Research Of Multimedia Document Retrieval

With the development and popularization of digital media technology and socialnetworking, network is full of huge amounts of multimedia documents which containvarious modalities. Traditional document retrieval can’t satisfy people’s needs. How tofind information that users need quickly and accurately from huge amounts ofmultimedia documents is an urgent problem to solve.The paper first makes a research on existing methods of multimedia informationretrieval. Based on various modalities in multimedia document, we present a newmethod that apply linear weighted sum of multimodal fusion to retrieval multimediadocuments. With this method, text retrieval resorts to full text search technology whichbases on Lucene. To image retrieval, it resorts to LIRE which applied visual featuressuch as texture and color. And hierarchy indexing was adopted to retrieval video.Weight is assigned to each modality in accordance with scale factor that results fromprecision rate of retrieval outcomes using single modality only. This method was provedto have high efficiency but need much more time. Merging Bag-of-words andBag-of-Features together is another way the paper provides to retrieval multimediadocument. And the BOW is improved BOW which classify words into differentcategories. Improved BOW reduces size of the codebook so greatly that it raises theefficiency enormously. Experiments proved that methods the paper present whichcomprehensively consider text and image have higher efficiency than those using singlemodality to retrieval multimedia documents.