Blind Super-Resolution Image Reconstruction Based on POCS Model
Wei Chau Xie, Feiyan Zhang, Hao Chen, Qianqing Qin · 2009
Due to the high cost and physical limitations of the high precision image sensors, it is not easy to directly obtain the desired high resolution (HR) images from sensors in many cases. A new iterative super-resolution (SR) algorithm based on projection onto convex sets (POCS) model is proposed. In this method, a new algorithm is adopted to estimate subpixel shift of the low-resolution (LR) images and do motion compensation. The main stage of the algorithm employs the iterative method of POCS model, utilizing some new constraint operators. The experiment shows the method is robust and effective. The disadvantage is that the convergence of the algorithm is not very fast.