Cascaded mode decision for MPEG-2 to H.264/AVC intra frame transcoding

Yinyi Lin, Jyun De Wu

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

The newest international video-coding standard H.264/AVC significantly achieves better coding performance compared to prior video coding standards such as MPEG-2, which has been widely used in today's digital video applications. To provide the interoperability between MPEG-2 and H.264/AVC, in this paper we propose an efficient cascaded mode decision for MPEG-2 to H.264/AVC intra frame transcoding, which is accomplished in two stages. In the first stage, the lowfrequency AC components of discrete cosine transform (DCT) block of a MB is used to select I4MB or I16MB mode prediction; while in the second stage sum of absolute Hadamard transform differences (SATD) coefficients, including SATD value and its variance, is used to skip improper modes for rate distortion optimization (RDO) mode decision. The experimental results, conducted on JM12.2 with fast C8MB mode decision, reveal that over 49% of computation time on average can be saved, while preserving good coding performance when compared with complex cascaded pixel domain transcoding (CCPDT); or 85% (a speed up factor of 7) when compared with CCPDT without considering fast C8MB. The coding performance is also compared with two other distinct algorithms [3], [4], and the results demonstrate that the proposed algorithm momentously outperforms these two algorithms in all three respects (PSNR, bit rate as well as computation time1

Original languageEnglish
Article number5373784
Pages (from-to)2172-2177
Number of pages6
JournalIEEE Transactions on Consumer Electronics
Volume55
Issue number4
DOIs
StatePublished - Nov 2009

Keywords

  • ρ-domain model
  • H.264/AVC
  • Intra mode decision
  • Rate distortion optimization (RDO)
  • Rate-distortion cost function

Fingerprint

Dive into the research topics of 'Cascaded mode decision for MPEG-2 to H.264/AVC intra frame transcoding'. Together they form a unique fingerprint.

Cite this