An Efficient Parallel Architecture for H.264/AVC Fractional Motion Estimation

  • Zhuo Zhao
  • Ping Liang
Conference paper
Part of the Smart Innovation, Systems and Technologies book series (SIST, volume 11)


This paper presents a new VLSI architecture for fractional motion estimation (FME) in H.264/AVC. Statistical characteristics of the motion vectors of different inter-prediction modes are analyzed. The FME architecture explored block-level parallelism and can process multiple blocks with the same prediction mode simultaneously, external memory data accesses are dramatically reduced. Simulation results show that the proposed architecture can support 1080P (1960x1088) at 30fps with a frequency of 80MHz.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Chen, T.C., Chien, S.Y., et al.: Analysis and architecture design of HDTV720P 30 frams/s H.264/AVC encoder. IEEE Trans. Circuits Systems. Video Technology 16(6), 673–688 (2006)CrossRefGoogle Scholar
  2. 2.
    Song, Y., Shao, M., et al.: H.264/AVC Fractional Motion E stimation Engine with ComputationReusing in HDTV1080P Real-Time Encoding Applications. In: 2007 IEEEWorkshop on Signal Processing Systems, October 17-19, pp. 509–514 (2007)Google Scholar
  3. 3.
    Tsung, P.K., Chen, W.Y., Ding, L.F., et al.: Single-iteration full-search fractional motion estimationfor quad full HD H.264/AVC encoding. In: ICME 2009, pp. 9–12 (2009)Google Scholar
  4. 4.
    Chen, T.C., Huang, Y.W., Chen, L.G.: Fully utilized and reusable architecture for fractional motion estimation of H.264/AVC. In: ICASSP 2004, May 17-21, vol. 5 (2004)Google Scholar
  5. 5.
    Zhang, L., Gao, W.: Reusable Architecture and Complexity-Controllable Algorithm for the Int eger/Fractional Motion Estimation of H.264. IEEE Transactions on Consumer Electronics 53(2), 749–756 (2007)CrossRefMathSciNetGoogle Scholar
  6. 6. Test Media,

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Zhuo Zhao
    • 1
  • Ping Liang
    • 1
  1. 1.University of CaliforniaRiversideUSA

Personalised recommendations