for invalid.

[0093] The specific steps are as follows:

[0094] Step 1: Prepare the "*.xml" input interface file, ensuring the file structure matches the attached file. Figure 3 The structures are identical; the StarFilePath field indicates the path to the input telemetry packet, and the WorkDir field indicates the path to the output file of the star-sensor attitude determination processing.

[0095] Step 2: Place the prepared "*.xml" input interface file in the same directory as "program".

[0096] Step 3: Execute the program command, one of the input parameters of which is the prepared input interface file.

[0097] Step 4: The output results are stored in the path recorded in WorkDir, and a StarMiner output XML file is also saved in the same directory.

[0098] The vacuum partition calibration method reduces the difference between the star sensor target imaging star map slice image at the center and periphery of the field of view. It includes: calibrating and preset the star sensor focal length, principal point and distortion parameters in a vacuum environment; dividing the imaging area into multiple regions at fixed intervals according to a preset number of partitions; calibrating the calibration points of each region separately; and calculating the correction function of each region.

[0099] In the vacuum calibration chamber, a single-star simulator and a two-dimensional turntable are fixed inside a vacuum container, with the rotation axis of the two-dimensional turntable coinciding with the parallel beam emitted from the single-star simulator, as shown in the attached diagram. Figure 4 As shown in the attached diagram. The star sensor is mounted and fixed on a two-axis rate position turntable. Figure 5 The star sensor shown is positioned with its X-axis at 0° to the turntable's X′ direction and its Z-axis at 0° to the turntable's Z′ direction, while its Y-axis aligns with the turntable's Y′ direction, directly below the turntable's rotation axis. Turn on the simulated star light source, set the magnitude, and adjust the gain until the star's centroid positioning accuracy is less than 0.05 pixels.

[0100] Vacuum partition calibration method: Mount the star sensor on a two-axis rate position turntable and power it on. Turn on the stellar simulation light source, set the magnitude, and adjust the gain until the centroid positioning accuracy of the star point is less than 0.05 pixels. Close the vacuum tank and evacuate it to maintain a vacuum pressure of ≤1×10-3 Pa. Set the vacuum tank temperature and allow it to stabilize. Power on the star sensor and record the lens temperature, detector temperature, and housing temperature. When the lens temperature change rate is less than 1℃ / hour, the star sensor is considered thermally stable, and parameter calibration is performed. Adjust the turntable so that the star point acquired by the star sensor from the stellar simulation light source reaches the center point. Based on the star sensor's field of view (18°×18°), considering the margin at the edge of the field of view, set the rotation axis and flip axis of the turntable to move within a range of ±8°. Gradually rotate the rotation axis and flip axis of the two-dimensional turntable. The star sensor's field of view is divided into n regions. The imaging area is then divided into multiple regions at fixed intervals, with the star sensor's principal point as the center of each region. Each region's calibration point is individually calibrated, and the correction function for each region is calculated to ensure that the calibration residual is independent of the distance from the sampling point to the center of the field of view. Star centroid data for each calibration point is collected 15 times, and the average value is calculated as the measured calibration value. The calibration coefficient is calculated from the turntable position and the measured calibration value. The vacuum chamber temperature is adjusted, and the calibration coefficients under different lens temperature conditions are calculated. The effect of temperature on the calibration coefficients is analyzed. After restoring the vacuum chamber to atmospheric pressure and the star sensor housing temperature to room temperature, the star sensor is powered off and the vacuum chamber is removed, ending the experiment.

[0101] Furthermore, the star sensor error correction method based on star point slice images obtains the centroid coordinates of each probe star in each frame by means of bilinear interpolation, gray-scale mean and threshold calculation, and star point centroid extraction, based on the star point slice image data.

