Object-oriented model for image information representation

Efficient management of non-alphanumeric data (such as images, videos, and graphs) is essential to a large number of applications. This paper discusses the issues in pictorial information modeling, examines the suitability of traditional data modeling approaches, and then proposes a new object-oriented data modeling approach, called Modeling Object-Oriented Data Semantics (MOODS). A key idea in MOODS is to integrate the image processing component of an image information management system with the data modeling technique. In addition, a user may provide additional assistance in defining semantics at various stages. We argue that the data modeling technique must capture the sequence of transformations and allow a variety of semantics to be associated with the same data to accommodate different application domains. Additional salient features of MOODS include a notion of abstract function classes and dynamic binding of functions to data.