MIR techniques and technologies

This chapter contains details of MIR systems for the analysis and content-based indexing of multimedia materials, and the phases and structures for search and retrieval. In general, the systems’ functioning processes can be divided into two major interrelated parts: operations relating to the creation of the database and document indexes, and operations relating to the process of search and retrieval. Paradigms and protocols experimented with so far in the design and implementation of MIR systems are all fairly similar, based on the direct treatment of the actual content of the document, even considering the role and the importance of traditional descriptive treatment. Fundamental to these operations is the ability to create a new type of index of data relating to documents, that in addition to traditional terminological elements is also enriched by contentual elements. In presenting the various operations the different methods of automatic, semi-automatic and manual treatment are compared, declaring the merits, defects and the necessary reciprocal integration of the capabilities of machines and humans. Finally, three examples of MIR systems are presented that manage Visual, Audio, Video and Multimedia documents.