[0102] All star sensor error correction methods based on star point slice images include predicted star point distortion correction. After the star sensor extracts the centroid coordinates of the detected star target, it needs to match them with the predicted navigation star point coordinates. Due to optical system distortion, if the distortion is not corrected, star points at the edge of the field of view may fail to match or mismatch due to larger errors. Therefore, distortion correction of the predicted navigation star point coordinates is necessary. After the star sensor successfully identifies the target star throughout the day, it switches to star tracking mode. The application software then uses the current quaternion q(t) to... k ), calculate the predicted value of the next quaternion. Based on predictive quaternions Calculate the attitude transition matrix A.

[0103] The coordinates of the star sensor's optical axis vector in the inertial frame can be calculated based on the attitude transfer matrix.

[0104] Finding and predicting the optical axis vector in the navigation star catalog The set of navigation stars S with an included angle less than half the field of view I Where N is the total number of navigation satellites within the field of view:

[0105] Let the coordinates of the navigation satellite in the inertial frame be... With a lens focal length of f, the attitude transfer matrix A corresponding to the predicted quaternion Q can be used to obtain the mapping coordinates of the navigation satellite on the detector target surface.

[0106] Let the projected coordinates of the navigation satellite be (x, y), in mm, and the principal point coordinates be (x0, y0), in mm. Distortion correction is then performed based on these coordinates. The vector of each satellite is calculated using the fitted probe trajectory and the corrected focal length. Attitude quaternions are calculated using the ESO2Q algorithm. The attitude quaternions are output, along with UTC time, second pulse time, attitude quaternion information, and valid flags, packaged and stored in TXT text format. An XML information table is also output.

[0107] In this system, the grayscale changes of each pixel in the star slice image of the target star detected by the star sensor are not continuous. To further improve the accuracy of star centroid extraction, in addition to the conventional grayscale mean and threshold calculations, a bilinear interpolation algorithm is used. Assume the source image of the star slice from the star sensor is m×n, and the interpolated image is a×b, with side length ratios of m / a and n / b. This ratio is usually not an integer and must be stored as a floating-point number. The (i,j)th pixel (row i, column j) of the interpolated image can be mapped back to the source image using the side length ratio. Its corresponding coordinates are (i×m / a, j×n / b). Obviously, these corresponding coordinates are generally not integers, and non-integer coordinates cannot be used on discrete data like images. Bilinear interpolation calculates the grayscale value of this point by finding the four pixels closest to these corresponding coordinates. Suppose we have four pixels located at (0,0), (1,0), (0,1), and (1,1), and we want to find the value at (0.3,0.4). A diagram illustrating the bilinear interpolation algorithm is shown below. Figure 6 .

[0108] This invention provides a star sensor error correction method based on star point slice images, which further includes star point centroid filtering. Multi-frame processing is used to perform polynomial fitting on the centroid trajectory of each detected star. As the star sensor flies with the satellite, stars move at a certain speed on the detector's image plane. Under normal circumstances, the satellite moves at a speed of 0.06 degrees / second, the star sensor's field of view is 18 degrees, the detector's image plane is 2048×2048 pixels, and the star sensor's data update rate is 10 Hz. Therefore, the number of sliding star pixels in each frame of the star point slice image is 0.06 / 18×2048 / 10 = 0.68 pixels. Currently, the star library design of star sensors generally includes stars up to magnitude 6.5. Based on the star sensor's optical system design and ground calibration, the imaging area of ​​the most numerous stars (5 to 6.5 magnitudes) on the star sensor's detector image plane is generally within a 3×3 or 4×4 window range. With the currently designed star point slice image gate size of 8×8, it can be ensured that the centroids of stars in adjacent frames fall within the prediction window.

