Guide star sensor image processing method and system and guide star processing board hardware framework

Abstract

The invention relates to a guide star sensor image processing method, a guide star sensor image processing system and a guide star processing board hardware framework. The guide star sensor image processing method comprises the following steps: confirming a working mode of a guide star sensor; performing calibration image processing or star point extraction processing according to the working mode of the guide star sensor, and if the working mode is not imaging, performing calibration image processing to obtain a background field correction image, a dark field correction image and a flat field correction image; if the working mode is imaging, performing calibration image correction on the imaging image by using the background field correction image, the dark field correction image and the flat field correction image to obtain a first star map; wherein if the imaged image is a first frame of full-image image, star point extraction processing is carried out subsequently, and star point information is obtained; if the imaging image is a second frame of full-image image, star point extraction, star point screening and star point matching are carried out subsequently; and if the imaging image is a windowing image, star point extraction and star point matching are carried out subsequently.

Description

Technical Field

[0001] This invention relates to the field of star guide sensor technology, and in particular to a star guide sensor image processing method, system, and star guide processing board hardware framework. Background Technology

[0002] A star guide sensor is a high-precision attitude measurement sensor for spacecraft that uses stars as an observation reference. It obtains the attitude information of a spacecraft by observing, identifying, and calculating stars. The working process of a star guide sensor mainly includes three processes: star point extraction, star map identification, and attitude calculation, among which star point extraction is the foundation of the star guide sensor's operation.

[0003] Existing star extraction methods have the following drawbacks: they lack processing to eliminate the effects of background noise, dark current, and pixel response inhomogeneity; and the fixed threshold method for segmenting star maps is ineffective in complex sky backgrounds or fields of view with gradient changes, and is prone to missed detections or false detections. Summary of the Invention

[0004] To address at least some of the problems mentioned above in the prior art, the present invention provides a star guide sensor image processing method, comprising: Based on the first row of frame header information of the guide star sensor image, confirm the working mode of the guide star sensor; The calibration image processing or star point extraction processing is performed according to the working mode of the star guide sensor, wherein: If the working mode is not imaging, then calibration image processing is performed to obtain a background field correction image, a dark field correction image, and a flat field correction image. If the working mode is imaging, the imaging image is calibrated using the background field correction image, dark field correction image, and flat field correction image to obtain the first star map; wherein: if the imaging image is the first full-frame image, star point extraction processing is performed to obtain star point information; if the imaging image is the second full-frame image, star point extraction, star point filtering, and star point matching are performed; if the imaging image is a windowed image, star point extraction and star point matching are performed.

[0005] Furthermore, the star point extraction process based on the first star map includes: Gaussian filtering is used to smooth the isolated point noise and stripe noise in the first star map to obtain the second star map. The second star map is segmented to obtain a binarized star map. The background pixels in the binarized star map have a value of 0, and the star pixels have a value of 1. The star pixels in the binarized star map are marked to obtain a star map with star marker numbers; and Star map with star markers is processed to locate sub-pixel star points, which yields star point information, including the star point marker number and the centroid coordinates of the star point.

[0006] Furthermore, the formula for calibrating and correcting the image acquired by the guide star sensor is as follows: , in For the first star map, For the original image, Corrected image for the background field. Correcting images for dark scenes. This is the standard flat field correction matrix. Image exposure time; Star map segmentation of the second star map based on dynamic thresholding of edge background: in, For detector coordinates The grayscale value of the pixel; T is the grayscale threshold, calculated as follows: , in, The standard deviation of grayscale for the edge background; These represent the number of rows and columns of pixels in the image sensor, respectively. The star pixels in the binary star map are labeled using the four-connected component analysis method, resulting in a star map with star pixel labels, including: The binarized star map is scanned pixel by pixel from left to right and top to bottom. Pixels with a grayscale value greater than zero are considered valid pixels. Valid pixels are marked, with the markings starting from 1. If the current pixel is a valid pixel, and there is exactly one valid pixel among the left and top neighbor pixels of the current pixel, take its mark as the mark number of the current pixel; If the current pixel is a valid pixel, and both its left and top neighbor pixels are valid pixels, then the label number of the left neighbor pixel is taken as the label number of the current pixel; if the label numbers of the left and top neighbor pixels are different, then the label numbers of the left and top neighbor pixels are recorded in the star label equivalence table, waiting for subsequent merging of equivalent label numbers. If the left and top neighbor pixels of the current pixel are both invalid pixels, then assign a new label to the current pixel; Repeat the above process until all pixels of the binarized star map are scanned to obtain a star map with preliminary star point markings. After the scan is complete, equivalent tags are merged and new tag numbers are assigned. Remove star targets with more than 10 constituent pixels and star targets with less than 3 constituent pixels, and redistribute the marker numbers according to the number of valid star targets.

[0007] Furthermore, if the operating mode is not imaging, then calibration image processing is performed to obtain a background field corrected image, a dark field corrected image, and a flat field corrected image, including: Determine whether the working mode is background field calibration. If so, perform background field calibration processing, average multiple background field images, and obtain a background field corrected image. If the working mode is not background field calibration, then determine whether it is dark field calibration. If so, perform dark field calibration processing, average multiple dark field images, and subtract the background field correction image to obtain the dark field correction image. If the working mode is not background field calibration and dark field calibration, then determine whether it is flat field calibration. If so, perform flat field calibration processing, average multiple flat field images, subtract the background field correction image and dark field correction image, and then normalize to obtain the flat field correction image.

