基于深度学习的SKA图像反卷积研究
Study on SKA Image Deconvolution Using Deep Learning
DOI: 10.12677/AAM.2022.112068, PDF,    国家自然科学基金支持
作者: 卢 梅, 张 利*, 王 蓓：贵州大学大数据与信息工程学院，贵州 贵阳；李丹宁：贵州省科学院，贵州 贵阳；米立功：黔南民族师范学院物理与电子科学学院，贵州 都匀；刘 祥, 张 明：中国科学院新疆天文台，新疆 乌鲁木齐；贺春林, 潘 伟：西华师范大学计算机学院，四川 南充
关键词: SKA；深度学习；图像反卷积；射电天文；SKA； Deep Learning； Image Deconvolution； Radio Astronomy
摘要: 干涉测量使得观测天文图像的分辨率显著提升，然而其阵型所带来的点扩展函数效应需要图像反卷积技术来消除。尽管传统CLEAN反卷积算法已经广泛应用于射电天文图像的点扩展函数消除，但仍然存在精度不高的问题。为了解决国际大科学工程——平方公里阵(SKA)的图像模糊问题，本文提出一种深度卷积神经网络来提升射电天文图像重建的质量。实验显示，相较于通用的方法，本文提出的方法能更好地重建弱源，并在整体图像质量上有明显提升。
Abstract: Radio interferometry makes the observing resolution of astronomical images significantly improved, but deconvolution is required to eliminate the effects of the point spread function (PSF). Although the traditional CLEAN-based deconvolution has been widely used to eliminate the observed PSF, it still has the problem of low accurate reconstruction. To solve the PSF problem of the square kilometer array (SKA), a deep convolutional neural network is proposed to improve the quality of radio image reconstruction. Experiments show that compared with the traditional method—CLEAN, the method proposed in this paper can better reconstruct weak sources and significantly improve the quality of an image.
文章引用：卢梅, 张利, 李丹宁, 米立功, 刘祥, 张明, 贺春林, 潘伟, 王蓓. 基于深度学习的SKA图像反卷积研究[J]. 应用数学进展, 2022, 11(2): 613-620. https://doi.org/10.12677/AAM.2022.112068

参考文献

[1] Pawsey, J.L., Payne-Scott, R. and McCready, L.L. (1946) Radio-Frequency Energy from the Sun. Nature, 157, 158-159. [Google Scholar] [CrossRef] [PubMed]
[2] Ryle, M. and Vonberg, D.D. (1948) An Investigation of Radio-Frequency Radiation from the Sun. Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences, 193, 98-120. [Google Scholar] [CrossRef] [PubMed]
[3] Thompson, A.R., Moran, J.M., Swenson, G.W. (2017) Interferometry and Synthesis in Radio Astronomy. 3rd Edition, Springer, Cham. [Google Scholar] [CrossRef] 
[4] Wu, X.P. (2019) China SKA Science Report. Science Press, Beijing.
[5] Cornwell, T.J., Golap, K. and Bhatnagar, S. (2005) Wide Field Imaging Problems in Radio Astronomy. Prodeeding of IEEE International Conference on Acoustics, Speech, and Signal Processing, Vol. 5, Philadelphia, 23 March 2005, 861-864. [Google Scholar] [CrossRef] 
[6] An, T. (2019) Science Opportunities and Challenges Associated with SKA Big Data. Science China Physics. Mechanics & Astronomy, 62, Article No. 989531. [Google Scholar] [CrossRef] 
[7] Zhang, L., Xu, L. and Zhang, M. (2020) Parameterized CLEAN Deconvolution in Radio Synthesis Imaging. Publications of the Astronomical Society of the Pacific, 132, Article ID: 041001. [Google Scholar] [CrossRef] 
[8] 张利, 肖一凡, 米立功, 卢梅, 赵庆超, 王蓓, 刘祥, 张明, 谢泉. 基于L_BFGS_B局部极小化的自适应尺度CLEAN算法[J]. 贵州大学学报(自然科学版), 2021, 38(1): 38-44.
[9] Högbom, J.A. (1974) Aperture Synthesis with a Non-Regular Distribution of Interferometer Baselines. Astronomy and Astrophysics Supplement, 15, 417-426.
[10] Cornwell, T.J. (2009) Hogbom’s CLEAN Algorithm. Impact on Astronomy and Beyond. Astronomy and Astrophysics, 500, 65-66. [Google Scholar] [CrossRef] 
[11] Cornwell, T.J. (2008) Multiscale CLEAN Deconvolution of Radio Synthesis Images. IEEE Journal of Selected Topics in Signal Processing, 2, 793-801. [Google Scholar] [CrossRef] 
[12] Bhatnagar, S. and Cornwell, T.J. (2004) Scale Sensitive Deconvolution of Interferometric Images. I. Adaptive Scale Pixel (ASP) Decomposition. Astronomy and Astrophysics, 426, 747-754. [Google Scholar] [CrossRef] 
[13] Zhang, L., Zhang, M. and Liu, X. (2016) The Adaptive-Loop-Gain Adaptive-Scale CLEAN Deconvolution of Radio Interferometric Images. Astrophysics and Space Science, 361, Article No. 153. [Google Scholar] [CrossRef] 
[14] Zhang, K., Zuo, W.M., Chen, Y.J., Meng, D.Y. and Zhang, L. (2017) Beyond Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising. IEEE Transactions on Image Processing, 26, 3142-3155. [Google Scholar] [CrossRef] 
[15] Ioffe, S. and Szegedy, C. (2015) Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift. Proceedings of Machine Learning Research, 37, 448-456.

为你推荐