High-resolution computational imaging method for optical pupil interferometry of chessboard constellations

Abstract

The present invention relates to the field of optoelectronic imaging, in particular to the field of ultra-high spatial resolution imaging. The present invention discloses a chessboard constellation optical pupil interferometry high-resolution computational imaging method. The method adopts a chessboard constellation high-density data acquisition system to achieve high-density sampling of the modulus G of the complex coherence coefficient μ(u, v) of the object light, and then combines the phase recovery (also known as phase recovery) algorithm and the image reconstruction algorithm to obtain a clear image of the object light. For ultra-large aperture optical telescopes that far exceed the rocket envelope, low, medium and high frequency cameras can be placed on different rocket satellite platforms, launched into orbit in batches, and reorganized in orbit to obtain an equivalent large optical aperture imaging optical system. Its equivalent aperture breaks through the rocket envelope limit and can be expanded to 10 meters, or even larger.

Description

Technical Field

[0001] The present invention relates to the field of photoelectric imaging, in particular to the field of ultra-high spatial resolution imaging. Background Art

[0002] To explore the mysteries of the universe and understand our own planet, scientists are tirelessly striving to increase the aperture of space optical telescopes and enhance their spatial resolution. Large-aperture optical telescopes have experienced rapid development over the past half century, continuously refining our understanding of the world. Numerous Nobel Prizes in physics have been awarded thanks to advanced optical telescope technology.

[0003] Space optical telescopes, shielded from the turbulence of Earth's atmosphere, offer unparalleled advantages in observational conditions. However, due to various limitations, such as the size of the launch vehicle envelope, their apertures have been limited. After 30 years of orbital operation of the 2.4-meter Hubble Telescope, scientists, through tireless efforts, successfully launched the spliced-aperture James Webb Telescope to the temperature-stable Lagrange L2 point in space on Christmas Day 2021. This marks the largest space optical telescope aperture to date, reaching 6.5 meters. Observations from these two telescopes have shed light on a range of fundamental cosmological questions. Astronomers eagerly anticipate that the aperture of space optical telescopes will exceed 10 meters, unlocking the mysteries of the universe through larger apertures and higher resolutions.

[0004] As astronomical space optical telescope technology advances, so too has the development of space optical telescopes for Earth observation, used to understand our blue planet. Following Russia's development of the 2.4-meter Kanopus-V1 and the United States' 2.4-meter Keyhole (KH) satellite, Japan is developing a 3.6-meter high-resolution Earth-observing optical telescope to meet the dual needs of rapid response and high-resolution Earth observation, thereby alleviating human suffering and damage after large-scale disasters. Due to the relatively harsh temperature environments in geostationary and sun-synchronous orbits, space optical telescopes currently utilize single, large-aperture optical structures. Publicly available literature indicates that the largest aperture is less than 4 meters.

[0005] In short, whether it's to peer deeply into the universe, explore the origins of humanity and the universe, or respond to emergencies on our planet and save humanity, the aperture of space optical telescopes urgently needs to exceed 10 meters. So, what technical solutions will be adopted for space optical telescopes exceeding 10 meters, or even reaching 100 meters? Will the James Webb spliced ​​aperture technology remain? If the spliced ​​aperture telescope is increased to 10 meters, how can it break through the limitations of the rocket fairing envelope and achieve a stable optical wavefront of tens of nanometers after in-orbit splicing? This goal is particularly difficult to achieve when placed in geostationary or sun-synchronous orbits, where the orbital temperature field is harsh. Can the technology of ground-based astronomical interferometers be borrowed? NASA and the European Space Agency (ESA) have pioneered the free-flyer interferometer program. The Space Interferometry Mission (SIM) is expected to be the first space-based long-baseline optical interferometer, with astronomical measurements far exceeding the capabilities of any other existing or under development projects. Research on this project continues, but progress is unclear. Therefore, we need to continue to broaden our thinking and explore more solutions.