[0109] Let the centroid coordinates of the star point in the k-th frame be (x... k y k The predicted position coordinates of the target within the k+1 frame image are: In the time interval (t1, t2, ..., tk), the centroid coordinate x has k measured values ​​(x1, x2, ..., xk). Due to the pixel fill factor effect of the star sensor, the grayscale of the image changes when the star point moves across a pixel, resulting in centroid calculation errors. Simultaneously, as the star enters and leaves the edge of the optical system's field of view, the attitude-fixing star used for star sensor attitude calculation changes, causing attitude calculation errors. Discontinuous extraction of the centroid of weak stars also contributes to attitude calculation errors. These random errors in the measurement data are determined based on the measurement time ti and the measured centroid data xi. If the satellite's angular velocity is constant, ideally, the star image trajectory is a straight line only when the star sensor's optical axis points within the orbital plane. If the optical axis points perpendicular to the orbital plane, the star image is a circle rotating around the optical axis. This is because, for optical images, a symmetrical optical system is generally used, and theoretically, the star image is stationary in the inertial coordinate system. However, when imaged onto the detector, due to the star sensor's on-orbit attitude rotating around the optical axis, the star image rotates around the optical axis, resulting in a circular trajectory. When the optical axis is in a direction other than either of these two, the trajectory of the star point is itself a curve. Simultaneously, due to distortion at the edges of the field of view, the trajectory of stars may jump or become discontinuous. Furthermore, when stars pass through the non-photosensitive area of ​​the detector pixels, uneven photoelectric conversion can cause coordinate shifts in the star point. Weaker stars may not be detected, resulting in trajectory interruptions. These factors cause fluctuations in the measurement attitude of the star sensor, such as... Figure 7 As shown.

[0110] Finally, exit the star-slice image processing program.

[0111] In summary, this invention provides a star sensor error correction method based on star point slice images, which improves the accuracy of star sensor attitude measurement, reduces low-frequency errors in noise equivalent angle and field of view period, and reduces the optical axis angle error between the star sensor and the payload. It can be used by remote sensing satellite users for ground processing and correction of payload images, improve the accuracy of nadir point positioning, realize high-precision topographic mapping and centimeter-level surface deformation detection, and also provide a reference for the design of new space photoelectric sensors.

[0112] Although the present invention has been described in detail through the preferred embodiments above, it should be understood that the above description should not be considered as a limitation of the present invention. Various modifications and substitutions to the present invention will be apparent to those skilled in the art after reading the above description. Therefore, the scope of protection of the present invention should be defined by the appended claims.

Claims

1. A method for correcting star sensor errors based on star point slice images, characterized in that, include: When the star sensor is working in orbit, it acquires star point slice image data, obtains the star sensor optical system parameters under vacuum conditions through the vacuum partition calibration method, and loads the optical system parameters into the star sensor default calibration parameters through a preset method. The star sensor uses default calibration parameters to perform on-orbit imaging and attitude measurement, obtains star sensor attitude measurement information, and outputs star point slice image data and transmits it to the ground. The ground server corrects the navigation star coordinates using a predictive star distortion correction method based on the star sensor attitude measurement information and star point slice image data. The star sensor error correction is completed by processing the star point coordinates through bilinear interpolation and star point centroid filtering based on the corrected navigation star point coordinates.

2. The star sensor error correction method based on star point slice images according to claim 1, characterized in that, The method of reducing the difference between the center and periphery of the star point slice image data through vacuum partition calibration to obtain the optical system parameters of the star sensor under vacuum conditions includes: The parameters of the star sensor optical system under vacuum conditions are obtained through a vacuum partition calibration method. The imaging area is divided into multiple regions at fixed intervals according to a preset number of partitions. The calibration points of each region are calibrated individually, and the parameters of the star sensor optical system in each region are calculated. The specific steps are as follows: Mount the star sensor on the two-axis rate position turntable, and turn on the collimator of the star simulation light source located opposite the star sensor after powering on; set the magnitude and adjust the gain until the centroid positioning accuracy of the star point is less than 0.05 pixels; The collimator and turntable of the stellar simulation light source are both placed inside a vacuum chamber; the vacuum chamber is closed, and a vacuum is drawn to maintain a vacuum level of ≤1×10⁻⁶. -3 Under a pressure of Pa, the vacuum tank temperature is set and stabilized. When the temperature change rate of the lens, detector, and housing of the star sensor is less than 1℃ / hour, the star sensor is deemed to be thermally stable, and parameter calibration begins. Adjust the two-axis rate position turntable so that the star point of the simulated star light source collected by the star sensor reaches the center point of the pixel plane of the star sensor detector. The star sensor's field of view is divided into n imaging regions. Each imaging region is further divided into multiple sub-imaging regions at fixed intervals, with the star sensor's main point as the center. Each sub-imaging region is then further divided into multiple unit regions. The calibration points of each unit region are calibrated individually. The rotation and flip axes of the two-axis rate-position turntable are set to move within each unit region, and the centroid data of the star points at the calibration points are collected. Each calibration point is collected 15 times, and the average value is calculated as the measured calibration value for each calibration point. Based on the position of the two-axis rate-position turntable and the measured calibration values ​​of each calibration point, the parameters of the star sensor's optical system are calculated.

