• DocumentCode
    2754587
  • Title

    A Computation and Power Reduction Technique for H.264 Intra Prediction

  • Author

    Adibelli, Yusuf ; Parlak, Mustafa ; Hamzaoglu, Ilker

  • Author_Institution
    Fac. of Eng. & Natural Sci., Sabanci Univ., Istanbul, Turkey
  • fYear
    2010
  • fDate
    1-3 Sept. 2010
  • Firstpage
    753
  • Lastpage
    760
  • Abstract
    H.264 intra prediction algorithm has a very high computational complexity. This paper proposes a technique for reducing the amount of computations performed by H.264 intra prediction algorithm. For each intra prediction equation, the proposed technique compares the pixels used in this prediction equation. If the pixels used in a prediction equation are equal, this prediction equation is simplified significantly. By exploiting the equality of the pixels used in prediction equations, the proposed technique reduces the amount of computations performed by 4x4 luminance prediction modes up to 78% with a small comparison overhead. The proposed technique does not affect the PSNR and bit rate. We also implemented an efficient 4x4 intra prediction hardware including the proposed technique using Verilog HDL. We quantified the impact of the proposed technique on the power consumption of this hardware on a Xilinx Virtex II FPGA using Xilinx XPower, and it reduced the power consumption of this hardware up to 13.7%.
  • Keywords
    computational complexity; hardware description languages; power aware computing; prediction theory; FPGA; H.264; Verilog HDL; Xilinx Virtex; Xilinx XPower; computational complexity; intra prediction algorithm; power reduction technique; prediction equation; Computer architecture; Equations; Hardware; Mathematical model; Pixel; Power demand; Prediction algorithms; Computation Reduction; FPGA; H. 264; Hardware Implementation; Intra Prediction; Low Power;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Digital System Design: Architectures, Methods and Tools (DSD), 2010 13th Euromicro Conference on
  • Conference_Location
    Lille
  • Print_ISBN
    978-1-4244-7839-2
  • Type

    conf

  • DOI
    10.1109/DSD.2010.115
  • Filename
    5615438