A fast mode decision algorithm from MPEG-2 to H.264 transcoding

Jianhua XIANG*, Junli LI, Yang LOU, Gangyi JIANG

*Corresponding author for this work

Research output: Book Chapters | Papers in Conference ProceedingsConference paper (refereed)Researchpeer-review

Abstract

In order to reduce the complexity of MPEG-2 to H.264 transcoding, an efficient fast mode decision algorithm is proposed. We apply machine learning principles by building a decision tree (the relationship between the information gathered during the MPEG-2 decoding stage and the H.264 coding modes of MBs) enabling the development of very low complexity transcoding mechanism. The decision tree is used to determine the coding modes of the P-frames MBs of the output H.264 encoded video sequences. Experimental results show that the proposed algorithm can dramatically reduce the transcoding time, with high quality and high coding efficiency of image.

Original languageEnglish
Title of host publicationAdvanced Research on Industry, Information System and Material Engineering, IISME2011
EditorsHelen ZHANG, Gang SHEN, David JIN
Place of PublicationSwitzerland
PublisherTrans Tech Publications Ltd
Pages985-988
Number of pages4
ISBN (Electronic)9783038135760
ISBN (Print)9783037850275
DOIs
Publication statusPublished - 2011
Externally publishedYes
Event2011 International Conference on Industry, Information System and Material Engineering - Guangzhou, China
Duration: 16 Apr 201117 Apr 2011

Publication series

NameAdvanced Materials Research
PublisherTrans Tech Publications Ltd.
NumberPart 1
Volume204-210
ISSN (Print)1022-6680

Conference

Conference2011 International Conference on Industry, Information System and Material Engineering
Abbreviated titleIISME2011
Country/TerritoryChina
CityGuangzhou
Period16/04/1117/04/11

Keywords

  • H.264
  • Machine learning
  • Mode decision
  • MPEG-2
  • Transcoding

Fingerprint

Dive into the research topics of 'A fast mode decision algorithm from MPEG-2 to H.264 transcoding'. Together they form a unique fingerprint.

Cite this