Method for generating star map of high-dynamic star sensor after proton irradiation

Abstract

The invention relates to a method for generating a star map of a high-dynamic star sensor after proton irradiation, which comprises the following steps of: converting an electron number into a gray value to obtain a proton radiation induced noise matrix, and combining simulation to obtain a gray value matrix of intrinsic dark current noise, photon shot noise and background noise; and adding with the proton radiation induced noise matrix to obtain a total noise gray value matrix, namely a background noise graph. A mathematical model of star point energy degradation under a high dynamic condition is utilized to obtain the gray level of each pixel when a star point centroid moves, a star point centroid track of the star sensor at a certain motion angular velocity is deduced, and the gray level value of the pixel in a star point energy influence range when the star sensor rotates is solved; and adding transient bright spots and star point image spot trailing extracted from the proton irradiation test acquisition image into the background noise map to obtain a high-dynamic star sensor star map after proton irradiation, and changing proton irradiation energy, fluence and star sensor rotation angular velocity to obtain a high-dynamic star sensor star map after proton irradiation. And obtaining a star map data set under different proton irradiation conditions and star sensor dynamic working conditions.

Description

Technical Field

[0001] This invention relates to the field of navigation, guidance and control, and specifically to a method for generating star maps of a high-dynamic star sensor after proton irradiation. Background Technology

[0002] A star sensor is a high-precision space attitude measurement device that uses stars as a reference frame and the starry sky as its working object. Due to its high precision, strong reliability, and good autonomy, it is widely used in various spacecraft. With the rapid development of my country's aerospace technology, agile satellites, long-range weapons, and space maneuvering platforms require high maneuverability, necessitating attitude navigation using high-dynamic star sensors.

[0003] The high angular velocity motion of agile satellites, long-range weapons, and space maneuvering platforms causes excessively rapid relative motion between the high-dynamic star sensor and the observed star, resulting in star trails on the image sensor target surface. Protons in the space environment incident on the pixel units of the CMOS image sensor cause cumulative radiation effects (total ionization dose effect, displacement damage effect) and single-event effects. The noise induced by cumulative radiation effects and the transient bright spots generated by single-event effects, overlapping with the star points, alter the grayscale distribution of the star trails, making them lose their fixed distribution characteristics. This increases the centroid positioning error of the star trails, easily causing redundant matching and mismatch problems during star image recognition, thus reducing the speed and success rate of star image recognition. The transient bright spots formed by protons incident at large angles on a CMOS image sensor are very similar in shape and grayscale distribution to star spot trails. Star image recognition programs cannot effectively distinguish between transient bright spots and star spot trails. These transient bright spots will be included in subsequent star image reconstruction, centroid localization, and recognition steps along with the star spot trails, severely affecting the attitude measurement accuracy of the star sensor. Therefore, it is urgent to study the impact mechanism of radiation damage to CMOS image sensors on the centroid localization of star spot trails in high-dynamic star sensors and to develop a centroid localization algorithm for star spot trails under space radiation conditions. However, since it is difficult to obtain star images captured by high-dynamic star sensors in orbit, computer simulation is generally used to generate star images during the development and functional debugging phases of star spot trail centroid localization algorithms. Existing star image generation methods do not incorporate radiation noise, and therefore cannot generate star images affected by space radiation. It is necessary to study the characteristics of radiation noise based on existing methods to generate dynamic star images with radiation noise, providing a dataset foundation for developing a centroid localization algorithm for star spot trails under space radiation conditions. This star map generation method can flexibly define complex optical and motion scenes, accurately simulating not only the energy degradation of star trails under different motion forms, but also realistically simulating the effect of proton irradiation on star map imaging. Summary of the Invention

