English

ANN queries: covering Voronoi diagram with hyperboxes

Computational Geometry 2014-06-18 v2 Data Structures and Algorithms

Abstract

Given a set SS of nn points in dd-dimensional Euclidean metric space XX and a small positive real number ϵ\epsilon, we present an algorithm to preprocess SS and answer queries that require finding a set SSS' \subseteq S of ϵ\epsilon-approximate nearest neighbors (ANNs) to a given query point qXq \in X. The following are the characteristics of points belonging to set SS': - sS\forall s \in S', \exists a point pXp \in X such that pqϵ|pq| \le \epsilon and the nearest neighbor of pp is ss, and - \exists a sSs' \in S' such that ss' is a nearest neighbor of qq. During the preprocessing phase, from the Voronoi diagram of SS we construct a set of box trees of size O(4dVδ(πϵ)d1)O(4^d\frac{V}{\delta}(\frac{\pi}{\epsilon})^{d-1}) which facilitate in querying ANNs of any input query point in O(1dlgVδ+(πϵ)d1)O(\frac{1}{d}lg \frac{V}{\delta} + (\frac{\pi}{\epsilon})^{d-1}) time. Here δ\delta equals to (ϵ2d)d(\frac{\epsilon}{2\sqrt{d}})^d, and VV is the volume of a large bounding box that contains all the points of set SS. The average case cardinality of SS' is shown to rely on SS and ϵ\epsilon.

Cite

@article{arxiv.1111.5893,
  title  = {ANN queries: covering Voronoi diagram with hyperboxes},
  author = {Rajasekhar Inkulu and Sanjiv Kapoor},
  journal= {arXiv preprint arXiv:1111.5893},
  year   = {2014}
}

Comments

This paper has been withdrawn by the authors. Not interesting to have it around

R2 v1 2026-06-21T19:41:20.986Z