3. The star sensor error correction method based on star point slice images according to claim 1, characterized in that, The star point slice image data includes the coordinates of the original slice image of the probe star, the grayscale of the diffuse spot pixels of the original slice image of the probe star, the centroid coordinates and grayscale values ​​of the original slice image of the probe star, the centroid coordinates, grayscale values ​​and star catalog numbers of the attitude-determining star and the tracking star, the number of tracking stars and the number of processing gates, the status word, the attitude quaternion and the UTC time.

4. The star sensor error correction method based on star point slice images according to claim 3, characterized in that, When the star point slice image data is transmitted to the ground, an asynchronous communication method is used, namely 1 start bit, 8 data bits, 1 odd parity bit, and 1 stop bit.

5. The star sensor error correction method based on star point slice images according to claim 4, characterized in that, When transmitting the star point slice image data to the ground, a multi-byte data transmission method is used, that is, the highest byte data is transmitted first, then the second highest byte data is transmitted, and finally the lowest byte data is transmitted; in the transmitted bit stream, the least significant byte comes first and the most significant byte comes last.

6. The star sensor error correction method based on star point slice images according to claim 5, characterized in that, The star point slice image data are arranged in ascending order of apparent star magnitude. Each star point slice image data is arranged from left to right along the X-axis direction of the detector image plane, and then from bottom to top along the Y-axis direction of the detector image plane. The positions of the first three pixels of each star point slice image data are used to store the x-coordinate of the lower left corner of the star point slice image, the y-coordinate of the lower left corner of the star point slice image, and the tracking star table serial number ID; the X-axis points to the horizontal axis of the detector image plane, and the Y-axis points to the vertical axis of the detector image plane.

7. The star sensor error correction method based on star point slice images according to claim 5, characterized in that, The ground server corrects the navigation star point coordinates based on star sensor attitude measurement information and star point slice image data using a predicted star point distortion correction method, including: Based on the attitude quaternion q(t) of the kth meteor sensor k ), calculate the predicted value of the next quaternion. Based on predictive quaternions The attitude transfer matrix A is calculated as follows: In the formula, the attitude transfer matrix A has 9 elements, which are a, ... 11 ,a 12 ,a 13 ,a 21 ,a 22 ,a 23 ,a 31 ,a 32 ,a 33 ; Based on the obtained attitude transfer matrix A, calculate the coordinates of the star sensor optical axis vector in the inertial frame. Find the coordinates of the star sensor's optical axis vector in the inertial frame within the navigation star catalog. The set of navigation stars S with an included angle less than half the field of view I : In the formula, N is the total number of navigation satellites within the field of view, and v l Let v be the coordinates of the i-th navigation satellite in the inertial frame. l1 v l2 v l3 For v l The three-axis components in the inertial frame; Let the coordinates of the l-th navigation satellite in the inertial frame be... If the focal length of the lens is f, then the mapped coordinates (x, y) of the navigation satellite on the detector target surface are: Let the principal point coordinates of the star sensor's optical lens on the detector target surface be (x0, y0). Based on this, distortion correction is performed, with radial distortion p1, p2 and tangential distortion q1, q2. Then the corrected navigation star point coordinates (x′, y′) are:

8. The star sensor error correction method based on star point slice images according to claim 1, characterized in that, The process of processing the star point coordinates based on the corrected navigation star point coordinates using bilinear interpolation and star point centroid filtering includes: The centroid coordinates of the star target are obtained based on the star point slice image data of each frame. The grayscale data of the star target image is uniformly distributed by bilinear interpolation to correct the grayscale and centroid of the non-photosensitive area of ​​the pixel. Then, the grayscale mean and threshold segmentation methods are used to segment the star target and the background image. Based on multi-frame star point slice image data, the centroid of the star points is filtered to realize the continuous motion of the star target on the image plane of the star sensor detector.

