Classical image restoration algorithm is based on point spread function (PSF) is known. However, it is difficult to determine PSF in many cases, we have to estimate the true image and PSF in the case of only knowing some of the information or no information of imaging systems, this process is called blind image restoration. For the Gaussian blurred image, it is very difficult to detect PSF, so blind restoration of the Gaussian blurred image has been a troublesome issue. In this paper, we suppose the initial PSF using the characteristics of Gaussian PSF, and do Wiener filtering based on Gaussian blurred image with noise, then gain the restoration image through the median filtering. Experiment shows that, restored image is better in subjective visual effect, and remarkable in recovery effect with good noise immunity. This method has some reference value to improve the image quality.