Efficient robust digital hyperplane fitting with bounded error

  • Authors:
  • Dror Aiger;Yukiko Kenmochi;Hugues Talbot;Lilian Buzer

  • Affiliations:
  • Université Paris-Est, Laboratoire d'Informatique Gaspard-Monge, France;Université Paris-Est, Laboratoire d'Informatique Gaspard-Monge, France;Université Paris-Est, Laboratoire d'Informatique Gaspard-Monge, France;Université Paris-Est, Laboratoire d'Informatique Gaspard-Monge, France

  • Venue:
  • DGCI'11 Proceedings of the 16th IAPR international conference on Discrete geometry for computer imagery
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

We consider the following fitting problem: given an arbitrary set of N points in a bounded grid in dimension d, find a digital hyperplane that contains the largest possible number of points. We first observe that the problem is 3SUM-hard in the plane, so that it probably cannot be solved exactly with computational complexity better than O(N2), and it is conjectured that optimal computational complexity in dimension d is in fact O(Nd). We therefore propose two approximation methods featuring linear time complexity. As the latter one is easily implemented, we present experimental results that show the runtime in practice.