Balanced aspect ratio trees: Combining the advantages of k-d trees and octrees

Citation
Ca. Duncan et al., Balanced aspect ratio trees: Combining the advantages of k-d trees and octrees, J ALGORITHM, 38(1), 2001, pp. 303-333
Citations number
39
Categorie Soggetti
Computer Science & Engineering
Journal title
JOURNAL OF ALGORITHMS
ISSN journal
01966774 → ACNP
Volume
38
Issue
1
Year of publication
2001
Pages
303 - 333
Database
ISI
SICI code
0196-6774(200101)38:1<303:BARTCT>2.0.ZU;2-W
Abstract
Given a set S of n points on R-d, we show, for fixed Li, how to construct i n O(n log n) time a data structure we call the balanced aspect ratio (BAR) tree. A BAR tree is a binary space partition tree on S that has O(log n) de pth in which every region is convex and "fat" (that is, has a bounded aspec t ratio). While previous hierarchical data structures such as k-d trees, qu adtrees, octrees, fair-split trees, and balanced box decompositions can gua rantee some of these properties, we know of no previous data structure that combines all of these properties simultaneously. The BAR tree data structu re has numerous applications ranging from geometric searching problems in f ixed dimensional space to the visualization of graphs and three-dimensional worlds, (C) 2001 Academic Press.