[0006] This paper proposes a method for ultra-high-resolution computational imaging using optical pupil interferometry for chessboard constellations. This method overcomes the limitations imposed by the rocket envelope on the aperture of optical telescopes, creating a technical solution for space optical telescopes with equivalent apertures of tens or even hundreds of meters. This method provides a new approach for high-resolution astronomical and Earth observations, and will contribute to the rapid development of space optical telescope technology. Summary of the Invention

[0007] According to the Van Sit-Zernike law, an incoherent light source with an arbitrary spatial object plane light intensity distribution I(ξ,η) can be expressed as a complex coherence coefficient μ(u,v) at any interference baseline aperture pair (x1, y1), (x2, y2) on the equivalent pupil plane of the optical system, where u and v are the spatial frequencies and satisfy: (x1, y1), (x2, y2) are the coordinates of the two apertures on the pupil plane, is the average operating wavelength, Δx = x1 - x2, Δy = y1 - y2, and Z0 is the object distance. The complex coherence coefficient μ(u, v) is the normalized Fourier transform of the incoherently extended light source distribution I(ξ, η). Therefore, if the complex coherence coefficient μ(u, v) of the light source can be measured, the Fourier spectrum of the light source can be obtained, and the light source intensity distribution I(ξ, η) can be calculated through the inverse Fourier transform.

[0008] The complex coherence coefficient μ(u, v) includes the modulus G and the phase angle φ (also known as phase φ), which can be expressed as μ = |G|e jφ In theory, it can be measured by the visibility of the interference fringes between the baseline aperture pair and the phase difference at zero optical path. Visibility is relatively easy to obtain, but measuring the argument φ at the wavelength level on a scale of tens, tens, or even hundreds of meters is not easy.

[0009] Common problems with existing large-scale long-baseline interferometers include: (1) low spatial spectrum coverage. For example, CHARA, which once obtained the image of Altair, combined the beams of six telescopes placed in a fixed position into CHARA's MIRC-X (the Michigan InfraRed Combiner-eXeter) beam combiner, which is one of the combiners in the world that can achieve the largest number of beams combined, and can achieve 15 baseline visibility and 10 closed phase observations. Due to the small number of baselines, the spatial spectrum coverage is improved only by the inter-satellite relative motion caused by the autonomous rotation of the earth. In an observation mission of Vega, only 25 spatial spectrum points were obtained after 6 nights. This type of imaging system is ultimately applicable to the imaging of simple stellar disks or binary star systems due to its low spatial spectrum coverage and severe limitations on low-frequency sampling. (2) It is difficult to measure the phase angle of the complex coherence coefficient. Astronomical interferometers often use channeled spectrum to disperse broadband optical signals and record them with imaging detectors. By observing the number of fringes at different optical path differences, the location of zero optical path difference is determined, thereby achieving measurement of the complex coherence coefficient angle. However, in complex space environments, at scales of tens, tens, or even hundreds of meters, the truss is affected by vibration, heat, and gravity gradients, causing deformation. This can cause the pre-calibrated zero optical path difference position to drift and lose its reference. During operation, the instrument needs to be repeatedly calibrated with the help of reference targets. In addition to the channeled spectrum method, the field of astronomical interferometry also commonly uses a closed phase measurement method. Three sets of relative phases are obtained by pairing three apertures in pairs. Although this method can theoretically eliminate the influence of atmospheric turbulence, because the number of closed phases is always less than the true phase, a specific algorithm needs to be designed to solve the true complex coherence coefficient angle.

[0010] When the complex coherence coefficient's amplitude and angle cannot be measured, scientists attempt to use an amplitude recovery algorithm to recover the amplitude and angle, thereby completing pupil interferometry image reconstruction. The GS algorithm provides a classic iterative algorithm for solving this type of problem, and was later improved by Fienup to form multiple variants. Because the GS algorithm is equivalent to the error reduction algorithm in the Fienup algorithm, it can be collectively referred to as the Fienup algorithm. This type of iteration often requires super-Nyquist spectral coverage, which limits its performance in existing long-baseline interferometers with low spatial frequency coverage.

