A new publication from Opto-Electronic Advances; DOI 10.29026/oea.2022.200082 reviews intelligent adaptive optics.
Adaptive optics technology can correct the dynamic aberrations of optical system at the processing speed of thousands of times per second in order to restore the distorted wavefront to plane wave. It has important applications in astronomical high-resolution imaging, laser transmission, biomedical imaging and other fields. Adaptive optics system mainly consists of three parts: wavefront sensor (like human’s eyes), wavefront controller (like human’s brain) and wavefront corrector (like human’s hands). The wavefront sensor is used to measure the optical aberration induced by atmospheric turbulence and the wavefront controller calculates the command sent to the wavefront corrector such as a deformable mirror to compensate the aberration in real time. Although adaptive optics has been developed more than 50 years and successfully driven the astronomical observation et al, scientists have more expectations for adaptive optics technology. They want the adaptive optics system to become smarter and more powerful.
At present, the research frontier of adaptive optics technology includes high-speed phase retrieval, wavefront prediction and end-to-end image post-processing et al. Traditional phase retrieval and image post-processing mainly rely on optimization algorithms requiring many iterations before convergence. Besides, these iterative algorithms tend to fall into local optimums. Therefore, the iterative-algorithm-based phase retrieval or image post processing have great defects in both speed and effect. The traditional wavefront prediction mainly relies on the accurate estimation of the linear model of atmospheric turbulence. When the estimation model does not match the fast-changing actual model, the prediction effect will deteriorate rapidly.
In recent years, with the rapid development of big data, machine learning algorithms and computing power, people have begun to use artificial intelligence (AI) technology to enable the transformation and upgrading of all walks of life. So, when AI and adaptive optics meet, what kind of spark can they produce? Is AI helpful to solve the above hot issues, and how can it promote the development of adaptive optics?
The authors of this article review the development of adaptive optics based on machine learning. This group has been engaged in the research of high-resolution adaptive optics imaging technology especially solar adaptive optics for many years. They have developed a series of solar adaptive optics systems including traditional single-conjugate adaptive optics, ground-layer adaptive optics and multi-conjugate adaptive optics systems. Besides, they have built one of the largest solar telescopes, the Chinese Large Solar Telescope. Recently, they began to focus on adaptive optics based on machine learning. It can be seen from publications that the Intelligent adaptive optics is developing rapidly and a review about this topic is essential and important for the academic community. The group has carried out research and discussion on related progress, focusing on the current key research direction and practical application difficulties of intelligent adaptive optics technology.
Although revolutionary achievements have not yet appeared, some encouraging results have been done by the scholars. The performance of main components in adaptive optics system is expected to be upgraded by AI. The team believes that in the aspect of wavefront detection, the data-driven deep learning technology is expected to make the phase retrieval get rid of the dependence on iterative algorithms, reduce the calculation time to sub millisecond level, and increase the speed by more than one order of magnitude compared with the existing level, to meet the speed requirements of dynamic atmospheric turbulence wavefront detection. In the aspect of wavefront prediction, the trained deep learning prediction model can accurately predict the wavefront distortion in non-stationary atmospheric turbulence with strong robustness. In image post-processing, deep learning is expected to achieve end-to-end real-time processing, improve the automation and intelligent level of image post-processing, and enable astronomers to obtain high-resolution astronomical images in time. However, the above research is still in their infancy, mainly focusing on the theoretical and laboratory level, and the generalization in the actual complex environment needs to be verified in practice. This is also an issue that the academic community of adaptive optics needs to focus on in the future.
.
Article reference:
Guo YM, Zhong LB, Min L, Wang JY, Wu Y et al. Adaptive optics based on machine learning: a review. Opto-Electron Adv 5, 200082 (2022). doi: 10.29026/oea.2022.200082
Keywords: adaptive optics / machine learning / deep learning
# # # # # #
Professor Rao is deputy director of IOE in CAS and deputy director of the Key Laboratory of adaptive optics, CAS. He is mainly engaged in large-aperture high-resolution optical imaging telescope and adaptive optics technology research. He has leaded over many national, provincial and ministerial level projects, including the major instrument program of NSFC and the key project of NSFC. In recent years, he has published more than 200 papers in academic journals and international conferences, including more than 100 papers in SCI. Three monographs have been published. He has won 1 first prize and 1 second prize of National Technological Invention Award, 4 first prizes and 2 second prizes of provincial and ministerial level scientific and technological progress award, and 1 outstanding scientific and technological achievement award of CAS. In addition, he also won the China Youth Science and technology award, the national 100 Excellent Doctoral Dissertation Award, the young scientists of the CAS, the top 10 outstanding young people of the CAS, and the outstanding contribution award of Wang Kuancheng, a western scholar of the CAS.
# # # # # #
Opto-Electronic Advances (OEA) is a high-impact, open access, peer reviewed monthly SCI journal with an impact factor of 8.933 (Journal Citation Reports for IF2021). Since its launch in March 2018, OEA has been indexed in SCI, EI, DOAJ, Scopus, CA and ICI databases over the time and expanded its Editorial Board to 36 members from 17 countries and regions (average h-index 49).
The journal is published by The Institute of Optics and Electronics, Chinese Academy of Sciences, aiming at providing a platform for researchers, academicians, professionals, practitioners, and students to impart and share knowledge in the form of high quality empirical and theoretical research papers covering the topics of optics, photonics and optoelectronics.
# # # # # #
More information: http://www.oejournal.org/oea
Editorial Board: http://www.oejournal.org/oea/editorialboard/list
All issues available in the online archive (http://www.oejournal.org/oea/archive).
Submissions to OEA may be made using ScholarOne (https://mc03.manuscriptcentral.com/oea).
ISSN: 2096-4579
CN: 51-1781/TN
Contact Us: oea@ioe.ac.cn
Twitter: @OptoElectronAdv (https://twitter.com/OptoElectronAdv?lang=en)
WeChat: OE_Journal
# # # # # #
Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.