[0008] Furthermore, the star point information includes the star point's marker number and the star point's centroid coordinates; The star selection includes selecting stars from each guide star sensor image that meet the following conditions based on the centroid coordinates of the stars: a) The distance from the edge of the guide sensor image is not less than 60 pixels; b) Stars with unsaturated grayscale in the image pixels of the guide sensor; c) Stars with a distance of more than 60 pixels between them; If the number of selected stars is greater than 8, select the 8 brightest stars; if the number of selected stars is less than 1, exit image processing. The star point matching includes: performing star point matching based on the selected star points, wherein: In each guide star sensor image, the star points selected in the current frame are compared one by one with the star points selected in the previous frame. Those with a row and column distance of less than 10 pixels are considered as matching pairs. One-to-many matching pairs are deleted to obtain one-to-one matching stars. If the number of matching stars is less than 1, the image processing is exited.

[0009] The present invention also provides a guide star sensor image processing system, including an FPGA chip and a CPU chip, wherein the FPGA chip is configured to perform the following actions: receive a guide star sensor image according to a control instruction sent by the CPU chip, and transmit the guide star sensor image to the CPU chip, or perform a calibration image processing step and a star point extraction preprocessing step in the guide star sensor image processing method; The CPU chip is configured to perform the following actions: Sending control commands to the FPGA chip; and The steps of acquiring the guide star sensor image forwarded by the FPGA chip and performing the guide star sensor image processing method include calibration image processing, star point extraction preprocessing, and star point extraction; or acquiring the preprocessed image of the FPGA chip and performing star point extraction to obtain star point information.

[0010] Furthermore, the calibration image processing step in the guide star sensor image processing method includes: Upon receiving the guide star sensor image, the system checks the first row of the image header to determine if the guide star sensor is in imaging mode. If not, it performs calibration image processing to obtain a background field correction image, a dark field correction image, and a flat field correction image, including: Determine whether the working mode is background field calibration. If so, perform background field calibration processing, average multiple background field images, and obtain a background field corrected image. If the working method is not the local field calibration, then determine whether it is dark field calibration. If so, perform dark field calibration processing, average multiple dark field images, and subtract the local field correction image to obtain the dark field correction image. If the working method is not background field calibration and dark field calibration, then determine whether it is flat field calibration. If so, perform flat field calibration processing, average multiple flat field images, subtract the background field correction image and dark field correction image, and then normalize to obtain the flat field correction image. The star point extraction preprocessing steps in the star sensor image processing method include: The imaging image acquired by the guide star sensor is calibrated and corrected using the background field correction image, dark field correction image, and flat field correction image to obtain the first star map; Gaussian filtering is used to smooth the isolated point noise and stripe noise in the first star map to obtain the second star map. Star point extraction includes the following steps: The second star map is segmented to obtain a binarized star map. The background pixels in the binarized star map have a value of 0, and the star pixels have a value of 1. The star pixels in the binarized star map are marked to obtain a star map with star marker numbers; and Star map with star markers is processed to locate sub-pixel star points, which yields star point information, including the star point marker number and the centroid coordinates of the star point.

[0011] The present invention also provides a guide star processing board hardware framework, comprising: FPGA chip; The CPU chip is connected to the FPGA chip; The data interaction terminal RAM is located in the FPGA chip and connected to the CPU chip; A first memory, which is connected to the FPGA chip; A calibration data storage device, which is connected to the FPGA chip; and CPU and memory.

[0012] Furthermore, the steps of the FPGA chip and CPU chip collaboratively executing the guide star sensor image processing method include: The FPGA chip receives the full star map or the window star map and stores it in the first memory. The FPGA chip reads the calibration image from the calibration data memory and performs calibration image processing on the calibration image to obtain the background field correction image, dark field correction image, and flat field correction image; The FPGA chip performs star point extraction preprocessing on the full-map star map or windowed star map; The preprocessed image data is transmitted to the CPU chip via dual-end RAM data exchange, where the CPU chip extracts star points to obtain star point information.

[0013] Furthermore, the steps of the CPU chip executing the guide star sensor image processing method include: The CPU chip prefetches the calibration image into the CPU memory; The FPGA chip receives the full star map or window star map, stores it in the first memory controlled by the FPGA chip, and copies it to the CPU memory through data interaction between the two-end RAM. The CPU chip performs calibration image processing and star point extraction.

[0014] The present invention has at least the following beneficial effects: The image processing method for the guide star sensor of the present invention integrates three calibration processes: detector background field, dark field, and flat field, forming a complete detector error correction chain, which can more thoroughly eliminate the effects of detector background noise, dark current, and pixel response inhomogeneity. In the calibration image processing stage, multiple images are averaged, which can effectively suppress random noise and generate high-quality corrected images. At the same time, the number of averaged images has an on-orbit update function to adapt to the needs of different precision and working scenarios. Compared with methods using fixed thresholds, the star map segmentation method based on dynamic thresholds of edge background in this invention can dynamically adjust the segmentation threshold according to the background brightness of different regions of the image. In complex sky backgrounds or fields of view with gradient changes, it can more reliably segment real stars and avoid missed or false detections, making it especially suitable for dynamic environments such as space observation. The image processing method of the present invention first determines the image type when receiving an image from a guide sensor, and performs calibration image processing or star point extraction processing based on the image category. Compared with the traditional instruction switching mode, this process has completely autonomous image type and working mode judgment and process control. The star guide sensor image processing system of the present invention can switch between FPGA+CPU image processing and CPU image processing to meet different needs. The FPGA chip selects to transmit the star guide sensor image to the CPU chip according to the control instructions sent by the CPU chip so that the CPU chip can independently complete the star guide sensor image processing, or execute the calibration image processing step and star point extraction preprocessing step in the star guide sensor image processing method. Attached Figure Description

[0015] To further illustrate the above and other advantages and features of the various embodiments of the present invention, a more specific description of the embodiments of the invention will be presented with reference to the accompanying drawings. It is to be understood that these drawings depict only typical embodiments of the invention and are therefore not intended to limit its scope. In the drawings, identical or corresponding parts will be indicated by identical or similar reference numerals for clarity.

