Chapter 17. Concepts and Techniques for Indexing Visual Semantics

  1. Vittorio Castelli and
  2. Lawrence D. Bergman
  1. Alejandro Jaimes and
  2. Shih-Fu Chang

Published Online: 7 APR 2002

DOI: 10.1002/0471224634.ch17

Image Databases: Search and Retrieval of Digital Imagery

Image Databases: Search and Retrieval of Digital Imagery

How to Cite

Jaimes, A. and Chang, S.-F. (2001) Concepts and Techniques for Indexing Visual Semantics, in Image Databases: Search and Retrieval of Digital Imagery (eds V. Castelli and L. D. Bergman), John Wiley & Sons, Inc., New York, USA. doi: 10.1002/0471224634.ch17

Editor Information

  1. IBM T.J. Watson Research Center

Author Information

  1. Columbia University, Department of Electrical Engineering, 500 West 120th Street, S.W. Mudd Bldg., Room 1312, New York, NY 10027

Publication History

  1. Published Online: 7 APR 2002
  2. Published Print: 7 DEC 2001

ISBN Information

Print ISBN: 9780471321163

Online ISBN: 9780471224631

SEARCH

Keywords:

  • visual information;
  • indexing;
  • content-based techniques;
  • object recognition;
  • visual object detectors

Summary

In this chapter, the authors discuss content-based retrieval, emphasizing that visual information can be indexed at multiple levels. In particular they discuss a conceptual framework to index visual content and analyze different content-based information retrieval (CBIR) approaches from the perspective of the conceptual framework. A distinction is made between the interface and indexing components of CBIR systems, the advantages and limitations of different interface modalities are outlined, and the benefits of different indexing mechanisms are discussed. The discussion, which is carried out in the context of the conceptual framework for indexing visual information, shows the importance of object-level indexing techniques in CBIR. As a result, a brief overview of some of the major object-recognition strategies is presented, particularly in relation to CBIR.