• DocumentCode
    2999619
  • Title

    Improved motion estimation algorithm based on ME-Skip

  • Author

    Liu, Zhifeng ; Jiang, Jinfeng ; Yang, Wentong ; Zhang, Aiping ; Wang, Jianhua

  • Author_Institution
    Machinery & Electron. Applic. Technol. Inst., Beijing Univ. of Technol., Beijing, China
  • fYear
    2009
  • fDate
    26-29 Nov. 2009
  • Firstpage
    2061
  • Lastpage
    2064
  • Abstract
    Based on ME-Skip algorithm proposed by H.S.Kang, an advanced motion estimation algorithm is presented to improve the real-time compression capability of H.264. The algorithm utilizes Zero-Block Detection theory to improve the skip macroblocks calculation of ME-Skip algorithm, which reduces the computing complexity. In the algorithm, non-skip macroblocks are divided into three motion types: slow motion, moderate motion and fast motion. Different motion type will be processed by different corresponding motion estimation algorithm. Experimental result shows that our advanced algorithm reduces the complexity of motion estimation, particularly the compression efficiency is higher in the slow motion of video sequence.
  • Keywords
    computational complexity; motion estimation; video coding; H.264; ME-Skip algorithm; computing complexity; fast motion video; moderate motion video; motion estimation algorithm; real-time compression capability; skip macroblocks calculation; slow motion video sequence; zero-block detection theory; Computational complexity; Encoding; Machinery; Motion detection; Motion estimation; Redundancy; Video coding; Video compression; Video sequences; Video surveillance; H.264; ME-Skip; MVFAST; Real-time; Zero-Block Detection;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Computer-Aided Industrial Design & Conceptual Design, 2009. CAID & CD 2009. IEEE 10th International Conference on
  • Conference_Location
    Wenzhou
  • Print_ISBN
    978-1-4244-5266-8
  • Electronic_ISBN
    978-1-4244-5268-2
  • Type

    conf

  • DOI
    10.1109/CAIDCD.2009.5375210
  • Filename
    5375210