Abstract: To cope with the uncertainty of the solution's uniqueness and convergence in the iterative blind deconvolution（IBD） method, a novel wavelet-based iterative blind image restoration algorithm is proposed.A singular value decomposition（SVD） and compression-based filtering method is used for pretreatment, discrete wavelet transform（DWT） is made after the initial image estimate, and local Gaussian model（LGM） is used as a priori restriction added to the IBD method.Simulation results showed that, the combination of these techniques can perform better about 3?dB(P_SNR) than the original IBD method in recovering images, with edge information well reserved.At the same time, the uniqueness and convergence properties of IBD are improved.