A Survey of Image Information Systems and Future Directions

This chapter gives a taxonomy of image indexing techniques, and presents a survey of image information systems. The chapter discusses research issues in data models. Some thoughts on active image information systems are given in the conclusion. The discussion of image indexing techniques could proceed in three directions: index representation, index organization, and index extraction. Image indexes are approximately represented and do not have an implicit order. Image index representations may have interrelated multiple attributes. The image index structures should also support similarity retrieval. Image information carries the following characteristics: The content of an image cannot be precisely described; spatial entities (objects) and relationships (image features) in images do not in themselves carry any semantic meanings; image-based information could be queried by pictures. The theory of symbolic projection, its applications, and systems incorporating advanced spatial reasoning and image information retrieval, may prove valuable when the design of the next generation of active image information systems is considered.