DSpace User Group Meeting 2007
  
  Home > Papers > Nicola Cavalli

    home
    overview
    programme
    call for papers
    submission
    papers
    registration
    organization
    links


  


  


    


  


 
Nicola Cavalli

Dspace Integration with Video Indexing and Summarization Systems

Nicola Cavalli
CPM - University Milano Bicocca

*Roberto Bisiani
CPM - University Milano Bicocca

Roberta Caccialupi
CPM - University Milano Bicocca

Georgia Conte
CPM - University Milano Bicocca

Valerio Minetti
CPM - University Milano Bicocca

     Full text: PDF
     Last modified: October 30, 2007

Abstract
The goal of the Dspace installation at the Centro di Produzione Multimediale (CPM) of Milan Bicocca University is to store, index and disseminate multimedia objects, such as image and video files. We have chosen to use the Dspace platform to offer this service even if we understand that, as of now, Dspace installations are more oriented to managing research output in the form of text files, e.g. pdf files. Databases of images and videos depend on efficient algorithms to enable fast browsing and access to the desired information, while the current release of Dspace does not offer any alternative to manual annotation of video contents.

Our solution to obtain a better video indexing and summarization is to integrate in our Dspace installation an application developed by the Visual and Imaging Laboratory at the University of Milano Bicocca, using the PluginManager of Dspace 1.4. This class creates and organizes plug-ins, and helps select a plug-in when there are several possible choices, as Don Gourley’s presentation at January 2007 DSUG highlighted . We plan to develop a plug-in for the filter-media task in order to add filtering support for video bitstreams, enabling the platform to create sets of significative keyframes describing the ingested contents.

Automatic video analysis is a complex process that can be divided into three main modules: shot boundary detection, key frame extraction, and summary post-processing. The Video Indexing and Summarization system developed by the Visual and Imaging laboratory at the University of Milano Bicocca uses a shot boundary detection algorithm based on the recognition of the editing effects (such as cuts, fades, dissolves, etc...) that identify the boundaries of the video sequences (shots). A keyframe extraction is then performed for each shot in the form of a still image. This keyframe acts as sort of bookmark and might also be used in the indexing process. A three-step post-processing algorithm then takes place to present users with easily accessible visual summaries that are exhaustive, but not redundant. In particular the third step identifies the default summary level that is shown to the users: starting from this set of keyframes, the users can then browse the video content at various levels of detail.

Integrating this system in Dspace means creating a more efficient infrastructure that is able on the one hand to manage and disseminate video contents and metadata through the OAI-PMH protocol and on the other to offer to the user an automatically created summary with the option of choosing different levels of detail.


Research
Support Tool
Capture Cite
View Metadata
Printer Friendly
Context
Author Bio
Define Terms
Related Studies
Media Reports
Google Search
Action



    Learn more
    about this
    publishing
    project...


Public Knowledge

  Website by AEPIC - CILEA, powered by OCS                                                     Top