Fuzzy signatures organized using S-Tree

In this paper we explore the possibility of an efficient organization of fuzzy signatures using the so-called S-Tree. We illustrate the usefullnes of the presented approach on a real-world example of a content-based image retrieval system. Images from the dataset are described using a fuzzy set of features. This description can be translated into a fuzzy signature and these signatures can be stored in a tree structure - similar to the B⁺tree - that allows efficient retrieval. Several variants are considered and evaluated.