INTRODUCTION TO CONTENT-BASED VISUAL PROCESSING - Video Object Extraction and Representation: Theory and Applications - page 18

Figure 1.9. MPEG-7 role as the language of video content: starting at the user, a
request for video content is issued. The query is then translated into an intermediate
MPEG-7 form and a search of similar MPEG-7 representations of video content is ini-
tiated. Similar MPEG-7 representations of objects are found and their corresponding
video objects are sent back.
lem in Chapter 4, we refer to the MPEG-4 standard for our test set,
application specifications, and ground truth results for comparison.
MPEG-7
Just as the MPEG-4 standard provided focus for extraction and syn-
thesis functionalities, the MPEG-7 standard explores issues of video ob-
ject representation and search. While the MPEG-4 standard was mainly
concerned with extraction of video objects, the MPEG-7 standard is con-
cerned with description and representation of these video objects. The
MPEG-7 standard defines a generic description language (Description
Definition Language, or DDL, for short) for the representation of varied
audio-visual objects such as 3-D object representation, scene descrip-
tion, or story structure of a video document. For each different type of
video object, there exists different types of description schemes (DS).
The MPEG-7 syntax (DDL) supports the multiplicity of these descrip-
tion schemes (DS). To support video object databases (see Figure 1.9),
MPEG-7 supports the standardization of representation and search of
video objects. Just as the MPEG-1/2 with encoder techniques and the
MPEG-4 standard with video object segmentation systems, the MPEG-
7 does not advocate any particular form of extraction. Instead, it sup-