A new reconstruction algorithm is proposed to resolve the problem of deconvolution of the point spread function during the imaging process. The algorithm is based on the decomposition of Gaussian function model, which can solve the problems from delta function model and can reconstruct extended sources and point-like sources well. The algorithm is tested with a simulated observation with VLA and compared with another 2 algorithms. The results indicate that the improved algorithm can effectively balance computational complexity and performances, make more accurate modeling of the image, and shows superior image reconstruction performances.

%U https://zkxb.jsu.edu.cn/EN/10.13438/j.cnki.jdzk.2020.04.008