Skip to Main Content
Although the topic of Super-Resolution Reconstruction (SRR) has recently received considerable attention within the traditional research community, the SRR estimations are based on LI or L2 statistical norm estimation. Therefore, these SRR methods are very sensitive to their assumed data and noise models, which limit their applications. The real noise models that corrupt the measure sequence are unknown; consequently, SRR algorithm using LI or L2 norm may degrade the image sequence rather than enhance it. The robust norm applicable to several noise and data models is desired in SRR algorithms. This paper proposes an alternate SRR approach based on the stochastic regularization technique of Bayesian MAP estimation by minimizing a cost function. The Lorentzian norm is used for measuring the difference between the projected estimate of the high-resolution image and each low resolution image, removing outliers in the data. Tikhonov regularization is used to remove artifacts from the final result and improve the rate of convergence. In order to cope with real sequences and complex motion sequences, the fast affine block-based registration is used in the registration step of SRR. The experimental results show that the proposed reconstruction can be applied on real sequences such as Suzie sequence and confirm the effectiveness of our method and demonstrate its superiority to other super-resolution methods based on LI and L2 norm for several noise models such as AWGN, Poisson and Salt & Pepper noise.