Low-bitrate video coding with third order geometric transformations

Cornelis H. Slump, Marcel A.J.A. van Veen, Frederik J. De Bruijn

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    Abstract

    This paper describes low-bitrate video compression based upon the characterization of the new frame as a set of geometric transformations of objects of the previous frame. Objects with motion are detected and the motion is estimated. The estimated motion (motion field) is used to obtain the parameters for the geometric transformations. The pertinent geometric transformations are rotation, translation, zooming and isotropic and anisotropic distortion. The motivation for choosing this set of third-order transformations is that we have at our disposal special ASICS for real-time video processing. We only want to transform moving objects and therefore the boundaries of the moving objects must be known. The boundaries of the objects are represented by closed contours.

    Original languageEnglish
    Title of host publicationSignal processing IX : theories and applications
    Subtitle of host publicationProceedings of EUSIPCO-98, Ninth European Signal Processing Conference, Rhodes, Greece, 8-11 September 1998
    Place of PublicationPatras
    PublisherEURASIP, European Association for Signal, Speech and Image Processing
    Number of pages4
    ISBN (Print)978-960-7620-06-4
    Publication statusPublished - 1998
    Event9th European Signal Processing Conference, EUSIPCO 1998 - Rhodes, Greece
    Duration: 8 Sep 199811 Sep 1998
    Conference number: 9

    Publication series

    NameProceedings European Signal Processing Conference
    PublisherIEEE
    Volume1998
    ISSN (Print)2219-5491

    Conference

    Conference9th European Signal Processing Conference, EUSIPCO 1998
    Abbreviated titleEUSIPCO
    CountryGreece
    CityRhodes
    Period8/09/9811/09/98

    Fingerprint

    Dive into the research topics of 'Low-bitrate video coding with third order geometric transformations'. Together they form a unique fingerprint.

    Cite this