On the architecture of H.264 to H.264 homogeneous transcoding platform

Abstract

For homogeneous transcoding, we usually transfer a compressed video into a lower bit-rate and/or video with a smaller size. In this paper we introduce the general architecture of a downing sizing transcoder and propose some novel techniques for its practical realization, including motion vector re-estimation, sub-pixel motion re-estimation, mode re-decision, etc. We then generalize the idea of transcoding to video enlargement and propose a simple framework for its re-encoding. The paper ends with some useful remarks on the formation of super-resolution videos via the transcoder frame work.