9. The star sensor error correction method based on star point slice images according to claim 8, characterized in that, The process involves calculating the centroid coordinates of the star target based on the star point slice image data of each frame, uniformly distributing the grayscale data of the star target image using bilinear interpolation to correct the grayscale and centroid effects of the non-photosensitive area of ​​the pixels, and then segmenting the star target and background image using grayscale mean and threshold segmentation methods, including: The image grayscale of each star point slice is obtained, and the grayscale mean m is calculated. Then, the star point extraction threshold T = m + offset is determined according to the preset threshold offset. Each pixel in the star point slice image is binarized according to the star point extraction threshold, with grayscale values ​​greater than the star point extraction threshold being 1 and those less than the threshold being 0. The gate image is labeled using the four-connected domain criterion to obtain the star point map slice image, and the centroid coordinates of the star point map slice image are calculated using the following formula: Where x i y i g i These represent the x-coordinate, y-coordinate, and grayscale value of each pixel that makes up the star point; Assume the source image of the star point slice of the star sensor is m×n in size, and the interpolated image is a×b. The side length ratios of the two images are m / a and n / b, respectively. The (i,j)th pixel of the interpolated image can be mapped back to the source image through the side length ratio, and its corresponding coordinates are (i×m / a,j×n / b). The grayscale value of the stellar target at image coordinates (x,y) is f(x,y); where R1 = (x, y1), R2 = (x, y2); Q 11 = (x1, y1), Q 12 = (x1, y2); Q 21 = (x2, y1), Q 22 = (x2, y2); P = (x, y); Then f(x,y) is:

10. A star sensor error correction method based on star point slice images according to claim 9, characterized in that, The step of filtering the centroid of star points based on multi-frame star point slice image data to achieve continuous motion of the stellar target on the image plane of the star sensor detector includes: Let the centroid coordinates of the star point in the k-th frame be (x... k y k The predicted position coordinates of the target within the k+1 frame image are (x... k+1 ,y k+1 ); in (t1,t2,…,t k During time, the centroid coordinate x has k measured values ​​(x1, x2, ..., xk). k ); Let the theoretical coordinates of the centroid of the star be... The deviation between the measured and theoretical values ​​of the centroid of the star is: The sum of squares of the deviations between the measured and theoretical values ​​of the centroids of k star points is: By fitting a quadratic function to the centroid data of the star points, and taking the partial derivatives of E with respect to a0, a1, and a2 respectively, and setting each partial derivative equal to 0, we obtain three equations about a0, a1, and a2: This system of linear equations can be written in matrix form as Ma = b, where the coefficient matrix M, the unknown vector a, and the constant vector b are respectively: According to the method for solving linear equation systems, when the coefficient matrix M is non-singular, the solution is a = M. -1 b; By calculating the product of the inverse of matrix M and vector b, we can obtain the expressions for a0, a1, and a2; the specific expressions are obtained by Cramer's rule and then simplified through the expansion of the determinant: The above equation is the general solution for fitting a quadratic function of x in the sense of minimum deviation, where Δ=k(∑t i 2 ∑t i 4 -(∑t i 3 ) 2 )+∑t i (∑t i 3 ∑t i -∑t i 2 ∑t i 2 )+∑t i 2 (∑t i ∑t i 3 -∑t i 2 ∑t i 2 Substituting the obtained values ​​of a0, a1, and a2 into the equation, the target value at time t is calculated. k+1 Predicted value at time

Images