• DocumentCode
    3612330
  • Title

    HEVC fast reference picture selection

  • Author

    Shih-Hsuan Yang ; Kai-Sheng Huang

  • Author_Institution
    Comput. Sci. & Inf. Eng., Nat. Taipei Univ. of Technol., Taipei, Taiwan
  • Volume
    51
  • Issue
    25
  • fYear
    2015
  • Firstpage
    2109
  • Lastpage
    2111
  • Abstract
    The new high efficiency video coding (HEVC) standard facilitates the realisation of ultrahigh-definition video applications. Although the compression efficiency of the HEVC standard is double that of the well-established H.264 standard, HEVC incurs a considerably higher encoding complexity. This Letter presents an algorithm aimed at expediting reference picture selection processes in HEVC encoding. The proposed algorithm is primarily based on a reference picture and block discovered using the advanced motion vector prediction (AMVP) algorithm. Two acceleration conditions are included in the proposed algorithm. The first early-decision condition entails examining the rate-distortion (RD) cost associated with AMVP, and if this cost is lower than a predefined threshold, motion estimation is confined to the AMVP-selected reference picture. Furthermore, for coding units of size 64 × 64 (N = 32) with a prediction-unit (PU) partitioning of 2N × N or N × 2N, the second PU uses the same reference picture as the first PU if the first PU satisfies the first early-decision condition. The experimental results revealed that the proposed algorithm considerably reduced the total encoding time with negligible RD degradation.
  • Keywords
    motion estimation; video coding; AMVP algorithm; H.264 standard; HEVC fast reference picture selection process; advanced motion vector prediction algorithm; encoding complexity; high efficiency video coding standard; motion estimation; prediction-unit partitioning; rate-distortion cost;
  • fLanguage
    English
  • Journal_Title
    Electronics Letters
  • Publisher
    iet
  • ISSN
    0013-5194
  • Type

    jour

  • DOI
    10.1049/el.2015.3094
  • Filename
    7355515