[0004] The purpose of this invention is to provide a method for generating a star map of a high-dynamic star sensor after proton irradiation. This method obtains the matrix of the increase in the number of dark current electrons caused by nuclear scattering and the matrix of the increase in the number of dark current electrons caused by Coulomb scattering in all pixels, and then converts the number of electrons into gray values ​​to obtain a proton radiation-induced noise matrix. Combined with simulation, the gray value matrices of intrinsic dark current noise, photon shot noise, and background noise are obtained. These matrices are added to the proton radiation-induced noise matrix to obtain the total noise gray value matrix, i.e., the background noise map. By using a mathematical model of star point energy degradation under high dynamic conditions, the grayscale values ​​of each pixel during the motion of the star point's centroid are obtained. Then, based on the rigid body motion attitude transformation relationship of the star sensor, the trajectory of the star point's centroid at a certain motion angular velocity is derived. The grayscale values ​​of pixels within the influence range of star point energy during star sensor rotation are calculated. The transient bright spots and star point image trails extracted from the proton irradiation test images are then added to the background noise map to obtain a high-dynamic star sensor star map after proton irradiation. By changing the proton irradiation energy, flux, and star sensor rotation angular velocity, star map datasets under different proton irradiation conditions and star sensor dynamic operating conditions can be obtained.

[0005] The method for generating a star map of a high-dynamic star sensor after proton irradiation, as described in this invention, is performed according to the following steps:

[0006] a. Based on the incident proton energy E, determine the nuclear elastic scattering cross section and nuclear inelastic scattering cross section between the proton and Si. Combined with the proton fluence Φ, obtain the exponential distributions F1(x) and F2(x) for the increase in dark current electron number caused by nuclear elastic scattering and nuclear inelastic scattering, respectively. Then, obtain the matrix M of the increase in dark current electron number caused by nuclear scattering in all pixels by sampling from this distribution. n (i,j);

[0007] λ i =Φσ i N atom V dep (1)

[0008]

[0009] λ i Φ represents the average number of times the i-th interaction occurs in a pixel, where Φ is the proton fluence and N is the mean number of times the i-th interaction occurs. atom σ is the atomic density. i V represents the cross section under the interaction of Coulomb elastic scattering, nuclear elastic scattering, and nuclear inelastic scattering. dep Let be the pixel volume, and n be the number of nuclear reactions occurring within the pixel volume. δ(x) is the probability density function of the increase in dark current of the pixel experiencing zero nuclear scattering. dark The mean of the exponential distribution is given by x, where x is the increase in the number of dark current electrons.

[0010] b. The distribution of PKA energy Er generated when the proton incident energy is E and the incident Si target thickness is d was simulated using the Monte Carlo program TRIM, and the Lindhard energy allocation function was used to allocate the PKA energy Er to the non-ionization energy loss E. n The energy distribution En' used by each PKA to cause atomic displacement is then obtained after adjustment using the NRT model. En' is then converted into the displacement damage dose D. dd And using the universal damage factor K dark This is converted into dark current electrons, yielding the distribution F1(x) of the increase in dark current electron number caused by Coulomb elastic scattering. Finally, combining the fluence Φ and the Coulomb elastic scattering cross section, the matrix M of the increase in dark current electron number caused by Coulomb scattering in all pixels is obtained by sampling on the F3(x) distribution. c (i,j);

[0011] c. Add the results from steps a and b to obtain the distribution matrix M of the increase in dark current electron number caused by protons with specific energies and specific flux. c (i,j)+M n (i,j), and convert the number of electrons into grayscale values ​​through the conversion gain to obtain the proton radiation-induced noise matrix M. p (i,j);

[0012] d. Simulate other noises in the CMOS image sensor, including intrinsic dark current noise, photon shot noise, and background noise. Then, sample the distributions of intrinsic dark current noise, photon shot noise, and background noise respectively to obtain the grayscale matrix M of other noises in the CMOS image sensor. r (i,j), and the proton radiation-induced noise matrix M in step c. p The sum of (i,j) values ​​yields the total noise grayscale matrix M(i,j), which is the background noise map;

[0013] e. After obtaining the background noise map in step d, simulate the star spot trailing effect formed by the star sensor under high dynamic conditions. Use a mathematical model of star energy degradation under high dynamic conditions to obtain the grayscale I of each pixel during the movement of the star centroid. ij Then, based on the rigid body motion attitude transformation relationship of the star sensor, the angular velocity of the star sensor is derived as (ω x ,ω y ,ω z The trajectory of the centroid of the star at time (x) c (t),y c (t)), then substitute it into I ij In the calculation formula, the grayscale value of the pixel within the influence range of the star point energy when the star sensor rotates is obtained;

[0014]

[0015] E dyn (x,y) represents the energy distribution on the star sensor image plane under dynamic conditions, Φ is the incident light flux, T is the integration time, and ρ is the defocus Gaussian radius. c ,y c Let (x) be the coordinates of the star's centroid, and let (x) be the trajectory of the star's centroid during the integration time. c (t),y c (t)), η QE For photoelectric quantum efficiency, K is the conversion gain, f is the focal length of the star sensor optical system, and the initial coordinates of the star's centroid are (x(0), y(0)).

