A novel restoration algorithm of the turbulence degraded images based on maximum likelihood estimation

The point spread function (PSF) of the turbulence degrading system, as well as the variance of the observation noise and the model of the original image, are unknown a priori in practical imaging processes when the aero-optic effect exists. Both of the PSF and the variance of the observation noise have to be identified from the turbulence degraded images before restoring them. An approach of identification and restoration for the turbulence degraded images based on the parameter estimation of the autoregressive moving average (ARMA) model are proposed. The turbulence degraded image is expressed as an autoregressive moving average process. In the algorithm proposed in this paper, the maximum likelihood (ML) approach is used to the identification of the ARMA parameters. The expectation maximization (EM) algorithm is employed to optimize the nonlinear likelihood function in an efficient way. This identification and restoration algorithm is used to the wind tunnel experimenting images, and the experimental results show that the restoration effect is improved obviously. The estimated version of the original image, and the PSF of the degrade process are simultaneously obtained from the degraded image obtained via the wind tunnel experiment in the process of the ARMA parameters being identified.