A large field-of-view automatic calibration device and calibration method for pre-calibration of star map arrays

Abstract

This invention discloses a large field-of-view automatic calibration device and method for pre-calibrating star arrays, solving the problem that existing star array pre-calibration methods cannot directly measure the relative spatial angle relationship between simulators of distant single stars. The calibration device includes a turntable, a calibration unit disposed at the working end of the turntable, and a host computer disposed outside the turntable. The calibration unit includes a mounting base and a starlight detection module, a data fusion module, and a relative spatial angle calculation module disposed within the mounting base. The mounting base is detachably connected to the working end of the turntable, and its side wall has a mounting through hole. The incident end of the starlight detection module is opposite to the mounting through hole, and its signal output end is electrically connected to the first signal input end of the data fusion module. The second signal input end of the data fusion module is connected to the signal output end of the turntable, and its output end is connected to the input end of the relative spatial angle calculation module. The relative spatial angle calculation module is electrically connected to the host computer.

Description

Technical Field

[0001] This invention relates to calibration devices and calibration methods, specifically to a large field-of-view automatic calibration device and calibration method for pre-calibration of star map arrays. Background Technology

[0002] Star arrays are common hardware-in-the-loop simulation testing devices for star sensors. They typically consist of multiple single-star simulators with identical parameter configurations. The relative spatial angles between the single-star simulators are based on the distribution of specific stars in a designated sky region. They simulate parallel light emitted by several stars at infinity and converge towards the entrance pupil of the star sensor, allowing the star sensor to image under dynamic conditions. The star sensor calculates simulated navigation information under the current hardware-in-the-loop simulation conditions based on the image information and its own motion attitude, providing test data for evaluating its actual dynamic performance.

[0003] Star chart arrays used for large dynamic range hardware-in-the-loop simulation testing generally require a large deployment space. For a standard 100mm aperture single-star simulator, a two-star array with a 0.5° angle requires a spacing of 11 meters, and a 1.0° angle requires a spacing of 5 meters. The star chart array must accommodate large-angle adjustments to the star sensor's entrance pupil; therefore, a standard star chart array needs to form a star sensor field of view of at least 30° × 30°, and must contain at least 50 single-star simulators. These 50 single-star simulators simulate the distribution of stars of different magnitudes in a specific region of the sky, and their relative spatial angles must be consistent with the equivalent relative spatial angles of stars at infinity. The installation accuracy of the single-star simulator in the star map array is crucial for the ground dynamic hardware-in-the-loop simulation test of star sensors and is the foundation for all dynamic performance evaluations of star sensors. Therefore, during the deployment of the star map array, it is necessary to pre-calibrate the relative spatial angle of the position of each single-star simulator. This serves as the basis for the installation adjustment of the single-star simulator and as a system installation error parameter for compensation and correction in the dynamic hardware-in-the-loop simulation of the star sensor.

[0004] Currently, theodolites are widely used in the pre-calibration of relative spatial angles in star chart arrays. First, the theodolite is set up in a fixed position. The reference position is confirmed by aiming at the emitting star image point of the reference single-star simulator in the star chart array. Then, the theodolite is used to aim at the emitting star image points of adjacent single-star simulators. The angular deviation between the two is the relative spatial angle relationship between them. This method relies on the operator's subjective aiming and reading. The operator needs to aim at the emitting star image point in the single-star simulator with their eyes and interpret the reading to measure the relative spatial angle between each single-star simulator and the reference position. Therefore, there will be differences in interpretation between testers with different aiming habits. On the other hand, the number of single-star simulators in the star chart array is large, and measuring them one by one can easily cause visual fatigue for the operator, making it difficult to ensure the consistency of measurement accuracy. In addition, the distribution angle of the single-star simulators in the star map array is large, but the theodolite is limited by the measurement range of the theodolite. The theodolite can only measure the relative spatial angle relationship between the single-star simulator and the single-star simulators in a smaller angular range around it. It cannot measure the relative spatial angle relationship between it and the single-star simulators that are far apart. Therefore, the relative spatial angle relationship between two single-star simulators that are far apart can only be calculated step by step based on the measurable relative spatial angle relationship. This will lead to the measurement error accumulating step by step, and finally introduce a large pre-calibration accuracy deviation for the ground dynamic semi-physical simulation test of the star sensor. Summary of the Invention

[0005] To address the technical problems of existing star array pre-calibration methods, which heavily rely on manual interpretation, have low calibration and measurement efficiency, and cannot directly measure the relative spatial angle relationship between distant single-star simulators, this invention provides a large-field-of-view automatic calibration device and method for star array pre-calibration, enabling fully automatic and accurate relative spatial angle measurement, calibration, and alignment between all single-star simulators in the star array.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A large field-of-view automatic calibration device for pre-calibration of star map arrays is characterized by comprising a turntable, a calibration unit disposed at the working end of the turntable, and a host computer disposed outside the turntable.

