• DocumentCode
    495326
  • Title

    An Improved MAD Prediction Method for H.264 Rate Control

  • Author

    Liu, Xiaodong ; Dai, Qionghai ; Fu, Chonggang

  • Author_Institution
    Grad. Sch. at Shenzhen, Broadband Network & Multimedia Res. Center, Tsinghua Univ., Shenzhen, China
  • Volume
    6
  • fYear
    2009
  • fDate
    March 31 2009-April 2 2009
  • Firstpage
    316
  • Lastpage
    319
  • Abstract
    H.264 is becoming the worldwide digital video standard for consumer electronics and personal computers. In particular, H.264 has already been selected as a key compression codec for the next generation of optical disc formats, HD-DVD and Blue-ray disc. Rate control (RC) plays an important role in allocating H.264 bits to adapt to the actual channel bandwidth and meanwhile achieve high-quality reconstructed images. In this paper, a novel achievement to improve the precision of mean absolute difference (MAD) prediction in H.264/AVC rate control is presented, which is different from the current standardized RC scheme (JVT-G012) in H.264 that uses a simple linear model, instead an formula is developed to dynamically calculate a ratio factor in the direct mean absolute difference method to optimize the bit allocation. Experimental results show that our method both improves the MAD prediction precision and decrease PSNR fluctuation between frames with less computing complexity.
  • Keywords
    computational complexity; prediction theory; video coding; Blue-ray disc; H.264/AVC rate control; MAD prediction method; bit allocation; channel bandwidth; computational complexity; consumer electronics; digital video standard; mean absolute difference prediction; personal computers; Automatic voltage control; Bandwidth; Codecs; Consumer electronics; Image coding; Image reconstruction; Microcomputers; Prediction methods; Radio control; Video compression;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Computer Science and Information Engineering, 2009 WRI World Congress on
  • Conference_Location
    Los Angeles, CA
  • Print_ISBN
    978-0-7695-3507-4
  • Type

    conf

  • DOI
    10.1109/CSIE.2009.37
  • Filename
    5170712