[0016] Figure 1 A schematic diagram of a guide processing board hardware framework according to an embodiment of the present invention is shown.

[0017] Figure 2 A flowchart of a guide star sensor image processing method according to an embodiment of the present invention is shown.

[0018] Figure 3 A flowchart of full-image processing according to an embodiment of the present invention is shown.

[0019] Figure 4 A flowchart of windowed image processing according to an embodiment of the present invention is shown. Detailed Implementation

[0020] It should be noted that the components in the accompanying drawings may be shown exaggerated for illustrative purposes and may not be to scale.

[0021] In this invention, the various embodiments are merely intended to illustrate the solutions of the invention and should not be construed as limiting.

[0022] In this invention, unless otherwise specified, the quantifiers “a” and “one” do not exclude scenarios involving multiple elements.

[0023] It should also be noted that, in the embodiments of the present invention, only a portion of the parts or components may be shown for clarity and simplicity. However, those skilled in the art will understand that, under the teachings of the present invention, the required parts or components can be added as needed for specific scenarios.

[0024] It should also be noted that within the scope of this invention, the terms "same", "equal", and "equal to" do not mean that the two values ​​are absolutely equal, but allow for a certain reasonable error. In other words, the terms also cover "substantially the same", "substantially equal", and "substantially equal to".

[0025] It should also be noted that in the description of this invention, the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not explicitly or implicitly suggest that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0026] Furthermore, the embodiments of the present invention describe the process steps in a specific order. However, this is only for the convenience of distinguishing each step, and is not a limitation on the order of each step. In different embodiments of the present invention, the order of each step can be adjusted according to the process.

[0027] In one embodiment, the guide star sensor image processing system includes an FPGA chip and a CPU chip, wherein the FPGA chip is configured to perform the following actions: receive a guide star sensor image according to a control instruction sent by the CPU chip, and transmit the guide star sensor image to the CPU chip, or perform a calibration image processing step and a star point extraction preprocessing step in the guide star sensor image processing method; The CPU chip is configured to perform the following actions: Sending control commands to the FPGA chip; and Acquire the star sensor image forwarded by the FPGA chip and execute the steps of the star sensor image processing method; or acquire the pre-processed image of the FPGA chip and extract star points to obtain star point information.

[0028] To meet the application requirements of image processing for the guide star sensor, a design approach that switches between two design architectures: "FPGA+CPU" and "CPU".

[0029] In the design, the hardware configuration consists of an FPGA chip and a CPU chip. The default software operation mode is FPGA and CPU working together. The FPGA performs calibration image processing and star point extraction preprocessing, while the CPU software performs more accurate star point extraction.

[0030] When the default processing algorithm cannot meet the processing requirements, the CPU software is reconstructed using S-band (S-band frequency band) uploading software. The software operation mode is switched to CPU processing, the FPGA only performs data reception, and the CPU software performs all image processing tasks, including calibration image processing, star point extraction preprocessing, and star point extraction.

[0031] The two design architecture switching methods can satisfy both high-performance processing under normal conditions and the idea of ​​software refactoring under abnormal conditions.

[0032] The FPGA hardware-level parallel pipelined processing is fast and can quickly complete image preprocessing; the CPU has complete floating-point arithmetic functions, which is beneficial for high-precision star point positioning and matching filtering, and can realize pipelined image reception, image processing, star point matching calculation and other functions at 4Hz and above. Taking into account the cost of on-orbit FPGA software updates, CPU image processing serves as a backup function for on-orbit FGS, enabling complete software refactoring and updates, which facilitates on-orbit debugging and testing.

[0033] Relying solely on a CPU chip to perform star sensor image processing results in low real-time performance (tens of seconds), high code complexity, and high flexibility. FPGA+CPU collaboration offers moderate real-time performance (seconds), but higher code complexity and lower flexibility.

[0034] Another approach involves using an FPGA chip to perform the star sensor image processing. While this method offers high real-time performance (tens of milliseconds), it suffers from high code complexity and low flexibility.

[0035] In one embodiment, the star guide sensor image processing method includes calibration image processing, preprocessing, and star point extraction.

[0036] Upon receiving the guide star sensor image, the operating mode of the guide star sensor is determined based on the first row of the image's header information. Then, calibration image processing or star extraction processing is performed according to the guide star sensor's operating mode. When the guide star sensor is in calibration mode, calibration image processing is performed to obtain a background field correction image, a dark field correction image, and a flat field correction image. When the guide star sensor is in imaging mode, star extraction processing is performed to acquire star information. Star extraction processing includes star extraction preprocessing and star extraction itself.