[0008] The turntable is used to drive the calibration unit to rotate;

[0009] The calibration unit includes a mounting base and a starlight detection module, a data fusion module, and a relative spatial angle calculation module installed within the mounting base;

[0010] The mounting base is detachably connected to the working end of the turntable, and its side wall is provided with a mounting through hole;

[0011] The incident end of the starlight detection module is opposite to the mounting through hole, and its signal output end is electrically connected to the first signal input end of the data fusion module. It is used to detect the image of the star map array to be calibrated and send the image to the data fusion module.

[0012] The second signal input terminal of the data fusion module is connected to the signal output terminal of the turntable, and its output terminal is connected to the input terminal of the relative spatial angle calculation module, which is used to fuse the correspondence between the images.

[0013] The relative spatial angle calculation module is electrically connected to the host computer and is used to calculate the relative spatial angle between each star point in the star map array to be calibrated and send it to the host computer.

[0014] Furthermore, the calibration unit also includes a coarse aiming module;

[0015] The coarse aiming module is located outside the mounting through hole of the mounting base and is detachably connected to the mounting base. Its optical axis is on the same straight line as the optical axis of the starlight detection module, which is used to characterize the direction of the incident end of the starlight detection module.

[0016] Furthermore, the starlight detection module includes a detector and a collimating optical lens disposed at the detection end of the detector;

[0017] The signal output terminal of the detector is electrically connected to the first signal input terminal of the data fusion module;

[0018] The collimating optical lens is detachably connected to the detector, with its input end corresponding to the mounting through hole and its output end corresponding to the detector's detection end.

[0019] Furthermore, the collimating optical lens includes a collimating frame detachably connected to the detector, a meniscus lens and a biconvex lens disposed within the collimating frame;

[0020] The meniscus lens is positioned close to the detector, with its central convex surface bent toward the biconvex lens;

[0021] The biconvex lens is positioned close to the mounting through hole.

[0022] Furthermore, the coarse aiming module includes a coarse aiming frame connected to the mounting base flange and a collimating lens, a light source, and a protective lens arranged sequentially within the coarse aiming frame in a direction away from the mounting base;

[0023] The exit end of the collimating lens corresponds to the entrance end of the starlight detection module through the mounting through hole.

[0024] A large field-of-view automatic calibration method for pre-calibration of star map arrays, based on the aforementioned large field-of-view automatic calibration device for pre-calibration of star map arrays, is characterized by including the following steps:

[0025] Step 1: Adjust the turntable so that the star array to be calibrated is within the field of view of the star detection module in the vertical direction;

[0026] Step 2: Turn off the power to the star array to be calibrated, and make the turntable rotate gradually from one end of the star array to the other in the horizontal direction at a preset speed. During the rotation, the background image of the star array to be calibrated is acquired frame by frame by the star detection module and sent to the data fusion module. At the same time, the turntable sends its first rotation angle information to the data fusion module. Then, the turntable is returned to the zero position.

[0027] Step 3: Turn on the power to the star array to be calibrated, and make the turntable rotate gradually from one end of the star array to the other in the horizontal direction at a preset speed. During the rotation, the original image of the star array to be calibrated is acquired frame by frame by the star detection module and sent to the data fusion module. At the same time, the turntable sends its output second rotation angle information to the data fusion module.

[0028] Step 4: The data fusion module performs time-aligned fusion of the first rotation angle information and the background image, and performs time-aligned fusion of the second rotation angle information and the original image. Then, based on the numerical relationship between the first rotation angle information and the second rotation angle information, it establishes the correspondence between the original image and the background image. After that, all image data is sent to the relative spatial angle calculation module.

[0029] Step 5: The relative spatial angle calculation module calculates the relative spatial angles between different single-star simulators in the star map array to be calibrated based on all received image data, thus completing the pre-calibration of the star map array.

[0030] Furthermore, step 1 is preceded by:

[0031] The above-mentioned large field-of-view automatic calibration device for pre-calibration of star map arrays is calibrated to obtain its bias level parameter B, projection intrinsic parameter K, normal correction coefficient matrix F(x,y), distortion calibration constants k1, k2, k3, p1, p2; and the prior information of the star map array to be calibrated is obtained.

