• DocumentCode
    3547839
  • Title

    Error sensitivity testing for the MC-EZBC scalable wavelet video coder

  • Author

    Shanableh, Tamer ; May, Tony

  • Author_Institution
    Dept. of Comput. Sci., American Univ. of Sharjah, United Arab Emirates
  • fYear
    2005
  • fDate
    23-26 May 2005
  • Firstpage
    6062
  • Abstract
    The paper looks at the error resiliency of the motion-compensated embedded zero block coder (MC-EZBC). The MC-EZBC is gaining momentum as a motion compensated temporal filtering solution. We propose to demultiplex the MC-EZBC bitstream into seven bit categories and examine the effect of bit errors on each bit category. It is shown that different bit categories have different immunity to bit errors. The significance map at quadtree level zero, the motion vectors and the coefficients sign bits are shown to be the most sensitive. These categories exceed 60% of the total bitrate. It is concluded that employing unequal error protection for the motion fields is sufficient. The coding algorithm need not be modified, hence preserving its bitstream embedding and universal scalability feature.
  • Keywords
    coding errors; motion compensation; quadtrees; transform coding; video coding; wavelet transforms; bit categories; bit errors; bitstream embedding; coding algorithm; coefficients sign bits; demultiplexing; error resiliency; error sensitivity testing; motion compensated temporal filtering solution; motion vectors; motion-compensated embedded zero block coder; quadtree level zero; scalable wavelet video coder; significance map; unequal error protection; universal scalability; Channel coding; Codecs; Computer errors; Error correction codes; Forward error correction; Partitioning algorithms; Protection; Scalability; Streaming media; Testing;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Circuits and Systems, 2005. ISCAS 2005. IEEE International Symposium on
  • Print_ISBN
    0-7803-8834-8
  • Type

    conf

  • DOI
    10.1109/ISCAS.2005.1466022
  • Filename
    1466022