Search pattern based on multi-direction motion vector prediction algorithm in H.264 and learning website

Tsung Han Tsai, Wei Chun Hsu, Yuan Chen Liu, Wan Chun Lee

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Motion estimation is the most important part that needs about 80% of computation in the procedure of image compression coding. So there are many fast algorithms for motion vector search proposed to reduce time consumed and keep good qualities. In this paper some traditional algorithms and algorithms using prediction values are investigated, including DS (diamond search), HEXBS (hexagonbased search) and UMHEXagonS. The mechanism of search pattern based on multi-direction motion vector prediction algorithm is proposed to develop a fast algorithm based on motion estimation and combining search patterns and directions. At the moment, we just speed up the encoding time and perform better high PSNR values in the proposed algorithm. But that is very similar with MDS or SPS, and the factors of software and hardware like the selection of algorithms or frame number and the phase of early termination or the difference of hardware are not discussed.

Original languageEnglish
Title of host publicationElectrical Engineering and Control - Selected Papers from the 2011 International Conference on Electric and Electronics, EEIC 2011
Pages217-224
Number of pages8
EditionVOL. 2
DOIs
StatePublished - 2011
Event2011 International Conference on Electric and Electronics, EEIC 2011 - Nanchang, China
Duration: 20 Jun 201122 Jun 2011

Publication series

NameLecture Notes in Electrical Engineering
NumberVOL. 2
Volume98 LNEE
ISSN (Print)1876-1100
ISSN (Electronic)1876-1119

Conference

Conference2011 International Conference on Electric and Electronics, EEIC 2011
Country/TerritoryChina
CityNanchang
Period20/06/1122/06/11

Keywords

  • motion estimation
  • motion vector
  • multi-direction
  • search pattern

Fingerprint

Dive into the research topics of 'Search pattern based on multi-direction motion vector prediction algorithm in H.264 and learning website'. Together they form a unique fingerprint.

Cite this