[0032] Furthermore, step 5 specifically includes:

[0033] Step 5.1: Identify candidate star point regions in the original image Iraw(x,y) based on the corresponding background image D(x,y), and obtain the coordinates of the star point center in the candidate star point region. ( , );

[0034] Step 5.2: Calculate the center coordinates of each star point in the original image of frame t. ( , The theoretical coordinates in the (t+1)th frame image and the (t+2)th frame image are used as the center, and the determination area is set based on the size of the area occupied by each star point in the star map array to be calibrated. If there are star point center coordinates in the determination area, the corresponding star point center coordinates in the original image of the tth frame are recorded as the real star point coordinates, and the number of real star point coordinates is recorded as M.

[0035] Step 5.3: Based on the prior information and the second rotation angle information, obtain the correspondence between the real star point coordinates and the single star simulator in the star map array to be calibrated;

[0036] Step 5.4: Calculate the direction vector of each real star point coordinate in the original image.

[0037]

[0038] in, Let R be the theoretical pixel coordinates of the m-th real star point coordinates projected from the single-star simulator onto the image plane of the starlight detection module at the j-th rotation angle of the turntable, where m∈[1,M]; j The rotation matrix is ​​composed of three Euler angle parameters when the turntable is at the j-th rotation angle; Let m be the actual coordinates of the star point at the j-th rotation angle of the turntable;

[0039] Step 5.5: Calculate the relative spatial angles between different single-star simulators in the star map array to be calibrated. :

[0040]

[0041] in, and Let a and b be the direction vectors of the a-th and b-th single-star simulators, respectively, where a∈[1,M]; b∈[1,M].

[0042] Complete the pre-calibration of the star map array.

[0043] Furthermore, step 5.1 specifically includes:

[0044] Step A: Obtain the background image D(x,y) corresponding to the original image Iraw(x,y) to calculate the corrected image I corresponding to the original image Iraw(x,y). correted (x,y):

[0045]

[0046] Step B: For the corrected image I correted Denoising is performed on (x,y), and the denoised corrected image I is obtained. corretedDivide the region (x, y) into Q regions s, where Q ≥ 6. Calculate the star detection threshold for each region s. :

[0047]

[0048] in, It is the average pixel value within the corresponding region image s. R is the standard deviation of pixels within the region image s, k is a calculation constant, k∈[-0.5,+0.5], and R is the standard deviation of the dynamic range;

[0049] Step C: Set the star detection threshold for each region image s. It compares the values ​​of each pixel it contains to obtain values ​​greater than the star detection threshold. The pixel values ​​are recorded as suspected star point regions;

[0050] Step D: For each suspected star point region, determine whether the adjacent pixels contain other suspected star point regions. If they do, then the suspected star point region is identified as a candidate star point region.

[0051] Step E: Calculate the coordinates of the star center in each candidate star region. ( , ).

[0052] Furthermore, step E specifically includes:

[0053] Step E1: Calculate the centroid coordinates (x0, y0) of each candidate star point region:

[0054]

[0055]

[0056] Where n is the number of pixels contained in the candidate star point region, i∈[1,n], and w is the gray value weight;

[0057] Step E2: Calculate the coordinates of the star center corresponding to the centroid coordinates (x0, y0). (x,y):

[0058]

[0059] in, ; Let be the standard deviation of the coordinate distribution of the i-th pixel in the x-direction; Let be the standard deviation of the coordinate distribution of the i-th pixel in the y-direction; To correct image I correted The corresponding star-shaped area;

[0060] Step E3: Set the center coordinates of each star point Distortion compensation is performed on (x,y) to obtain the coordinates of the star center. ( , ):

[0061]

[0062]

[0063] in, .

[0064] The beneficial effects of this invention are:

[0065] 1. The present invention provides a large field-of-view automatic calibration device and calibration method for pre-calibration of star arrays, which completely eliminates manual interpretation. After the large field-of-view automatic calibration device is installed and deployed, all single-star simulators in the star array to be calibrated can be observed in a single exposure. Moreover, the geometric constraints of star points in multi-frame images are strong, and the pre-calibration of the relative spatial angle relationship between all single-star simulators in the star array to be calibrated can be completed in one go, which greatly improves the efficiency of star array pre-calibration.

[0066] 2. The present invention provides a large field-of-view automatic calibration method for pre-calibration of star map arrays. It is a global optimization method that uses all observation data from all positions to solve all parameters. The system error is evenly distributed, so the relative spatial angle relationship obtained is very reliable and stable, and can achieve sub-arcsecond accuracy.