[0011] Based on the above theoretical and technical foundations, the present invention discloses a chessboard constellation optical pupil interferometry high-resolution computational imaging method. This method is based on the principle of optical pupil interferometry computational imaging and adopts a chessboard constellation high-density data acquisition system to achieve high-density sampling of the modulus G of the object light complex coherence coefficient μ(u, v). Then, it combines the amplitude angle recovery algorithm (also known as phase recovery) and the image reconstruction algorithm to obtain a clear object light image that meets certain constraints, such as Figure 1 As shown, the chessboard constellation high-density data acquisition system ( Figure 1 1) is generally composed of a single-aperture camera and multiple chessboard imagers with different minimum baselines, which can achieve high-density sampling of the modulus G of the object-light complex coherence coefficient μ(u, v). The constraints in the process of this method ( Figure 1 2) is generally determined by the specific structural parameters of the chessboard constellation high-density data acquisition system and the imaging object distance. Figure 1 3) maximizes the use of high-density modulo G data of the complex coherence coefficient μ(u, v), and obtains a clear image with high spatial resolution through an iterative optimization algorithm. During the initial iterative optimization of the amplitude recovery and image reconstruction algorithm, a random image is introduced into the spatial domain. The amplitude φ of the complex coherence coefficient μ(u, v) in the spatial frequency domain is obtained through the physical transmission model of the chessboard constellation high-density data acquisition system. This is combined with the modulo G of the measured high-density object light complex coherence coefficient μ(u, v). The iterative optimization process continues until the reconstructed image satisfies the constraints in the spatial frequency domain or frequency domain. The optimization ends and a clear image is output.

[0012] In order to obtain a high sampling rate of the complex coherence coefficient modulus in the spatial frequency domain of the object light, the chessboard constellation high-density data acquisition system of the present invention ( Figure 1 1) of the invention patent 201711000143.2 refers to the chessboard-type rectangular aperture array arrangement structure and sampling method. This sampling method can achieve continuous, non-redundant, and non-omission spatial frequency continuous sampling within the spatial frequency range limited by the longest baseline. The information acquisition system using the chessboard imager is detailed in the invention patent 202010965700.X. Figure 2As shown, the object light is collected by the chessboard array (101) through apertures of different baseline lengths, and is successively transmitted through the waveguide grating splitter (102) and the single-mode optical fiber (103) in pairs to control the coherence. The coherent information envelope is scanned by adjusting the phase delay (104), and the photoelectric detection and readout system (105) obtains the modulus G of the complex coherence coefficient μ(u, v) of different spatial frequencies. During the signal acquisition process, the paired apertures of different baseline lengths in the chessboard imager adjust the phase of the phase delay of the subsequent transmission light path, thereby scanning the position of zero optical path difference of the paired light path, collecting and storing the contrast of the interference fringes, that is, obtaining the modulus G of the complex coherence coefficient μ(u, v) of the full spatial spectrum within the maximum baseline limit. However, the optical pupil interferometric imaging system of a single chessboard imager has a system field of view limited by a single small aperture size D. Among them, λ is the observation wavelength, corresponding to the spatial frequency domain characteristic period u o =1 / FoV, corresponding to one times the Nyquist sampling frequency. According to the sampling theorem, for the extended target, the spatial frequency domain sampling period u c =B min / λ should be smaller than the characteristic period u corresponding to the field of view o , where the minimum interval of the sampling baseline is Obviously, the above-mentioned single chessboard aperture array cannot meet the sampling requirement. In order to meet the sampling requirement and avoid spatial physical overlap, the chessboard constellation high-density data acquisition system ( Figure 1 1) consists of a traditional single-aperture camera and multiple chessboard aperture pairs with different minimum baselines. Taking the realization of 2 times the Nyquist frequency as an example, the high-density acquisition system of the modulus G of the spatial frequency domain complex coherence coefficient μ(u, v) consists of 5 optical systems, including 4 chessboard imagers and 1 single-aperture camera, forming a constellation, as shown in Figure 3 As shown, the minimum baseline length B corresponding to the chessboard imager 1 (106) min_1 =D; the minimum baseline length corresponding to the chessboard imager 2 (107) Minimum baseline length corresponding to chessboard imager 3 (108) Minimum baseline length corresponding to chessboard imager 4 (109) Due to the corresponding baseline length B min The modulus G of the spatial frequency domain complex coherence coefficient μ(u, v) smaller than D cannot be obtained through aperture pair interferometry, so a traditional single-aperture camera (110) is required to obtain a spatial domain image and then invert to obtain the modulus G of the frequency domain complex coherence coefficient μ(u, v). Because the corresponding aperture or baseline length is small, the collected data corresponds to low-frequency information in the spatial frequency domain.

