|
|
 |
The driving force is to disassociate content search from the availability of laboriously annotated metadata databases. Thus, algorithms to be developed will provide an alternative and complementary path for metadata based audio/video content search. The proposed approach advocates the automatic extraction of content features directly from the compressed content (“fingerprints” or “thumbnails”), thus greatly accelerating the search process. Further, through an associated classification, search databases will be rendered compact and suitable for binary search techniques, providing fast searches over huge content databases.
The search engine will be applicable to a number of different use cases ranging from "similarity" searches for video and audio web content, to information harvesting and data mining. Moreover the DIVAS approach enables “stream searchers” suitable for DRM resolution, advertisement time tracking, etc.
DIVAS implements algorithms for compressed video and audio characterisation, fingerprint extraction, and segmentation applied on compressed content, all being essential for efficient search and result correlation in audio/video search engines. This opens the way for the introduction and seamless integration of audiovisual searching to ANY web search engine, and the location of video content ANYWHERE, irrespective of transformations and annotations, thus adding true direct multimedia search capability to ambient intelligence.
|