[0067] 3. This invention provides a large field-of-view automatic calibration device for pre-calibration of star arrays. It employs a large field-of-view optical system, where a single frame image can cover all single-star simulators within the elevation range of the star array to be calibrated, and adjacent frames contain numerous overlapping observation areas. These overlapping observation areas provide rich and strongly constrained overlapping observation information, enabling the large field-of-view automatic calibration method proposed in this invention to effectively separate the variables coupled with the turntable angle positioning error and the actual star position. This avoids the turntable angle positioning error being completely transferred to the star coordinate calculation results, significantly reducing the impact of the turntable angle positioning error on the final calibration result, and lowering the requirements and cost of the turntable hardware. Attached Figure Description

[0068] Figure 1 This is a schematic diagram of an embodiment of the large field-of-view automatic calibration device for pre-calibration of star map arrays according to the present invention;

[0069] Figure 2 This is a schematic diagram of the overall structure of the calibration unit in an embodiment of the present invention;

[0070] Figure 3 This is a partial structural schematic diagram of the calibration unit in an embodiment of the present invention;

[0071] Figure 4 This is a partial cross-sectional view of the calibration unit in an embodiment of the present invention.

[0072] The attached figures are labeled as follows:

[0073] 0. Turntable; 1. Calibration Unit; 11. Mounting Base; 111. Mounting Through Hole; 12. Starlight Detection Module; 121. Detector; 122. Collimating Optical Lens; 1221. Collimating Frame; 1222. Meniscus Lens; 1223. Biconvex Lens; 13. Data Fusion Module; 14. Relative Spatial Angle Calculation Module; 15. Coarse Aiming Module; 151. Coarse Aiming Frame; 152. Collimating Lens; 153. Light Source; 154. Protective Mirror; 2. Host Computer. Detailed Implementation

[0074] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0075] This invention provides a large field-of-view automatic calibration device for pre-calibration of star map arrays, such as... Figure 1 As shown, the large field-of-view automatic calibration device includes a turntable 0, a calibration unit 1 disposed at the working end of the turntable 0, and a host computer 2 disposed outside the turntable 0.

[0076] Turntable 0 is used to drive calibration unit 1 to rotate so that calibration unit 1 can traverse all single-star simulators in the star map array to be calibrated from left to right;

[0077] like Figure 1 , Figure 3 and Figure 4 As shown, the calibration unit 1 includes a mounting base 11 and a starlight detection module 12, a data fusion module 13 and a relative spatial angle calculation module 14 disposed within the mounting base 11, as well as a coarse aiming module 15 installed outside the mounting base 11.

[0078] The mounting base 11 is detachably connected to the working end of the turntable 0, and its side wall is provided with a mounting through hole 111. The bottom of the mounting base 11 is firmly integrated with a high-precision mounting position determination module. This module is a metal mounting base with precision positioning pin holes and a high parallel mounting surface. Its shape meets the mechanical interface of the turntable, ensuring that the large field of view automatic calibration device is stably connected to the center of the turntable through this base, and there is no relative displacement between it and the turntable during the entire experiment.

[0079] The incident end of the starlight detection module 12 is opposite to the mounting through-hole 111, and its signal output end is electrically connected to the first signal input end of the data fusion module 13. It is used to detect the image of the star map array to be calibrated and send the image to the data fusion module 13. Figure 4 As shown, the starlight detection module 12 includes a detector 121 and a collimating optical lens 122 disposed at the detection end of the detector 121; the signal output end of the detector 121 is electrically connected to the first signal input end of the data fusion module 13; the collimating optical lens 122 is a large-aperture, short-focal-length, large-field-of-view optical lens; specifically, the collimating optical lens 122 includes a collimating frame 1221 detachably connected to the detector 121, a meniscus lens 1222 disposed within the collimating frame 1221, and a biconvex lens 1223; the meniscus lens 1222 is disposed close to the detector 121, with its central convex surface bent toward the biconvex lens 1223; the biconvex lens 1223 is disposed close to the mounting through hole 111. The radius of curvature r of the front surface of the meniscus lens 1222 is... 21 =28.41mm, radius of curvature of the rear surface r 22 =17.2mm, lens center thickness d2=6mm; radius of curvature r of the front surface of the biconvex lens 1223 11 =84.34mm, rear surface radius of curvature r 12 =-100.5mm, lens center thickness d1=8mm; the front surface is the surface close to the coarse aiming module 15.

[0080] The second signal input terminal of the data fusion module 13 is connected to the signal output terminal of the turntable 0, and its output terminal is connected to the input terminal of the relative spatial angle calculation module 14, used to fuse the correspondence between various images. The data fusion module 13 is a multi-interface electronic processing unit with a miniaturized square circuit board package.

[0081] The relative spatial angle calculation module 14 is electrically connected to the host computer 2 and is used to calculate the relative spatial angles between various star points in the star array to be calibrated and send them to the host computer 2. The relative spatial angle calculation module 14 is integrated into the body of the large field of view automatic calibration device in the form of an embedded dedicated information processing board. It is connected to the data fusion module 13 through a high-speed data bus, receives all data images, and runs the core algorithm based on the bundle adjustment method to solve the relative spatial angles between the light output angles of each single star simulator.

