1999 IEEE Region 10 Conference, TENCON 1999, pp.337 - 340
Abstract
Efficient query processing in multi-dimensional indexing structures is an important issue for effective employment of multimedia data applications. A considerable number of studies, to date, have been conducted in this area. However the efficiency of the proposed solutions generally deteriorates as the dimension of the data increases. We introduce a filtering method for efficient processing of k-nearest neighbor queries in multi-dimensional indexing structures. The proposed method is based on the R*-tree, and uses a vantage point for effective similarity searches. Through the use of a vantage point we are abler to filter out data objects and decrease the distance computation time. Experimental results show that the k-nearest neighbor search that uses the proposed method consistently outperforms the search performance that uses the existing method for the R*-tree.