An iterative deconvolution algorithm is presented in detail which utilizes regularization to combine maximum-likelihood (ML) estimate of convolution error and several physical constraints to build error function. The physical constraints used in this algorithm include positivity, band-limit information and the information of multiple frames. By minimizing the combined error metric of individual ones, the object can be expected to be recovered from the noisy data. In addition, numerical simulation of Phase Screen distorted by atmospheric turbulence following the Kolmogorov spectrum is also made to generate the PSFs which are

used to simulate the degraded images.