In this thesis, two methods for the Maximum A Posteriori (MAP) based super-resolution are proposed. All the degradation parameters, namely, additive noise, blur and sub-pixel motion are considered unknown. The study focuses on the simultaneous estimation of the unknown parameters and the underlying high-resolution image. Two types of image priors have been considered, the Gaussian Simultaneous Autoregressive (SAR) and the Huber Markov Random Field (HMRF), and the results have been compared. Special focus is laid on the estimation of the PSF blurring and two methods have been proposed for the modeling of PSF blur and its estimation. Mathematical derivations and analytical proofs support the algorithms. Conclusions are drawn on the basis of the performance evaluation of the proposed algorithms with each other and with the existing techniques. It is shown that the two methods proposed achieve stable and desired solution for the super-resolution problem. |