[0082] The coarse aiming module 15 is a cylindrical laser pointer-like structure, which can be connected to the relative spatial angle calculation module 14 for unified control via the interface of the large field-of-view automatic calibration device. The coarse aiming module 15 is located outside the mounting through-hole 111 of the mounting base 11 and is detachably connected to the mounting base 11. Its optical axis is on the same straight line as the optical axis of the starlight detection module 12, enabling it to characterize the pointing of the incident end of the starlight detection module 12. Figure 4 As shown, the coarse aiming module 15 includes a coarse aiming frame 151 connected to the flange of the mounting base 11, and a collimating lens 152, a light source 153, and a protective mirror 154 sequentially arranged within the coarse aiming frame 151 in a direction away from the mounting base 11; the exit end of the collimating lens 152 corresponds to the incident end of the starlight detection module 12 through the mounting through hole 111; the radius of curvature r of the front surface of the collimating lens 152 is... 31 =3.98mm, rear surface radius of curvature r 32 =-21.3mm, lens center thickness d3=2mm.

[0083] In the static state, each module is centered around the star detection module 12, supported by the mounting base 11. The data fusion module 13 and the relative spatial angle calculation module 14 are interconnected via cables or an internal bus, while the coarse aiming module 15 is fixed to the front end of the collimating optical lens 122 as an optical accessory, forming a fully functional and structurally stable large field-of-view automatic calibration device. To suppress the influence of temperature changes on the optical axis pointing, the mounting base 11, collimating frame 1221, and coarse aiming frame 151 are all made of low-expansion titanium alloy material with a coefficient of thermal expansion matching that of the optical elements. This ensures that the line-of-view offset does not exceed 0.1″ within a temperature range of 20±10℃, greatly improving the temperature adaptability of the large field-of-view automatic calibration device and ensuring the accuracy and stability during the pre-calibration process.

[0084] This invention provides a large field-of-view automatic calibration method for pre-calibration of star map arrays, comprising the following steps:

[0085] Step 0: Calibrate the above-mentioned large field-of-view automatic calibration device for pre-calibration of star map array, and obtain its bias level parameter B, projection intrinsic parameter K, normal correction coefficient matrix F(x,y), distortion calibration constants k1, k2, k3, p1, p2; and obtain the prior information of the star map array to be calibrated.

[0086] Step 1: Adjust the turntable 0 so that the star map array to be calibrated is within the field of view of the star detection module 12 in the vertical direction;

[0087] Step 2: Turn off the power to the star array to be calibrated, and make the turntable 0 rotate gradually from one end of the star array to the other in the horizontal direction at a speed of 2° per second. During the rotation, the background image of the star array to be calibrated is acquired frame by frame by the star detection module 12 and sent to the data fusion module 13. At the same time, the turntable 0 sends its first rotation angle information to the data fusion module 13. Then, the turntable 0 is restored to the zero position.

[0088] Step 3: Turn on the power supply of the star map array to be calibrated, and make the turntable 0 rotate gradually from one end of the star map array to the other in the horizontal direction at a speed of 2° per second. During the rotation, the original image of the star map array to be calibrated is acquired frame by frame by the star light detection module 12 and sent to the data fusion module 13. At the same time, the turntable 0 sends its output second rotation angle information to the data fusion module 13.

[0089] During the rotation of turntable 0, because all the single-star simulators in the star map array to be calibrated emit light simultaneously, the large field of view automatic calibration device can continuously capture all star maps during the rotation process, i.e., the original images, and the images contain clear star points generated by all the single-star simulators in the current field of view. During this process, the data fusion module 13 inside the large field of view automatic calibration device receives all the images collected by the star detection module 12 and the rotation angle information of turntable 0 (i.e., the first rotation angle information or the second rotation angle information) in real time.

[0090] Step 4: The data fusion module 13 performs time-aligned fusion of the first rotation angle information and the background image, and performs time-aligned fusion of the second rotation angle information and the original image. Then, based on the numerical relationship between the first rotation angle information and the second rotation angle information, it establishes the correspondence between the original image and the background image. After that, all image data is sent to the relative spatial angle calculation module 14.