[0037] The preprocessing for star point extraction includes image calibration and Gaussian filtering. The specific steps are as follows: The imaging image acquired by the guide sensor is calibrated and corrected using the background field correction image, dark field correction image, and flat field correction image to obtain the first star image. Gaussian filtering is then applied to smooth isolated point noise and stripe noise in the first star image to obtain the second star image. The imaging image can be a full-image or a windowed image.

[0038] Star point extraction includes: performing star image segmentation on the second star image to obtain a binarized star image, in which the background pixels have a value of 0 and the star pixels have a value of 1; marking the star pixels in the binarized star image to obtain a star image with star point markers; and performing sub-pixel localization processing on the star image with star point markers to obtain star point information, wherein the star point information includes the star point marker number and the centroid coordinates of the star point.

[0039] When the image is the first full-frame image, star point extraction is performed; if the image is the second full-frame image, star point extraction, star point filtering, and star point matching are performed; if the image is a windowed image, star point extraction and star point matching are performed subsequently.

[0040] Figure 1 A schematic diagram of a guide processing board hardware framework according to an embodiment of the present invention is shown.

[0041] like Figure 1 As shown, a guide processing board hardware framework includes: an FPGA chip 101, a CPU chip 102, a data interaction terminal RAM (random access memory) 103, a first memory 104, a calibration data memory 105, and a CPU memory 106.

[0042] FPGA chip 101 is connected to CPU chip 102. FPGA chip 101 has a CPU control interface 107, which is connected to CPU chip 102. Data interaction terminal RAM 103 is located in FPGA chip 101 and is connected to CPU chip 102.

[0043] Based on existing model experience, 1.5K is processed using FPGA. For a 1.5K image, extracting more than two star points requires a time performance of less than 100ms, approximately 80ms. The estimated computational scale for guide processing necessitates the use of an XQR2V3000 FPGA with 3 million gates and aerospace-grade V capability.

[0044] The guidance process is similar to that of the star sensor. The Jennifer Optonik star sensor used in the DH satellite employs an FPGA+CPU hardware structure, a Sparc V8 AT697F processor, and an output frequency of 10Hz. The Chengguang star sensor used in the XX5 satellite also uses an FPGA+CPU, with the CPU being a TI SMJ320C6701.

[0045] In this plan, the AT697F processor, which has high performance and radiation resistance, will be used with a working frequency of 64MHz.

[0046] The FPGA chip 101 is connected to the first memory 104 and the calibration data memory 105. The first memory 104 is a synchronous dynamic random access memory (SDRAM). The calibration data memory 105 stores the calibration image.

[0047] The first memory 104 is equivalent to the FPGA's main memory. The memory space required for FPGA operations (including verification functions) includes: The storage space for the two images during image processing (approximately 4MB, plus frame header information, slightly larger), and the size of the calibration data used in the calculation (possibly 1KB) 1K sizeof(float)).

[0048] CPU memory 106 is synchronous dynamic random access memory (SDRAM).

[0049] The memory space required for CPU operation (including verification functions) includes: In the FPGA+CPU collaborative working mode, the CPU requires less memory; the main consideration is the memory space required when only the CPU is operating. The program execution space is 2MB, and the data space is similar to that of the FPGA.

[0050] The FPGA chip 101 has an LVDS bus control interface 108. The LVDS bus control interface 108 connects to two independent VT-CCD interface circuits 109, one of which is connected to the host three-wire synchronous serial port LVDS 110, and the other is connected to the standby three-wire synchronous serial port LVDS 111.

[0051] The LVDS bus control interface 108 is used to receive external LVDS data, which includes full-map star charts and window star charts.

[0052] When the FPGA chip and CPU chip work together, the FPGA chip receives external LVDS data (full-image star map or window star map) and stores it in the first memory 104. At the same time, it reads the calibration image from the calibration data memory 105, performs calibration image processing on the calibration image to obtain the background field correction image, dark field correction image, and flat field correction image. Then, it performs star point extraction preprocessing on the full-image star map or window star map. The preprocessed image data is exchanged to the CPU chip through the data interaction dual-end RAM 103, and the CPU chip performs star point extraction to obtain star point information.

[0053] When only CPU chip 102 is working, the CPU chip prefetches the calibration image into CPU memory 106. FPGA chip 101 receives external LVDS data (full-image star map or window star map), stores it in the first memory 104 controlled by FPGA chip, and copies it to CPU memory managed by CPU through data interaction dual-end RAM 103. The CPU chip 102 then performs calibration image processing and star point extraction processing.

[0054] In one embodiment, the star guide processing board hardware framework further includes: a power-on reset module 112, a watchdog circuit 113, a dynamic refresh chip 114, a CAN bus interface circuit 115, a CAN communication controller 116, a CAN transceiver 117, a 1-channel CAN bus 118, an image data control interface 119, an internal LVDS 120, a telemetry output module 121, a CPU crystal oscillator 122, a program memory 123, a boot program memory 124, an RS422 interface circuit 125, an RS422 debug serial port 126, and a JTAG interface 127.

[0055] The watchdog circuit 113 is located in the FPGA chip 101. The power-on reset module 112 connects the CPU chip 102 and the watchdog circuit 113.

[0056] The dynamic refresh chip 114 is connected to the FPGA chip 101.

[0057] The CAN bus interface circuit 115 is located in the FPGA chip 101. The CAN bus interface circuit 115 is connected to the CAN communication controller 116, the CAN communication controller 116 is connected to the CAN transceiver 117, and the CAN transceiver 117 is connected to one CAN bus. The CAN communication controller 116 is an SJA1000T, and the CAN transceiver 117 is an 82c250.

