Representing concave objects using Virtual Images [chapter]

Timothy Arndt, Gennaro Petraglia, Monica Sebillo, Genoveffa Tortora
1995 Visual Database Systems 3  
Recent research in the design of image database systems emphasizes the necessity of defining an indexing methodology which is not based on form features but rather on invariant numerical, topological or geometrical characteristics of the images. In this paper, we concentrate on the spatial data and relationships which represent the most important method for recognizing images. In particular, we focus on iconic indexing methodologies based on order theory and symbolic projections and we address
more » ... he problem of ambiguity arising from the description of concave objects with these representations. Towards this end, we extend the cutting mechanism of the 2D C-string iconic index to perform cutting lines whenever a concave point (split point) might lead to ambiguity during the reconstruction process. The iconic index obtained according to this set of rules can be managed by the Atomic Relation Extraction Method to derive the corresponding virtual images which represent the normalized pictorial information. Keywords Iconic indexing, spatial reasoning, image data model, content-based query S. Spaccapietra et al. (eds.), Visual Database Systems 3
doi:10.1007/978-0-387-34905-3_3 fatcat:if46x2qoc5enlgxmjyre5buloa