• DocumentCode
    2698582
  • Title

    Multi-octagon-grid search algorithm for fast motion estimation

  • Author

    Duanmu, C.J. ; Zhang, Yu ; Chen, Xing ; Zhou, Shuihong

  • Author_Institution
    Dept. of Inf. Eng., Zhejiang Normal Univ., Jinhua
  • fYear
    2008
  • fDate
    20-23 June 2008
  • Firstpage
    284
  • Lastpage
    289
  • Abstract
    The UMHexagonS algorithm has been accepted for the possible implementation of the block motion estimation process in the H.264 video coding standard due to its high accuracy and low computational requirement. In this paper, a new algorithm is proposed, which improves on the UMHexagonS algorithm in three aspects. Firstly, a new prediction vector is added for avoiding being trapped in a local minimum. Secondly, the 5times5 full-search is replaced with an octagon search and possibly two small diamond searches to reduce the computational requirement. Thirdly, the multi-hexagon-grid search is changed to a multi-octagon-grid search to both reduce the computational complexity and better catch the direction of the optimal vector. Simulation results demonstrate that the proposed algorithm can reduce five to ten percent of the computational complexity of the UMHexagonS algorithm without loss of its accuracy.
  • Keywords
    computational complexity; motion estimation; video coding; H.264 video coding standard; UMHexagonS algorithm; block motion estimation process; computational complexity; multioctagon-grid search algorithm; prediction vector; Automation; Computational complexity; Discrete cosine transforms; Discrete transforms; Discrete wavelet transforms; Motion compensation; Motion estimation; Video coding; Video compression; Video sequences; Block motion estimation; H.264; Multi-octagon-grid search algorithm; UMHexagonS algorithm;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Information and Automation, 2008. ICIA 2008. International Conference on
  • Conference_Location
    Changsha
  • Print_ISBN
    978-1-4244-2183-1
  • Electronic_ISBN
    978-1-4244-2184-8
  • Type

    conf

  • DOI
    10.1109/ICINFA.2008.4608012
  • Filename
    4608012