A study of spatial data models and their application to selecting information from pictorial databases

People have always used visual techniques to locate information in the space surrounding them. However with the advent of powerful computer systems and user-friendly interfaces it has become possible to extend such techniques to stored pictorial information. Pictorial database systems have in the past primarily used mathematical or textual search techniques to locate specific pictures contained within such databases. However these techniques have largely relied upon complex combinations of numeric and textual queries in order to find the required pictures. Such techniques restrict users of pictorial databases to expressing what is in essence a visual query in a numeric or character based form. What is required is the ability to express such queries in a form that more closely matches the user's visual memory or perception of the picture required. It is suggested in this thesis that spatial techniques of search are important and that two of the most important attributes of a picture are the spatial positions and the spatial relationships of objects contained within such pictures. It is further suggested that a database management system which allows users to indicate the nature of their query by visually placing iconic representations of objects on an interface in spatially appropriate positions, is a feasible method by which pictures might be found from a pictorial database. This thesis undertakes a detailed study of spatial techniques using a combination of historical evidence, psychological conclusions and practical examples to demonstrate that the spatial metaphor is an important concept and that pictures can be readily found by visually specifying the spatial positions and relationships between objects contained within them.