The integrating effect of the Point Spread Function （PSF） of a satellite remote sensing system significantly blurred the acquired image and reduced the spatial resolution of the data. This paper firstly analyzed the main factors which result in the image blur. We then discussed the derivative of Linear Spread Function （LSF） by the use of linear step structures that are recorded in the image. The PSF of an image can be approximated by interpolating two linearly separable orthogonal LSFs. Deconvolution filters are derived specifically for each image from its PSF using a Weiner filter in the frequency domain and applying an invert Fourier transform to convert into space domain. The use of Wiener filter could guarantee the achievement of the cleared image while reducing the noise influence. Finally

the deconvolution filters were applied in the 3rd channel of a full scene CBERS 1 image so that it can restore the original signal.A significant improvement of the image quality was achieved.