[0016] f. Conduct proton irradiation experiments with different energies and angles on the CMOS image sensor, acquire images online during the irradiation process, and extract transient bright spots in the images using a dual threshold segmentation method;

[0017] g. Add the star spot tail obtained in step e and the transient bright spot obtained in step f to the background noise map obtained in step d to obtain the high dynamic star sensor star map after proton irradiation. By setting different proton energies E, fluence Φ, and star sensor rotation angular velocity (ω), the high dynamic star sensor star map can be obtained. x ,ω y ,ω z By repeating steps a to g, we can obtain star map datasets under different proton irradiation conditions and dynamic working conditions of star sensors.

[0018] This invention discloses a method for generating star maps using a high-dynamic star sensor after proton irradiation. Building upon existing methods, this approach studies radiation noise characteristics to generate dynamic star maps with added radiation noise. It allows for flexible definition of complex optical and motion scenarios, accurately simulating the energy degradation of star trails under different motion conditions and realistically simulating the impact of proton irradiation on star map imaging. This provides a dataset foundation for developing algorithms to locate the centroid of star trails in space radiation environments. The star maps generated by this method realistically recreate star maps under space radiation environments and motion conditions, facilitating the development of subsequent star map recognition and star centroid localization algorithms. Attached Figure Description

[0019] Figure 1 This is the construction process of the simulated star map dataset of the present invention;

[0020] Figure 2 The nuclear elastic scattering cross section of protons at different energies;

[0021] Figure 3 The nuclear inelastic scattering cross section for protons at different energies;

[0022] Figure 4An exponential distribution representing the increase in the number of dark current electrons caused by nuclear elastic scattering and nuclear inelastic scattering;

[0023] Figure 5 The PKA energy distribution for 10MeV and 80MeV protons;

[0024] Figure 6 The Coulomb elastic scattering cross section of protons at different energies;

[0025] Figure 7 The extracted transient bright spot image;

[0026] Figure 8 This is a star map generated under dynamic conditions. Detailed Implementation

[0027] The present invention will now be described in further detail with reference to the accompanying drawings.

[0028] Example

[0029] The method for generating a star map of a high-dynamic star sensor after proton irradiation, as described in this invention, is specifically performed according to the following steps:

[0030] The irradiation experiment used a domestically produced scientific-grade 4T CMOS image sensor with a resolution of 2048×2048, a pixel size of 11μm×11μm, and a total pixel array size of 22mm×22mm.

[0031] a. Based on the incident proton energy E, determine the nuclear elastic scattering cross section and nuclear inelastic scattering cross section between the proton and Si. Combined with the proton fluence Φ, obtain the exponential distributions F1(x) and F2(x) for the increase in dark current electron number caused by nuclear elastic scattering and nuclear inelastic scattering, respectively. Then, obtain the matrix M of the increase in dark current electron number caused by nuclear scattering in all pixels by sampling from this distribution. n (i,j);

[0032] λ i =Φσ i N atom V dep (1)

[0033]

[0034] λ i This represents the average number of times the i-th interaction occurs in a pixel. Φ is the proton fluence, and N... atom σ is the atomic density. i V represents the cross section under the interaction of Coulomb elastic scattering, nuclear elastic scattering, and nuclear inelastic scattering. depLet be the pixel volume, n be the number of nuclear reactions occurring within the pixel volume, and δ(x) be the probability density function of the increase in dark current of the pixel with zero nuclear scattering. dark The mean of the exponential distribution is given by x, where x is the increase in the number of dark current electrons.

