Content-based Image Retrieval and Characterization on Specific Web Collections

Content-based image and video retrieval is a fast growing and increasing relevant research area. The research community recognizes the following main challenges in this field [8]: the bridging of the semantic gap (understanding the meaning behind the query), the content-based retrieval of videos (finding a video similar to another one), and the increasing huge amount of digital data, produced by digital consumer devices (e.g. digital cameras) and computational devices (hard disks, CD-ROMs, etc.), which needs a semantic understanding and also produces a scale problem. In addition to that, we believe that the content-based retrieval of images and videos in the web is an important and challenging area where less research has been done, probably because of technical and practical reasons. As all of us know, popular search engines allow the retrieval of images on the web using only text queries. This situation should be improved and we think we should start to develop methods and strategies for the content-based retrieval of information on the web, the largest and most used multimedia database in the world...

Download PDF