Object-Based Indexing of Compressed Video Content: From SD to HD Video

In the vast domain of digital multimedia libraries, the efficient and universal access to the stored data becomes possible due to content description and indexing metadata. Their automatic production is a challenging task addressing classical problems of objects extraction and description. In this framework, ever growing quantity of content is already available in compressed form. This holds for SD video encoded in MPEG standards and for quickly coming HD video for which recent standards as MJPEG2000 are used. In this paper we propose a general framework for efficient extraction of moving objects in SD and HD video with a focus on the hierarchical video signal representation by wavelets pyramids.