[0091] Step 5: The relative spatial angle calculation module 14 calculates the relative spatial angles between different single-star simulators in the star map array to be calibrated based on all received image data;

[0092] Step 5.1: Identify candidate star point regions in the original image Iraw(x,y) based on the corresponding background image D(x,y), and calculate the coordinates of the star center in the candidate star point region. ( , Specifically, this includes:

[0093] Step A: Obtain the background image D(x,y) corresponding to the original image Iraw(x,y) to calculate the corrected image I corresponding to the original image Iraw(x,y). correted (x,y):

[0094]

[0095] Step B: Correct the image I correted Denoising is performed on (x,y), and the denoised corrected image I is obtained. correted The region (x, y) is divided into Q regions s, where Q=9 in this embodiment. The star detection threshold for each region s is calculated. Considering the low contrast of low magnitude images, an improvement was made to the conventional NiBlack local thresholding method, and the calculation formula is as follows:

[0096]

[0097] in, It is the average pixel value within the corresponding region image s. is the standard deviation of pixels within the region image s, k∈[-0.5,+0.5] is a calculation constant, in this embodiment k=-0.2, and R is the standard deviation of the dynamic range, which is generally taken as R=128 in 8-bit images.

[0098] Step C: Set the star detection threshold for each region image s. It compares the values ​​of each pixel it contains to obtain values ​​greater than the star detection threshold. The pixel values ​​are recorded as suspected star point areas, thereby eliminating the interference of single pixel noise.

[0099] Step D: For each suspected star point region, determine whether the adjacent pixels contain other suspected star point regions. If they do, then the suspected star point region is identified as a candidate star point region.

[0100] Step E: Calculate the coordinates of the star center in each candidate star region. ( , Specifically, this includes:

[0101] Step E1: Calculate the centroid coordinates (x0, y0) of each candidate star point region:

[0102]

[0103]

[0104] Where n is the number of pixels contained in the candidate star point region, i∈[1,n], and w is the gray value weight;

[0105] Step E2: Calculate the coordinates of the star center corresponding to the centroid coordinates (x0, y0). (x,y):

[0106]

[0107] in, A two-dimensional Gaussian model of star point distribution; Let be the standard deviation of the coordinate distribution of the i-th pixel in the x-direction; Let be the standard deviation of the coordinate distribution of the i-th pixel in the y-direction; To correct image I corretedThe corresponding star-shaped area;

[0108] Step E3: Because the large field of view of the automatic calibration device is large, the coordinates of the center of the star point located at the edge of the field of view are... (x,y) needs distortion compensation to obtain the coordinates of the star center. ( , ):

[0109]

[0110]

[0111] in, ;

[0112] Step 5.2: Since the large field of view of the automatic calibration device is large, adjacent frames may contain the same star points. Therefore, it is necessary to calculate the center coordinates of each star point in the original image of frame t. ( , The theoretical coordinates in the (t+1)th frame and the (t+2)th frame are determined. Then, the determination area is set based on the theoretical coordinates and the size of the area occupied by each star point in the star array to be calibrated. If there are star point center coordinates in the determination area, the corresponding star point center coordinates in the original image of the tth frame are recorded as the real star point coordinates. The number of real star point coordinates is recorded as M.

[0113] Step 5.3: Based on prior information and second rotation angle information, obtain the correspondence between the real star point coordinates and the single star simulator in the star map array to be calibrated;

[0114] Step 5.4: Calculate the direction vector of each real star point coordinate in the original image.

[0115] =

[0116] in, Let R be the theoretical pixel coordinates of the m-th real star point coordinates corresponding to the single-star simulator projected onto the image plane of starlight detection module 12 at the j-th rotation angle of turntable 0, where m∈[1,M]; j Let be the rotation matrix consisting of three Euler angle parameters when turntable 0 is at the j-th rotation angle; Let m be the actual coordinates of the star point at the j-th rotation angle of turntable 0;

[0117] The direction vector of a single-star simulator is fixed in its "body coordinate system" and is denoted as a three-dimensional direction vector. When turntable 0 is at the j-th rotation angle, it has a rotation matrix R consisting of three Euler angle parameters.j This rotation matrix R j The vectors in the global coordinate system are transformed to the coordinate system of the large field-of-view automatic calibration device. Therefore, at the j-th rotation angle of the turntable, the direction vector of the ray from the m-th single-star simulator in the coordinate system of the large field-of-view automatic calibration device is: Furthermore, this direction vector is projected onto the image plane through a large field-of-view automatic calibration device model that includes intrinsic parameters and distortion, yielding the theoretical pixel coordinates. The Levenberg-Marquardt algorithm (a nonlinear least squares optimization algorithm) is used for nonlinear iterative optimization, and adjustments are made iteratively. and R j This minimizes the total reprojection error. When the convergence condition is met, the optimal single-star simulator direction vector is obtained. Because the large field of view of the automatic calibration device is large, adjacent frames will contain the same star points, which constitute a large overlapping observation area. For the star points in these overlapping areas, their true orientation in inertial space is fixed. Even if the turntable 0 has a certain positioning accuracy error, during iterative adjustment calculations, the global optimal solution will be perfected by fine-tuning the error value of the turntable angle or fine-tuning the three-dimensional coordinates of all star points, thus achieving the optimal fit for all observed image point data under the constraint of overlapping observation.

