Fast approximate nearest-neighbor queries in metric feature spaces by buoy indexing

Abstract

An indexing scheme for solving the problem of nearest neighbor queries in generic metric feature spaces for content-based retrieval is proposed aiming to break the “dimensionality curse'. The basis for the proposed method is the partitioning of the feature dataset into a fixed number of clusters that are represented by single buoys. Upon submission of a query request, only a small number of clusters whose buoys are close to the query object are considered for the approximate query result, cutting down the amount of data to be processed effectively. Results from extensive experimentation concerning the retrieval accuracy are given. The influence of control parameters is investigated with respect to the tradeoff between retrieval accuracy and query execution time.