As the most popular simplified model of the optical imaging system, the acquisition of the Gaussian point spread function (PSF) parameter is one of the hotspots and key points on which people do research in the field of image restoration. Based on the idea by which there exists deterministic mathematical relationship between Gaussian OTF feature points as well as its parameter and the frequency representation of the image in an existed literature, we proposed an automatic, accurate, stable, and improved approach. This method is able to give prominence to the related calculation feature by a Gaussian convolution and degeneration operation and finally realize the automatic estimation of PSF parameter of a microscopic image. Experiments have proved that a good restoration result can be achieved utilizing the estimated PSF by the present method, which is of considerable application and reference value in restoration of other sorts with Gaussian approximate PSF model or 3D microscopic image restoration .