Mapping high dimensional features onto Hilbert curve: Applying to fast image retrieval

The use of high-dimension features is unavoidable in many applications of image retrieval and techniques of dimension reductions are not always efficient. The space-filling curve reduces the number of dimensions to one while preserving the neighborhood relation. In this paper, Hilbert curve, the most neighborhood preserving space-filling curve, is used in shape retrieval. The retrieving is accelerated by ordering the shapes in 1-D dynamical data structure, which enables rapid insertions of new images without changing existing data. A proposal of fast mapping facilitate the computing of 1-D Hilbert indexes from high dimensional features.