[0058] Image data control interface 119 is located in FPGA chip 101. Internal LVDS 120 is connected to image data control interface 119. The satellite guide processing board hardware framework directly receives image data from LVDS and inputs it into the FPGA chip. On the one hand, it performs image processing and star point extraction; on the other hand, (via command switch) it outputs a copy of the image data through internal LVDS 120 to the satellite's large-capacity storage for downloading and checking the original image data on the ground.

[0059] The telemetry output module 121 is used for remote measurement of circuit status.

[0060] The CPU crystal oscillator 122, program memory (EEPROM) 123, boot program memory (Boot PROM) 124, RS422 interface circuit 125, and JTAG interface 127 are all connected to the CPU chip 102. RS422 interface circuit 125 is connected to the RS422 debug serial port 126. Program memory 123 is an EEPROM with 2MB of memory and includes EDAC verification functionality. Boot program memory 124 (radiation-hardened): Boot PROM with 32KB of memory.

[0061] The CPU chip contains pilot-side processing CPU software, which is divided into boot code and application layer code. The boot code, located at address 0 of the PROM, is used to boot and load the application code. Before loading, it performs a verification function on the code stored in the boot program memory 124. After the application code passes the verification, it is loaded into memory for execution. The application layer code implements the following functions: CAN communication is established with the platform's satellite computer and camera control box; Image calibration processing, star point extraction preprocessing, and precise star point extraction algorithm; Camera calibration mode processing; Software refactoring.

[0062] The functions implemented by the FPGA software for guiding satellites include: Receive LVDS serial data; Identify the image frame header information and store the two images into the first memory 104 respectively; The current operating mode (FPGA+CPU or CPU only) is analyzed through the CPU control interface 107 that interacts with the CPU. It receives control information from the CPU and exchanges data with the CPU through the dual-end RAM103 area; When working collaboratively, it performs calibration image processing and star point extraction preprocessing operations; Camera calibration mode processing; The image data is forwarded to the multiplexing board via the internal LVDS bus for multiplexing.

[0063] The image processing algorithms in the guide star processing CPU software and guide star processing FPGA software should have at least some of the following functions: In working mode, after receiving a full-image or a windowed image, it performs corresponding preprocessing, star point extraction, centroid calculation, and star point filtering and matching according to the image type.

[0064] In calibration mode, the system receives calibration images, completes on-orbit calibration processing of the background field, dark field, and flat field, and generates corresponding image correction matrices which are stored in the calibration memory.

[0065] Figure 2 A flowchart of a guide star sensor image processing method according to an embodiment of the present invention is shown.

[0066] A method for processing images of a guide star sensor includes the following steps: Based on the first row of frame header information in the guide star sensor image, the operating mode of the guide star sensor is confirmed.

[0067] Calibration image processing or star point extraction processing is performed based on the working mode of the guide star sensor; wherein: If the working mode is not imaging, then calibration image processing is performed to obtain the background field correction image, dark field correction image, and flat field correction image. If the working mode is imaging, the imaging image is calibrated and corrected using the background field correction image, dark field correction image, and flat field correction image to obtain the first star map. If the imaging image is the first full-frame image, star point extraction processing is performed to obtain star point information. If the imaging image is the second full-frame image, star point extraction, star point filtering, and star point matching are performed. If it is a windowed image, star point extraction and star point matching are performed.

[0068] The first line of the frame header information also includes information such as the guide sensor image type, window size, number of windows, window position, window exposure time, and imaging configuration. The guide sensor image type includes either a full-image or a windowed image.

[0069] The process of calibrating the background field, dark field, and flat field to obtain the corrected background field image, dark field image, and flat field image includes the following steps: Step 1: Determine whether the working mode is background field calibration. If so, perform background field calibration processing, average multiple background field images to obtain a background field corrected image. Step 2: Determine if the working mode is dark field calibration. If so, perform dark field calibration processing, average multiple dark field images, and subtract the background field correction image to obtain the dark field correction image. Step 3: Determine whether the working mode is flat field calibration. If so, perform flat field calibration processing, average multiple flat field images, subtract the background field correction image and dark field correction image, and then normalize to obtain the flat field correction image.

[0070] The requirements for baseline field calibration are that the accuracy of on-orbit baseline measurements should be less than 1 / 3 of the readout noise.

[0071] Background field calibration is performed using a background field image exposed at 0s. Ten background field images B1-B10 (BiasImage, measured in DN values) are acquired, along with the operating mode and temperature. Calibration can be performed under different configuration parameters, operating modes, and operating temperatures to obtain calibration-corrected images for different conditions. Based on the actual operating state of the imaging star map, calibration-corrected images under the same conditions are selected for preprocessing. Operating modes include full-field attitude acquisition mode, window mode, and tracking mode, among others.

[0072] Average the 10 background field images: , in Corrected background image (standard background image).

[0073] The requirements for dark field calibration are: the accuracy of CCD on-orbit dark field measurement should be better than 1 / 3 of the standard deviation of the maximum dark current.

[0074] Dark-field calibration method: Thirteen unexposed dark-field images D1-D were acquired under integration times of 0.5s, 0.8s, and 2.5s. 13 (Dark Image, measured in DN) / Operating Mode / Temperature.

