• DocumentCode
    2432445
  • Title

    Scalable and low cost design approach for variable block size motion estimation (VBSME)

  • Author

    Parandeh-Afshar, H. ; Brisk, P. ; Ienne, P.

  • Author_Institution
    Sch. of Comput. & Commun. Sci., Ecole Polytech. Federate de Lausanne (EPFL), Lausanne, Switzerland
  • fYear
    2009
  • fDate
    28-30 April 2009
  • Firstpage
    271
  • Lastpage
    274
  • Abstract
    Variable block size motion estimation (VBSME) in state-of-the-art video coding standards is one of the key features which improves the coding efficiency significantly compared to the previous standards. VBSME hardware design is a challenging task due to its complexity. The processing power requirement for VBSME depends on many factors such as frame size, frame rate and search area. In video coding standards these features are allowed to vary, depending on the requirements of the application. In this paper, a scalable and low cost approach is proposed for designing the VBSME which allows us to tailor the architecture for different applications requirements and implementation targets efficiently. This approach can be used in redesigning of current VBSME architectures to improve their scalability and reduce their design costs. Moreover, as this technique is not block size dependent, it can be employed in designing future coding standards with different block sizes.
  • Keywords
    block codes; motion estimation; video coding; VBSME hardware design; variable block size motion estimation; video coding; Adders; Automatic voltage control; Communication standards; Computer architecture; Costs; Hardware; Motion estimation; Registers; Surface-mount technology; Video coding;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    VLSI Design, Automation and Test, 2009. VLSI-DAT '09. International Symposium on
  • Conference_Location
    Hsinchu
  • Print_ISBN
    978-1-4244-2781-9
  • Electronic_ISBN
    978-1-4244-2782-6
  • Type

    conf

  • DOI
    10.1109/VDAT.2009.5158147
  • Filename
    5158147