[0013] As can be seen from the above working principle, the chessboard imager in the chessboard constellation high-density data acquisition system realizes discrete acquisition of the complex coherence coefficient μ(u, v) module G of the target light on the space surface through the aperture pair arrays with different baseline lengths within the maximum baseline limit. Therefore, when the equivalent ultra-large aperture optical telescope of the chessboard constellation high-density data acquisition system far exceeds the rocket envelope, that is, when the maximum baseline length of the chessboard imager in the chessboard constellation high-density data acquisition system exceeds the rocket envelope, it can be divided into aperture pair array low-frequency cameras with short baseline length, aperture pair array medium-frequency cameras with medium baseline length, and aperture pair array high-frequency cameras with long baseline length according to the aperture pair baseline length in the chessboard imager. The cameras can be constructed by placing aperture pair array cameras with different baseline lengths on different satellite platforms and launching them in batches for on-orbit assembly. Figure 4 As shown, for example, the aperture pair array low-frequency camera (111) with a short baseline length is placed on the satellite platform rocket_1, and the aperture pair array medium-frequency camera (112) with a medium baseline length is placed on the satellite platform rocket_2. The aperture pair array high-frequency cameras (113, 114) with a long baseline length can be split and placed on the satellite platform rocket_3 and the satellite platform rocket_4 according to space requirements. After being launched into orbit, they are reassembled to obtain an equivalent ultra-large aperture optical telescope. Among them, the aperture pair array low-frequency camera with a short baseline length can be replaced by a traditional single-aperture camera to obtain low-spatial spectrum information of the target to be measured. Therefore, the low-frequency camera is generally a traditional single-aperture camera.

[0014] When detecting and imaging non-transient targets such as uninhabited islands and national borders, the chessboard constellation high-density data acquisition system consists of relatively moving aperture pairs, such as Figure 5 As shown, the aperture of satellite A is relatively stationary, while the aperture of satellite B moves step by step relative to the aperture of satellite A along a predetermined path. With each step of aperture B's movement, the system completes a sampling of the modulus G of the complex coherence coefficient μ(u, v) of the object light. When aperture B completes the predetermined trajectory and sampling, high-density sampling of the modulus G of the complex coherence coefficient μ(u, v) of the object light is completed within the maximum baseline limit. The sampling of the low spatial spectrum region covered by the relative motion equivalent short baseline length aperture pair array can be replaced by a traditional single aperture camera. That is, the motion range of the relative motion aperture pair covers the mid- and high-frequency regions in addition to the low-frequency region covered by the traditional single aperture camera.

[0015] In summary, the present invention discloses a chessboard constellation optical pupil interferometry high-resolution computational imaging method. The method is based on the principle of optical pupil interferometry computational imaging, and adopts a chessboard constellation high-density data acquisition system composed of a traditional single-aperture camera and a plurality of chessboard aperture arrays with unequal minimum baselines to achieve high-density sampling of the modulus G of the complex coherence coefficient μ(u, v) of the object light, and then combines the phase angle recovery algorithm and the image reconstruction algorithm to obtain a clear image of the object light that meets certain constraints. When building an ultra-large aperture optical telescope that far exceeds the rocket envelope, that is, when its maximum baseline length far exceeds the rocket envelope, a technical route of multi-satellite batch launch and on-orbit assembly and construction is implemented. A low-frequency camera with a short baseline length, a medium-frequency camera with a medium baseline length, and a high-frequency camera with a long baseline length are placed on different rocket satellite platforms, launched into orbit in batches, and reassembled on-orbit to obtain an imaging optical system with an equivalent ultra-large optical aperture, so that its equivalent optical aperture breaks through the rocket envelope limit and can be expanded to 10 meters, or even larger.

