Abstract
In ground-based astronomy, images of objects in outer space are captured using ground-based telescopes. However, the imaging system is generally interfered with by atmospheric turbulence and hence images are blurred. Atmospheric turbulence results in distortions in the wavefront, further leading to perturbed propagation and ultimately image degradation. This degradation is typically characterized by the point spread function, which is crucial for obtaining sharp, clear images. Unfortunately, no direct information on the point spread function is available.To address this issue and correct blurred images, the Adaptive Optics system is a commonly introduced technique in telescopes. Adaptive Optics serves to correct the distortions in the observed image caused by atmospheric turbulence, thereby enhancing the clarity and resolution of celestial observations. Adaptive Optics systems can directly detect the wavefront gradient and further estimate the wavefront phase. Finally, the distorted wavefront becomes flat and improves the image. The estimated wavefront information can also be translated to the point spread function by the Fourier optics model, which ultimately reconstructs the image. This step has been well investigated, and translating from the wavefront phase to the point spread function is also well modeled. Therefore, in this thesis, we focus on the most critical variables in the above reconstruction process, the wavefront phase.
We start from the background related to ground-based astronomy and explain the macroscopic model as well as related data characteristics and data acquisition in Chapter 1. In Chapter 2, we proposed the process of modeling multiple wavefront sensors model under the assumption of a single layer and detection condition with multiple time steps. The relationship between the detected wavefront gradient and the reconstructed target wavefront was constructed. Specifically, turbulence statistics, the Taylor freezing hypothesis, and the wind speed concept were incorporated into the modeling. In Chapter 3, we propose the core model throughout this thesis, the H2L2 model. We explain the motivation and rationale of the model. Explicitly, we conduct a series of experiments to verify the superior performance of the H2L2 model and validate the advantages of using multi-WFS modeling. In Chapter 4, we extend the single-layer model to the multiple-layer scene. Accordingly, we propose a tomography H2L2 model for this case and give numerical analysis. In Chapter 5, we focus on the single moment scene of the H2L2 model and propose a target object reference model for more realistic modeling needs. Finally, a series of experiments are conducted to demonstrate the effectiveness of our algorithm.
Especially for the experiments, our research emphasizes the comprehensiveness of real-world applications. Two primary operational modes in Adaptive Optics systems are considered: open-loop for predetermined corrections and closed-loop for real-time corrections. The experiments in Chapters 3 and 4 are related to open-loop, while in Chapter 5 are conducted in both. For algorithm evaluation, we consider L2 relative error, H1 relative error, and phase residual error. In our experiments, numerical results are presented to illustrate that our model can provide better phase reconstruction in all mentioned scenarios.
In the ongoing and future directions and perspective, we plan to continue our ongoing project related to adaptive wind model and apply our approaches to an advanced wavefront sensor, the Pyramid Wavefront Sensor, which is known for high-precision and high-sensitivity wavefront measurements capable of detecting subtle distortions for real-time correction. Besides, a more efficient and accurate optical flow estimation would be considered to replace the traditional local all-pass filter for blind reconstruction.
| Date of Award | 28 Aug 2024 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Jean-Michel Henri Olivier MOREL (Supervisor) & Hon Fu, Raymond CHAN (External Co-Supervisor) |