• DocumentCode
    239736
  • Title

    A low complexity and high performance interpolation filter for MPEG IVC

  • Author

    Hao Lv ; Ronggang Wang ; Yangang Cai ; Huizhu Jia ; Xiaodong Xie ; Wen Gao

  • Author_Institution
    Sch. of Electron. & Comput. Eng., Peking Univ., Shenzhen, China
  • fYear
    2014
  • fDate
    20-23 Aug. 2014
  • Firstpage
    141
  • Lastpage
    145
  • Abstract
    Fractional-pel motion compensation is widely adopted in the modern video coding standards such as H.264/AVC, AVS, and HEVC etc. The interpolation filter is a critical factor that influences the coding efficiency. In this paper, a generation algorithm of interpolation filter coefficients is utilized. Based on the coefficients generation algorithm, three different tap filters, namely 6 tap, 8 tap, 10tap, are tested. A combination of 6 tap and 8 tap interpolation filters is proposed and proved to be optimal scheme considering both performance and computational complexity. And it is beneficial to make software optimization more effective, especially when SIMD-like (Single Instruction Multiple Data) operation is used. Experiments show that the average BD-rate gains on luma Y, chroma U and V are 8.01%, 5.08% and 4.98% for CS1 (Constraint set 1), and 9.21%, 7.53% and 7.63% for CS2 (Constraint set 2) in MPEG IVC reference software ITM5.0, compared with the traditional IVC interpolation method. The coding efficiency gains are significant for some video sequences and can reach up to 28.7%. With the merits of high performance and low complexity, our proposed method is formally adopted by MPEG IVC.
  • Keywords
    computational complexity; filtering theory; interpolation; motion compensation; parallel processing; video coding; AVS; Fractional pel motion compensation; H.264/AVC; HEVC; MPEG IVC; SIMD; coding efficiency; coefficients generation algorithm; computational complexity; generation algorithm; interpolation filter coefficients; optimal scheme; single instruction multiple data; software optimization; video coding; Complexity theory; Digital signal processing; Encoding; Finite impulse response filters; Interpolation; Transform coding; Video coding; IVC; MPEG; fractional-pel; interpolation; motion estimation;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Digital Signal Processing (DSP), 2014 19th International Conference on
  • Conference_Location
    Hong Kong
  • Type

    conf

  • DOI
    10.1109/ICDSP.2014.6900817
  • Filename
    6900817