[0016] The sampling density of the modulus G of the object-light complex coherence coefficient μ(u, v) for the high-density sampling mentioned above is higher than the Nyquist frequency. However, after the performance of the amplitude and angle recovery algorithm and the image reconstruction algorithm are improved, the sampling density can be lower than the Nyquist frequency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 :The working principle diagram of the chessboard constellation optical pupil interferometry high-resolution computational imaging method

[0018] Figure 2 :Chessboard imager working principle diagram

[0019] Figure 3 : A constellation of single-aperture cameras and checkerboard imagers with varying minimum baseline lengths.

[0020] Figure 4 : Concept diagram of an equivalent ultra-large aperture optical telescope obtained by launching in batches and reassembling in orbit.

[0021] Figure 5 : Schematic diagram of the relative motion path of two apertures.

[0022] Figure 6 :Georgeostationary orbit Earth imaging simulation effect diagram.

[0023] In the figure, 1-chessboard constellation high-density data acquisition system, 2-constraints, 3-angle recovery and image reconstruction algorithm; among them: Figure (a) is the simulation input original image, Figure (b) is the modulus G distribution of the complex coherence coefficient μ(u, v) corresponding to the baseline length range of 18m in the spatial frequency domain of the original image, Figure (c) is the simulation effect diagram of the imaging of a single aperture camera with a diameter of 3.5m, and Figure (d) is the result of sampling the array at 1 times the Nyquist sampling frequency using a chessboard aperture combined with direct Figure (e) shows the inverted and reconstructed image when the array is sampled at 2 times the Nyquist frequency by a single aperture camera with a diameter of 3.5m, and the data is sampled at 2 times the Nyquist frequency by a checkerboard aperture. Figure (f) shows the inverted and reconstructed image when the array is sampled at 3 times the Nyquist frequency by a single aperture camera with a diameter of 3.5m, and the data is sampled at 3 times the Nyquist frequency by a single aperture camera with a diameter of 3.5m.

[0024] 101-Aperture pair checkerboard array with different baseline lengths, 102-Waveguide grating spectrometer array, 103-Single-mode fiber paired transmission array, 104-Phase retarder array, 105-Photoelectric detection and readout system;

[0025] 106-Minimum baseline length B min_1 =D chessboard chessboard imager 1,107-minimum baseline length Checkerboard imager 2, 108-minimum baseline length Checkerboard imager 3,109-minimum baseline length Checkerboard imager 4,110 - traditional single aperture camera;

[0026] 111 - a low-frequency camera with a short baseline length, 112 - an aperture-pair-array medium-frequency camera with a medium baseline length, 113 - an aperture-pair-array high-frequency camera with a long baseline length, 114 - an aperture-pair-array high-frequency camera with a long baseline length. DETAILED DESCRIPTION