[0075] Average the 13 background field images, then subtract the background field correction image: , in These are the dark-field corrected images (standard dark-field images) with integration times of 0.5s, 0.8s, and 2.5s.

[0076] The requirements for flat-field calibration are: the final flat-field calibration accuracy within each filter should be better than 0.5%.

[0077] Leveling calibration method: Acquire 22 flat-field images F1-F near half the charge of the full well. 22 (Flat Field Image, measured in DN) / Operating Mode / Temperature. This refers to 22 flat field images of LED diffuse reflection after the shutter is closed.

[0078] The 22 flat-field images are averaged, then the background field correction image and the dark field correction image are subtracted, and finally normalized to obtain the flat-field correction image.

[0079] A flat-field corrected image (standard flat-field image) can be represented by the following formula: , in and Representing the background field corrected image and the dark field corrected image, This is a flat-field corrected image under the current conditions.

[0080] The standard flat-field correction matrix can then be expressed as: , Where m and n represent the number of rows and columns of the flat field correction matrix, respectively. It is the (x, y)th element of the standard flat field correction matrix. To correct the DN value of the (x, y)th pixel in the flat-field image, The average DN value of the flat-field corrected image.

[0081] Star point extraction processing is performed to obtain star point information, including preprocessing and star point extraction. Preprocessing includes calibrating the image to obtain a first star map; and applying Gaussian filtering to the first star map to obtain a second star map. The image can be a full-image or a windowed image.

[0082] The first star map is obtained by calibrating the imaging image acquired by the guide star sensor using the background field correction image, dark field correction image, and flat field correction image.

[0083] The formula for calibrating and correcting the image acquired by the guide star sensor is as follows: , in For the first star map, For the original image, Corrected image for the background field. Correcting images for dark scenes. This is the standard flat field correction matrix. This refers to the image exposure time.

[0084] The specific steps for star point extraction based on the first star map are as follows: Step 1: Use Gaussian filtering to smooth the isolated point noise and stripe noise in the first star map to obtain the second star map.

[0085] Step 2: Perform star map segmentation on the second star map to obtain a binarized star map. The background pixels in the binarized star map have a value of 0, and the star pixels have a value of 1.

[0086] Star map segmentation of the second star map based on dynamic thresholding of edge background: in, For detector coordinates The grayscale value of the pixel; T is the grayscale threshold, calculated as follows: , in, These represent the number of rows and columns of pixels in the image sensor, respectively.

[0087] The average gray value of pixels at the image edges (first row, first column, last row, and last column): Standard deviation of gray level of image edge pixels: .

[0088] Step 3: Mark the star pixels in the binarized star map to obtain a star map with star marker numbers.

[0089] The binarized star map is scanned pixel by pixel from left to right and top to bottom. Pixels with a grayscale value greater than zero are considered valid pixels. Valid pixels are marked, with the markings starting from 1. If the current pixel is a valid pixel, and there is exactly one valid pixel among the left and top neighbor pixels of the current pixel, take its mark as the mark number of the current pixel; If the current pixel is a valid pixel, and both its left and top neighbor pixels are valid pixels, then the label number of the left neighbor pixel is taken as the label number of the current pixel; if the label numbers of the left and top neighbor pixels are different, then the label numbers of the left and top neighbor pixels are recorded in the star label equivalence table, waiting for subsequent merging of equivalent label numbers. If the left and top neighbor pixels of the current pixel are both invalid pixels, then assign a new label to the current pixel; Repeat the above process until all pixels of the binarized star map are scanned to obtain a star map with preliminary star point markings. After the scan is complete, equivalent tags are merged and new tag numbers are assigned. Remove star targets with more than 10 constituent pixels and star targets with less than 3 constituent pixels, and redistribute the marker numbers according to the number of valid star targets.

[0090] Step 4: Perform sub-pixel localization processing on the star map with star markers to obtain star information, which includes the star marker number and the centroid coordinates of the star.

[0091] Sub-pixel localization of star points is performed using the thresholded centroid method. The thresholded centroid method uses grayscale values ​​subtracted from the star map background threshold as weights. Among them, the pixel gray value of the background threshold is... , Indicates the X-coordinate of the centroid of the star point. N represents the Y-coordinate of the centroid of the star point, and N represents the total number of pixels in the connected domain of the star point. This represents the gray level of the i-th pixel in the connected domain of the star points. This represents the x-coordinate of the i-th pixel in the connected domain of the star point. This represents the y-coordinate of the i-th pixel in the connected domain of the star point.

[0092] After obtaining star information through sub-pixel positioning, determine whether the filtering flag is 1. If it is 1, star filtering is performed; otherwise, image processing is exited.

[0093] Step 5: Filter star points based on their centroid coordinates.

[0094] Based on the centroid coordinates of the stars, select star points from each guide sensor image that meet the following conditions: a) The distance from the edge of the guide sensor image is not less than 60 pixels; b) Stars with unsaturated grayscale in the image pixels of the guide sensor; c) Stars with a distance of more than 60 pixels between them; If the number of stars selected is greater than 8, then select the 8 brightest stars. If the number of selected stars is less than 1, exit image processing.

[0095] If the number of selected stars is greater than 0 and the matching flag is 1, proceed to the star matching stage; otherwise, save the star selection results, set the matching flag to 1, and exit image processing.

[0096] Step 6: Perform star point matching based on the selected stars: In each guide star sensor image, the star points selected in the current frame are compared one by one with the star points selected in the previous frame. Those with a row and column distance of less than 10 pixels are considered as matching pairs. One-to-many matching pairs are deleted to obtain one-to-one matching stars.

