Transformation-Based Spatial Partition Join

Abstract

Spatial joins find all pairs of spatial objects that satisfy a given spatial relationship. In this paper, we present the Transformation-Based Spatial Partition Join algorithm (TSPI), a new spatial join algorithm that performs join in the transform space without using indexes. Since the existing algorithms deal with extents of spatial objects in the original space, they either need to replicate the spatial objects or have a relatively complex partition structure - resulting in degrading performance. In contrast, the Transformation-Based Spatial Partition join transforms objects in the original space into points in the transform space and deals only with points having no extents. The transformation does not incur any additional overhead. Thus, Our algorithm has advantages over existing ones in that (1) it obviates the need for replicating spatial objects, and (2) its partition structure is simple. As a result, it always has better performance compared to existing algorithms. Extensive experiments show that the Transformation-Rased Spatial Partition. join improves performance by 19.4-38.0% over the existing algorithms compared.