An improved scaled gradient projection (ISGP) for astronomical images restruction

With regard to ground-based astronomy observations, due to the interference of atmospheric turbulence, noise and other factors, the observed images are degraded, which makes it difficult to obtain high-resolution astronomical images. In such case, astronomical image restoration is essential. Of all the image deconvolution algorithms, the one based on scaled gradient projection (SGP) is highly effective while equipping a high precision and a fast convergence. But the SGP algorithm requires an accurate point spread function (PSF) estimation as a prerequisite, which limits the application flexibility of SGP. In order to solve this problem, in this paper, influences of different PSFs on SGP are analyzed and an improved SGP algorithm combined with PSF estimation is proposed. Furthermore, so as to analyze the performance, the developed algorithm is placed in the system which includes all approaches of image processing. Simulation results show that the PSF estimation and the improved algorithm gain brilliant performance on astronomical image restoration.