Super-Resolution from Compressed Video

Abstract

The problem of recovering a high-resolution frame from a sequence of low-resolution and compressed images is considered. The presence of the compression system complicates the recovery problem, as the operation reduces the amount of frequency aliasing in the low-resolution frames and introduces a non-linear noise process. Increasing the resolution of the decoded frames can still be addressed in a recovery framework though, but the method must also include knowledge of the underlying compression system. Furthermore, improving the spatial resolution of the decoded sequence is no longer the only goal of the recovery algorithm. Instead, the technique is also required to attenuate compression artifacts.