[0097] If the number of matching stars is less than 1, exit image processing.

[0098] Optionally, in step 7, select a navigation star from the matching stars: In each guide star sensor image, matching stars are sorted from brightest to darkest, and the four brightest stars are selected as guide stars; if the number of matching stars is less than four, all matching stars are selected as guide stars.

[0099] The satellite navigation system outputs guidance data containing the position of the navigation satellite's center of mass to the satellite computer. Based on the current satellite attitude and the navigation satellite's center of mass position, the satellite computer packages information such as the FGS operating mode, number of windows, position, size, and exposure time, and sends it to the VT camera. Simultaneously, the satellite computer calculates the satellite attitude quaternions based on measurements from other onboard sensors and changes in the navigation satellite's center of mass position.

[0100] The following sections describe the full-image processing workflow and the windowed image processing workflow, respectively. Figure 3 A flowchart of full-image processing according to an embodiment of the present invention is shown. Figure 4 A flowchart of windowed image processing according to an embodiment of the present invention is shown.

[0101] like Figure 3 As shown, the full-image processing workflow includes: Extract the first row of frame header information of the guide star sensor image to determine the working mode of the guide star sensor; if the working mode of the guide star sensor is imaging, further determine whether the guide star sensor image is a full image. If it is, perform full image processing; otherwise, switch to windowed image processing.

[0102] The steps of full-image processing include: image calibration correction, Gaussian filtering, star marking, sub-pixel star location, star filtering, star matching, and guide star selection. The details of these steps have been described above and will not be repeated here.

[0103] After completing the navigation star selection, set the filter flag to 0 and the matching flag to 0, then end the image processing.

[0104] like Figure 4 As shown, after confirming that the guide sensor image is a windowed image, the filter flag is set to 0, followed by calibration image correction, Gaussian filtering, star point marking, star point sub-pixel localization, and star point matching. The details of these steps have been described above and will not be repeated here.

[0105] If the number of matching stars is greater than 0, the matching star information is saved and output, and the image processing ends. If the number of matching stars is less than 1, the image processing exits.

[0106] While some embodiments of the present invention have been described in this application, those skilled in the art will understand that these embodiments are merely illustrative. Numerous variations, alternatives, and improvements will arise in those skilled in the art under the teachings of this invention without departing from its scope. The appended claims are intended to define the scope of the invention and thereby cover methods and structures within the scope of the claims themselves and their equivalents.

Claims

1. A method for processing images of a guide star sensor, characterized in that, include: Based on the first row of frame header information of the guide star sensor image, confirm the working mode of the guide star sensor; The calibration image processing or star point extraction processing is performed according to the working mode of the star guide sensor, wherein: If the working mode is not imaging, then calibration image processing is performed to obtain a background field correction image, a dark field correction image, and a flat field correction image. If the working mode is imaging, the imaging image is calibrated using the background field correction image, dark field correction image, and flat field correction image to obtain the first star map; wherein: if the imaging image is the first full-frame image, star point extraction processing is performed to obtain star point information; if the imaging image is the second full-frame image, star point extraction, star point filtering, and star point matching are performed; if the imaging image is a windowed image, star point extraction and star point matching are performed.

2. The star guide sensor image processing method according to claim 1, characterized in that, Star point extraction based on the first star map includes: Gaussian filtering is used to smooth the isolated point noise and stripe noise in the first star map to obtain the second star map. The second star map is segmented to obtain a binarized star map. The background pixels in the binarized star map have a value of 0, and the star pixels have a value of 1. The star pixels in the binarized star map are marked to obtain a star map with star marker numbers; and Star map with star markers is processed to locate sub-pixel star points, which yields star point information, including the star point marker number and the centroid coordinates of the star point.

3. The star guide sensor image processing method according to claim 2, characterized in that, The formula for calibrating and correcting the image acquired by the guide star sensor is as follows: ， in For the first star map, For the original image, Corrected image for the background field. Correcting images for dark scenes. This is the standard flat field correction matrix. Image exposure time; Star map segmentation of the second star map based on dynamic thresholding of edge background: in, For detector coordinates The grayscale value of the pixel; T is the grayscale threshold, calculated as follows: ， in, The standard deviation of grayscale for the edge background; These represent the number of rows and columns of pixels in the image sensor, respectively. The star pixels in the binary star map are labeled using the four-connected component analysis method, resulting in a star map with star pixel labels, including: The binarized star map is scanned pixel by pixel from left to right and top to bottom. Pixels with a grayscale value greater than zero are considered valid pixels. Valid pixels are marked, with the markings starting from 1. If the current pixel is a valid pixel, and there is exactly one valid pixel among the left and top neighbor pixels of the current pixel, take its mark as the mark number of the current pixel; If the current pixel is a valid pixel, and both its left and top neighbor pixels are valid pixels, then the label number of the left neighbor pixel is taken as the label number of the current pixel; if the label numbers of the left and top neighbor pixels are different, then the label numbers of the left and top neighbor pixels are recorded in the star label equivalence table, waiting for subsequent merging of equivalent label numbers. If the left and top neighbor pixels of the current pixel are both invalid pixels, then assign a new label to the current pixel; Repeat the above process until all pixels of the binarized star map are scanned to obtain a star map with preliminary star point markings. After the scan is complete, equivalent tags are merged and new tag numbers are assigned. Remove star targets with more than 10 constituent pixels and star targets with less than 3 constituent pixels, and redistribute the marker numbers according to the number of valid star targets.

