The acquisition of point spread function varying with space, the calculation model of deconvolution, the morbidness of inverse problem, and the renatured algorithm are solved using point interpolation method, cyclic matrix model, Laplace regularization and conjugate gradient iteration. Upon which, the image restoration algorithm with varying space is established and the basic model of image restoration is analyzed. At last, the image restoration algorithms with and without varying space are compared with simulation, the result shows that the image restoration algorithm with varying space is better than spatial invariant algorithm.