[0027] Example 1: To achieve an imaging effect of 360m wide and 0.5m resolution for a visible light camera with a central wavelength of 500nm operating in a 36,000km geostationary orbit, the maximum baseline of the space optical telescope needs to be B max A single aperture camera with a diameter of 3.5m is used as a low-frequency camera to perform low-frequency information imaging. Within the range of 3.5m to 18m from the baseline, a chessboard aperture with a unit size of 100mm is used to collect high-frequency information from the array camera. Assume that the chessboard constellation imaging system is located in the geostationary orbit. Figure 6The original image shown in (a) is imaged. The modulus G distribution of the complex coherence coefficient μ(u, v) corresponding to the baseline length range of 18m in the spatial frequency domain is shown as follows: Figure 6 As shown in (b), when only a single-aperture camera with a diameter of 3.5m is used for imaging, the imaging effect is as follows: Figure 6 As shown in (c), when the array is sampled at 1 times the Nyquist sampling frequency using a chessboard aperture, combined with the 1 times Nyquist sampling frequency data of a single aperture camera with a diameter of 3.5m, the inverted and reconstructed image is as follows: Figure 6 As shown in (d), when the chessboard aperture samples the array at 2 times the Nyquist sampling frequency, combined with the 3.5m diameter single aperture camera imaging 2 times the Nyquist sampling frequency data, the inverted reconstructed image is as follows Figure 6 As shown in (e), when sampling is performed at 3 times the Nyquist frequency, combined with the 3.5m diameter single aperture camera imaging 3 times the Nyquist sampling frequency data, the inverted reconstructed image is as follows Figure 6 (f). Figure 6 The simulation results show that the constellation high-resolution imaging system can accurately invert and reconstruct the object-side scene information when adopting 2 times the Nyquist frequency sampling, and the image quality is further improved when adopting 3 times the Nyquist sampling frequency. Based on the simulation results, it is determined that the chessboard constellation high-resolution imaging system adopts a 2 Nyquist frequency sampling configuration, including the above-mentioned 3.5m single-aperture camera and a four-chessboard aperture array pupil interferometer system. The baseline span of the four chessboard aperture array pupil interferometer systems ranges from 3.5m to 18m, and the minimum baselines are 100mm, 125mm, 150mm, and 175mm, respectively, realizing high-density sampling beyond the Nyquist sampling frequency. The chessboard constellation data acquisition and image reconstruction optimization iterative process uses the HIO (continious Hybrid Input Output) algorithm in the Fienup algorithm family for phase recovery and image reconstruction.

[0028] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A chessboard constellation optical pupil interferometry high-resolution computational imaging method, characterized by: A high-density data acquisition system based on a chessboard constellation is used to achieve high-density sampling of the modulus G of the complex coherence coefficient μ(u, v). This is then combined with an image reconstruction algorithm and an amplitude-angle recovery algorithm, also known as a phase recovery algorithm, to obtain a clear image of the object light that meets certain constraints. The chessboard constellation high-density data acquisition system consists of a single-aperture camera and multiple chessboard-type imagers with unequal minimum baselines, realizing high-density sampling of the modulus G of the object-light complex coherence coefficient μ(u, v).

2. The chessboard constellation optical pupil interferometry high-resolution computational imaging method according to claim 1, characterized in that: When the ultra-large aperture optical telescope equivalent to the chessboard constellation high-density data acquisition system far exceeds the rocket envelope, that is, when the maximum baseline length of the chessboard imager in the chessboard constellation high-density data acquisition system exceeds the rocket envelope, the chessboard imager is divided into aperture pair array low-frequency cameras with short baseline length, aperture pair array medium-frequency cameras with medium baseline length, and aperture pair array high-frequency cameras with long baseline length according to the aperture pair baseline length. The aperture pair array cameras with different baseline lengths are placed on different satellite platforms and launched in batches for on-orbit assembly.

3. The chessboard constellation optical pupil interferometry high-resolution computational imaging method according to claim 2, characterized in that: The aperture pair array low-frequency camera with a short baseline length can be replaced by a traditional single-aperture camera to obtain low spatial spectrum information of the target to be measured.

4. The chessboard constellation optical pupil interferometry high-resolution computational imaging method according to claim 1, characterized in that: The chessboard constellation high-density data acquisition system is composed of a pair of relatively moving apertures, that is, the aperture of star A is relatively stationary, and the aperture of star B moves step by step relative to the aperture of star A along a predetermined path. With each step of the aperture of star B, the system completes a sampling of the modulus G of the object-light complex coherence coefficient μ(u, v). When the aperture of star B completes the predetermined trajectory movement and sampling, high-density sampling of the modulus G of the object-light complex coherence coefficient μ(u, v) is completed within the maximum baseline limit.

5. The chessboard constellation optical pupil interferometry high-resolution computational imaging method according to claim 1, characterized in that: The density of the high-density sampling is higher than the Nyquist frequency, but after the performance of the argument recovery algorithm and the image reconstruction algorithm are improved, the sampling density can be lower than the Nyquist frequency.

Images