| 其他摘要 | The goal of astronomers has always been to build larger ground-based solar telescopes for in-depth research on solar activity. However, atmospheric turbulence on Earth is the main factor limiting the resolution of ground-based solar telescope imaging systems, and wavefront distortion leads to blurring, degradation, geometric displacement, and other issues in solar speckle images. In order to overcome the influence of the atmosphere and achieve high-resolution imaging of ground-based solar telescopes, adaptive optics technology and high-resolution image reconstruction methods are usually used. The adaptive optics system detects and corrects the wavefront at the pupil of the telescope in real time, but the corrected field of view is limited and there is a corrected phase residual. To truly achieve or approach the diffraction limit imaging of the telescope, high-resolution image reconstruction processing is still required.The high-resolution image reconstruction method can effectively overcome the influence of atmospheric turbulence at the algorithmic level and efficiently achieve high-resolution imaging without an adaptive optical system. The 1m New Vacuum Solar Telescope (NVST), located on the banks of Fuxian Lake in Chengjiang, Yunnan, is one of the pioneers in high-resolution imaging of solar telescopes in China. It has released a massive amount of high-resolution imaging data through image reconstruction methods. When the solar speckle image is severely blurred, degraded, and geometrically displaced due to atmospheric influence, the high-resolution image reconstructed by NVST using speckle masking method often produces artifacts, and some reconstructed images do not meet scientific research requirements, wasting valuable collection data and observation time.The main content of this paper is to discuss the reconstruction algorithm of solar speckle images under the influence of atmospheric turbulence. The reconstruction process of speckle images is divided into two parts: correcting geometric displacement of speckle images based on non-rigid alignment technology and reconstructing target images based on wavefront estimation through deconvolution. Research analysis and algorithm design are conducted for these two parts respectively. And the rationality and feasibility of the algorithm principle were verified through simulation experiments and actual data reconstruction experiments.The overall work content mainly includes three aspects:Firstly, analyze the principles and characteristics of speckle masking technology and deconvolution method in solar image reconstruction methods. By introducing the principles and processes of these two classic solar image reconstruction algorithms, as well as the results of NVST measured data reconstruction, the difficulties in the research of solar speckle images reconstruction are discussed.Secondly, research and design of image reconstruction algorithms that combine non-rigid alignment and wavefront estimation. Research and implementation of non rigid alignment technology for correcting geometric displacement in short exposure images. Through non-rigid alignment, pixel by pixel alignment and registration of short exposure images are completed, expanding the halo area. On this basis, the lucky image frame selection technique was optimized, selecting images with more high-frequency information and stable structure as input images for multi frame blind deconvolution. By using the corrected average frame image as the target image prior and substituting it into the phase iteration solution process of the cost functional, the solution interval of the optimization search is effectively limited, thus avoiding premature local optima in the phase iteration process. The instantaneous transfer function of the atmospheric telescope integrated system was restored using wavefront estimation information, thereby reconstructing the target image. Based on the above steps, a multi frame blind deconvolution solar image reconstruction algorithm combining non rigid alignment technology has been successfully constructed, which we refer to as the MFBD-CNRA algorithm. The simulation dataset reconstruction results indicate that the image quality of MFBD-CNRA reconstruction is also closer to the real image in terms of structural similarity and peak signal-to-noise ratio.Thirdly, analysis and discussion of NVST measured data. Constructing NVST telescope observation datasets under different visual conditions, comparing the image results of various classic post image reconstruction algorithms with the MFBD-CNRA algorithm, and demonstrating the good robustness and effectiveness of the MFBD-CNRA algorithm through quantitative indicators such as power spectrum curve, structural similarity, and intensity distribution curve.The research work and corresponding results of this article can provide a complete experimental technique for high-resolution reconstruction of speckle images of ground-based solar telescopes, and provide reference solutions for further optimization of other image reconstruction methods. The research algorithm is applicable to both phase diversity technology and multi-channel systems, providing a research foundation for the subsequent development of multi-channel reconstruction technology for ground-based telescopes. |
修改评论