Super resolution using a simple and practical sequence alignment

This paper presents a new method which aligns two video sequence images both in space and in time for super resolution. The two sequences are recorded by different cameras which are moved jointly and uncalibrated with unknown but fixed internal and external parameters. The advantage of this method is that it is very simple and practical because of a simple transformation between images within each sequence and much more accurate motion estimation in space domain. This approach solves the issue of severe aliasing using super resolution to images recorded at different times. Experiment has been done with video records that are generated from the same one to examine this approach, great results are obtained and shown in this paper later.