[0035] b. The distribution of PKA energy Er generated when the proton incident energy is E and the incident Si target thickness is d was simulated using the Monte Carlo program TRIM, and the Lindhard energy allocation function was used to allocate the PKA energy Er to the non-ionization energy loss E. n After adjustment using the NRT model, the energy distribution En' used by each PKA to cause atomic displacement is obtained, and then En' is converted into displacement damage dose D. dd And using the universal damage factor K dark This is converted into dark current electrons, yielding the Coulomb elastic scattering dark current electron number increase distribution F3(x); then, combining the fluence Φ and the Coulomb elastic scattering cross section, sampling on the F1(x) distribution yields the matrix M of dark current electron number increase caused by Coulomb scattering in all pixels. c (i,j);

[0036] c. Add the results from steps a and b to obtain the distribution matrix M of the increase in dark current electron number caused by protons with specific energies and specific flux. c (i,j)+M n (i,j), and convert the number of electrons into grayscale values ​​through the conversion gain to obtain the proton radiation-induced noise matrix M. p (i,j);

[0037] d. Simulate other noises in the CMOS image sensor, including intrinsic dark current noise, photon shot noise, and background noise. Then, sample the distributions of intrinsic dark current noise, photon shot noise, and background noise respectively to obtain the grayscale matrix M of other noises in the CMOS image sensor. r (i,j), and then combined with the proton radiation-induced noise matrix M in step c. p The sum of (i,j) values ​​yields the total noise grayscale matrix M(i,j), which is the background noise map;

[0038] e. After obtaining the background noise map in step d, simulate the star spot trailing formed by the star sensor under high dynamic conditions, and use the mathematical model of star point energy degradation under high dynamic conditions to obtain the gray level I of each pixel when the centroid of the star point moves. ij Then, based on the rigid body motion attitude transformation relationship of the star sensor, the angular velocity of the star sensor is derived as (ω x ,ω y ,ω z The trajectory of the centroid of the star at time (x) c (t),yc (t)), then substitute it into I ij In the calculation formula, the grayscale value of the pixel within the influence range of the star point energy when the star sensor rotates is obtained;

[0039]

[0040] E dyn (x,y) represents the energy distribution on the star sensor image plane under dynamic conditions, Φ is the incident light flux, T is the integration time, and ρ is the defocus Gaussian radius. c ,y c Let (x) be the coordinates of the star's centroid, and let (x) be the trajectory of the star's centroid during the integration time. c (t),y c (t)), η QE For photoelectric quantum efficiency, K is the conversion gain, f is the focal length of the star sensor optical system, and the initial coordinates of the star's centroid are (x(0), y(0)).

[0041] f. Conduct proton irradiation experiments with different energies and angles on the CMOS image sensor, acquire images online during the irradiation process, and extract transient bright spots in the images using a dual threshold segmentation method;

[0042] g. Add the star spot tail obtained in step e and the transient bright spot obtained in step f to the background noise map obtained in step d, thus obtaining the high-dynamic star sensor star map after proton irradiation. This can be achieved by setting different proton energies E, fluence Φ, and star sensor rotation angular velocity (ω). x ,ω y ,ω z By repeating steps a to g, we can obtain star map datasets under different proton irradiation conditions and dynamic working conditions of star sensors.

[0043] See attached diagram:

[0044] a. Based on the incident proton energy E, determine the nuclear elastic scattering cross section and nuclear inelastic scattering cross section between the proton and Si. Combined with the proton fluence Φ, obtain the exponential distributions F1(x) and F2(x) for the increase in dark current electron number caused by nuclear elastic scattering and nuclear inelastic scattering, respectively. Then, obtain the matrix M of the increase in dark current electron number caused by nuclear scattering in all pixels by sampling from this distribution. n (i,j);

[0045] λ i =Φσ i N atom V dep (1)

[0046]

[0047] λi This represents the average number of times the i-th interaction occurs in a pixel. Φ is the proton fluence, and N... atom σ is the atomic density. i V represents the cross section under the interaction of Coulomb elastic scattering, nuclear elastic scattering, and nuclear inelastic scattering. dep Let be the pixel volume, and n be the number of nuclear reactions occurring within the pixel volume. δ(x) is the probability density function of the increase in dark current of the pixel experiencing zero nuclear scattering. dark The mean of the exponential distribution is given by x, where x is the increase in the number of dark current electrons.