4. The star guide sensor image processing method according to claim 1, characterized in that, If the operating mode is not imaging, then calibration image processing is performed to obtain a background field corrected image, a dark field corrected image, and a flat field corrected image, including: Determine whether the working mode is background field calibration. If so, perform background field calibration processing, average multiple background field images, and obtain a background field corrected image. If the working mode is not background field calibration, then determine whether it is dark field calibration. If so, perform dark field calibration processing, average multiple dark field images, and subtract the background field correction image to obtain the dark field correction image. If the working mode is not background field calibration and dark field calibration, then determine whether it is flat field calibration. If so, perform flat field calibration processing, average multiple flat field images, subtract the background field correction image and dark field correction image, and then normalize to obtain the flat field correction image.

5. The star guide sensor image processing method according to claim 1, characterized in that, The star point information includes the star point's marker number and the star point's centroid coordinates; The star selection includes selecting stars from each guide star sensor image that meet the following conditions based on the centroid coordinates of the stars: a) The distance from the edge of the guide sensor image is not less than 60 pixels; b) Stars with unsaturated grayscale in the image pixels of the guide sensor; c) Stars with a distance of more than 60 pixels between them; If the number of selected stars is greater than 8, select the 8 brightest stars; if the number of selected stars is less than 1, exit image processing. The star point matching includes: performing star point matching based on the selected star points, wherein: In each guide star sensor image, the star points selected in the current frame are compared one by one with the star points selected in the previous frame. Those with a row and column distance of less than 10 pixels are considered as matching pairs. One-to-many matching pairs are deleted to obtain one-to-one matching stars. If the number of matching stars is less than 1, the image processing is exited.

6. A star guide sensor image processing system, characterized in that, It includes an FPGA chip and a CPU chip, wherein the FPGA chip is configured to perform the following actions: receive a guide star sensor image according to the control instructions sent by the CPU chip, and transmit the guide star sensor image to the CPU chip or perform the calibration image processing step and the star point extraction preprocessing step in the guide star sensor image processing method; The CPU chip is configured to perform the following actions: Sending control commands to the FPGA chip; and The steps of acquiring the guide star sensor image forwarded by the FPGA chip and performing the guide star sensor image processing method include calibration image processing, star point extraction preprocessing, and star point extraction; or acquiring the preprocessed image of the FPGA chip and performing star point extraction to obtain star point information.

7. The star guide sensor image processing system according to claim 6, characterized in that, The calibration image processing steps in the guide star sensor image processing method include: Upon receiving the guide star sensor image, the system checks the first row of the image header to determine if the guide star sensor is in imaging mode. If not, it performs calibration image processing to obtain a background field correction image, a dark field correction image, and a flat field correction image, including: Determine whether the working mode is background field calibration. If so, perform background field calibration processing, average multiple background field images, and obtain a background field corrected image. If the working method is not the local field calibration, then determine whether it is dark field calibration. If so, perform dark field calibration processing, average multiple dark field images, and subtract the local field correction image to obtain the dark field correction image. If the working method is not background field calibration and dark field calibration, then determine whether it is flat field calibration. If so, perform flat field calibration processing, average multiple flat field images, subtract the background field correction image and dark field correction image, and then normalize to obtain the flat field correction image. The star point extraction preprocessing steps in the star sensor image processing method include: The imaging image acquired by the guide star sensor is calibrated and corrected using the background field correction image, dark field correction image, and flat field correction image to obtain the first star map; Gaussian filtering is used to smooth the isolated point noise and stripe noise in the first star map to obtain the second star map. Star point extraction includes the following steps: The second star map is segmented to obtain a binarized star map. The background pixels in the binarized star map have a value of 0, and the star pixels have a value of 1. The star pixels in the binarized star map are marked to obtain a star map with star marker numbers; and Star map with star markers is processed to locate sub-pixel star points, which yields star point information, including the star point marker number and the centroid coordinates of the star point.

8. A hardware framework for a satellite guide processing board, characterized in that, include: FPGA chip; The CPU chip is connected to the FPGA chip; The data interaction terminal RAM is located in the FPGA chip and connected to the CPU chip; A first memory, which is connected to the FPGA chip; A calibration data storage device, which is connected to the FPGA chip; and CPU and memory.

9. The star guide processing board hardware framework according to claim 8, characterized in that, The steps involved in the collaborative execution of the guide star sensor image processing method by the FPGA chip and CPU chip include: The FPGA chip receives the full star map or the window star map and stores it in the first memory. The FPGA chip reads the calibration image from the calibration data memory and performs calibration image processing on the calibration image to obtain the background field correction image, dark field correction image, and flat field correction image; The FPGA chip performs star point extraction preprocessing on the full-map star map or windowed star map; The preprocessed image data is transmitted to the CPU chip via dual-end RAM data exchange, where the CPU chip extracts star points to obtain star point information.

10. The star guide processing board hardware framework according to claim 8, characterized in that, The steps involved in the CPU chip executing the guide sensor image processing method include: The CPU chip prefetches the calibration image into the CPU memory; The FPGA chip receives the full star map or window star map, stores it in the first memory controlled by the FPGA chip, and copies it to the CPU memory through data interaction between the two-end RAM. The CPU chip performs calibration image processing and star point extraction.

Images