06

2024

-

07

Tsinghua University has made new progress in the direction of computational imaging.

Author:


At the beginning of the 17th century, humans began to point their observation instruments to the distant universe, hoping to capture photons that traveled through thousands of years and receive messages from distant galaxies. However, atmospheric turbulence is like a transparent ghost floating in the sky, interfering with the progress of photons and obscuring the secrets of the early universe. In 1964, the American physicist Richard Feynman (Richard Feynman) pointed out that "turbulence is one of the most important unsolved problems in classical physics". Atmospheric turbulence, a highly chaotic system, is one of the most difficult to be captured in turbulence, and its motion mode is highly random, which is difficult to accurately model, detect and predict.

Fang Lu's team from the Department of Electronic Engineering of Tsinghua University has carried out cross-cooperation with Academician Dai Qionghai and Associate Professor Wu Jiamin of the Department of Automation. They have proposed a new principle of calculating optical field, established a digital adaptive optics model, developed a wide-area wavefront calculation sensor chip (Wide-field WavefrontSensor, WISE), and realized real-time detection and prediction of atmospheric turbulence in the range of over 1100 angular seconds (diagonal). The imaging technology has the advantages of large field of view, high resolution and strong robustness, and the sensing range is nearly a thousand times higher than the widely used Shack-Hartmann wavefront sensor. The detection field of view of the WISE chip is equivalent to the sum of hundreds of wavefront sensors, which can be widely used in existing optical systems, enabling wide-area detection and prediction of atmospheric turbulence, correcting atmospheric turbulence disturbances, and realizing efficient acquisition and accurate reconstruction of large-scale optical signals.

WISE helps high-precision turbulence prediction In the one-way propagation of optical signals, accurate turbulence detection is sufficient to eliminate errors, but in the two-way interaction, the rapid evolution of turbulence brings new challenges. A typical interaction process, such as space optical communication, consists of a downlink detection link and an uplink compensation link, and due to the time difference between the two links, compensation cannot be performed directly according to the detection result, but compensation needs to be performed by predicting the turbulence distribution at a future moment, I .e., pre-compensation. At this time, the accuracy of turbulence prediction is very important. The video (above) shows the dynamic evolution of the turbulence distribution, from spatially uniform turbulence in a small field of view to spatially non-uniform turbulence in a wide area. When we only observe the turbulence in a small field of view, it is difficult to find its temporal evolution, which is the difficulty of turbulence prediction based on traditional adaptive optics technology. When the field of view expands, the evolution of turbulence becomes traceable. As stated in the Taylor frozen flow hypothesis, the overall flow of atmospheric turbulence can be clearly observed in a wide range of observational data, which will provide strong support for the accurate prediction of turbulence. Based on the WISE chip and the time-space neural network model, the team achieved high-precision turbulence prediction (under video) in a large field of view, with the predicted wavefront error reduced from 224nm to 109nm (the index used is RMSE), a significant improvement over traditional adaptive optics. The WISE chip explores a new path for the study of the spatiotemporal dynamic evolution of atmospheric turbulence.

From scanning light field element imaging to WISE chips, photon ghosts are no longer mysterious, and the telescope's field of view can penetrate the atmosphere. The team continues to innovate in the field of computational imaging to computationally empower astronomy and open a new chapter in computational astronomical imaging. When the field of view is infinite, the field of view will also be boundless. In the future, the team will further give full play to the advantages of meta-imaging wide-area wavefront sensing to help the new generation of wide-field high-resolution ground-based optical sky survey, Lingyun far-sighted and far-sighted.

The related research results, entitled "Direct Observation of Atmospheric Turbulence with a Video-rate Wide-field Wavefront Sensor" (based on wide-area wavefront sensor), were published in Nature Photonics on July 1.

The Department of Electronic Engineering and the National Research Center of Tsinghua University are the first units of the paper. Fang Lu, Dai Qionghai and Wu Jiamin are the corresponding authors of the paper. Guo Yuduo, a doctoral student in the 2021-level electronics department, Hao Yuhan, an undergraduate student in the 2021-level automation department, and Wan Sen, an assistant researcher in the automation department, are the first authors of the paper. Postdoctoral Zhang Hao and assistant researcher Zhu Laiyu participate in the work. The research is supported by the 2030 Major Project of the Ministry of Science and Technology, the Basic Science Center Project, and the Beijing National Research Center for Information Science and Technology.

Chongqing Chongfan Technology Co Ltd