• DocumentCode
    509433
  • Title

    Robust Frame Error Concealment Algorithm for H.264/AVC Wireless Video Applications

  • Author

    Li, Min ; Wang, Bo

  • Author_Institution
    Sch. of Comput. Sci., Wuhan Univ. of Sci. & Eng., Wuhan, China
  • Volume
    1
  • fYear
    2009
  • fDate
    18-20 Nov. 2009
  • Firstpage
    675
  • Lastpage
    679
  • Abstract
    This paper develops a new error concealment algorithm for the whole-frame losses in wireless video transmission based on H.264/AVC. At first, we propose a forward motion vector extrapolation algorithm based on variable-size block motion vector estimation to obtain the motion vector field of the loss frame without yielding the hole problems. Then, we partition the missing frame into several object regions by clustering the motion vectors, furthermore, we apply median filter to reconstruct the motion vectors for hole areas based on the reliabilities of neighboring motion vectors. Finally, all macroblocks in the lost frame are estimated by using the extrapolated motion vectors. Experimental results show that the proposed algorithm provides better image quality than conventional methods both objectively and subjectively.
  • Keywords
    extrapolation; image motion analysis; median filters; multimedia communication; radiocommunication; video coding; video streaming; H.264/AVC wireless video application; median filter; motion vector extrapolation; motion vector field; robust frame error concealment; variable-size block motion vector estimation; whole-frame losses; wireless video transmission; Automatic voltage control; Clustering algorithms; Extrapolation; Filters; Image reconstruction; Motion estimation; Partitioning algorithms; Propagation losses; Robustness; Yield estimation; error concealment; extrapolation; median filter; motion vector filed;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Multimedia Information Networking and Security, 2009. MINES '09. International Conference on
  • Conference_Location
    Hubei
  • Print_ISBN
    978-0-7695-3843-3
  • Electronic_ISBN
    978-1-4244-5068-8
  • Type

    conf

  • DOI
    10.1109/MINES.2009.241
  • Filename
    5370375