A Very High Rate Super Resolution Algorithm Unified on A Leclerc Stochastic Estimation SR and A Single Image Super Resolve Algorithm

Abstract

Super-resolution (SR) algorithm, which intends to compute a finer resolution from a single inferior image or many inferior images, plays a vital aspect in several leading-edge digital image processing in both theory and implementation perspectival. To provide a very high ratio rate of spatial enhanced image, we proposed a super resolution algorithm, which is unified on a Leclerc stochastic estimation super resolution algorithm and a single image super resolve algorithm. At the first process, a Leclerc stochastic estimation super resolution algorithm is used to apply on a pact of inferior images, which are usually a low spatial resolution (so called LR) and are corrupted by noise to provide a higher spatial 4x image with finer quality. At the second process, a single image super resolve algorithm, which is unified on a pre-resolving of upper frequency and a outlier control function is used to provide a higher spatial 16x image. In the analytic testing part, the tested results are demonstrated that the proposed super resolution algorithm can effectively create a finer and higher spatial 16x image from a pact of inferior images under distinct noise forms.