In order to solve the current local blur degree, resulting in poor effect of image restoration, this paper recovered from image feature point detection and filtering angle, proposed image restoration algorithm SIFT feature point detection and based on Wiener filter. First of all, according to the characteristics of the scale space extrema, key point positioning and direction of distribution, the design of feature descriptor, fuzzy image feature point distribution, to establish the coordinates of the center of the purpose of disc restoration model. Then, based on the characteristics of the point spread disk function, coupled with Fourier transform and least squares filtering, an unconstrained Wiener filter operator is designed to achieve the purpose of image restoration. Finally, according to the characteristics of the corner location to guide the restoration of the filter disk function to calculate the starting position, based on software engineering to achieve the restoration algorithm. The experimental results show that compared with the current restoration algorithm, the restoration technique has higher accuracy and stability.