Abstract: In ground-based astronomy, blind deconvolution has been the key issue in?restoration of atmospheric turbulence-degraded images, challenged by lack of accurate information of PSF （Point Spread Function）, and image restorations that are ill-conditioned or ill-posed. Estimation of PSF based on?sparse?representation provides a way to solve the problem. Sensitive to dimension parameter, the sparsity estimation algorism of PSF is improved through predicting PSF dimensions by edge measurement in this paper. Accurate PSF estimation is thus achieved that leads to clearer astronomic image reconstruction. Case study shows that the modified estimation algorithm is applicable to astronomic imaging in improving the accuracy of PSF dimensions and sharp of images reconstructed.

Key words: ground-based astronomy, sparsity measure, Point Spread Function（PSF） size pre-estimation, image restoration

摘要： 地基天文观测中，对大气湍流造成的天文模糊图像进行复原时，由于缺少准确的PSF（Point spread function）信息，加之复原过程的病态性，天文图像的盲反卷积运算一直是难点。针对天文图像的稀疏特性，基于稀疏测度的PSF估计算法可有效地解决PSF信息缺省的问题，该算法受PSF尺寸参数选择的影响较大。通过基于边界测度的PSF尺寸预估计算法对稀疏测度PSF估计算法进行改进。优化算法首先对PSF尺寸有效预估计，得到精确的PSF估计值，最终提高了图像复原的精度。仿真结果表明，优化后的算法应用到天文图像复原过程中后，可以更为准确地估计出模糊图像的PSF大小，改善天文模糊图像的复原效果。

关键词: 地基天文, 稀疏测度, 点扩散函数尺寸估计, 图像复原