[0048] Nuclear elastic scattering cross section and nuclear inelastic scattering cross section at different proton energies, such as Figure 2 and Figure 3 As shown, the exponential distribution of the increase in dark current electron number caused by nuclear elastic scattering and nuclear inelastic scattering is as follows: Figure 4 As shown;

[0049] b. The distribution of PKA energy Er generated when the proton incident energy is E and the incident Si target thickness is d was simulated using the Monte Carlo program TRIM, and the Lindhard energy allocation function was used to allocate the PKA energy Er to the non-ionization energy loss E. n After adjustment using the NRT model, the energy distribution En' used by each PKA to cause atomic displacement is obtained, and then En' is converted into displacement damage dose D. dd And using the universal damage factor K dark This is converted into dark current electrons, yielding the distribution F1(x) of the increase in dark current electron number caused by Coulomb elastic scattering. Then, combining the fluence Φ and the Coulomb elastic scattering cross section, the matrix M of the increase in dark current electron number caused by Coulomb scattering in all pixels is obtained by sampling on the F1(x) distribution. c (i,j);

[0050] The PKA energy distributions of 10 MeV and 80 MeV protons are as follows: Figure 5 As shown, the proton Coulomb elastic scattering cross section at different energies is as follows: Figure 6 As shown;

[0051] c. Add the results from steps a and b to obtain the distribution matrix M of the increase in dark current electron number caused by protons with specific energies and specific flux. c (i,j)+M n (i,j), and convert the number of electrons into grayscale values ​​through the conversion gain to obtain the proton radiation-induced noise matrix M. p (i,j);

[0052] d. Simulate other noises in the CMOS image sensor, including intrinsic dark current noise, photon shot noise, and background noise. Then, sample the distributions of intrinsic dark current noise, photon shot noise, and background noise respectively to obtain the grayscale matrix M of other noises in the CMOS image sensor. r (i,j), and then combined with the proton radiation-induced noise matrix M in step c. p The sum of (i,j) values ​​yields the total noise grayscale matrix M(i,j), which is the background noise map;

[0053] e. After obtaining the background noise map in step d, simulate the star spot trailing formed by the star sensor under high dynamic conditions, and use the mathematical model of star point energy degradation under high dynamic conditions to obtain the gray level I of each pixel when the centroid of the star point moves. ij Then, based on the rigid body motion attitude transformation relationship of the star sensor, the angular velocity of the star sensor is derived as (ω x ,ω y ,ω z The trajectory of the centroid of the star at time (x) c (t),y c (t)), then substitute it into I ij In the calculation formula, the grayscale value of the pixel within the influence range of the star point energy when the star sensor rotates is obtained;

[0054]

[0055] E dyn (x,y) represents the energy distribution on the star sensor image plane under dynamic conditions, Φ is the incident light flux, T is the integration time, and ρ is the defocus Gaussian radius. c ,y c Let (x) be the coordinates of the star's centroid, and let (x) be the trajectory of the star's centroid during the integration time. c (t),y c (t)). η QE For photoelectric quantum efficiency, K is the conversion gain, f is the focal length of the star sensor optical system, and the initial coordinates of the star's centroid are (x(0), y(0)).

[0056] f. Conduct proton irradiation experiments with different energies and angles on the CMOS image sensor. Images are acquired online during the irradiation process, and transient bright spots are extracted from the images using a dual-threshold segmentation method. The extracted transient bright spots are shown below. Figure 7 As shown;

[0057] g. Add the star spot tail obtained in step e and the transient bright spot obtained in step f to the background noise map obtained in step d, thus obtaining the high-dynamic star sensor star map after proton irradiation. This can be achieved by setting different proton energies E, fluence Φ, and star sensor rotation angular velocity (ω). x ,ωy ,ω z By repeating steps a to g, we obtain star map datasets under different proton irradiation conditions and dynamic operating conditions of the star sensor. x =ω y =ω=7° / s, the exposure time for each star chart is between 70ms and 110ms, and the generated star charts under dynamic conditions are as follows: Figure 8 As shown.

[0058] The above description is merely a specific implementation of the method for generating star maps of high-dynamic star sensors after proton irradiation according to the present invention. However, the scope of protection of the present invention is not limited thereto. Any substitutions or additions that can be understood by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of the present invention.