[0118] Step 5.5: Calculate the relative spatial angles between different single-star simulators in the star map array to be calibrated. :

[0119]

[0120] in, and Let a and b be the direction vectors of the a-th and b-th single-star simulators, respectively, where a∈[1,M]; b∈[1,M].

[0121] Complete the pre-calibration of the star map array.

[0122] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present invention should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A large field of view auto-calibration apparatus for star pattern array pre-calibration, characterized in that: It includes a turntable (0), a calibration unit (1) set at the working end of the turntable (0), and a host computer (2) set outside the turntable (0). The turntable (0) is used to drive the calibration unit (1) to rotate; The calibration unit (1) includes a mounting base (11) and a starlight detection module (12), a data fusion module (13), and a relative spatial angle calculation module (14) installed in the mounting base (11). The mounting base (11) is detachably connected to the working end of the turntable (0), and its side wall is provided with a mounting through hole (111). The incident end of the starlight detection module (12) is opposite to the mounting through hole (111), and its signal output end is electrically connected to the first signal input end of the data fusion module (13) for detecting the image of the star map array to be calibrated and sending the image to the data fusion module (13). The second signal input terminal of the data fusion module (13) is connected to the signal output terminal of the turntable (0), and its output terminal is connected to the input terminal of the relative spatial angle calculation module (14) for fusing the correspondence between the images; The relative spatial angle calculation module (14) is electrically connected to the host computer (2) and is used to calculate the relative spatial angle between each star point in the star map array to be calibrated and send it to the host computer (2).

2. The large field-of-view automatic calibration device for pre-calibration of star map arrays according to claim 1, characterized in that: The calibration unit (1) also includes a coarse aiming module (15); The coarse aiming module (15) is located outside the mounting through hole (111) of the mounting base (11) and is detachably connected to the mounting base (11). Its optical axis is on the same straight line as the optical axis of the starlight detection module (12) and is used to characterize the direction of the incident end of the starlight detection module (12).

3. The large field-of-view automatic calibration device for pre-calibration of star map arrays according to claim 2, characterized in that: The starlight detection module (12) includes a detector (121) and a collimating optical lens (122) disposed at the detection end of the detector (121). The signal output terminal of the detector (121) is electrically connected to the first signal input terminal of the data fusion module (13); The collimating optical lens (122) is detachably connected to the detector (121), with its input end corresponding to the mounting through hole (111) and its output end corresponding to the detection end of the detector (121).

4. The large field-of-view automatic calibration device for pre-calibration of star map arrays according to claim 3, characterized in that: The collimating optical lens (122) includes a collimating frame (1221) detachably connected to the detector (121), a meniscus lens (1222) and a biconvex lens (1223) disposed within the collimating frame (1221). The meniscus lens (1222) is positioned close to the detector (121), with its central convex surface bent toward the biconvex lens (1223); The biconvex lens (1223) is positioned close to the mounting through hole (111).

5. The large field-of-view automatic calibration device for pre-calibration of star map arrays according to claim 2, 3, or 4, characterized in that: The coarse aiming module (15) includes a coarse aiming frame (151) connected to the flange of the mounting base (11) and a collimating lens (152), a light source (153), and a protective lens (154) arranged sequentially in the coarse aiming frame (151) in a direction away from the mounting base (11). The exit end of the collimating lens (152) corresponds to the entrance end of the starlight detection module (12) through the mounting through hole (111).

6. A large field of view automatic calibration method for star map array pre-calibration based on the large field of view automatic calibration device for star map array pre-calibration according to any one of claims 1-5, characterized in that, Includes the following steps: Step 1: Adjust the turntable (0) so that the star map array to be calibrated is within the field of view of the star detection module (12) in the vertical direction; Step 2: Turn off the power supply of the star array to be calibrated, and make the turntable (0) rotate gradually from one end of the star array to the other in the horizontal direction at a preset speed. During the rotation, the background image of the star array to be calibrated is acquired frame by frame by the star detection module (12) and sent to the data fusion module (13). At the same time, the turntable (0) sends its first rotation angle information to the data fusion module (13). Then, the turntable (0) is restored to the zero position. Step 3: Turn on the power supply of the star map array to be calibrated, so that the turntable (0) rotates gradually from one end of the star map array to the other end in the horizontal direction at a preset speed. During the rotation, the original image of the star map array to be calibrated is acquired frame by frame by the star light detection module (12) and sent to the data fusion module. At the same time, the turntable (0) sends the second rotation angle information it outputs to the data fusion module (13). Step 4: The data fusion module (13) performs time-aligned fusion of the first rotation angle information and the background image, and performs time-aligned fusion of the second rotation angle information and the original image. Then, based on the numerical relationship between the first rotation angle information and the second rotation angle information, it establishes the correspondence between the original image and the background image. After that, it sends all image data to the relative spatial angle calculation module (14). Step 5: The relative spatial angle calculation module (14) calculates the relative spatial angle between different single-star simulators in the star map array to be calibrated based on all received image data, and completes the pre-calibration of the star map array.

