Article Preview
TopThe volume of TV content is currently increasing which makes its management more problematic both for the producer and the consumer. For the spectator, the main goal is to focus on interesting TV broadcast data in timely. To achieve such an outline, many researchers have focused on many fields such as indexing, retrieving and also summarization. Some other works have focused on how to assist users in browsing large TV broadcast collection. All works tend to define an effectiveness access scheme to the audiovisual data that integrate all these elementary techniques. Based on different modalities (text, audio, and image), existing works extracted the low-level features to describe video shot.
In Campanella, Leonardi, and Migliorati (2009) the proposed graphical framework for video content goal tends to facilitate a quick understanding and an annotation of the semantic content of a video sequence. The basic idea is to present data in a 2D feature space in which the shots of the considered video sequence are located. The temporal position and the specific content of each shot can be displayed and analyzed in more detail. The user can directly select features to be used during the navigation session. The proposed tool also offers functionalities for automatically and semi-automatically retrieving and annotating the shots. Such a tool emphasizes on the user’s fundamental role when annotating and accessing multimedia corpora.
In Cees, Snoek, Worring, Koelma, and Smeulders (2007), the MediaMill search engine helps the user in accessing a large video scale due to a semantic indexing process. The system proposes a lexicon of 436 feature. The user can define a textual or, for example, a query when searching information. Some advanced graphical interfaces are then proposed to the user in order to enhance the navigation process in retrieved results (CroosBrowser, GalaxyBrowser, or SphereBrowser). Each visualization tool uses a semantic measurement between retrieved video documents. The clustering is also employed in a 3D space in order to make the data distribution clear.
Mesh “Multimedia Semantic Syndication for Enhanced News Services” Mesh Project (n.d.) proposed a complete framework for video, most often news, management adapted by any user. MESH aims at extracting, comparing and combining content from multiple multimedia news sources, automatically creating advanced personalized multimedia summaries, syndicating summaries and content based on the extracted semantic information, and providing end users with a “multimedia mesh” news navigation system. Mesh project also focuses on the personalization outline by recognizing the user’s preferences while searching the web and automatically suggests useful information.