Low Complexity Motion Refinement for DCT-Domain Video Transcoder

In this paper, we address in the topic of spatial-downscaling video transcoder in DCT domain. The proposed techniques include the hierarchical fast motion resampling (HFMR) with accuracy motion resampling, and the fast refinement for non-integral-MV (FRNI) with low complexity motion vector refinement. For FRNI, it can be applied not only in our proposed algorithm but also combined with other video transcoders which have MC-DCT (motion compensation in DCT domain) and reduce overall the computational complexity. Furthermore, in combination of FRNI and HFMR, it will get the best performance compared with the conventional approaches. Experimental results show that our proposed algorithms can improve the entire quality and also reduce the complexity in MC-DCT