J Oint I Mage R Egistration and E Xample - B Ased S Uper - R Esolution a Lgorithm

Supper-resolution (SR) methods are classified into two different methods: image registration (IR)-based methods and example-based methods. The proposed joint SR method is focused on estimating highresolution (HR) video sequences from low-resolution (LR) ones by combining the two different methods. The IR-based SR method collects information from adjacent frames to reconstruct HR images in the video sequence. Example-based SR methods give good textures and strong edges in the result HR video. In this paper, IR-based and example-based SR methods are fused based on the gradient features. The proposed joint SR method gives smaller peak signal to noise ratio than the example-based method, however it shows better reconstruction results on high-level features such as characters in images. Experimental result of the proposed joint SR method shows less noise and higher contrast than the example-based method.