Claims

1. A method for generating a star map of a high-dynamic star sensor after proton irradiation, characterized in that... Follow these steps: a. Based on the incident proton energy E, determine the nuclear elastic scattering cross section and nuclear inelastic scattering cross section between the proton and Si. Combined with the proton fluence Φ, obtain the exponential distributions F1(x) and F2(x) for the increase in dark current electron number caused by nuclear elastic scattering and nuclear inelastic scattering, respectively. Then, obtain the matrix M of the increase in dark current electron number caused by nuclear scattering in all pixels by sampling from this distribution. n (i,j); l i =Fs i N atom V dep (1) λ i This represents the average number of times the i-th interaction occurs in a pixel. Φ is the proton fluence, and N... atom σ is the atomic density. i V represents the cross section under the interaction of Coulomb elastic scattering, nuclear elastic scattering, and nuclear inelastic scattering. dep Let v be the pixel volume, n be the number of nuclear reactions occurring within the pixel volume, and δ(x) be the probability density function of the increase in dark current of the pixel with zero nuclear scattering. dark The mean of the exponential distribution is given by x, where x is the increase in the number of dark current electrons. b. The distribution of PKA energy Er generated when the proton incident energy is E and the incident Si target thickness is d was simulated using the Monte Carlo program TRIM, and the Lindhard energy allocation function was used to allocate the PKA energy Er to the non-ionization energy loss E. n After adjustment using the NRT model, the energy distribution En' used by each PKA to cause atomic displacement is obtained, and then En' is converted into displacement damage dose D. dd And using the universal damage factor K dark This is converted into dark current electrons, yielding the Coulomb elastic scattering dark current electron number increase distribution F1(x); then, combining the fluence Φ and the Coulomb elastic scattering cross section, sampling on the F3(x) distribution yields the matrix M of dark current electron number increase caused by Coulomb scattering in all pixels. c (i,j); c. Add the results obtained in steps a and b to obtain the distribution matrix M of the increase in the number of dark current electrons caused by protons with specific energies and specific flux. c (i,j)+M n (i,j), and convert the number of electrons into grayscale values ​​through the conversion gain to obtain the proton radiation-induced noise matrix M. p (i,j); d. Simulate other noises in the CMOS image sensor, including intrinsic dark current noise, photon shot noise, and background noise. Then, sample the distributions of intrinsic dark current noise, photon shot noise, and background noise respectively to obtain the grayscale matrix M of other noises in the CMOS image sensor. r (i,j), and then combined with the proton radiation-induced noise matrix M in step c. p The sum of (i,j) values ​​yields the total noise grayscale matrix M(i,j), which is the background noise map; e. After obtaining the background noise map in step d, simulate the star spot trailing formed by the star sensor under high dynamic conditions, and use the mathematical model of star point energy degradation under high dynamic conditions to obtain the gray level I of each pixel when the centroid of the star point moves. ij Then, based on the rigid body motion attitude transformation relationship of the star sensor, the angular velocity of the star sensor is derived as (ω x ,ω y ,ω z The trajectory of the centroid of the star at time (x) c (t),y c (t)), then substitute it into I ij In the calculation formula, the grayscale value of the pixel within the influence range of the star point energy when the star sensor rotates is obtained; E dyn (x,y) represents the energy distribution on the star sensor image plane under dynamic conditions, Φ is the incident light flux, T is the integration time, and ρ is the defocus Gaussian radius. c ,y c Let (x) be the coordinates of the star's centroid, and let (x) be the trajectory of the star's centroid during the integration time. c (t),y c (t)), η QE For photoelectric quantum efficiency, K is the conversion gain, f is the focal length of the star sensor optical system, and the initial coordinates of the star's centroid are (x(0), y(0)). f. Conduct proton irradiation experiments with different energies and angles on the CMOS image sensor, acquire images online during the irradiation process, and extract transient bright spots in the images using a dual threshold segmentation method; g. Add the star spot tail obtained in step e and the transient bright spot obtained in step f to the background noise map obtained in step d, thus obtaining the high-dynamic star sensor star map after proton irradiation. This can be achieved by setting different proton energies E, fluence Φ, and star sensor rotation angular velocity (ω). x ,ω y ,ω z By repeating steps a to g, we can obtain star map datasets under different proton irradiation conditions and dynamic operating conditions of star sensors.

Images