Object and query transformation: supporting multi-dimensional queries through code reuse
Proceedings of the ninth international conference on Information and knowledge management
Image indexing and similarity retrieval based on spatial relationship model
Information Sciences—Informatics and Computer Science: An International Journal - Special issue: Introduction to multimedia and mobile agents
Hi-index | 0.00 |
Many spatial applications deal with data characterized by extremely high dimensionality. Unfortunately, other than the fact that contemporary spatial access methods (SAMs) are inadequate to handle large sets of high dimensional data, we know little about the underlying causes of their inadequacy. The paper investigates the performance of spatial access methods that approximate extended objects by minimum bounding rectangles (MBRs). It exposes the conceptual problems of traditional MBR based SAMs resulting in their inability to handle high dimensional data. The paper shows that a new MBR based structure, called the QSF-tree, which avoids the problems of traditional SAMs, gracefully adapts to accommodate the increasing dimensionality of the spatial data. The experimental evidence demonstrates that QSF-trees outperform three popular MBR based SAMs in both low- and high-dimensional spaces. As the number of dimensions grows, the improvements have a tendency to increase.