一种基于压缩感知框架的射电天文图像复原算法

A radio astronomy image restoration algorithm based on compressed sensing framework

射电天文图像去卷积是射电天文学中的一项关键数据处理技术,其主要目标是去除天空图像中由天文观测仪器引入的效应,从而复原原始的天空图像.射电望远镜阵列采用稀疏干涉阵列,成像原理与光学望远镜有所不同.如果UV空间中的采样点不足够密集,将会导致在图像重建时无法获得足够高分辨率的信息,传统的射电天文图像重建算法未能根本解决UV空间欠采样的问题.本文基于压缩感知理论框架,并结合射电天文图像稀疏性的先验知识,研究了一种新的射电天文图像去卷积算法,该算法将脏图的去卷积过程转化为一个旨在求解全局最小化的凸优化问题,即基于IUWT-CS的射电干涉图像重建算法.为了评估该算法的重建性能,研究使用了射电天文学仿真软件包OSKAR对SKA1-low射电望远镜阵列进行模拟观测,并对观测得到的点源和扩展射电源进行了去卷积处理.实验结果表明,与HOGBOM-CLEAN,MS-CLEAN和IUWT-FISTA方法相比,IUWT-CS方法显著提高了射电图像的重建质量,实现了更加精细的去噪和复原效果.

Deconvolution of radio astronomy images is a key data processing technique.Its main goal is to remove the effects introduced by the instrument from the observed sky images to recover the original sky images.However,radio interferometer arrays employ sparse interferometric arrays,whose imaging principles differ from those of optical telescopes.If the sampling points in the UV space are not sufficiently dense,this will lead to insufficient high-resolution information during image reconstruction.Traditional radio astronomy image reconstruction algorithms fail to fundamentally solve the problem of UV space undersampling.This paper adopts the compressed sensing theoretical framework,combines prior knowledge of the sparsity of radio astronomy images,and studies a new radio astronomy image deconvolution algorithm,namely the IUWT-CS-based radio interferometric image reconstruction algorithm.This algorithm transforms the dirty image deconvolution process into a convex optimization problem to find the global minimum.To evaluate the reconstruction performance of this algorithm,we used the OSKAR radio astronomy simulation software package to simulate low SKA1 observations and performed deconvolution processing on the extended radio sources obtained.Experimental results show that,compared with the HOGBOM-CLEAN,MS-CLEAN,and IUWT-FISTA methods,the IUWT-CS method significantly improves the reconstruction quality of radio images and achieves finer denoising and restoration effects.