7. The large field of view auto-calibration method for star pattern array pre-calibration according to claim 6, characterized in that, Step 1 includes the following: The large field-of-view automatic calibration device for pre-calibration of star map arrays as described in any one of claims 1-5 is calibrated to obtain its bias level parameter B, projection intrinsic parameter K, normal correction coefficient matrix F(x,y), distortion calibration constants k1, k2, k3, p1, p2; and prior information of the star map array to be calibrated is obtained.

8. The large field of view auto-calibration method for star pattern array pre-calibration according to claim 7, characterized in that, Step 5 specifically includes: Step 5.1, identify the candidate star point region in the original image Iraw(x, y) based on the corresponding background image D(x, y), and obtain the star point center coordinates of the candidate star point region ( , ) Step 5.2: Calculate the center coordinates of each star point in the original image of frame t. ( , The theoretical coordinates in the (t+1)th frame image and the (t+2)th frame image are used as the center, and the determination area is set based on the size of the area occupied by each star point in the star map array to be calibrated. If there are star point center coordinates in the determination area, the corresponding star point center coordinates in the original image of the tth frame are recorded as the real star point coordinates, and the number of real star point coordinates is recorded as M. Step 5.3: Based on the prior information and the second rotation angle information, obtain the correspondence between the real star point coordinates and the single star simulator in the star map array to be calibrated; Step 5.4: Calculate the direction vector of each real star point coordinate in the original image. : ； in, Let m be the theoretical pixel coordinates of the m-th real star point coordinates corresponding to the turntable (0) at the j-th rotation angle, projected onto the image plane of the starlight detection module (12), where m∈[1,M]; R j Let be the rotation matrix consisting of three Euler angle parameters when the turntable (0) is at the j-th rotation angle; Let m be the actual coordinates of the turntable (0) at the j-th rotation angle; Step 5.5: Calculate the relative spatial angles between different single-star simulators in the star map array to be calibrated. : ； in, and Let a and b be the direction vectors of the a-th and b-th single-star simulators, respectively, where a∈[1,M]; b∈[1,M]. Complete the pre-calibration of the star map array.

9. The large field-of-view automatic calibration method for pre-calibration of star map arrays according to claim 8, characterized in that, Step 5.1 specifically includes: Step A, obtain the background image D(x,y) corresponding to the original image Iraw(x,y) to calculate the correction image I correted (x,y): ； Step B: For the corrected image I correted Denoising is performed on (x,y), and the denoised corrected image I is obtained. correted Divide the region (x, y) into Q regions s, where Q ≥ 6. Calculate the star detection threshold for each region s. : ； in, It is the average pixel value within the corresponding region image s. R is the standard deviation of pixels within the region image s, k is a calculation constant, k∈[-0.5,+0.5], and R is the standard deviation of the dynamic range; Step C: Set the star detection threshold for each region image s. It compares the values ​​of each pixel it contains to obtain values ​​greater than the star detection threshold. The pixel values ​​are recorded as suspected star point regions; Step D: For each suspected star point region, determine whether the adjacent pixels contain other suspected star point regions. If they do, then the suspected star point region is identified as a candidate star point region. Step E: Calculate the coordinates of the star center in each candidate star region. ( , ).

10. The large field-of-view automatic calibration method for pre-calibration of star map arrays according to claim 9, characterized in that, Step E specifically includes: Step E1: Calculate the centroid coordinates (x0, y0) of each candidate star point region: ； ； Where n is the number of pixels contained in the candidate star point region, i∈[1,n], and w is the gray value weight; Step E2: Calculate the coordinates of the star center corresponding to the centroid coordinates (x0, y0). (x,y): ； in, A two-dimensional Gaussian model of star point distribution; Let be the standard deviation of the coordinate distribution of the i-th pixel in the x-direction; Let be the standard deviation of the coordinate distribution of the i-th pixel in the y-direction; To correct image I correted The corresponding star-shaped area; Step E3: Set the center coordinates of each star point Distortion compensation is performed on (x,y) to obtain the coordinates of the star center. ( , ): ； ； in, .

Images